Uploaded byafrozkustor
18 views

Data Science Applied to Sustainability Analysis Jennifer B. Dunn

Data Science Applied to Sustainability Analysis Jennifer B. Dunn Data Science Applied to Sustainability Analysis Jennifer B. Dunn Data Science Applied to Sustainability Analysis Jennifer B. Dunn

Embed presentation

Download to read offline
Download the full version and explore a variety of ebooks or textbooks at https://ebookmass.com Data Science Applied to Sustainability Analysis Jennifer B. Dunn _____ Follow the link below to get your download now _____ https://ebookmass.com/product/data-science-applied-to- sustainability-analysis-jennifer-b-dunn/ Access ebookmass.com now to download high-quality ebooks or textbooks
Here are some recommended products for you. Click the link to download, or explore more at ebookmass.com Genomic Data Sharing Jennifer B. Mccormick https://ebookmass.com/product/genomic-data-sharing-jennifer-b- mccormick/ Applied Epistemology Jennifer Lackey https://ebookmass.com/product/applied-epistemology-jennifer-lackey/ Applied Data Analysis and Modeling for Energy Engineers and Scientists https://ebookmass.com/product/applied-data-analysis-and-modeling-for- energy-engineers-and-scientists/ Intelligent Data Analysis: From Data Gathering to Data Comprehension Deepak Gupta https://ebookmass.com/product/intelligent-data-analysis-from-data- gathering-to-data-comprehension-deepak-gupta/
Data Science With Rust: A Comprehensive Guide - Data Analysis, Machine Learning, Data Visualization & More Van Der Post https://ebookmass.com/product/data-science-with-rust-a-comprehensive- guide-data-analysis-machine-learning-data-visualization-more-van-der- post/ Applied Longitudinal Analysis 2nd Edition https://ebookmass.com/product/applied-longitudinal-analysis-2nd- edition/ Foundations of Calculus for Data Science: An Foundational Guide to Understanding the Application of Calculus to Data Science Bisette https://ebookmass.com/product/foundations-of-calculus-for-data- science-an-foundational-guide-to-understanding-the-application-of- calculus-to-data-science-bisette/ Introduction to Python for Econometrics, Statistics and Data Analysis Kevin Sheppard https://ebookmass.com/product/introduction-to-python-for-econometrics- statistics-and-data-analysis-kevin-sheppard/ (eBook PDF) Intro to Python for Computer Science and Data Science: Learning to Program with AI, Big Data and The Cloud https://ebookmass.com/product/ebook-pdf-intro-to-python-for-computer- science-and-data-science-learning-to-program-with-ai-big-data-and-the- cloud/
DATA SCIENCE APPLIEDTO SUSTAINABILITY ANALYSIS
DATA SCIENCE APPLIEDTO SUSTAINABILITY ANALYSIS Edited by JENNIFER B. DUNN Northwestern-Argonne Institute of Science and Engineering, Evanston, IL, USA; Center for Engineering Sustainability and Resilience, Northwestern University, Evanston, IL, USA; Chemical and Biological Engineering, Northwestern University, Evanston, IL, USA PRASANNA BALAPRAKASH Northwestern-Argonne Institute of Science and Engineering, Evanston, IL; Math and Computer Science Division, Argonne National Laboratory, Lemont, IL, USA
Elsevier Radarweg 29, PO Box 211, 1000 AE Amsterdam, Netherlands The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom 50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States Copyright © 2021 Elsevier Inc.All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions. This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein). Notices Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary. Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods,compounds,or experiments described herein.In using such information or methods they should be mindful of their own safety and the safety of others,including parties for whom they have a professional responsibility. To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress ISBN: 978-0-12-817976-5 For Information on all Elsevier publications visit our website at https://www.elsevier.com/books-and-journals Publisher: Candice Janco Acquisitions Editor: Marisa LaFleur Editorial Project Manager: Charlotte Kent Production Project Manager: BharatwajVaratharajan Cover Designer:Victoria Pearson Typeset by Aptara, New Delhi, India
v Contributors ix 1. Overview of data science and sustainability analysis 1 Prasanna Balaprakash, Jennifer B. Dunn Data science is central to advances in sustainability 1 Types of sustainability analyses 6 Data science tools 7 Overview of case studies in data science in sustainability 10 References 13 PART 1 Environmental Health and Sustainability 15 2. Applying AI to conservation challenges 17 Niraj Swami Introduction 17 What is adaptive management? 19 Creating value with AI 21 The impact technology canvas: synchronizing AI initiatives 24 Summary 26 References 27 3. Water balance estimation in Australia through the merging of satellite observations with models 29 Valentijn R.N. Pauwels, Ashley Wright, Ashkan Shokri, Stefania Grimaldi Introduction 29 Case studies 32 Conclusion 38 References 40 4. Machine learning in the Australian critical zone 43 Elisabeth N. Bui What is CZ science? 43 Machine learning techniques used 45 Questions addressed and data used 51 Key results, insights, and why they matter to the field of sustainability 65 Future development and application of data science in CZ science 69 Glossary terms 72 References 73 CONTENTS
Contents vi PART 2 Energy and Water 79 5. A clustering analysis of energy and water consumption in U.S. States from 1985 to 2015 81 Evgenia Kapousouz, Abolfazl Seyrfar, Sybil Derrible, Hossein Ataei Introduction 81 Materials and methods 82 Results 87 Conclusion and discussion 99 Acknowledgments 100 Appendix 100 References 106 6. Exploring rooftop solar photovoltaics deployment and energy injustice in the US through a data-driven approach 109 Sergio Castellanos, Deborah A. Sunter, Daniel M. Kammen Introduction 109 A focus on distributed solar PV 110 Assessing rooftop PV potential across cities 112 Evaluating equity in rooftop PV deployment: a case study for the United States 117 Concluding remarks 124 References 125 7. Data-driven materials discovery for solar photovoltaics 129 Leon R. Devereux, Jacqueline M. Cole Introduction 129 Fundamentals of photovoltaics 131 Data-driven materials discovery schemes 140 Case studies of solar materials discovery 146 Future outlook 161 References 162 PART 3 Sustainable Systems Analysis 165 8. Machine learning in life cycle assessment 167 Mikaela Algren, Wendy Fisher, Amy E. Landis Introduction to life cycle assessment (LCA) 168 LCAs role in sustainability 168 General methods for process LCA 171 Impact assessment 173 Uses and limitations of LCA 175 Tools and data sources for LCA 175
Contents vii Introduction to machine learning (ML) 176 Types of ML 177 Choosing a machine learning algorithm 179 Tools for ML 182 ML in LCA 182 ML for surrogate LCAs 183 ML in LCI 184 ML in LCIA 185 ML in interpretation and analysis 186 Conclusion 188 References 188 9. Industry sustainable supply chain management with data science 191 Deboleena Chakraborty, Richard K. Helling Introduction to LCA 191 Today’s limitations 195 LCA applications 196 Vision/Needs 200 References 201 PART 4 Society and Policy 203 10. Deep learning with satellite imagery to enhance environmental enforcement 205 Cassandra Handan-Nader, Daniel E. Ho, Larry Y. Liu Introduction 205 The methodological evolution of remote sensing 208 Using deep learning to identify CAFOs 216 Discussion 221 References 224 11. Towards achieving the UNs data revolution: combining earth observation and socioeconomic data for geographic targeting of resources for the sustainable development goals 229 Gary R. Watmough, Charlotte L.J. Marcinko Why social remote sensing? 229 Using EO data for understanding socioeconomic conditions 233 How can we socialize the pixel? 240 A socio-ecologically informed approach to linking EO and socioeconomic data 242 Future directions for EO and social data 248 Conclusions 252 References 252
Contents viii 12. An indicator-based approach to sustainable management of natural resources 255 Esther S. Parish, Virginia H. Dale, Maggie Davis, Rebecca A. Efroymson, Michael R. Hilliard, Henriette Jager, Keith L. Kline, Fei Xie Introduction 255 Selecting and prioritizing indicators 258 Indicator datasets and data science considerations 259 Data analytics 264 Implications for society & policy 274 Conclusion 275 References 276 PART 5 Conclusion 281 13. Research and development for increased application of data science in sustainability analysis 283 Jennifer B. Dunn, Prasanna Balaprakash Introduction 283 Needs for data to enable data science in sustainability 283 Data sources 286 Data science advances 287 Conclusion 290 References 291 Index 293
ix Mikaela Algren Colorado School of Mines, Golden, CO, USA Hossein Ataei Department of Civil, Materials, and Environmental Engineering, University of Illinois at Chicago, Chicago, IL, USA Prasanna Balaprakash Northwestern-Argonne Institute of Science and Engineering, Evanston, IL; Math and Computer Science Division,Argonne National Laboratory, Lemont, IL, USA Elisabeth N. Bui CSIRO Land and Water, Canberra,ACT,Australia Sergio Castellanos The University of Texas at Austin, Austin,TX, USA Deboleena Chakraborty Dow, Midland, MI, USA Jacqueline M. Cole Cavendish Laboratory, Department of Physics, University of Cambridge, J. J.Thomson Avenue, Cambridge, UK; ISIS Neutron and Muon Facility, STFC Rutherford Appleton Laboratory, Harwell Science and Innovation Campus, Oxfordshire, Didcot, UK; Department of Chemical Engineering and Biotechnology, University of Cambridge,West Cambridge Site, Philippa Fawcett Drive, Cambridge, UK Virginia H. Dale University of Tennessee Knoxville,TN, USA Maggie Davis Oak Ridge National Laboratory, Oak Ridge,TN, USA Sybil Derrible Department of Civil, Materials, and Environmental Engineering, University of Illinois at Chicago, Chicago, IL, USA Leon R. Devereux Cavendish Laboratory, Department of Physics, University of Cambridge, J. J.Thomson Avenue, Cambridge, UK Jennifer B. Dunn Northwestern-Argonne Institute of Science and Engineering, Evanston, IL, USA; Center for Engineering Sustainability and Resilience, Northwestern University, Evanston, IL, USA; Chemical and Biological Engineering, Northwestern University, Evanston, IL, USA Rebecca A. Efroymson Oak Ridge National Laboratory, Oak Ridge,TN, USA Contributors
Contributors x Wendy Fisher Colorado School of Mines, Golden, CO, USA Stefania Grimaldi Monash University, Department of Civil Engineering, Clayton,VIC,Australia Cassandra Handan-Nader Department of Political Science,Stanford University,Stanford,CA,USA;Regulation,Evaluation, and Governance Lab,Stanford,CA,USA Richard K. Helling Dow, Midland, MI, USA Michael R. Hilliard Oak Ridge National Laboratory, Oak Ridge,TN, USA Daniel E. Ho Department of Political Science,Stanford University,Stanford,CA,USA;Stanford Institute for Human-Centered Artificial Intelligence,Stanford,CA,USA;Regulation,Evaluation,and Governance Lab,Stanford,CA,USA;Stanford Institute for Economic Policy Research,Stanford, CA,USA;Stanford Law School,Stanford,CA,USA Henriette Jager Oak Ridge National Laboratory, Oak Ridge,TN, USA Daniel M. Kammen University of California, Berkeley, CA, USA Evgenia Kapousouz Department of Public Administration, University of Illinois at Chicago, Chicago, IL, USA Keith L. Kline Oak Ridge National Laboratory, Oak Ridge,TN, USA Amy E. Landis Colorado School of Mines, Golden, CO, USA Larry Y. Liu U.S. Court of Appeals for the Eleventh Circuit, Birmington,AL, USA Charlotte L.J. Marcinko School of Engineering, University of Southampton, Southampton, UK Esther S. Parish Oak Ridge National Laboratory, Oak Ridge,TN, USA Valentijn R.N. Pauwels Monash University, Department of Civil Engineering, Clayton,VIC,Australia Abolfazl Seyrfar Department of Civil, Materials, and Environmental Engineering, University of Illinois at Chicago, Chicago, IL, USA Ashkan Shokri Monash University, Department of Civil Engineering, Clayton,VIC,Australia Deborah A. Sunter Tufts University, Medford, MA, USA
Contributors xi Niraj Swami Senior Director, Conservation Technology Strategy & Enablement at The Nature Conservancy, Founding Partner, SCADVentures, Chicago, IL, USA Gary R.Watmough School of Geosciences, University of Edinburgh, Edinburgh, UK Ashley Wright Monash University, Department of Civil Engineering, Clayton,VIC,Australia Fei Xie Oak Ridge National Laboratory, Oak Ridge,TN, USA
1 Data Science Applied to Sustainability Analysis © 2021 Elsevier Inc. DOI: 10.1016/C2018-0-02415-9 All rights reserved. CHAPTER 1 Overview of data science and sustainability analysis Prasanna Balaprakasha,b , Jennifer B. Dunna,c,d a Northwestern-Argonne Institute of Science and Engineering, Evanston, IL, USA b Math and Computer Science Division,Argonne National Laboratory, Lemont, IL, USA c Center for Engineering Sustainability and Resilience, Northwestern University, Evanston, IL, USA d Chemical and Biological Engineering, Northwestern University, Evanston, IL, USA Chapter Outlines Data science is central to advances in sustainability 1 Types of sustainability analyses 6 Data science tools 7 Supervised learning 8 Unsupervised learning 9 Reinforcement learning 10 Tools 10 Overview of case studies in data science in sustainability 10 Data science is central to advances in sustainability The frequently-used term sustainability is often defined per the Brundtland Report’s definition of sustainable development: “Sustainable development is development that meets the needs of the present without com- promising the ability of future generations to meet their own needs.” (World Commission on Environment and Our Common, 1987) For the purposes of this book, we intend the term sustainability to mean the poten- tial for achieving a high quality of life in human, social, environmental, and economic systems, both today and in the future. For this potential to be realized, society must reach a point where air quality, water quality,and soil health are robust and do not pose a threat to ecosystem or human health. Air quality remains unsatisfactory globally, and is the fifth risk factor for global mortal- ity in 2017, associated with 4.9 deaths (Fig. 1.1) and 147 million healthy life years lost. (Health Effects Institute 2019) In addition, water quality globally is a challenge, with only half of water bodies exhibiting good quality per United Nations Sustainability
Data science applied to sustainability analysis 2 Development Goal monitoring initiatives (Fig. 1.2). (UN Environment, 2018) In addi- tion to air and water quality, soil quality has significant implications for human health yet can suffer from pollution from industry, mining, or waste disposal. In Europe, sites with likely soil contamination number 340,000 with only one-third of these under- going detailed study and only 15 percent of those remediated. (Food and Agriculture Organization of the United Nations 2015) In the US, the Environmental Protection Agency has remediated 9.3 million ha of contaminated land with 160 contaminated sites on the Superfund National Priorities List remaining to be evaluated and 49 new sites proposed to be added to the list. In addition to contamination, other challenges to soil sustainability include erosion and loss of organic carbon. (Food and Agriculture Organization of the United Nations 2015) Similarly, the abundance of energy, preferably produced from minimally polluting and renewable resources, and clean water is essential for society’s survival. Like the quality of air, water, and soil, energy and water use are closely tied to human activity.The world has nearly tripled its energy consumption since 1971 (Fig.1.3).(International Energy Agency. IEA, 2020) Fossil fuels (coal, oil, and natural gas) continue to dominate the production of energy with the attendant impacts from mining and extraction and subsequent combus- tion of these sources which releases greenhouse gas emissions into the air which contain carbon that had been long-sequestered in the earth.Combustion of fossil fuels diminishes air quality. In addition to considering total energy production, it is important to evaluate energy efficiency,which has only improved by 1.2 percent from 2017 to 2018.Improving energy efficiency is one of the best strategies available to cutting energy consumption Fig. 1.1 Number of deaths attributable to air pollution in 2017. Data source: Global Burden of Disease Study 2017. IHME, 2018. (Health Effects Institute 2019).
Overview of data science and sustainability analysis 3 and associated pollution. (International Energy Agency 2019) The production of energy, along with many other activities,including agriculture consumes water.Correspondingly, as the population has increased and clean water supplies have diminished, water scarcity is a reality for approximately one-half of the global population. (Boretti and Rosa, 2019) Fig. 1.2 Proportion of water bodies with good ambient water quality (percent) in 2017 (UN Environ- ment 2018). 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Overall Rivers Proportion of water bodies with good ambient water quality Open Water Bodies Groundwater Fig. 1.3 World total energy supply by source (million tons of oil equivalent). (International Energy Agency. IEA, 2020). 16,000 14,000 12,000 10,000 8,000 6,000 4,000 2,000 0 1971 1976 Energy (million tons oil equivalent) 1981 1986 1991 1996 2001 2006 2011 2016 Coal Oil Natural gas Nuclear Hydro Biofuels and waste Other
Data science applied to sustainability analysis 4 Challenges faced in achieving sustainability include improving food and water secu- rity, maintaining biodiversity, reducing air and water pollution, reducing greenhouse gas emissions, increasing reuse and recycling, and increasing system-level efficiencies in energy, urban, agricultural and industrial systems. Furthermore, extracting and disposing of the materials we need to make the equip- ment, devices, and food we need to run our society can be unsustainable, generating pollution and operating without concern for the long-term availability of critical materials. In fact, the very technologies society is relying on to address climate change, including wind turbines, solar panels, and lithium-ion batteries, rely on metals (cobalt, nickel, copper, rare earths) that are mined, often in developing countries where envi- ronmental regulations are often insufficient to protect populations from exposure to pollution in the air, water, and soil. (Sovacool et al., 2020) Conserving natural lands is an important part of ensuring a healthy and productive future for human society. Natural lands such as grasslands, wetlands, and forests provide innumerable ecosystems services such as mitigating floods, sequestering carbon, and enhancing biodiversity.Targeted conservation initiatives are required to slow the pace and extent of extinction, improves environmental quality, and retain the inspirational value of nature. (Balvanera, 2019) Economic and consumer preference drivers often can favor technology and soci- etal developments that advance towards sustainability, but law and policy are important drivers as well. (Ashford and Hall, 2011) For example, one reason energy efficiency gains have faltered (Fig. 1.4) is a lack of clear policy to advance energy efficiency. (International Energy Agency 2019) Finally, social well-being, in part as indicated by the portion of the world’s popula- tion that has can viably provide food and other basic needs for themselves and their Fig. 1.4 Global Improvements in Primary Energy Intensity 2000–20186 . 3.5% 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% 2000-2009 2010-2014 2015 2016 2017 2018
Overview of data science and sustainability analysis 5 families, is an important element of sustainability. Global levels of poverty between 2013 and 2015 declined through all regions of World Bank analysis yet the percent of people living at the International Poverty Line of $1.90 per day stayed relatively constant in many of these regions, showing a decrease most notably in South Asia (Fig. 1.5).The over 700 million people globally living below this poverty line in 2015 is an unignorable indication that sustainability has not yet been attained. Undoubtedly, the breadth of earth, industrial, and societal systems that contribute to sustainability is immense. Developing technology, societal, and policy approaches to address each facet of sustainability can be guided by analyses that point the way, for example, towards pollution or water scarcity hotspots, the most impactful energy efficiency technologies, or regionally-specific conservation strategies. These analyses can make use of ever-growing volumes of data including satellite imagery, continuous sensor data from industrial processes, social media data, and environmental sensors, to name only a few.As a result, data science techniques have become central to addressing sustainability challenges and this role will only expand in the future. Accordingly, we have assembled this book with the contribution of co-authors who are addressing sustainability challenges in the spheres of environmental health, energy and water, sustainable industrial systems, and society and policy. Our intention is to provide a well-rounded set of case studies addressing different challenges using varying types of sustainability analysis and data to serve as a reference for analysts who seek to employ data science in their work and for data scientists looking to apply their skills to sustainability challenges.Another audience for this book will be policy makers who rely on sustainability analyses as a decision making tool to evaluate how governments Fig. 1.5 Number of people and percent of population at the International Poverty Line of $1.90/day (2011 PPP). (World Bank 2018) 900 800 700 600 500 400 300 200 100 0 East Asia and Pacific Million People Europe and Central Asia Latin America and the Carribean Percent of People Population 2013 Population 2015 % 2013 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% % 2015 Middle East and North Africa South Asia Sub-Saharan Africa World Total
Data science applied to sustainability analysis 6 could collect data that would support these efforts and use the results of these analyses in policymaking.Additionally, this book could be used in data science and systems analysis classrooms to provide case study examples, especially at the graduate student level. In the remainder of this introductory chapter, we review different types of analy- ses that guide our understanding of and action towards increasing sustainability. We provide an overview of data science tools that can be used in sustainability analyses. Finally, we introduce the different case studies readers will encounter in the remaining chapters.We note that the concluding chapter of this book summarizes data gaps and research needs for the further building of data science applications in sustainability analysis. Types of sustainability analyses The term “sustainability analysis” is meant to be broad for this book to capture a wide range of analyses that can address or evaluate society’s advancement towards sustainable resource management and wellbeing. We summarize some examples of analysis types that fall under the term sustainability analysis in Fig.1.6.As one example,natural systems modeling improves our understanding of the geoscience, bioscience, and social science that underpins natural systems relies on data analysis and modeling, with an effort to move towards prediction. Hydrologic modeling, for example, based on land character- istics and precipitation data, can help predict the location and effects of flooding from major precipitation events. Soil carbon modeling that explores the influence of agricul- tural management practices on levels of carbon storage in soils is another example of analysis that could be grouped under “sustainability analysis.” Furthermore, modeling of air pollution dispersion could also be categorized under this umbrella. All of these natural systems models contain parameters that must be estimated based on evaluation of data sets. Furthermore, many types of analyses can enhance the design of industrial, energy, and water systems that offer sustainability improvements over the status quo. As one primary example, machine learning can be used to speed up the design of new mate- rials that can be used in any number of important sustainability applications from Fig. 1.6 Examples of sustainability analysis types that are increasingly using data science techniques.
Overview of data science and sustainability analysis 7 designing membranes that exhibit less fouling in water treatment applications thereby reducing energy and chemicals used in wastewater treatment to exploring next genera- tion lithium-ion battery chemistries. Additionally, as the Industrial Internet of Things continues to expand, analysts will apply data science techniques to identify opportuni- ties to improve the energy, water, and material efficiency of industrial processes. Finally, evaluating the progress of consumers’ adoption of technology that will be more energy or water efficient, for example, is another important type of sustainability analysis.This type of analysis could be based on earth observation data in the case of adoption of large infrastructure or based on social media posts that indicate shifts in technology use in the home, on the road, or in the workplace. Two mainstays of sustainable systems analysis are life cycle assessment (LCA) and materials flow analysis (MFA). Whereas LCA evaluates the environmental effects of a product or process – from fuels to electronics to foods – MFA tracks the flows of com- modities within a system boundary, which could be a city, a region, or a nation. LCA and MFA are at the very beginning of applying data science techniques, in general because datasets are often insufficiently large to allow data science approaches to offer value.As the data revolution continues,these two analysis types have many opportunities to leverage data science techniques. Finally, evaluations of social well-being are another important pillar of sustainability analyses because sustainability is often described as having three pillars – economic, social, and environmental. One expanding enabler of using data science approaches in social well-being evaluations is satellite imagery, which provides a bird’s eye view of living conditions for Earth’s inhabitants.While these data can show us these conditions, they cannot identify what has caused them.This second and critical step will require the linkage of image interpretation and causal analysis. Regardless of analysis type, data availability is a cornerstone of all of these analyses. In some instances that remain data sparse, the use of data science techniques in these areas is anticipatory rather than widespread. Furthermore, the examples we provide here are not all-encompassing and the list of types of sustainability analyses that benefit from data science approaches today and into the future will evolve and grow. Data science tools Broadly speaking, data science is an inter-disciplinary field that adopts data collection, pre-processing, meaning/useful feature extraction methods, data exploration meth- ods, and predictive models to extract knowledge from a wide range of structured and unstructured data.Given the structure,size,heterogeneity,and complexity of the data sets, a wide range of data science tools and techniques have been developed. Among them, statistical machine learning is a prominent class of methods that are used and adopted for many data science task. Next, we will review three widely used subclass of ML methods.
Data science applied to sustainability analysis 8 Supervised learning It is used to model the functional relationship between the output variables and one or more independent input variables. Typically, the original function relationship is unknown and/or hard to derive in an analytical form.The approach starts with a set of training data given as a large set of input-output pair.The goal is to find a surrogate function for original function relationship such that the difference between prediction from the surrogate function and the observed value is minimal for all input-output pair in the training data and the unseen testing data. Several supervised learning algorithms exist in the ML literature. Based on the functionality, one can group them as follows: regularization, instance-based methods, recursive partitioning, kernel-based methods, artificial neural networks, bagging, and boosting methods. Often, the best method depends on the data and type of the modeling task, such as volume of data, variety of data, and speed required for training and inference. Here, we cover several widely adopted algorithms to cover different groups. We will review them from regression perspective (predicting a scalar value).Without loss of generality, most of these methods also handle classification (predicting a class). Multivariate linear regression (Bishop, 2006) is one of the most simple methods for modeling the functional relationship between inputs and output. It models the functional relationship using a linear equation.This is given by the sum of product of each input with a scaling factor. A bias factor is also added to the equation.The mul- tivariate linear regression involves finding the scaling factors and the bias. It is one of the well understood method and often preferred for interpretability and simplicity. It is important that data science practitioners try and adopt this method as a baseline and comparison to other methods. Ridge regression (Hoerl and Kennard, 1970) is a regularization algorithm that is designed to reduce the model complexity so that the model does not overfit the train- ing data. This overfitting occurs in supervised learning when the model learns small variations and/or noise in the training set and consequently loses prediction accuracy on the testing data.To do so, in addition to minimizing the error between predicted and actual observations, the method penalizes the training objective with respect to input coefficients and achieves tradeoff between minimizing the error and minimizing the sum of the square of the coefficients. k-nearest-neighbor regression (Bishop, 2006) belongs to the class of instance- based methods, where the training data is stored in memory and the model is built only during testing. Given a testing point, the method first finds k nearest input points in the training data and returns the prediction as the average of k outputs.Typically, k and the nearest distance metric are user defined hyperparameters. Support vector machine (Drucker et al., 1996) is a widely-used kernel-based method. It uses a kernel function to project the input space onto a higher-dimensional feature space; a linear regression is performed in the transformed space.The training is
Overview of data science and sustainability analysis 9 formulated as a convex quadratic optimization problem, for which efficient optimiza- tion algorithms are utilized.The effectiveness of this method depends on a good choice of kernel type and their hyperparameters. Decision tree regression (Breiman, 1984) belongs to the class of recursive parti- tioning methods. It recursively splits the multidimensional input space of training points into regions such that inputs with similar outputs fall within the same region.The splits give rise to a set of if-else rules. For each region, an average over the output values is computed and stored at the end of each rule. Given a new testing point, the decision tree employs the if-else rule to return the stored value as the predicted value. Random forest (Breiman, 2001) is a bagging approach that considers random subsamples of the training dataset and builds a decision tree on each subsample. Given a new test data point, the prediction from each tree is averaged to obtain the predicted value. Gradient boosting regression (Friedman, 2002) is similar to random forest but the trees are constructed sequentially on each random subsample. The key idea is to build each tree to minimize the error of the previous tree. Deep neural networks (Goodfellow et al., 2016) belong to the class of artificial neural networks.They are characterized by stacked layers, where each layer is composed of a number of units. Each unit receives inputs from units from previous layers, which are combined in a weighted linear fashion and passed through a nonlinear function. The first layer receives the training points and the predictions are obtained from the last layer of the stack.The training phase consists of modifying the weights of the stacked layers to minimize the prediction error on the training data set.This is typically done by stochastic gradient descent optimization method that computes the gradients of the objective function with respect to all the weights in the network and uses them to update the weights. Unsupervised learning Traditionally, unsupervised learning methods were used for exploratory analysis. (Bishop,2006) Notably,clustering and dimension reduction methods were adopted for a wide range of data science tasks.The former computes the distances between the points in the given data using a distance metric, which is then used to group similar points. The latter is often employed to project the high dimensional data into low dimensional embedding space for visualization. In recent years, auto encoders, a class of deep neu- ral networks, have received significant attention for dimension reduction method due to their ability to perform effective nonlinear dimension reduction and handle large amount of data.Another key advancement in the area of unsupervised learning is gener- ative modeling, which has potential to understand and explain the underlying structure of the input data when there are few–or even no–labels.A promising generative model- ing approach that has received much recent attention is generative adversarial networks
Data science applied to sustainability analysis 10 (GAN). (Goodfellow et al., 2014) The basic idea in GAN is to train two deep neural networks simultaneously and capture the domain-specific features and representations from the unlabeled data and deploy them as labeled data becomes available.For example, GANs can produce high quality synthetic images of real-world objects without having any explicit labels of what those objects are. By automatically extracting the underlying structure of the inputs without labels, GANs can empower supervised learning methods to understand the context of the domain in which they operate. Reinforcement learning It is an approach that is concerned with is concerned with training agents for autono- mous design and control. (Sutton and Barto, 2018) The agents interact within an envi- ronment, receive rewards, and use them to improve the actions iteratively using training settings.The agents once trained can be deployed for control in test settings. Tools The data science software ecosystem is quite vibrant has a wide range of software tools and many of them are open-source. Scikit-learn (Pedregosa et al., 2018) is one of the widely used package for numerous data science tasks. It has implementation of prepro- cessing, unsupervised, and supervised learning methods that are integral part of many data science pipelines. Similarly, R project for statistical computing (R Core Team, 2021) provides a number of libraries to build data science pipelines with minimal effort. Jupyter notebook (Kluyver et al., 2016) and R studio (Allaire, 2012) are productivity centric integrated development editors for interactive data science code development. Tensorflow (Abadi et al., 2016) and Pytorch (Paszke et al., 2019) are packages for dif- ferentiable computing and are widely used for the design and development of deep neural network models.Python and R ecosystem provides a number of libraries for data visualization (for example, Matplotlib (Hunter, 2007) and ggplot2 (Wickham, 2011)). RapidMiner, (Mierswa and Klinkenberg, 2018) Weka (Hall et al., 2009), and KNIME (Berthold et al., 2009) software tools designed for users with minimal programming experience.They provide easy to use interfaces to build data science pipelines but do not provide flexibility and configurability as programming-intensive software stack. Overview of case studies in data science in sustainability Data science techniques have been applied to numerous domains within the sustainabil- ity field.For example,social media data have been analyzed with data science techniques to inform an understanding of urban sustainability including aspects like mobility and economic development (Ilieva and McPhearson, 2018) and even waste minimization in beef supply chain. (Mishra and Singh, 2018) In general, the agricultural sector holds much promise for applications of data science to improve farming sustainability such as
Overview of data science and sustainability analysis 11 reducing use of fertilizer and irrigation. (Kamilaris et al., 2017) Considering the social side of sustainability, predictive analytics and data visualization have been used to study and improve the humanitarian supply chain. (Gupta et al., 2019) With such an expansive space of intersection between data science and sustainability, this book covers only a subset of an ever-growing field.We have focused on the broad top- ics of environmental quality and sustainability,energy and water,sustainable systems analy- sis, and society and policy. Fig. 1.7 places each chapter in this book in one of these topics. Environmental Quality and Sustainability focuses on how we can better under- stand natural ecosystems and design strategies to protect them and improve air, water, and soil quality. Swami (Chapter 2) examines the many ways artificial intelligence can contribute to conservation efforts. Bui (Chapter 3) describe the application of machine learning techniques including supervised pattern recognition, random forests, support vector machines, and deep learning to investigate spatial patterns such as species distri- bution, streamflow, and land use within Australian Critical Zones. For this application, machine learning techniques have proven helpful to predict spatial patterns, identify regions vulnerable to factors such as erosion or soil organic carbon loss, and to find the drivers of spatial patterns. Pauwels et al. (Chapter 4) describe several methods that have been used to improve hydrologic modeling to inform water resources for Australia. The methods described include Bayesian techniques and Monte Carlo methods that demonstrate improvements in parameter estimation over other methods and can better predict flooding. As described in Section 1.1, energy use and water consumption continue to rise. Kapousouz et al. (Chapter 5) use clustering to explore spatial and temporal use patterns Fig. 1.7 Organization of case studies in this book.
Data science applied to sustainability analysis 12 in energy and water consumption at the state level. Identifying patterns can lead to technology and policy development to reduce resource consumption. Developing and deploying technology to harness renewable energy is one important approach to mini- mizing the influence of energy production. Devereux and Cole (Chapter 7) explore avenues for using machine learning to develop solar photovoltaic cells that move this technology to optimal performance. One example is to use machine learning to predict property based on structure.Another relevant application of machine learning is to carry out high-throughput computational screening.A final option is to automatically gener- ate property databases. One helpful tool in this case is using natural language processing to mine the technical literature for property information that can be include in such a database.As new technology develops, it is helpful to evaluate how it is being deployed and used in the real world. One reality of solar photovoltaic deployment is that it tends to be limited in areas of lower socio-economic status. Castellanos et al. (Chapter 6) integrated Google Project Sunroof and United States Census data and carried out regression analysis and bootstrapping to explore racial and ethnic disparities of rooftop solar voltaic technology. With this information in hand, interventions can be better designed to increase solar PV deployment. In quantifying sustainability, it is important to remember that no one technology, policy, region, or other subdivision of a full system acts in isolation. For example, rede- signing a product with new materials to achieve a lighter weight could reduce the fuel consumed in shipping it, but the new materials themselves could be more energy inten- sive to produce than the original materials. For this reason, it is important to consider full systems when developing new technologies, considering changes to existing tech- nologies, or considering policies that could reduce environmental burdens in one por- tion of the supply chain only to increase them elsewhere. LCA is one analysis approach, standardized by the International Standards Organization, (ISO 2006; ISO 2006) that has been used to evaluate how changes in one step of a life cycle (e.g., manufactur- ing) alter the overall environmental effects of a product or process. Algren and Landis (Chapter 8) consider how machine learning can be best applied in LCA and provide several examples.Chakraborty and Helling (Chapter 9) describe how LCA in the indus- trial sector in particular has the power to guide supply chains to enhanced sustainability especially as companies gain access to data required to build comprehensive LCAs.They provide a vision for what this future state could look like. In addition to understanding the natural world’s health and how it functions, how technology influences sustainability through energy, water consumption and supply chain effects, focusing on societal behavior and policy development can also yield insights into enhancing overall societal sustainability. Some policies designed to limit pollution can be challenging to enforce. Handan-Nader et al. (Chapter 10) describe the use of machine learning with high-resolution satellite imagery to direct environmental regulation enforce- ment.They provide examples in detecting oil spills, deforestation, and air pollutant emis- sions.Subsequently,they describe their work with identifying concentrated animal feeding
Overview of data science and sustainability analysis 13 operations that are of interest to environmental regulators but may be difficult to locate because there is no public database of their locations. Earth observation (EO) data such as satellite imagery is in general showing increasing promise for supporting decision making and policy development in the sustainability space, including when relationships can be drawn between information from EO and data from other sources such as air pollutant monitors and economic databases, for example.Watmough and Marcinko (Chapter 11) explore how EO data can be used to target, develop, and evaluate policy to reduce pov- erty. Finally, Parish et al. (Chapter 12) describe how multiple field measurements and EO data can be integrated to inform sustainability assessments through the use of indicators, including developing techniques to monetize ecosystem services. The final chapter of this book outlines research needs that must be addressed to con- tinue to expand the application of data science techniques to sustainability challenges. As the intersection between data science and sustainability continues to develop and mature, the science of understanding earth system, technologies that are energy- and water-efficient, and policies that encourage the ongoing reduction of pollution, includ- ing greenhouse gas emissions, and poverty will benefit. References Abadi, M., Barham, P., Chen, J., Davis,A., Dean, J., Devin, M., Ghemawat, S., Irving, G., Isard, M., Kudlur, M., Levenberg, J., Monga, R., Moore, S., Murray, D.G., Steinter, B.,Tucker, P.,Vasudevan,V.,Warden, P., Wicke, M.,Yu,Y., Zheng, X., 2016.Tensorflow: a system for large-scale machine learning, Proceedings of the 12th USENIX Symposium on Operating Systems Design and Implementation, 265–283 https://www.usenix.org/conference/osdi16/technical-sessions/presentation/abadi. Allaire, J., 2012. RStudio: integrated development environment for R. in vol. 770, 394. Ashford, N.A., Hall, R.P., 2011.The Importance of regulation-induced innovation for sustainable develop- ment. Sustainability 3, 270–292. Balvanera, P., 2019. Societal burdens of nature loss. Science 366, 184–185. Berthold, M., Cebron, N., Dill, F., Gabriel,T.R., Kotter,T., Meinl,T., Ohl, P.,Thiel, K.,Wiswedel, B., 2009. KNIME-the Konstanz information miner: version 2.0 and beyond. AcM SIGKDD explorations Newsletter. https://doi.org/10.1145/1656274.1656280. Bishop, C.M., 2006. Pattern Recognition and Machine Learning. Springer, NewYork, NY. Boretti,A., Rosa, L., 2019. Reassessing the projections of theWorldWater Development Report. npj Clean Water 2, 1–6. Breiman, L., 1984. Classification and Regression Trees. Chapman Hall/CRC, NewYork, NY. Breiman, L., 2001. Random Forests. Mach. Learn. 45, 5–32. Drucker, H., Burges, C.J.C., Kaufman, L., Smola, A.,Vapnik,V, 1996. Support vector regression machines, Proceedings of the 9th International Conference on Neural Information Processing Systems. MIT Press, pp. 155–161. Food and Agriculture Organization of the United Nations. Status of the world’s soil resources. (2015). Friedman, J.H., 2002. Stochastic gradient boosting. Comput. Stat. Data Anal. 38, 367–378. Goodfellow, I., Bengio,Y., Courvile,A., 2016. Deep Learning. MIT Press. http://www.deeplearningbook. org/. Goodfellow, I.J. et al. Generative adversarial networks. arXiv:1406.2661[cs, stat] (2014). Gupta, S.,Altay, N., Luo, Z., 2019. Big data in humanitarian supply chain management: a review and further research directions.Ann Oper Res 283, 1153–1173. Hall, M., Eibe, F., Holmes, G., Pfahringer, B., Reutemann, P.,Witten,A.H., 2009.The WEKA data mining software: an update.ACM SIGKDD explorations newsletter 11, 10–18 https://www.kdd.org/explora- tion_files/p2V11n1.pdf.
Data science applied to sustainability analysis 14 Health Effects Institute, 2019. State of Global Air 2019. Special Report. https://www.stateofglobalair.org/ sites/default/files/soga_2019_report.pdf. Hoerl, A.E., Kennard, R.W., 1970. Ridge regression: biased estimation for nonorthogonal problems. Technometrics 12, 55–67. Hunter, J.D., 2007. Matplotlib: a 2D graphics environment. Comput. Sci. Eng. 9, 90–95. I. Mierswa R. Klinkenberg. RapidMiner Studio (9.1) [Data science, machine learning, predictive analyt- ics]. (2018). Ilieva, R.T., McPhearson,T., 2018. Social-media data for urban sustainability. Nat. Sustain. 1, 553–565. International Energy Agency. IEA, 2020.World total energy supply by source, 1971-2018. https://www.iea. org/data-and-statistics/charts/world-total-energy-supply-by-source-1971-2018. International EnergyAgency.Energy efficiency.2019.https://www.iea.org/reports/energy-efficiency-2019 (2019). ISO. Environmental management: life cycle assessment; Principles and framework. (2006). ISO. Environmental management: life cycle assessment; Requirements and guidelines. (2006). Kamilaris,A., Kartakoullis,A., Prenafeta-Boldú, F.X., 2017.A review on the practice of big data analysis in agriculture. Comput. Electron.Agric. 143, 23–37. Kluyver, T., et al., 2016. Jupyter Notebooks – a Publishing Format for Reproducible Computational Workflows, 87–90. Mishra, N., Singh, A., 2018. Use of Twitter data for waste minimisation in beef supply chain. Ann. Oper. Res. 270, 337–359. Paszke, A., et al., 2019. PyTorch: an imperative style, high-performance deep learning library. In:Wallach, H. et al (Ed.), Advances in Neural Information Processing Systems 32. Curran Associates, Inc., pp. 8026–8037. Pedregosa, F. et al. Scikit-learn: machine learning in python. arXiv:1201.0490[cs] (2018). R Core Team. http://www.R-project.org/. Sovacool, B.K, et al., 2020. Sustainable minerals and metals for a low-carbon future. Science 367, 30–33. Sutton, R.S., Barto,A.G., 2018. Reinforcement Learning:An Introduction.The MIT Press. UN Environment. Progress on ambient water quality, piloting the monitoring methodology and intitial findings for SDG indicator 6.3.2, 2018. (2018). Wickham, H., 2011. ggplot2.WIREs Computational Statistics 3, 180–185. World Bank. Poverty and shared prosperity 2018. Piecing together the poverty puzzle. https://www.world- bank.org/en/publication/poverty-and-shared-prosperity (2018). World Commission on Environment and Development. Our common future. (1987).
15 PART 1 Environmental Health and Sustainability 2. Applying AI to conservation challenges 17 3. Water balance estimation in Australia through the merging of satellite observations with models 29 4. Machine learning in the Australian critical zone 43
17 Data Science Applied to Sustainability Analysis © 2021 Elsevier Inc. DOI: 10.1016/C2018-0-02415-9 All rights reserved. CHAPTER 2 Applying AI to conservation challenges Niraj Swami Senior Director, Conservation Technology Strategy Enablement at The Nature Conservancy, Founding Partner, SCADVentures, Chicago, IL, USA Chapter Outlines Introduction 17 What is adaptive management? 19 Creating value with AI 21 Planning and implementing with AI 21 Monitoring evaluating with AI 22 Learning adapting with AI 23 The impact technology canvas: synchronizing AI initiatives 24 Summary 26 Introduction We live in a world where we’re surrounded by experiences,insights and tasks that are enriched by Artificial Intelligence (AI). From smart speakers in our living room that can distinctively understand commands from different voices to emails that basically write themselves based on your context. Conservation is no stranger to the breadth of capabilities AI enriches. A quick Google search on “AI in Conservation” will reveal numerous examples of how AI has been used in the conservation space.We did a quick survey of innovative AI-based initiatives at The Nature Conservancy and the greater sustainability technol- ogy community to highlight some intriguing areas of work: • Leveraging image-classification technologies to decipher land types from satellite imagery to further inform flood insurance policy schemes (Fig. 2.1). • Leveraging AI to extract patterns from user-generated social media posts to under- stand impact and economic significance of eco-tourism (near coral reefs) (Mapping Ocean Wealth na). • Using machine learning to forecast future water runoffs by modeling the physics of snowfall and snowmelt in a given region (Fig. 2.2). • Detecting and tagging specific development attributes from satellite imagery, such as plotting solar and wind energy plants in India and identify off-highway vehicle activ- ity in Mojave desert (Fig. 2.3). • Unmanned marine drones (Fig. 2.4) applying AI to self-navigate harsh ocean condi- tions using an array of on-device sensors, 360° cameras and GPS data, while alerting targets of interest remotely and also collecting hydrological data for on-shore analysis.
Data science applied to sustainability analysis 18 With data being collected on-device,these drones perform advanced image and time series data analysis, machine learning and other post-processing tasks on the edge. The use cases are aplenty. However, a notable pattern in these examples is the reli- ance on one particular data source - remote sensing data (images and other layers of insights) from satellites orbiting our planet. In 2014, the Wildlife Conservation Society shared “Ten ways remote sensing can contribute to conservation” (Rose et al., 2014) - primarily identifying use cases in monitoring conditions, understanding and predicting environmental changes. A quick Google search on “remote sensing datasets” will give you dozens of imagery datasets that can be leveraged for building AI solutions. While the readily available satellite imagery dataset is a great candidate for data sci- ence, what other scenarios could AI practitioners and data scientists pursue? How can we strategically look for conservation-related challenges best positioned for a boost Fig. 2.1 Community Rating System by Federal Emergency Management Agency (Open Space na). Fig. 2.2 Rapid progression of snowmelt and vegetation growth in the South Payette River Basin at three selected times in the spring (South Payette River Case Study na).
Applying AI to conservation challenges 19 from cognitive technologies? In this chapter, we explore how AI-based solutions may be applied at various stages of a conservation project by looking through an innovator’s lens at Adaptive Management. What is adaptive management? Conservation is science guiding action, with the goal to prevent wasteful use of a natu- ral resource. Let’s explore the funnel of information in a typical conservation project. Fig. 2.3 Google Earth images of Off-Highway Vehicle (OHV) route proliferation, courtesy of John Zablocki and Michael Clifford at The Nature Conservancy. Fig. 2.4 Wind-Powered Ocean Drones created by SailDrone (Redefining Ocean Data Collection na).
Data science applied to sustainability analysis 20 A theory of change starts from a fundamental hypothesis. Formulated by research, a claim is then reviewed by peers, revised and published for a potential practitioner to implement.We then scope out a conservation project, detailing the business case, stake- holders and intended impact. Once implemented, we collect insights and analyze it to understand if our theory is actually working.Based on those insights,we learn and adapt accordingly.As work moves along this spectrum, the actors shift and so do stakeholders. This information funnel leans on the principles of adaptive management. Grounded in systems theory, experimental science and business, adaptive manage- ment is a systemic approach for improving natural resource management by learn- ing from management outcomes (as per the US Department of the Interior https:// www.doi.gov/sites/doi.gov/files/migrated/ppa/upload/Chapter1.pdf). Conservation programs often lean on this approach to drive strategic decision-making via a cyclical “always learning” mindset (Fig. 2.5). However, conservation is a team sport - one where scientists, conservationists, field workers, management professionals, engineers, analysts, policy makers, business leaders and citizens must play together to drive value for our planet.As we look at the broader scope of information needs, knowledge tasks and gaps in day-to-day workflows of these various stakeholders through the lens of the Adaptive Management approach, we start exposing new AI-ready potential. Fig. 2.5 The Adaptive Management Cycle (West, na).
Applying AI to conservation challenges 21 Each stage of the plan-do-evaluate-learn-adjust cycle sparks new contexts to explore AI solutions for and the specific stakeholder relationships to solve for - how scientists and conservationists engage with management leaders, how can policy makers learn from field surveyors, how can we allocate enterprise value, private capital and citizen engagement to meet our sustainability goals. The following sections dive deeper into each of the stages of the cycle. Creating value with AI As we traverse the various stages of the Adaptive Management Cycle, we can expose AI-ready challenges and opportunities by looking for avenues to accelerate, augment, assist or automate sub tasks. Planning and implementing with AI Planning conservation projects often begins with science and outcome measures.A key task in this stage involves researching and collecting raw information from surveys, local reports,census data,stakeholder data,geospatial imagery and other unstructured content representing ground truth. Some potential areas for AI to aid are: • Natural language tools: extracting structured content from unstructured data sources can catalyze how scientists and policy makers can model and plan conserva- tion interventions by leaning on technologies like OCR (optical character recogni- tion), translation and natural language processing (to extract keywords and conser- vation-centric named entities). AI enthusiasts have also explored sentiment analysis of Twitter content (Fig. 2.6) to capture near real-time insights on how people and nature intersect (floods in Chennai India). Fig. 2.6 Sample social media content (Blog, 2017).
Data science applied to sustainability analysis 22 • Search and knowledge technologies: organizing and sharing information with science and management peers relies on contextually relevant knowledge - an area of search that has gained increased AI-based enrichment by services like Microsoft’s Azure Cognitive Search and Allen Institute’s Semantic Scholar (a free, AI-powered research tool for scientific literature, semanticscholar.org (AI-Powered Research Tool. Semantic Scholar na)).A growing area of opportunity is around analysis and extraction of structured data from multilingual content (e.g. handwritten survey and historical research reports in local languages). A majority of data collection in the field,for instance in regions like Africa and South Asia,has historically been done us- ing paper-based surveys and images. Scientists spend hours gleaning scanned images for insights they can use in their research and analysis.AI technologies like Optical Character Recognition (Matei et al., 2013) can help automate and speed up this digitization task, freeing up precious time for science teams. In addition, structured content from these handwritten corpora of knowledge can be used for more in- depth analysis (such as similarity analysis, entity and conceptual relationships). • Digital engagement: enterprises, government initiatives and citizens are increas- ingly accessing information and alerts via a wide spectrum digital channels - from social networks to smart assistants in their living rooms.AI-powered innovations such as Amazon Alexa Skills and notifications on smart watches enable personalized and engaging ways to connect real world data (e.g. weather forecasts, local water quality) to individual contexts (e.g. impact of drought to their businesses, water footprint of their purchases and behaviors).The City ofVirginia Beach has one such Alexa Skill (water sense na) that provides information on water levels and road closures for citizen users. • Data governance needs:downstream questions around data quality,availability and governance allow for experts in AI and data science to help guide what outcome measures are incorporated in the planning and implementation stages. Monitoring evaluating with AI Monitoring evaluating conservation projects entails engaging with field staff, remote technologies,analysis tools on the edge and a wide spectrum of monitoring methodolo- gies. Some potential areas for AI to aid are: • Edge assistance for field work:iNaturalist is a great example of how field data collec- tion can be combined with the image processing AI to give instant insights (iNaturalist na).Innovations in machine learning deployment,such asTensorflow Lite (Dokic et al., 2020) allow practitioners to expose prediction tools (e.g. to assess soil conditions from images) on observation data from mobile and embedded devices. SailDrone, a wind- powered unmanned sailing device (saildrone.com/technology), leverages edge analysis of solar-powered meteorological and oceanographic sensors and on-device cameras to autonomously navigate harsh ocean conditions and report data.
Applying AI to conservation challenges 23 • Anomaly detection of time series data: in projects where monitoring of condi- tion is performed using Internet of Things (IoT) devices and sensors, there is op- portunity to tap into cognitive technologies to monitor data quality,sensor status and detecting patterns on the fly.Data science approaches used in other industries,such as fraud detection (banking transactions),fault monitoring (utilities and manufacturing) and intelligent alerts (healthcare) may transfer to conservation projects with stream- ing or time series data (e.g. monitoring soil nutrient levels, water quality, pH, etc.). • Developing new remote sensing datasets: access to high quality labeled data- sets is key for supervised learning. Data science tools allow us to generate broader geospatial representations from highly specific conservation projects with on-the- ground sensors. For instance, a biometrician at The Nature Conservancy developed novel AI-based approaches to analyze sound recordings from bioacoustics sensors in a forest and produced a model that represented characteristics of the ecological soundscape across measured forest conditions (Geospatial Conservation Atlas na). • Data science for evaluation: in cases where remote sensing imagery and monitoring datasets are readily available, data science can help assess condition and generate in- sights.The U.S.Geological Survey’s Land Cover Monitoring,Assessment,and Project dataset (Land Change Monitoring,Assessment, and Projection na) is one such can- didate dataset that may be leveraged for land change monitoring,land parcel tracking and building monitoring and alert systems by applying machine learning models and image analysis algorithms. Learning adapting with AI The final two stages of Adaptive Management Cycle demand synthesis of the conser- vation project’s decision-making process (from planning and implementation) and the evidence of its effects (from monitoring and evaluating) for two primary reasons. First, to report and share our approach and outcomes with stakeholders that can help scale impact to other geographies and contexts. Second, to feed actionable insights back into the next iteration of planning, (re)prioritization and adjustments of strategies. Cognitive technologies give us tools to understand, represent and model for these tasks in a variety of ways: • Unraveling connections from impact: knowledge representation is a key step in capturing the nuances of an AI problem’s domain. Representing decision-models, field knowledge, evidence and evaluation of outcomes in a knowledge graph al- low advanced data analyses, such as clustering and similarity detection. Social net- works and digital media platforms (like Netflix) use knowledge graphs to understand patterns of human behavior with the outcome to drive action (post something or consume a clip). In conservation projects, a knowledge graph can help us identify similarity in evidence sets and evaluation methodologies between different projects to assess, audit and adjust the underlying decision-making process.
Data science applied to sustainability analysis 24 • Selection and prioritization: another application of conservation-related knowl- edge graphs is for leveraging our learning’s, constraints (costs, labor and time) and contextual knowledge (like policy and other enabling factors) to select and prioritize work. Fig. 2.7 highlights an example scenario. Imagine you have limited funds to deploy for a conservation project - one in India and another in Argentina.What if you didn’t have to pick one project? What if you could pick a component, or two, from each project that advances the overall learning and insights on work in India and Argentina? Breaking down the knowledge about each region into enabling fac- tors,tools,policy support and other attributes allows us to compare common themes, focal areas and subsequently draw insights from graph-based analyses, such as com- monalities, linkages and cluster detection. • Creating digital twins:according to an article published on Nature.com (Tao and Qi, 2019), there is an increasing number of scenarios where a “digital twin” of a complex system has been used to help detect risks, optimize choices and model outcomes. For instance, NASA uses a digital copy of its spacecraft to monitor status (Glaessgen and Stargel,2012).Singapore uses a digital twin to monitor and improve utilities (Tomorrow. Mag.2019).What if we were to create a digital twin of a conservation project? The impact technology canvas: synchronizing AI initiatives In the entrepreneurial community, we have seen two key mindsets that have cata- lyzed powerful products - a ‘build-learn-iterate’ mindset (e.g. the Lean Canvas (1-Page Fig. 2.7 Using a Knowledge Graph to represent and analyze relationships between key project com- ponents.
Applying AI to conservation challenges 25 Business Plan | LEANSTACK na)) and the user-centric mindset (e.g.Jobs-to-be-done by Clayton Christensen (JobsTo Be Done - Christensen Institute na)).In conservation, Adaptive Management offers a similar “continuous learning” mindset and a common lens for AI practitioners, researchers and innovators to understand and communicate their work. As we see contributions from all over the globe, we recommend a tool to capture, organize, share and collaborate on their AI-enabled efforts - the Impact Technology Canvas (Fig. 2.8). The Impact Technology Canvas systematically breaks down climate and sustainabil- ity work into utility (AI “jobs to be done”) and leverage (people, assets, resources, risks and capital) - regardless of the scale/scope of the solution (moonshot or roofshot). It enriches Adaptive Management’s continuous learning cycle with shared context: target personas, stakeholders, assumptions, partners, datasets, tools and other enabling contex- tual parameters (like location, outcome measures, focus areas, socio-economic factors). The canvas gives us a foundational structure and framing to capture problems, people, science, local knowhow, constraints and solutions behind specific climate or sustainability challenges.As the community of AI practitioners maps out the various AI initiatives using the canvas, we can more efficiently and effectively (i) identify reusable Fig. 2.8 The Impact Technology Canvas.
Data science applied to sustainability analysis 26 patterns and learning in technology, data and methods (sensor AI, machine learning models and economies of scope), and (ii) optimally analyze upstream/downstream risks gaps (and source new ideas, low-hanging fruit, research gaps and challenges), as illustrated in Fig. 2.9. Summary Our planet is a unique social network - the natural resources on which all life depends in turn are dependent on life itself. As we look to science, data, AI and technology to effectively mitigate climate change and drive sustainability,we can’t ignore the intercon- nectedness of this social network.Embedded in this interconnectedness is a key nuance - a constantly evolving, complex peoplenature relationship, which takes datasets beyond ecological domains and into human and societal factors.Take, for instance, the factors (human, ecological and socio-economic) you’d have to consider for a successful climate change intervention (say, a solar farm) in rural Northern India.These factors might be radically different for rural California! An electric vehicle might be a better fit for the same challenge. The interconnectedness of this complex peoplenature system and the noise (the uncertainties complexities of our actions and their effects) is a ripe environment for AI-shaped problem sets and innovation. Fig. 2.9 Unraveling scalability and transferability in the AI community using the canvas.
Applying AI to conservation challenges 27 With increased computing capacity (cloud infrastructure), advanced data manage- ment (sensor networks and edge technologies) and our ability to collaborate without borders,AI practitioners can catalyze how fast we understand, learn, apply and traverse the dynamic people and nature relationship. References 1-Page Business Plan | LEANSTACK. na https://leanstack.com/leancanvas/. AI-Powered Research Tool. Semantic Scholar. na http://semanticscholar.org/. Blog, G., 2017.Tapping Twitter Sentiments: a Complete Case-Study on 2015 Chennai Floods. Analytics Vidhya https://www.analyticsvidhya.com/blog/2016/07/capstone-project/ Published May 29. Dokic, K., Martinovic, M., Mandusic, D., 2020. Inference speed and quantisation of neural networks with TensorFlow Lite for Microcontrollers framework, 2020 5th South-East Europe Design Automation, Computer Engineering, Computer Networks and Social Media Conference. SEEDA-CECNSM. doi:10.1109/seeda-cecnsm49515.2020.9221846. Glaessgen, E., Stargel, D., 2012. The Digital Twin Paradigm for Future NASA and U.S. Air Force Vehicles, 53rd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials ConferenceBR20th AIAA/ASME/AHS Adaptive Structures ConferenceBR14th. AIAA. doi:10.2514/6.2012-1818. iNaturalist. na https://iNaturalist.org/. Jobs To Be Done - Christensen Institute. na https://www.christenseninstitute.org/jobs-to-be-done/. Land Change Monitoring, Assessment, and Projection. na https://www.usgs.gov/core-science-systems/ eros/lcmap. Mapping Ocean Wealth. na http://oceanwealth.org/. Matei, O., Pop, P.C.,Vălean, H., 2013. Optical character recognition in real environments using neural net- works and k-nearest neighbor.Applied Intelligence 39 (4), 739–748. doi:10.1007/s10489-013-0456-2. Open Space. Coastal resilience. na https://coastalresilience.org/project/open-space/. “Redefining Ocean Data Collection.” Saildrone, saildrone.com/. na. Rose, R.A., Byler, D., Eastman, J.R., et al., 2014.Ten ways remote sensing can contribute to conservation. Conserv. Biol. 29 (2), 350–359. doi:10.1111/cobi.12397. Singapore experiments with its digital twin to improve city life Tomorrow.Mag. https://www.smartcitylab. com/blog/digital-transformation/singapore-experiments-with-its-digital-twin-to-improve-city-life/. Published May 30. South Payette River Case Study. na https://hydroforecast.com/case-study-south-payette/. Tao, F., Qi, Q., 2019. Make more digital twins. Nature News. https://www.nature.com/articles/d41586- 019-02849-1. Published September 25. The science of sound: acoustic soundscapes of mature forests in the temperate northern triangle of Myanmar. Geospatial Conservation Atlas. na https://geospatial.tnc.org/datasets/4899be47d4d34068 ac53035ac32bc7b3. “water sense”.naAlexa Skill.https://www.amazon.com/City-of-Virginia-Beach-storm/dp/B078K9F953. West, S. Meaning and action in sustainability science: interpretive approaches for social-ecological systems research. na 10.13140/RG.2.2.32127.10406.
29 Data Science Applied to Sustainability Analysis © 2021 Elsevier Inc. DOI: 10.1016/C2018-0-02415-9 All rights reserved. CHAPTER 3 Water balance estimation in Australia through the merging of satellite observations with models Chapter Outlines Introduction 29 Importance of water balance estimation in Australia 29 Problems with current methods for water balance estimation 30 Case studies 32 Inversion of rainfall rates from streamflow records 32 Estimation of flood forecast-effective river bathymetry through inverse modeling 34 Estimation of terrestrial water storage through the merging of Gravity Recovery and Climate Experiment (GRA CE) data with a hydrologic model 37 Conclusion 38 Introduction Importance of water balance estimation in Australia Due to its geographical properties,Australia faces a number of unique challenges related to its water management. It is the driest of all inhabited continents, while at the same time it is prone to devastating floods. According to the Queensland government, the average annual cost of floods is $377 Million (measured in 2008 Australian dollars). Especially disastrous were the 2010–11 floods, which costed the Australian economy an estimated $30 Billion. An estimate from Geoscience Australia resulted in a cost of $6.64 Billion (2013 Australian dollars) for the floods in the LockyerValley and Brisbane in January 2011 alone.This calculation included deaths and injuries but excluded most indirect losses. 35 deaths were confirmed, and 20,000 people were made homeless.The most important and cost-effective tools for the limitation and mitigation of the impact of floods are improved flood warning systems and community awareness, which are topics that are currently studied extensively in Australia. At the same time, droughts continue to damage Australia’s economy. Between 2006 and 2009 the drought reduced national GDP by roughly 0.75 percent. Between 1997 and 2002, the contribution of agriculture to the Australian economy averaged 2.9 per- cent. Between 2003 and 2009, the years of the millennium drought, this percentage Valentijn R.N. Pauwels, Ashley Wright, Ashkan Shokri, Stefania Grimaldi Monash University, Department of Civil Engineering, Clayton,VIC,Australia
Data science applied to sustainability analysis 30 reduced to 2.4 percent. During the two worst years of the drought, 2003 and 2007, this contribution further reduced to 2.1 percent [ABS, 2011, van Dijk et al., 2013]. Further, climate change is likely to worsen the impact of droughts in the South West and South East of Australia [Climate Council of Australia, 2018a]. It is already evident that droughts cause difficulties in the water security of Australia [Climate Council of Australia, 2018b]. For this reason, groundwater is an important contributor to the Australian economy.According to Deloitte Access Economics [2013], the economic contribution of groundwater use to the Australian Gross Domestic Product (GDP) is between $3.0 and $11.1 Billion,with households using approximately $410 Million worth of groundwater each year. These issues clarify the need for a reliable estimation of the water resources of Australia.A number of methods are currently available for this purpose, but each suffers from a number of issues.The remainder of this section provides an overview of these methods and their drawbacks, after which a number of studies are described that focus on these issues. Problems with current methods for water balance estimation Soil moisture is a key variable in the hydrologic cycle, as it determines the partitioning of the available energy into latent, sensible, and ground heat fluxes, and the net pre- cipitation into infiltration and surface runoff. Consequently, numerous satellite missions have been devoted to estimating this variable at large spatial scales through remote sens- ing. Relatively recent examples are the Sentinel-1 [Attema et al., 2007], the Advanced Scatterometer (ASCAT) [Figa-Saldana et al., 2002], the Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E) [Njoku et al., 2003], the Soil Moisture Ocean Salinity (SMOS) [Kerr et al., 2010], the Soil Moisture Active Passive (SMAP) [Entekhabi et al.,2010],and the Global Change Observation Mission (GCOM) [Imaoka et al., 2010] missions.Throughout the last decades, one major improvement in satellite remote sensing-derived soil moisture products is their temporal resolution.This has improved from 35 days, for the Synthetic Aperture Radar on European Remote Sensing satellites (SAR/ERS), to 3 days, for the ASCAT AMSR/E, SMOS, and SMAP missions [Li et al., 2018]. However, remaining issues with large-scale remote sensing of soil moisture contents are the potential to only provide soil moisture estimates of the upper centimeters of the soil, difficulties in the validation of the products, and errors and uncertainties in the inversion of the raw signal into soil moisture contents [Peng and Loew, 2017]. The only mission that attempts to estimate the water storage of the entire soil profile is the Gravity Recovery and Climate Experiment (GRACE), which provides infor- mation on the change of the terrestrial water storage (TWS) over time. Even though this mission has been proven extremely useful for a number of different objectives, the coarse temporal (monthly) and spatial (∼150,000 km2) resolutions, and problems with
Water balance estimation in Australia through the merging of satellite observations with models 31 the conversion of the storage changes into volumes continue to complicate the use of the GRACE TWS retrievals [Girotto et al., 2016]. Remote sensing can also provide information on inundation extent and level. As water surfaces tend to be very smooth,a radar signal reaching open water will be reflect- ed away from the satellite sensor and a low backscatter will be observed. Conversely, rough dry areas will return part of the energy to the sensor resulting in higher back- scatter [Henderson and Lewis, 2008]. This is the principle of radar-based flood map- ping [Chini et al., 2017; Shen et al., 2019]. However, the interpretation of backscatter signal can be very difficult in complex environments, such as urban areas and areas with emerging vegetation [Pierdicca et al., 2018]. For this reason, increasing research efforts are being dedicated to the definition of accurate algorithms for the detection of floods using radar data.An overlay of the flooded area onto a Digital Elevation Model (DEM) can then provide an estimate of the flood water level [Hostache et al., 2009]. Both remote sensing-derived inundation extent and level can be used to evaluate the predic- tive skill of hydraulic flood forecasting models, and the selection of the most effective dataset is being debated in the literature. [Shen et al., 2019]. A common method to estimate the surface water balance at large spatial scales is through land surface modeling. These models simulate all processes related to the partitioning of precipitation into streamflow, evapotranspiration and change of stor- age, and aim to provide an answer to the fundamental question “What happens to the rain” [Penman, 1961]. Despite strong advances in computational power, the quality of the meteorological forcing and input data, model parameter estimation algorithms, and the representation of the physical processes, errors and uncertainties in each of these fields continue to lead to erroneous model predictions [Boughton et al., 2016]. Errors in the model prediction are propagated at each time step, which can severely affect the accuracy of the final model water balance estimates [Ellett et al., 2006].With respect to accurate flood modeling at large spatial scales, the most important obstacles are the lack of an accurate three-dimensional representation of the morphology of the floodplain and of the rivers [Garcia-Pintado et al., 2015;Wood et al., 2016]. Increasing efforts have been dedicated to the delivery of corrected satellite-derived DEMs for the implementa- tion of hydraulic models [Yamazaki et al., 2017, 2019]. However, DEMs usually provide information on floodplain morphology only; conversely, information on the river chan- nel depth and shape (i.e.river bathymetry) is rarely or not available for many catchments in the world [Domeneghetti, 2016].This problem can be overcome by measuring the river bathymetry, or depth, in-situ, but this is not practical even at the catchment scale. Furthermore, river bathymetry can be prone to significant changes over short time scales [Soar et al., 2017]. This chapter describes three studies tackling the above problems using large data sets. The methodologies were developed and tested using Australian catchments as test sites. The left hand side of Fig. 3.1 shows the location of the test sites.The right hand side of
Data science applied to sustainability analysis 32 Fig. 3.1 shows and overview of the required data, methods, and modeling techniques of these studies. Case studies Inversion of rainfall rates from streamflow records It is widely known that the estimation of spatially averaged rainfall rates is extremely challenging. An especially common method to achieve this goal is the interpolation of point scale rainfall observations, which suffers from the lack of spatial coverage of the stations.Weather radars and satellite observations can also be used for this purpose, but these suffer from uncertainties in the inversion algorithms and inconsistencies between the rainfall generated at the cloud level and the amounts reaching the ground [El Kenawy et al., 2019]. Inversion algorithms are used in environmental modeling to estimate independent variables, in non-linear systems, such as rainfall rates and model parameters from the measured dependent variables such as reflectivity and streamflow.A more recent approach to estimate spatially average rainfall rates makes use of the mobile phone network [Overeem et al., 2013], by relating the attenuation of the electromag- netic signal from the transmitting to the receiving antenna to the rain rate.However,the required commercial data are frequently quite difficult to acquire. Based on these issues and on the fact that discharge observations represent the catchment-integrated response to rainfall events, a different approach is to invert observed streamflow data into rainfall time series [Kirchner et al., 2009]. Because of the non-linear response of streamflow to rainfall advanced inverse modeling techniques may be required for this purpose. A methodology to estimate rainfall rates from streamflow records using inverse modeling techniques was developed by Wright et al. [2017b].To do so it is necessary to parameterize the current best estimate of areal rainfall which is derived from imperfect observations. The algorithm started by noting that most rainfall values at the hourly Fig. 3.1 Left hand side: Overview of the test sites for this chapter. Right hand side: Data, methods, and modelling techniques used in the three studies.
Other documents randomly have different content
devout demeanour at the stake, 161, 164, 170, 175, 199; when in the flames begged the Cross to be held before her, 161, 175, 195; Jesus her dying utterance, 161, 176, 273, 301, 305; pity excited by her execution, 191, 192, 255; contrition of her executioner, 161, 163, 194; exact place of execution, 170, 175; death desired by the English, 186; her ashes cast into the Seine, 193, 207, 301, 302, 305; her appearance in June, 1429, 30; no authentic portrait known, 49; her abstemious diet, 237, 243, 296; prison diet, 15, 16; pious and simple life, xiii; physical hardihood, xiii; her presence controlled vice and raised tone of French army, xii- xiii, 243, 245, 249, 250, 251, 264, 268, 270; hospitable to poor, 221, 224, 272; problem as to her knowledge of logic and theology, xix; testimony to virtue and courage, xxvi, 319; eloquent and forensic, yet prudent and simple in answers, xxvii, 177, 179; Charles VII. orders posthumous enquiry nearly twenty years later, v, xxi, 371; abortive, 372; enquiry ordered by Pope Nicholas V., xxii, 372; no definite result, 373; Pope Calixtus, on petition of Jeanne’s mother, Isabella, causes solemn enquiry at Paris, xxii, 373–376; sworn information of events in the last days of Jeanne’s life, 147– 8, 150; official Latin text of trial and rehabilitation, xxv; sentence of rehabilitation xxiii, 321–328, 376 Jeanne d’Arc family, see d’Arc Jhesus Maria on banner, 31, 91, 361; on letters, 35, 36, 349, 350, 352, 369 Josephine, Empress, 249
Joyart, Mengette, 222 Jumièges, Abbot of, 127 La Basque, standard-bearer, 316, 317 La Charité sur Loire, 53, 73, 317, 352, 361 Lacloppe, Bertrand, 218 Ladies’ Tree, see Tree Ladvenu, Br. Martin, 148, 150, 168, 170, 175, 191, 193–5, 328, 338, 372 Lagny, 29, 52, 78 La Hire, Maréchal, 233, 235, 250, 263, 264, 277, 279, 293, 308, 311, 312, 314 La Macée, Lady, 305 Lambert or Lombart, Jean, 306 Lancaster, House of, xvii Lapse, The, 121–134, 326 Lapau, Mme., 260 La Rose, Philippe, 373 La Rousse, woman, 9, 217, 219, 344 La Saussaye in diocese of Evreux, 209 Laval, Guy and André de, 30 Laverdy, M., 331 Laxart, Durand, 218, 225, 230 Laxart, Jean de, 6, 215 Lebouchier, Maître Guillaume, 100, 106 Lebouchier, Messire Pierre, 198 Lebuin, Michael, 225 Lecamus, Canon Jacques, 151
Lecomte, Denis, 332, 374 Ledoux, Maître Jean, 100, 118 Le Drapier, Perrin, 218 Lefevre, Maître Jean, Bishop of Démétriade, 101, 205, 210, 338 Lefumeux, Messire Jean, 231 Leguise, Bishop Jean, 240 Le Marie, Guillaume, 244, 306 Lemaître, Maître, Jean, Vice-Inquisitor of Beauvais, 8, 66, 67, 94, 95, 98, 99, 105, 129, 132, 133, 140, 141, 142, 168, 182, 193, 194, 322 Lenozolles, Maître Jean de, 290, 340 Leparmentier, Maugier, 300 Le Renard, see Thérouanne Leroyer, Catharine, 21, 227 Leroyer, Henry, 223, 227, 228 Le Royer, Thévenin, 218 L’Esbahy, Jacques, 248 Letters to English, 36–8, 235, 246–7, 286–7 Letter of Duke of Burgundy, 335 Letter of Count d’Armagnac, 34–5, 351; Jeanne’s reply, 35 Ligny, Count de, 178, 294, see Luxembourg, Jean Limoges, ix Lingué, Jean, 6 Lisle, 177 Loches, 268, 270 Lohier, Maître Jean, xix, xx, 166–7, 254, 257, 339 Loire, camps on the, 237
Longueville, Lord Prior of, 100 Loré, Sieur Ambroise de, 275, 279 Lorraine, its traditions, viii, ix, 9 Lorraine, Charles, Duke of, 11, 30, 214, 218, 226, 272, 230 Lorraine, Réné of Anjou, Duke of, 11 Louis XI., vii, 245, 275, 289 Louviers, siege proposed, 157, 158 Loyseleur, Nicolas, 56, 117, 118, 130, 134, 148, 152, 165, 166, 167, 169, 179, 182, 183, 202, 258, 298–9, 340, 341 Lude, Sieur de, 278 Luxembourg, Jean de, 58, 334, 335, 336; his wife, Jeanne de Bethune, 46; see also Ligny, Count de Luxembourg, Count Waleran de, 46, 334 Luxembourg, Messire Louis de, Archbp. of Rouen, 163, 294; see Thérouanne Machet, Gerard, Bishop of Castres, 238 Maçon, Maître Jean, 248 Maçon, Robert le, 238 Mailly, Jean de, Bishop of Noyon, 255–6 Manchon, Guillaume, 56, 68, 77, 95, 126, 136, 146, 165, 172, 178–187, 188, 197, 212, 298, 331, 340, 372, 374 Mandrakes, 42 Manuel, Pierre, 304 Margaret of Anjou, 11, 275 Margaret of Bavaria, 272 Margaret of Scotland, 223
Marguerie, André, 101, 117, 118, 121, 192, 202, 208–9 Marie, Messire Thomas, 211 Marigny, 335 Marriage, action against Jeanne, 62, 64, 344 Martel, Charles, 27 Martin V., Pope, 34 Mary of Anjou, Queen, 46 Massieu, Maître Jean, 56, 68, 85, 117, 135, 141, 170, 171–176, 198, 339, 372 Maugier, Pierre, 373 Maurice, Maître Pierre, 56, 89, 95, 96, 101, 103, 121, 122, 148, 149, 166, 171, 180, 209, 302 Maxey-sur-Vays, 19, 225 Meaux, Bishop of, see Versailles Meaux, Viscountess de, 46 Mehun, 237 Melun, 57, 73 Melville, Lord, xi Merlin, prophecy, 21, 188, 241 Message, Mathieu, 244 Metz, Jean de Novelemport, called Jean de, x, 12, 136, 218, 223–5, 226, 228, 230, 265, 291, 301 Meung, 249, 263 Meung-sur-Loire, 237, 278, 289 Meung-sur-Yèvre, 245, 317 Midi, Maître Nicolas, 56, 61, 64, 67, 69, 74, 77, 79, 85, 89, 94, 95, 96, 100, 103,
106, 109, 119, 121, 134, 142, 166, 171, 176, 177, 207, 255, 258, 295, 300 Milan, Duke of, 6 Milet, Colette, 295 Milet, Pierre, 295, 296 Minet, Messire Jean, 6 Minier, Pierre, 209 Moen, Jean, 216 Monnet, Maître Jean, 258–9, 340 Montargis, battle, 232, 242 Moreau, Jean, 303 Morel, Maître Aubert, 101, 117, 118, 119 Morel, Jean, 6, 215 Morin, Maître Jourdin, 274, 282 Mortemer, Abbot of, 127 Mugot, see Contes, L. de Musée de Trocadéro, Paris, 49 Musnier, Simonin, 221 Muton, Guillaume, 103 Naples, vii Napoleon I., v Neufchâteau, 9, 10, 212, 214, 216, 218, 220, 344 Newman, Cardinal, xxii Nibat, Jean de, 100 Nicholas V., Pope, xxi, 372 Normandy, 371
Norwich, Bishop of, 127 Novelomport, Jean de, 12; see Metz Noyon, Bishop and Diocese of, 4, 121, 127, 142 Oath, administration of, 5, 6, 7 Olivier, Alain, 328 Orient, Pierre, 97 Orleans family, xv Orleans, Charles, Duke of, 72, 280 Orleans, Duke of, 12, 65, 307, 353 Orleans held by patriots, ix; its siege, vii, 31, 32, 35; relief ordered, x; siege raised by Jeanne d’Arc, 233–237, 242, 245, 246–7, 249, 260–263, 266–270, 275–6, 284–289, 292, 293, 296, 297, 307, 300–317 Ourches, Albert d’, 228 Paris, vii, xv, 181, 352; its assault, 14, 28, 73, 78, 353; Church of Notre Dame, 373; National Library, 331, 332; Parliament of, 336; University of, xxi, 119, 120, 122, 138, 177, 336 Paris, Guillaume Chartier, Bishop of, 321, 373 Partada, Alphonse de, 314 Pasquerel, Brother Jean, 32, 281, 282 Patay, Battle of, xii, 266, 280, 289, 293 Peter of Pomfret, ix Petit, Gerard, 229
Philip II, 273 Philip the Fair, xxii Picard ravages, 9 Pigache, Maître Jean, 209 Pinchon, Jean, 99 Poitiers, vii, ix, xi, xviii, xix, 24, 136, 265, 305; book at, 25; Church of, 116; Clergy of, 201, 244 Pole, William de la, Earl of Suffolk, 36, 241, 248, 263, 265, 277, 278 Pollichon, see Poulengey Pont l’Evêque, 73, 361 “Pontiffs, Three,” 34 Pope and Empire, xvi Pope Calixtus, v, xxii, 178, 373 Pope Nicholas V., 372, 373 Pope of Rome, 33, 36, 91, 114, 116, 128, 131, 145, 159, 160, 189, 210 Poulengey, Bertrand de, 12, 136, 218, 224, 226, 228–231, 265; see Pollichon, B. Poulnois, Hauves, 283 Pouthon, the Burgundian, 335 Preaux, Abbot of, 127 Pressy, Sieur Jean de, 47 Preston Manor, Sussex, 42 Prévosteau, Guillaume, 374 Process or transcript of trial, 180, 181, 187, 188, 196, 197, 200, 210, 325, 326, 327 Prophecy—France lost by a woman, saved by a maid, 226, 227
Prophecy of d’Avignon, 269 Queen of Charles VII, 270, 271 Quesnay, Maurice de, 106 Quicherat, xxv, 242 Rabateau, Jean, 24, 243, 265, 269, 306 Raguier, Hemon, Treasurer, 215, 275, 283 Raiguesson, Jean, 6, 215 Rais, Gilles de Laval, Seigneur de, 233 Raymond, 260 Regnault de Chartres, xiv; see Rheims Rehabilitation Sentence, 321–328, 371 Relapse, 135–146, 326 Reynel, Maître Jean de, 257 Rheims xii, 215, 237, 239, 244, 245, 264, 292, 304, 361; Jeanne’s house, 51 Rheims Cathedral xii, xiv, 51 Rheims, Jean Jouvenal des Ursins, Archbishop of, 321, 373 Rheims Reginald, Archbp. of 324 Rheims, Regnault de Chartres, Archbp. of, xi, xiv, xix, 24, 115, 118, 201, 233, 240, 305 Ricarville, Guillaume de, 245 Richard II, ix Richard, the Archer, 224, 226, 228, 230 Richard, Brother, 42, 48, 53 Richelieu, 289
Ricquier, Jean, 301 Rose, Philippe de, 178 Roncessey-sous-Neufchâteau, 216 Rouel, Jean de, 257 Rouen, v, xix, 119, 372; Castle, 3, 110, 117, 119, 121, 290; the Ornament Room, 8, 340, 351; Archiepiscopal Chapel, 120, 138, 139, 328; Trial of Jeanne d’Arc, 3, 99, 121, 132, 181, 252; Jeanne’s Imprisonment, 96, 106, 192, 195, 199, 205, 255, 294, 299, 303, 305; Cemetery of Saint Ouen, 127, 132; Market Place, 170, 327 Roussel, Guillaume, 328 Roussel, Jean, 328 Roussel, Raoul, 118 Rouvray, Battle, 11 Royal Letters of Jeanne’s Surrender, 3 Saint Bernard, 119 St. Catherine, x, 23, 24, 26, 39, 40, 47, 60, 62, 65, 69, 71, 74, 75, 77, 79, 84, 88, 90, 92, 109, 115, 131, 137, 345, 352, 355, 357, 358, 363, 366, 368, 370 St. Catherine de Fierbois, village, 12, 27, 28, 89, 349 St. Charles, 234, 235 St. Denis, 13, 14, 29, 62, 88, 349, 353, 354, 361; war cry, 89 St. Eusebius, Cardinal of England, 127, 161, 187, 209 St. Gabriel, 118, 357, 366, 371 St. Lo of Rouen, Prior of, 127 St. Louis, 234, 235
St. Margaret, x, 23, 24, 26, 39, 40, 47, 60, 62, 65, 71, 74, 77, 84, 88, 90, 92, 109, 115, 131, 137, 345, 352, 354, 357, 358, 363, 366, 368, 370 Saint Mesmin, Amian de, 248 St. Michael, 25, 39, 42, 44, 45, 64, 84, 85, 199, 255, 344, 355, 357, 358, 366, 368, 370 St. Michel au Peril-de-la-Mer, Abbot of, 127 St. Nicholas du Port, 226, 277–8, 229 St. Ouen of Rouen, Abbot of, 127 St. Ouen, Jeanne preached to at, 172, 187, 191, 255, 259, 295, 376 St. Peter and his Successors, 124, 131 St. Pierre-le-Moustier, 246, 318 St. Quentin, Burgundy’s letter to people, 335 St. Thomas, 160 St. Urbain, town of, 12 Scales, Lord, 37, 248, 278 Scotland, King of, vii Séguin, Pierre, 244 Selles-en-Berry, 30, 271 Senlis, Bishop of, 51, 78 Shakespeare, ix Sicily, Yolande, Queen of, 275, 309 Soissons, 4, 54 Sologne, the, 260, 284 Stafford, Earl of, 183, 294 Suffolk, Earl of, 21; see Pole Surname, girls take mother’s, 95
Surreau, Laurent, receiver-general, 257, 328 Sword obtained at St. Catherine de Fierbois, 28, 89, 349, 353 Talbot, John, Earl of Shrewsbury, 37, 234, 235, 248, 266, 279, 280 Talbot, William, 7, 338 Taquel, Nicolas, 68, 146, 185, 195–198, 298, 340 Teresa, viii Thermes, Simon de, 229 Thérouanne, Bishop of (Cardinal de Luxembourg), 121, 127, 142, 208, 334 Thévenin, Jeannette, 6, 213, 215, 216 Thierry, Maître Reginald, 245 Thierry, Wautrin, 375 Thiesselin, Jeannette, 6, 213, 215, 217 Thou, Jacques de, 248 Tilly, Janet de, 232 Torcenay, Jean de, Bailly of Chaumont, 225, 229 Toul, 344 Touraine, Jacques de, 56, 89, 103, 106, 119, 166, 171, 183, 252, 257 Touroulde, Dame Marguerite la, 270–272 Tours, 28, 268, 283, 284 Torture, 117, 119, 126, 339 Toutmouillé, Brother Jean, 148, 150, 157, 372 Tree at Domremy, Ladies’ or Fairies’, 20, 214, 217, 219, 221, 229, 343, 344, 366 Tremouille, Seigneur de la, 60, 71, 78, 115 Tressart, Maître Jean, 192 Trèves, Sieur de, 238
Trèves, Lady de, 282 Trial, see Process Troyes, xii, 48, 49, 292; treaty of, viii, 226 Turquetil, Maître Eustace, 172 Turrelure, Pierre, 306 Vallée, Maître Guillaume, 190 Vatican, 331 Vaucouleurs, x, 9, 214, 225, 226, 227, 230; remains of castle, 11, 12, 65 Vaucouleurs, Alain de, 226 Vaux, Pasquier, de, 61, 64, 66 Venderès, Nicolas de, 99, 117, 118, 121, 135, 139, 147 Vendôme, Count de, 74 Verneuil, battle of, 277 Versailles, Pierre de, Abbot of Talmont, 243, 265, 269, 270, 274, 282 Vienne, Colet de, 12, 218, 224, 226, 228, 230 Villars, Sieur de, 232, 314 Viole, Maître Aignan, 297 Wandonne, Lionel Bastard de, 335 Ward, Jean Dieu-le-, 218 Warwick, Earl of, 106, 162, 164, 169, 174, 179, 183, 186, 189, 190, 197, 200, 212, 232, 254, 294, 299, 335, 338 Washington, George, xiv Waterin, Jean, 220 Well Sunday, 20, 214, 217, 219 William of Worcester, 336
Woman’s dress, Jeanne and, 21, 46, 47, 95, 114–115, 134, 136, 138, 159, 169, 174, 186, 204, 228, 347, 348, 367, 368 Ysambard de la Pierre, 66, 67, 69, 77, 79, 85, 89, 118, 134, 135, 159– 162, 164, 168, 188, 190, 194, 340
1. Had there been any desire on the part of the French King to rescue Jeanne from captivity, a ‘King’s ransom,’ which was later paid for her by Cauchon, could scarcely have been refused in those days for a prisoner of war, however renowned. Unhappily for the memory of Charles, she was left to the tender mercies of the English without any offer being made for her release, or any attempt at rescue. There existed a bitter feeling of jealousy towards Jeanne in consequence of her great successes in the field. This was notably shown during her attack upon Paris, where she was thwarted in every direction, and all possibility of victory was taken from her by the conduct of the King. Whether or not Flavy, the Governor of Compiègne, who was completely under the control of the King, betrayed Jeanne at Compiègne, by shutting the gates and closing the drawbridge at her approach, will never be known, but suspicion has always pointed to his betrayal of the Maid. Alain Bouchard states that, in the year 1488, he heard from two aged men of Compiègne, who had themselves been present, that a few days before her capture, the Maid was attending Mass in the Church of St. Jacques. After communicating and spending some time in devotion, she turned to the assembled congregation, and, leaning against a pillar, uttered this prediction: “My good friends, my dear little children, I am sold and betrayed. Soon I shall be given up to death. Pray to God for me, for I can no longer serve the King and the Kingdom of France.”—Grandes Annales de Bretagne, also Miroir des Femmes Vertueuses. 2. The House of Lancaster was fervidly orthodox. Persecution of heretics begins with Henry IV. The “Cardinal of England” (Beaufort Bishop of Winchester) was the malleus hereticorum at home and abroad. He spoke against the Hussites at the Council of Basle, and he planned Crusades against both heretics and “Saracens.” 3. The court before which Jeanne was brought to trial at Rouen was not a court of the Holy Office or Inquisition, neither was it,
as the English courts for the trial of heresy were in Lancastrian times, a statutable court of ecclesiastical jurisdiction on whose decision, certified by the bishop, the sheriff was bound to act. It was a composite tribunal. The Bishop of Beauvais claimed and exercised jurisdiction as Ordinary. But the Deputy Inquisitor was joined with him as co-ordinate judge with officers of his own. The Inquisition arose out of the troubles in Spain and South France, where heresy was to some extent necessarily a kind of treason to the polity of Christian Europe. Men were punished for heretical opinions, but these heretical opinions were in most cases lapses from allegiance at a time of national peril. The later Inquisition has no such excuse. 4. The Great Schism arose out of the Babylonian captivity at Avignon (1306–1376). Popes and anti-Popes contended for 40 years (1378–1418). France was on the side of the Avignon Popes, while the Empire and England supported the Popes in Rome. Philip the Fair, by arrangement with the Pope, changed the Papal chair to Avignon. During the seventy years of the captivity, when the Church was ruled by French Popes, France underwent the disasters of Crecy and Poitiers, and became almost a province of England. 5. It is agreed by all authorities that Jeanne was not captured in the Diocese of Beauvais, which ended at the Bridge of Compiègne. Jeanne was taken north of the Bridge, on the right bank of the river, and either in the Diocese of Noyon or Soissons, which of the two has not been determined. The Bishop’s assertion is distinctly untrue. 6. On January 6th, 1412. “In nocte Epiphiniarum Domini.” (Letter from Boulainvilliers to the Duke of Milan. Quicherat, vol. v., 116.) 7. The Font and Holy water stoup in the old Church at Domremy are said to be those in use in the 15th century.
8. Jeanne appears to have had a great many godparents. In the Enquiry made at Domremy in 1455, eight are mentioned, viz.: Jean Morel, Jean Barrey, Jean de Laxart, and Jean Raiguesson, as godfathers; and Jeannette Thévenin, Jeannette Thiesselin, Beatrix Estellin, and Edith Barrey, as godmothers. 9. John Gris, or Grey, a gentleman in the Household of the Duke of Bedford, afterwards knighted. He was appointed chief guardian to the Maid, with two assistants, all members of the King’s Body Guard. They appear to have left her entirely in the hands of the common soldiers five of whom kept constant watch over her. 10. There is no certain date for this event. By some it is placed between the first and second visits to Vaucouleurs, in 1428; by others, earlier, at the time of the Picard ravages of the neighbourhood in the September of 1426. 11. Robert de Baudricourt, Squire, Captain of Vaucouleurs in 1428; afterwards knighted and made Councillor and Chamberlain to the King and Bailly of Chaumont, 1454. 12. Of the ancient château the “Porte de France” alone survives. From this gate Jeanne rode out with her escort to visit the King at Chinon. The crypt of the chapel remains, where Jeanne constantly prayed. 13. This is said to have been on account of the impression produced on him by Jeanne’s prediction, on February 12th: “To-day the gentle Dauphin hath had great hurt near the town of Orleans, and yet greater will he have if you do not soon send me to him.” This “great hurt” proved to be the Battle of Rouvray, in which the French and Scottish troops were defeated by the English under Sir John Fastolf. 14. Charles I., the reigning Duke de Lorraine in 1428, was in very bad health, and, having no son, the succession was a matter of some anxiety. He died in 1431, and was succeeded by his son- in-law, Réné of Anjou, who had married his only daughter,
Isabella. This Réné was a brother of Queen Mary, wife of Charles VII., and father of our own Queen Margaret, married in 1441 to Henry VI. 15. Jean de Novelomport, called de Metz, Bertrand de Poulengey, Colet de Vienne, the King’s Messenger, and three servants. 16. March 22nd, 1428. 17. This letter appears later, p. 36. Jeanne may have forgotten its contents, as both these expressions occur; or the Clerics who acted as her amanuenses may have inserted them without her knowledge. 18. Jeanne was entertained by command of the King in a small room on the first floor of the Tour de Coudray, within the Castle walls. Her room was approached by a staircase outside the tower. The vaulted roof of the room has fallen in and the fireplace is in ruins, but the room could easily be restored. Jeanne stayed here from March 8th to April 20th, 1429. She was two days at Chinon before she obtained access to the King. 19. Charles de Bourbon, Count de Clermont, Governor of the Duchy of the Bourbonnais and the Comté of Auvergne, during the captivity of his father in England. 20. On September 8th, 1429. 21. Up to the end of her life, Jeanne spoke of the Bishop as the person responsible for her trial and death. “Bishop, I die through you,” was her last speech to him, on May 30th, the day of her martyrdom. 22. This, and a subsequent enquiry, on February 27th, as to Jeanne’s habit of fasting, would seem to suggest a desire on the part of the questioner to prove that her visions had a more or less physical cause in a weak bodily state resulting from abstinence. As Jeanne’s usual food consisted of a little bread dipped in wine and water, and as she is reported to have had when at home (not in war) but one meal a day, it need hardly
be supposed that she suffered much from the results of a Lenten Fast. 23. The fifteen days’ respite would coincide with the first Examination held in the Prison, May 10th, the first day on which the Allegory of the Sign was given. 24. Gérardin of Epinal, to whose child Jeanne was godmother, is probably the person alluded to; he gave witness in 1455 that Jeanne had called him “Burgundian.” 25. A small fortress in an island formed by two arms of the Meuse, nearly opposite the village of Domremy. 26. According to local tradition, this tree stood to within the last 50 years, and was struck by lightning; another has been planted in its place. The house, in which Jeanne was born, remained in the possession of the De Lys family till the 16th Century, when it passed into the hands of the Count de Salm, Seigneur of Domremy. In the 18th Century it became the property of Jean Gerardin, whose grandson, Nicolas, gave it up in 1818 to the Department of Vosges; so that it is now preserved as National property. 27. This is probably a survival of the Fontinalia, an old Latin festival. The custom of decorating the wells and springs was kept up in England until the last century, and still exists in a few remote villages. The name ‘Well Sunday’ survives, though the processions of youths and maidens have long passed away. The ‘fontaine aux Groseilliers’ is still in existence. It is an oblong tank of water, with the original spring flowing through it. The great beech tree stood close by. 28. Pierre de Bourlement, Head of the ancient house of Bassigny, and Lord of the Manor of Bourlement. He was the last of his race. 29. Merlin had foretold the coming of a maiden out of an Oak- wood from Lorraine; and a paper containing a prophecy to this
effect had been sent, at the beginning of Jeanne’s career, to the English Commander, the Earl of Suffolk. There was also an old prophecy (quoted by Jeanne herself to Catharine Leroyer) that France, which had been “lost by a woman, should be saved by a Maid.” The conduct of Isabeau of Bavaria, wife of Charles VI., might certainly be said to have fulfilled the first half of this prophecy; and a tradition in the eastern counties that “deliverance should come from a maid of the Marches of Lorraine” must have directed many hopes to the mission of the Maiden from Domremy, though she herself does not seem to have known of the last prediction until some time later. The Oak-wood covers the hills above Domremy to this day. 30. This is the first identification of the “revelations” with any name; Jeanne had always spoken of her “Voices” or her “Counsel.” 31. This Examination at Poitiers had taken place in the Chapel attached to the Palace of the Counts of Poitou, which still exists and adjoins the ‘Salle des Pas Perdus,’ now the Great Hall of the Palais de Justice. It was conducted under the direction of the Archbishop of Rheims during the months of March and April, 1429, and extended over three weeks. At the conclusion, the assembly sent, as the result of their inquiries, a resolution to the King to the effect that he should follow the Maid’s guidance, and seek for the sign she promised him in the relief of Orleans, as a proof of the Divine origin of her mission, “for,” they added, “to doubt or forsake her without any appearance of evil would be to vex the Holy Spirit, and to make himself unworthy of the help of God: so saith Gamaliel in the Council of the Jews with regard to the Apostles.” Unfortunately, no trace of this Examination has been found: the ‘Book of Poitiers’ is referred to several times in the Trial; but it was not forthcoming at the time of the Rehabilitation. It was probably lost or destroyed by Jeanne’s enemies among her own party. The Archbishop of Rheims would have had it in his charge: and he was consistently opposed to Jeanne throughout.
During her stay at Poitiers the Maid lodged in the house of Jean Rabatier. 32. According to local tradition, this Church was originally founded by Charles Martel in 732, after his victory over the Saracens, whom he here ceased to pursue, and deposited his sword as an offering. This is by some supposed to have been the sword which later Jeanne sent for; but the legend is not of an early date, and there is no suggestion of the kind in contemporary writings. According to one authority, the Greffier de la Rochelle, the sword was found in a reliquary, which had not been opened for twenty years or more. The Chronique de la Pucelle and the Journal of the Siege of Orleans state that it was one of many votive offerings, and was recognized by Jeanne’s description of the five crosses on the blade, possibly a Jerusalem Cross. Some of the old Chronicles say that Jeanne told the King she had never been at Fierbois: but this statement is disproved by her own words in this answer. The suggestion that, having been to three Masses in the Church, she might easily have seen the sword, is to some extent answered by the alleged difficulty of the Priests to find, among the many swords there, the one she had specially described. Of the ultimate fate of this sword there are many versions, and no two agree exactly as to date. It was certainly broken in striking a camp-follower, one of a class the Maid had forbidden to enter the Camp; but whether this was just after the retreat from Paris or earlier, it does not seem possible to decide. Jeanne herself says she “had it up to Saint-Denis” and “Lagny,” both of which dates would imply the autumn of 1429: but most witnesses tell the story of its being broken in the July preceding, though several different places are mentioned as the scene of the incident. 33. On September 13th, 1429. 34. A small town near Auxerre. In this neighbourhood some of the chronicles place the incident referred to of the breaking of the
sword. The question may, therefore, have been intended to elicit the story. 35. The armour offered at Saint-Denis was the “blanc harnois” she wore during the earlier part of her career. When the church was pillaged by the English troops shortly after, this armour was sent to the King of England; but no further trace of it is known to exist. 36. Jeanne appears to have been a good horse-woman; she rode “horses so ill-tempered that no one would dare to ride them.” The Duke de Lorraine, on her first visit to him, and the Duke d’Alençon, after seeing her skill in riding a course, each gave her a horse; and we read also of a gift of a war-horse from the town of Orleans, and “many horses of value” sent from the Duke of Brittany. She had entered Orleans on a white horse, according to the Journal du Siège d’Orléans; but seems to have been in the habit of riding black chargers in war; and mention is also made by Châtelain of a “lyart” or grey. A story, repeated in a letter from Guy de Laval, relates that, on one occasion (June 6th, 1428), when her horse, “a fine black war-horse” was brought to the door, he was so restive that he would not stand still. “Take him to the Cross,” she said; and there he stood, “as though he were tied,” while she mounted. This was at Selles; and local tradition says that, from her lodging (a Dominican Monastery now the Lion d’Or hotel) the old iron town-cross was visible. It stood until about a century ago some fifteen paces in front of the north door of the Church, and was removed when the cemetery was converted into a market place. The Monastery was the property of the monks of Glatigny. The writers of the letter referred to above, Guy and André de Laval, were grandsons of Bertrand du Guesclin: the letter was dated Selles, June, 1429. The following are extracts: “... On Monday (June 6th) I left the King to go to Selles en Berry, four leagues from Saint Aignan. The King had summoned the Maid to come before him from Selles, where she then was, and many said this was much in my favour, so that I might see her. The said Maid treated my brother and me with
great kindness: she was armed at all points, save the head, and bore lance in hand. After we had arrived at Selles, I went to her lodging to see her, and she called for wine for me and said she would soon have me drink it in Paris. She seemed to me a thing divine, in all she did and all I saw and heard. “On Monday evening she left Selles to go to Romorantin.... I saw her mounting her horse armed all in white, save the head, a little axe in her hand.... And then, turning to the door of the Church, which was quite near, she said in a gentle woman’s voice, ‘You priests and clergy, make processions and prayers to God.’ Then she turned again on her way saying, ‘Draw on, draw on!’ her standard flying, borne by a gracious page, and her little axe in her hand. One of her brothers who arrived eight days since, left also with her, armed all in white.” 37. The banner was painted at Tours, while Jeanne was staying there, before her march to the relief of Orleans. The account for payment, in the “Comptes” of the Treasurer of War, gives: “À Hauvres Poulnoir, paintre, demourant à Tours, pour avoir paint et baillé estoffes pour une grand estandart et ung petit pour la Pucelle ... 25 livres tournois.” The description of this banner varies in different authors. The following account is compiled from them. “A white banner, sprinkled with fleur-de-lys; on the one side, the figure of Our Lord in Glory, holding the world, and giving His benediction to a lily, held by one of two Angels who are kneeling on each side: the words ‘Jhesus Maria’ at the side; on the other side the figure of Our Lady and a shield with the arms of France supported by two Angels” (de Cagny). This banner was blessed at the Church of Saint-Sauveur at Tours (Chronique de la Pucelle and de Cagny). The small banner or pennon had a representation of the Annunciation. There was also a third banner round which the priests assembled daily for service, and on this was depicted the Crucifixion (Pasquerel).
Another banner is mentioned by the Greffier de la Rochelle, which Jeanne is said to have adopted as her own private pennon. It was made at Poitiers; and represented on a blue ground a white dove, holding in its beak a scroll, with the words, “De par le Roy du Ciel.” 38. May 7th, 1429. 39. This prophecy is recorded in a letter written, April 22nd, 1429, a fortnight before the event, by a Flemish diplomatist, De Rotslaer, then at Lyons. Her chaplain, Pasquerel, also states, in his evidence given in 1455, that she had told him of the coming injury on the previous day. 40. June 11th, 1429. 41. Gallicè: “en leur petite cotte,” i.e., with only the light clothing worn under their armour. 42. The “three Pontiffs” referred to are Martin V. (Colonna), the real and acknowledged Pope; the schismatic, Clement VIII.; and a mere pretender, Benedict XIV., who was supported only by one Cardinal. The Schism was practically at an end at the time of this letter, as Clement had abdicated a month earlier (July 26th). Clement VIII. is the true title, though called Clement VII. in Count d’Armagnac’s letter. 43. The English lost Paris in 1436. 44. Compiègne was relieved early in November; Saint Martin’s Day is November 11th. 45. The mandrake was a part of the accepted paraphernalia of a sorcerer. It was kept wrapped in a silk or linen cloth, and was supposed to preserve its owner from poverty. Brother Richard had recently preached a sermon against them (April, 1429); and many had been burned in consequence. 46. The balance was a frequent accessory to Saint Michael in the French stained glass windows of the 13th and 14th centuries. A
noted example in the Cathedral at Arles represents him weighing the souls of the departed in a balance as big as himself. One of the earliest examples in England is that in a fresco-painting at Preston Manor, Sussex, said to be of the reign of Edward I., in which Saint Michael appears weighing the souls of the faithful, accompanied by Jeanne’s saints, Saint Catherine and Saint Margaret. 47. Mary of Anjou, wife of Charles VII., daughter of Louis, Duke of Anjou and Yolande of Arragon. 48. Jeanne was taken from Beaurevoir early in August, and removed from there, when the negotiations for selling her were complete, about the middle of November. 49. Jeanne, Countess de Saint-Pol et Ligny, sister to Count Waleran de Luxembourg and aunt to Jean de Luxembourg. 50. Jeanne de Bethune, Viscountess de Meaux, wife of Jean de Luxembourg. Both these ladies were at Beaurevoir during Jeanne’s captivity, and shewed her great kindness, even interceding for her that she should not be sold to the English. 51. The Sieur de Pressy, in Artois. Present in the Burgundian camp when Jeanne was taken prisoner, and afterwards at Arras, where she was imprisoned on her way from Beaurevoir to Rouen. The questions seem to suggest that Beaupère had before him some information which has not come down to us. 52. This may perhaps refer to a popular belief in a halo, as of a Saint, surrounding the Maid’s head. 53. Brother Richard, a Mendicant Friar; some say, Augustan; some, Cordelier. He was preaching in Paris and the neighbourhood in 1428–9; and said, amongst other things, in a sermon at Sainte Géneviève, April 16th, 1419, that “strange things would happen in 1430.” He professed to have been in Jerusalem; and his sermons were so popular that congregations were found to listen to him for 10 or 11 hours,
from 5 o’clock in the morning! He was driven out of Paris by the English and went to Troyes, where he joined the Maid. 54. No absolutely authentic portraits of Jeanne are known. A head of fine work, the portrait of a young girl wearing a casque and of Jeanne’s time, is at the Musée Historique at Orleans. Tradition asserts that when Jeanne entered Orleans in triumph with the relieving force a sculptor modelled the head of his statue of St. Maurice from Jeanne herself. This head is a portion of the statue which formerly stood in the church at Orleans dedicated to St. Maurice. The church was demolished in 1850. A photograph from the head is given as the frontispiece to this book, and an admirable copy maybe seen at the Musée du Trocadéro in Paris. It should have been stated on the frontispiece that the original is at Orleans, the copy in Paris. 55. Latin text adds: “dum rex suus consecraretur.” Tradition asserts that at the Coronation Jeanne stood on the left and slightly in front of the altar, coming direct from the sacristy of the cathedral. The coronation throne stood in front of the high altar. The cathedral and its painted glass exist as at the Coronation, with the exception of some comparatively recent stone work surrounding the choir. The Coronation of the Kings of France has taken place at Rheims Cathedral since the twelfth century. The King was not to all intents King of France until he had been anointed by the Holy Oil, brought in great state to the cathedral from the more ancient church of St. Remy. An inscription on the front of the Hotel Maison Rouge, situated near the west entrance of the cathedral, states that the town entertained Jeanne’s father and mother in that house during the Coronation. 56. About £200. 57. November 9th, 1429. 58. The Minute adds: “and I should be cured.”
59. Surrendered July 22nd. 60. In spite of this assertion, the Bishop was present at four out of the nine Examinations. 61. On May 23rd, 1430. 62. In the Minute only. 63. Not in the Minute. Latin text reads: “quod dedit regi suo dum venit ad eum.” 64. The “sign,” i.e. the appearance of “the White Lady.” 65. Jean, Duke d’Alençon: son of the Duke killed at Agincourt. He was of the blood-royal of France, and had married a daughter of the Duke d’Orléans. Jeanne was on very friendly terms with him, and always called him her “Beau Duc.” 66. The allegory of the Angel sent with a crown, here first given to avoid “perjury,” i.e., breaking her promise to preserve the King’s secret, is explained by Jeanne herself, on the last day of her life, to mean her own mission from Heaven to lead Charles to his crowning. 67. In the Minute: “et l’admener en trois ans”: not in the Latin Text. 68. The Minute reads: “la laissant faire de prisonniers.” 69. March 8th, 1428; it was before Easter, which in that year fell on March 7th. 70. The house in which Jeanne lodged at Chinon is said to have belonged to a certain Regnier de la Barrier, whose widow or daughter is the “worthy woman” referred to. Jeanne was afterwards lodged in the Tower of Coudray, where her room may still be seen. It is approached by a staircase outside the tower. The vaulted roof has fallen in, and the fireplace is damaged, but the walls are intact, and the room could easily be
Welcome to our website – the perfect destination for book lovers and knowledge seekers. We believe that every book holds a new world, offering opportunities for learning, discovery, and personal growth. That’s why we are dedicated to bringing you a diverse collection of books, ranging from classic literature and specialized publications to self-development guides and children's books. More than just a book-buying platform, we strive to be a bridge connecting you with timeless cultural and intellectual values. With an elegant, user-friendly interface and a smart search system, you can quickly find the books that best suit your interests. Additionally, our special promotions and home delivery services help you save time and fully enjoy the joy of reading. Join us on a journey of knowledge exploration, passion nurturing, and personal growth every day! ebookmasss.com

More Related Content

PDF
Data Science Applied to Sustainability Analysis Jennifer B. Dunn
byrhbmgxbpyr2027
 
PDF
Data Science Applied to Sustainability Analysis 1st edition - eBook PDF
bygirdypabisc7
 
PDF
Data Science Applied to Sustainability Analysis 1st edition - eBook PDF
byamulukaotar
 
PDF
Data Science Applied to Sustainability Analysis 1st edition - eBook PDF
byibrhemobonye
 
PDF
Data Science Applied to Sustainability Analysis 1st edition - eBook PDF
bysilveodienox
 
PDF
Data Science Applied to Sustainability Analysis 1st edition - eBook PDF
byamyotteilozs
 
PDF
Data Science Applied to Sustainability Analysis Jennifer B. Dunn
bysymsimou
 
PDF
Data Science Applied to Sustainability Analysis 1st edition - eBook PDF
bybahadamagbo
 
Data Science Applied to Sustainability Analysis Jennifer B. Dunn
byrhbmgxbpyr2027
 
Data Science Applied to Sustainability Analysis 1st edition - eBook PDF
bygirdypabisc7
 
Data Science Applied to Sustainability Analysis 1st edition - eBook PDF
byamulukaotar
 
Data Science Applied to Sustainability Analysis 1st edition - eBook PDF
byibrhemobonye
 
Data Science Applied to Sustainability Analysis 1st edition - eBook PDF
bysilveodienox
 
Data Science Applied to Sustainability Analysis 1st edition - eBook PDF
byamyotteilozs
 
Data Science Applied to Sustainability Analysis Jennifer B. Dunn
bysymsimou
 
Data Science Applied to Sustainability Analysis 1st edition - eBook PDF
bybahadamagbo
 

Similar to Data Science Applied to Sustainability Analysis Jennifer B. Dunn

PDF
Climate Change Science Causes Effects And Solutions For Global Warming 1st Ed...
byardaysanico
 
PDF
Engineering for Sustainable Development: Theory and Practice Wahidul K. Biswas
bydtrlimuel
 
PDF
Data Science In Agriculture And Natural Resource Management 1st Edition G P O...
bytmlgvrdtb722
 
PDF
Engineering for Sustainable Development: Theory and Practice Wahidul K. Biswas
byznvvpcn4884
 
PDF
Artificial Intelligence and Data Science in Environmental Sensing Mohsen Asadnia
byvaldonfiros33
 
PDF
Big Data Artificial Intelligence And Data Analytics In Climate Change Researc...
byalfteranaa
 
PDF
The Material Basis of Energy Transitions 1st Edition Alena Bleicher (Editor)
byhasensporlr0
 
PDF
Data Science in Agriculture and Natural Resource Management 1st Edition G. P....
bymyruptevinkl
 
PDF
Integrating Data Science And Earth Science Challenges And Solutions Laurens B...
byndetydudai
 
PDF
Life Cycle Sustainability Assessment For Decisionmaking Methodologies And Cas...
byhamiccorzozy
 
PDF
Life Cycle Sustainability Assessment for Decision-Making: Methodologies and C...
bytphsssdib082
 
PDF
Data Science in Agriculture and Natural Resource Management 1st Edition G. P....
bykhaterjarne
 
PDF
Environmental Sustainability, Research, IOT, and Future Development
byijtsrd
 
PDF
Sustainability and Analytics Mohammed Brueckner
byMohammed Brueckner
 
PDF
Advanced Computational Techniques For Sustainable Computing Megha Rathi
byalanizanolo9
 
PDF
Renewables based technology sustainability assessment 1st Edition Jo Dewulf
bysekdfvke885
 
PDF
Empowering Sustainable Development Reporting through Data Science Services.pdf
byTekLink International LLC
 
PDF
Leveraging Earth Observation Data for Sustainability Marketing Enhancing Busi...
byRyan Laird
 
PPTX
Principles of green data mining
byKarthikKaricheti
 
PDF
Where's the value in energy data science? Finding energy savings opportuniti...
byAdrian Friday
 
Climate Change Science Causes Effects And Solutions For Global Warming 1st Ed...
byardaysanico
 
Engineering for Sustainable Development: Theory and Practice Wahidul K. Biswas
bydtrlimuel
 
Data Science In Agriculture And Natural Resource Management 1st Edition G P O...
bytmlgvrdtb722
 
Engineering for Sustainable Development: Theory and Practice Wahidul K. Biswas
byznvvpcn4884
 
Artificial Intelligence and Data Science in Environmental Sensing Mohsen Asadnia
byvaldonfiros33
 
Big Data Artificial Intelligence And Data Analytics In Climate Change Researc...
byalfteranaa
 
The Material Basis of Energy Transitions 1st Edition Alena Bleicher (Editor)
byhasensporlr0
 
Data Science in Agriculture and Natural Resource Management 1st Edition G. P....
bymyruptevinkl
 
Integrating Data Science And Earth Science Challenges And Solutions Laurens B...
byndetydudai
 
Life Cycle Sustainability Assessment For Decisionmaking Methodologies And Cas...
byhamiccorzozy
 
Life Cycle Sustainability Assessment for Decision-Making: Methodologies and C...
bytphsssdib082
 
Data Science in Agriculture and Natural Resource Management 1st Edition G. P....
bykhaterjarne
 
Environmental Sustainability, Research, IOT, and Future Development
byijtsrd
 
Sustainability and Analytics Mohammed Brueckner
byMohammed Brueckner
 
Advanced Computational Techniques For Sustainable Computing Megha Rathi
byalanizanolo9
 
Renewables based technology sustainability assessment 1st Edition Jo Dewulf
bysekdfvke885
 
Empowering Sustainable Development Reporting through Data Science Services.pdf
byTekLink International LLC
 
Leveraging Earth Observation Data for Sustainability Marketing Enhancing Busi...
byRyan Laird
 
Principles of green data mining
byKarthikKaricheti
 
Where's the value in energy data science? Finding energy savings opportuniti...
byAdrian Friday
 

Recently uploaded

PDF
Cut off for Asst Professor Recruitment by HPSC i.e. Haryana Public Service Co...
byRajesh Verma
 
PDF
Result Analysis of the Pre Board 2025 .pdf
byDirectorate of Education Delhi
 
PPTX
Cardiovascular System or The Details of Heart.pptx
byAshish Umale
 
PPTX
CHAPTER NO. 05 BIOLOGICAL SOURCE,CHEMICAL CONSTITUENTS AND THERAPEUTIC EFFICA...
byGanu Gavade
 
PDF
Plant Respiration.pdf Remedial Biology Unit-5 B.Pharm 1st Year
bySaurabh Dubey
 
PPTX
1. Importance, scope and constraints of vegetable and spice crop production i...
byUmeshTimilsina1
 
PDF
Renewable Energy Resources Unit 2 Easy notes (EduShine Classes).pdf
byaaquib913
 
PPTX
Salesforce Spring 26` Release Key Features.pptx
byMehediHasan939830
 
PDF
ADIEU, BRIGITTE BARDOT - THE ACTRESS WHO CHANGED THE WORLD.pdf
byMilanStankovic19
 
PDF
Life-Processes-in-Animals PPT/7TH CLASS NEW NCERT /CURIOSITY SCIENCE /BY K SA...
bySandeep Swamy
 
PPTX
Structure and function of Integumentary system (skin).pptx
byAshish Umale
 
PPTX
What are the Custom lot_serial number in Odoo 19
byCeline George
 
PPTX
Aplastic_Anaemia_PA17.1_Presentation.pptx
byDibyajyoti Prusty
 
PPTX
Centrifugation pharmaceutical engineering
byShraddha Bandgar
 
PPTX
Acid Base Titration First Year B Pharm.pptx
byMs. Ashatai Patil
 
PPTX
Cultivation Practice of Major Cole crops in Nepal.pptx
byUmeshTimilsina1
 
PDF
• Reinforcing the Human Firewall: Moving From Awareness to Applied Skill
byAdityarajsinh Gohil
 
PDF
Digital_Empathy_Toolkit_Netiquette_and_Responsible_Communication
byNidhi Pandya
 
PDF
India Cybercrime Threats & Response 2025: Digital Arrests, 1930 Helpline & Ze...
bySanjay Rathod
 
PPTX
Non aqueous titration First Year B. Pharm.pptx
byMs. Ashatai Patil
 
Cut off for Asst Professor Recruitment by HPSC i.e. Haryana Public Service Co...
byRajesh Verma
 
Result Analysis of the Pre Board 2025 .pdf
byDirectorate of Education Delhi
 
Cardiovascular System or The Details of Heart.pptx
byAshish Umale
 
CHAPTER NO. 05 BIOLOGICAL SOURCE,CHEMICAL CONSTITUENTS AND THERAPEUTIC EFFICA...
byGanu Gavade
 
Plant Respiration.pdf Remedial Biology Unit-5 B.Pharm 1st Year
bySaurabh Dubey
 
1. Importance, scope and constraints of vegetable and spice crop production i...
byUmeshTimilsina1
 
Renewable Energy Resources Unit 2 Easy notes (EduShine Classes).pdf
byaaquib913
 
Salesforce Spring 26` Release Key Features.pptx
byMehediHasan939830
 
ADIEU, BRIGITTE BARDOT - THE ACTRESS WHO CHANGED THE WORLD.pdf
byMilanStankovic19
 
Life-Processes-in-Animals PPT/7TH CLASS NEW NCERT /CURIOSITY SCIENCE /BY K SA...
bySandeep Swamy
 
Structure and function of Integumentary system (skin).pptx
byAshish Umale
 
What are the Custom lot_serial number in Odoo 19
byCeline George
 
Aplastic_Anaemia_PA17.1_Presentation.pptx
byDibyajyoti Prusty
 
Centrifugation pharmaceutical engineering
byShraddha Bandgar
 
Acid Base Titration First Year B Pharm.pptx
byMs. Ashatai Patil
 
Cultivation Practice of Major Cole crops in Nepal.pptx
byUmeshTimilsina1
 
• Reinforcing the Human Firewall: Moving From Awareness to Applied Skill
byAdityarajsinh Gohil
 
Digital_Empathy_Toolkit_Netiquette_and_Responsible_Communication
byNidhi Pandya
 
India Cybercrime Threats & Response 2025: Digital Arrests, 1930 Helpline & Ze...
bySanjay Rathod
 
Non aqueous titration First Year B. Pharm.pptx
byMs. Ashatai Patil
 

Data Science Applied to Sustainability Analysis Jennifer B. Dunn

  • 1.
    Download the fullversion and explore a variety of ebooks or textbooks at https://ebookmass.com Data Science Applied to Sustainability Analysis Jennifer B. Dunn _____ Follow the link below to get your download now _____ https://ebookmass.com/product/data-science-applied-to- sustainability-analysis-jennifer-b-dunn/ Access ebookmass.com now to download high-quality ebooks or textbooks
  • 2.
    Here are somerecommended products for you. Click the link to download, or explore more at ebookmass.com Genomic Data Sharing Jennifer B. Mccormick https://ebookmass.com/product/genomic-data-sharing-jennifer-b- mccormick/ Applied Epistemology Jennifer Lackey https://ebookmass.com/product/applied-epistemology-jennifer-lackey/ Applied Data Analysis and Modeling for Energy Engineers and Scientists https://ebookmass.com/product/applied-data-analysis-and-modeling-for- energy-engineers-and-scientists/ Intelligent Data Analysis: From Data Gathering to Data Comprehension Deepak Gupta https://ebookmass.com/product/intelligent-data-analysis-from-data- gathering-to-data-comprehension-deepak-gupta/
  • 3.
    Data Science WithRust: A Comprehensive Guide - Data Analysis, Machine Learning, Data Visualization & More Van Der Post https://ebookmass.com/product/data-science-with-rust-a-comprehensive- guide-data-analysis-machine-learning-data-visualization-more-van-der- post/ Applied Longitudinal Analysis 2nd Edition https://ebookmass.com/product/applied-longitudinal-analysis-2nd- edition/ Foundations of Calculus for Data Science: An Foundational Guide to Understanding the Application of Calculus to Data Science Bisette https://ebookmass.com/product/foundations-of-calculus-for-data- science-an-foundational-guide-to-understanding-the-application-of- calculus-to-data-science-bisette/ Introduction to Python for Econometrics, Statistics and Data Analysis Kevin Sheppard https://ebookmass.com/product/introduction-to-python-for-econometrics- statistics-and-data-analysis-kevin-sheppard/ (eBook PDF) Intro to Python for Computer Science and Data Science: Learning to Program with AI, Big Data and The Cloud https://ebookmass.com/product/ebook-pdf-intro-to-python-for-computer- science-and-data-science-learning-to-program-with-ai-big-data-and-the- cloud/
  • 6.
  • 8.
    DATA SCIENCE APPLIEDTO SUSTAINABILITYANALYSIS Edited by JENNIFER B. DUNN Northwestern-Argonne Institute of Science and Engineering, Evanston, IL, USA; Center for Engineering Sustainability and Resilience, Northwestern University, Evanston, IL, USA; Chemical and Biological Engineering, Northwestern University, Evanston, IL, USA PRASANNA BALAPRAKASH Northwestern-Argonne Institute of Science and Engineering, Evanston, IL; Math and Computer Science Division, Argonne National Laboratory, Lemont, IL, USA
  • 9.
    Elsevier Radarweg 29, POBox 211, 1000 AE Amsterdam, Netherlands The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, United Kingdom 50 Hampshire Street, 5th Floor, Cambridge, MA 02139, United States Copyright © 2021 Elsevier Inc.All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions. This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein). Notices Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary. Practitioners and researchers must always rely on their own experience and knowledge in evaluating and using any information, methods,compounds,or experiments described herein.In using such information or methods they should be mindful of their own safety and the safety of others,including parties for whom they have a professional responsibility. To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress ISBN: 978-0-12-817976-5 For Information on all Elsevier publications visit our website at https://www.elsevier.com/books-and-journals Publisher: Candice Janco Acquisitions Editor: Marisa LaFleur Editorial Project Manager: Charlotte Kent Production Project Manager: BharatwajVaratharajan Cover Designer:Victoria Pearson Typeset by Aptara, New Delhi, India
  • 10.
    v Contributors ix 1. Overviewof data science and sustainability analysis 1 Prasanna Balaprakash, Jennifer B. Dunn Data science is central to advances in sustainability 1 Types of sustainability analyses 6 Data science tools 7 Overview of case studies in data science in sustainability 10 References 13 PART 1 Environmental Health and Sustainability 15 2. Applying AI to conservation challenges 17 Niraj Swami Introduction 17 What is adaptive management? 19 Creating value with AI 21 The impact technology canvas: synchronizing AI initiatives 24 Summary 26 References 27 3. Water balance estimation in Australia through the merging of satellite observations with models 29 Valentijn R.N. Pauwels, Ashley Wright, Ashkan Shokri, Stefania Grimaldi Introduction 29 Case studies 32 Conclusion 38 References 40 4. Machine learning in the Australian critical zone 43 Elisabeth N. Bui What is CZ science? 43 Machine learning techniques used 45 Questions addressed and data used 51 Key results, insights, and why they matter to the field of sustainability 65 Future development and application of data science in CZ science 69 Glossary terms 72 References 73 CONTENTS
  • 11.
    Contents vi PART 2 Energyand Water 79 5. A clustering analysis of energy and water consumption in U.S. States from 1985 to 2015 81 Evgenia Kapousouz, Abolfazl Seyrfar, Sybil Derrible, Hossein Ataei Introduction 81 Materials and methods 82 Results 87 Conclusion and discussion 99 Acknowledgments 100 Appendix 100 References 106 6. Exploring rooftop solar photovoltaics deployment and energy injustice in the US through a data-driven approach 109 Sergio Castellanos, Deborah A. Sunter, Daniel M. Kammen Introduction 109 A focus on distributed solar PV 110 Assessing rooftop PV potential across cities 112 Evaluating equity in rooftop PV deployment: a case study for the United States 117 Concluding remarks 124 References 125 7. Data-driven materials discovery for solar photovoltaics 129 Leon R. Devereux, Jacqueline M. Cole Introduction 129 Fundamentals of photovoltaics 131 Data-driven materials discovery schemes 140 Case studies of solar materials discovery 146 Future outlook 161 References 162 PART 3 Sustainable Systems Analysis 165 8. Machine learning in life cycle assessment 167 Mikaela Algren, Wendy Fisher, Amy E. Landis Introduction to life cycle assessment (LCA) 168 LCAs role in sustainability 168 General methods for process LCA 171 Impact assessment 173 Uses and limitations of LCA 175 Tools and data sources for LCA 175
  • 12.
    Contents vii Introduction tomachine learning (ML) 176 Types of ML 177 Choosing a machine learning algorithm 179 Tools for ML 182 ML in LCA 182 ML for surrogate LCAs 183 ML in LCI 184 ML in LCIA 185 ML in interpretation and analysis 186 Conclusion 188 References 188 9. Industry sustainable supply chain management with data science 191 Deboleena Chakraborty, Richard K. Helling Introduction to LCA 191 Today’s limitations 195 LCA applications 196 Vision/Needs 200 References 201 PART 4 Society and Policy 203 10. Deep learning with satellite imagery to enhance environmental enforcement 205 Cassandra Handan-Nader, Daniel E. Ho, Larry Y. Liu Introduction 205 The methodological evolution of remote sensing 208 Using deep learning to identify CAFOs 216 Discussion 221 References 224 11. Towards achieving the UNs data revolution: combining earth observation and socioeconomic data for geographic targeting of resources for the sustainable development goals 229 Gary R. Watmough, Charlotte L.J. Marcinko Why social remote sensing? 229 Using EO data for understanding socioeconomic conditions 233 How can we socialize the pixel? 240 A socio-ecologically informed approach to linking EO and socioeconomic data 242 Future directions for EO and social data 248 Conclusions 252 References 252
  • 13.
    Contents viii 12. An indicator-basedapproach to sustainable management of natural resources 255 Esther S. Parish, Virginia H. Dale, Maggie Davis, Rebecca A. Efroymson, Michael R. Hilliard, Henriette Jager, Keith L. Kline, Fei Xie Introduction 255 Selecting and prioritizing indicators 258 Indicator datasets and data science considerations 259 Data analytics 264 Implications for society & policy 274 Conclusion 275 References 276 PART 5 Conclusion 281 13. Research and development for increased application of data science in sustainability analysis 283 Jennifer B. Dunn, Prasanna Balaprakash Introduction 283 Needs for data to enable data science in sustainability 283 Data sources 286 Data science advances 287 Conclusion 290 References 291 Index 293
  • 14.
    ix Mikaela Algren Colorado Schoolof Mines, Golden, CO, USA Hossein Ataei Department of Civil, Materials, and Environmental Engineering, University of Illinois at Chicago, Chicago, IL, USA Prasanna Balaprakash Northwestern-Argonne Institute of Science and Engineering, Evanston, IL; Math and Computer Science Division,Argonne National Laboratory, Lemont, IL, USA Elisabeth N. Bui CSIRO Land and Water, Canberra,ACT,Australia Sergio Castellanos The University of Texas at Austin, Austin,TX, USA Deboleena Chakraborty Dow, Midland, MI, USA Jacqueline M. Cole Cavendish Laboratory, Department of Physics, University of Cambridge, J. J.Thomson Avenue, Cambridge, UK; ISIS Neutron and Muon Facility, STFC Rutherford Appleton Laboratory, Harwell Science and Innovation Campus, Oxfordshire, Didcot, UK; Department of Chemical Engineering and Biotechnology, University of Cambridge,West Cambridge Site, Philippa Fawcett Drive, Cambridge, UK Virginia H. Dale University of Tennessee Knoxville,TN, USA Maggie Davis Oak Ridge National Laboratory, Oak Ridge,TN, USA Sybil Derrible Department of Civil, Materials, and Environmental Engineering, University of Illinois at Chicago, Chicago, IL, USA Leon R. Devereux Cavendish Laboratory, Department of Physics, University of Cambridge, J. J.Thomson Avenue, Cambridge, UK Jennifer B. Dunn Northwestern-Argonne Institute of Science and Engineering, Evanston, IL, USA; Center for Engineering Sustainability and Resilience, Northwestern University, Evanston, IL, USA; Chemical and Biological Engineering, Northwestern University, Evanston, IL, USA Rebecca A. Efroymson Oak Ridge National Laboratory, Oak Ridge,TN, USA Contributors
  • 15.
    Contributors x Wendy Fisher Colorado Schoolof Mines, Golden, CO, USA Stefania Grimaldi Monash University, Department of Civil Engineering, Clayton,VIC,Australia Cassandra Handan-Nader Department of Political Science,Stanford University,Stanford,CA,USA;Regulation,Evaluation, and Governance Lab,Stanford,CA,USA Richard K. Helling Dow, Midland, MI, USA Michael R. Hilliard Oak Ridge National Laboratory, Oak Ridge,TN, USA Daniel E. Ho Department of Political Science,Stanford University,Stanford,CA,USA;Stanford Institute for Human-Centered Artificial Intelligence,Stanford,CA,USA;Regulation,Evaluation,and Governance Lab,Stanford,CA,USA;Stanford Institute for Economic Policy Research,Stanford, CA,USA;Stanford Law School,Stanford,CA,USA Henriette Jager Oak Ridge National Laboratory, Oak Ridge,TN, USA Daniel M. Kammen University of California, Berkeley, CA, USA Evgenia Kapousouz Department of Public Administration, University of Illinois at Chicago, Chicago, IL, USA Keith L. Kline Oak Ridge National Laboratory, Oak Ridge,TN, USA Amy E. Landis Colorado School of Mines, Golden, CO, USA Larry Y. Liu U.S. Court of Appeals for the Eleventh Circuit, Birmington,AL, USA Charlotte L.J. Marcinko School of Engineering, University of Southampton, Southampton, UK Esther S. Parish Oak Ridge National Laboratory, Oak Ridge,TN, USA Valentijn R.N. Pauwels Monash University, Department of Civil Engineering, Clayton,VIC,Australia Abolfazl Seyrfar Department of Civil, Materials, and Environmental Engineering, University of Illinois at Chicago, Chicago, IL, USA Ashkan Shokri Monash University, Department of Civil Engineering, Clayton,VIC,Australia Deborah A. Sunter Tufts University, Medford, MA, USA
  • 16.
    Contributors xi Niraj Swami SeniorDirector, Conservation Technology Strategy & Enablement at The Nature Conservancy, Founding Partner, SCADVentures, Chicago, IL, USA Gary R.Watmough School of Geosciences, University of Edinburgh, Edinburgh, UK Ashley Wright Monash University, Department of Civil Engineering, Clayton,VIC,Australia Fei Xie Oak Ridge National Laboratory, Oak Ridge,TN, USA
  • 18.
    1 Data Science Appliedto Sustainability Analysis © 2021 Elsevier Inc. DOI: 10.1016/C2018-0-02415-9 All rights reserved. CHAPTER 1 Overview of data science and sustainability analysis Prasanna Balaprakasha,b , Jennifer B. Dunna,c,d a Northwestern-Argonne Institute of Science and Engineering, Evanston, IL, USA b Math and Computer Science Division,Argonne National Laboratory, Lemont, IL, USA c Center for Engineering Sustainability and Resilience, Northwestern University, Evanston, IL, USA d Chemical and Biological Engineering, Northwestern University, Evanston, IL, USA Chapter Outlines Data science is central to advances in sustainability 1 Types of sustainability analyses 6 Data science tools 7 Supervised learning 8 Unsupervised learning 9 Reinforcement learning 10 Tools 10 Overview of case studies in data science in sustainability 10 Data science is central to advances in sustainability The frequently-used term sustainability is often defined per the Brundtland Report’s definition of sustainable development: “Sustainable development is development that meets the needs of the present without com- promising the ability of future generations to meet their own needs.” (World Commission on Environment and Our Common, 1987) For the purposes of this book, we intend the term sustainability to mean the poten- tial for achieving a high quality of life in human, social, environmental, and economic systems, both today and in the future. For this potential to be realized, society must reach a point where air quality, water quality,and soil health are robust and do not pose a threat to ecosystem or human health. Air quality remains unsatisfactory globally, and is the fifth risk factor for global mortal- ity in 2017, associated with 4.9 deaths (Fig. 1.1) and 147 million healthy life years lost. (Health Effects Institute 2019) In addition, water quality globally is a challenge, with only half of water bodies exhibiting good quality per United Nations Sustainability
  • 19.
    Data science appliedto sustainability analysis 2 Development Goal monitoring initiatives (Fig. 1.2). (UN Environment, 2018) In addi- tion to air and water quality, soil quality has significant implications for human health yet can suffer from pollution from industry, mining, or waste disposal. In Europe, sites with likely soil contamination number 340,000 with only one-third of these under- going detailed study and only 15 percent of those remediated. (Food and Agriculture Organization of the United Nations 2015) In the US, the Environmental Protection Agency has remediated 9.3 million ha of contaminated land with 160 contaminated sites on the Superfund National Priorities List remaining to be evaluated and 49 new sites proposed to be added to the list. In addition to contamination, other challenges to soil sustainability include erosion and loss of organic carbon. (Food and Agriculture Organization of the United Nations 2015) Similarly, the abundance of energy, preferably produced from minimally polluting and renewable resources, and clean water is essential for society’s survival. Like the quality of air, water, and soil, energy and water use are closely tied to human activity.The world has nearly tripled its energy consumption since 1971 (Fig.1.3).(International Energy Agency. IEA, 2020) Fossil fuels (coal, oil, and natural gas) continue to dominate the production of energy with the attendant impacts from mining and extraction and subsequent combus- tion of these sources which releases greenhouse gas emissions into the air which contain carbon that had been long-sequestered in the earth.Combustion of fossil fuels diminishes air quality. In addition to considering total energy production, it is important to evaluate energy efficiency,which has only improved by 1.2 percent from 2017 to 2018.Improving energy efficiency is one of the best strategies available to cutting energy consumption Fig. 1.1 Number of deaths attributable to air pollution in 2017. Data source: Global Burden of Disease Study 2017. IHME, 2018. (Health Effects Institute 2019).
  • 20.
    Overview of datascience and sustainability analysis 3 and associated pollution. (International Energy Agency 2019) The production of energy, along with many other activities,including agriculture consumes water.Correspondingly, as the population has increased and clean water supplies have diminished, water scarcity is a reality for approximately one-half of the global population. (Boretti and Rosa, 2019) Fig. 1.2 Proportion of water bodies with good ambient water quality (percent) in 2017 (UN Environ- ment 2018). 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Overall Rivers Proportion of water bodies with good ambient water quality Open Water Bodies Groundwater Fig. 1.3 World total energy supply by source (million tons of oil equivalent). (International Energy Agency. IEA, 2020). 16,000 14,000 12,000 10,000 8,000 6,000 4,000 2,000 0 1971 1976 Energy (million tons oil equivalent) 1981 1986 1991 1996 2001 2006 2011 2016 Coal Oil Natural gas Nuclear Hydro Biofuels and waste Other
  • 21.
    Data science appliedto sustainability analysis 4 Challenges faced in achieving sustainability include improving food and water secu- rity, maintaining biodiversity, reducing air and water pollution, reducing greenhouse gas emissions, increasing reuse and recycling, and increasing system-level efficiencies in energy, urban, agricultural and industrial systems. Furthermore, extracting and disposing of the materials we need to make the equip- ment, devices, and food we need to run our society can be unsustainable, generating pollution and operating without concern for the long-term availability of critical materials. In fact, the very technologies society is relying on to address climate change, including wind turbines, solar panels, and lithium-ion batteries, rely on metals (cobalt, nickel, copper, rare earths) that are mined, often in developing countries where envi- ronmental regulations are often insufficient to protect populations from exposure to pollution in the air, water, and soil. (Sovacool et al., 2020) Conserving natural lands is an important part of ensuring a healthy and productive future for human society. Natural lands such as grasslands, wetlands, and forests provide innumerable ecosystems services such as mitigating floods, sequestering carbon, and enhancing biodiversity.Targeted conservation initiatives are required to slow the pace and extent of extinction, improves environmental quality, and retain the inspirational value of nature. (Balvanera, 2019) Economic and consumer preference drivers often can favor technology and soci- etal developments that advance towards sustainability, but law and policy are important drivers as well. (Ashford and Hall, 2011) For example, one reason energy efficiency gains have faltered (Fig. 1.4) is a lack of clear policy to advance energy efficiency. (International Energy Agency 2019) Finally, social well-being, in part as indicated by the portion of the world’s popula- tion that has can viably provide food and other basic needs for themselves and their Fig. 1.4 Global Improvements in Primary Energy Intensity 2000–20186 . 3.5% 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% 2000-2009 2010-2014 2015 2016 2017 2018
  • 22.
    Overview of datascience and sustainability analysis 5 families, is an important element of sustainability. Global levels of poverty between 2013 and 2015 declined through all regions of World Bank analysis yet the percent of people living at the International Poverty Line of $1.90 per day stayed relatively constant in many of these regions, showing a decrease most notably in South Asia (Fig. 1.5).The over 700 million people globally living below this poverty line in 2015 is an unignorable indication that sustainability has not yet been attained. Undoubtedly, the breadth of earth, industrial, and societal systems that contribute to sustainability is immense. Developing technology, societal, and policy approaches to address each facet of sustainability can be guided by analyses that point the way, for example, towards pollution or water scarcity hotspots, the most impactful energy efficiency technologies, or regionally-specific conservation strategies. These analyses can make use of ever-growing volumes of data including satellite imagery, continuous sensor data from industrial processes, social media data, and environmental sensors, to name only a few.As a result, data science techniques have become central to addressing sustainability challenges and this role will only expand in the future. Accordingly, we have assembled this book with the contribution of co-authors who are addressing sustainability challenges in the spheres of environmental health, energy and water, sustainable industrial systems, and society and policy. Our intention is to provide a well-rounded set of case studies addressing different challenges using varying types of sustainability analysis and data to serve as a reference for analysts who seek to employ data science in their work and for data scientists looking to apply their skills to sustainability challenges.Another audience for this book will be policy makers who rely on sustainability analyses as a decision making tool to evaluate how governments Fig. 1.5 Number of people and percent of population at the International Poverty Line of $1.90/day (2011 PPP). (World Bank 2018) 900 800 700 600 500 400 300 200 100 0 East Asia and Pacific Million People Europe and Central Asia Latin America and the Carribean Percent of People Population 2013 Population 2015 % 2013 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% % 2015 Middle East and North Africa South Asia Sub-Saharan Africa World Total
  • 23.
    Data science appliedto sustainability analysis 6 could collect data that would support these efforts and use the results of these analyses in policymaking.Additionally, this book could be used in data science and systems analysis classrooms to provide case study examples, especially at the graduate student level. In the remainder of this introductory chapter, we review different types of analy- ses that guide our understanding of and action towards increasing sustainability. We provide an overview of data science tools that can be used in sustainability analyses. Finally, we introduce the different case studies readers will encounter in the remaining chapters.We note that the concluding chapter of this book summarizes data gaps and research needs for the further building of data science applications in sustainability analysis. Types of sustainability analyses The term “sustainability analysis” is meant to be broad for this book to capture a wide range of analyses that can address or evaluate society’s advancement towards sustainable resource management and wellbeing. We summarize some examples of analysis types that fall under the term sustainability analysis in Fig.1.6.As one example,natural systems modeling improves our understanding of the geoscience, bioscience, and social science that underpins natural systems relies on data analysis and modeling, with an effort to move towards prediction. Hydrologic modeling, for example, based on land character- istics and precipitation data, can help predict the location and effects of flooding from major precipitation events. Soil carbon modeling that explores the influence of agricul- tural management practices on levels of carbon storage in soils is another example of analysis that could be grouped under “sustainability analysis.” Furthermore, modeling of air pollution dispersion could also be categorized under this umbrella. All of these natural systems models contain parameters that must be estimated based on evaluation of data sets. Furthermore, many types of analyses can enhance the design of industrial, energy, and water systems that offer sustainability improvements over the status quo. As one primary example, machine learning can be used to speed up the design of new mate- rials that can be used in any number of important sustainability applications from Fig. 1.6 Examples of sustainability analysis types that are increasingly using data science techniques.
  • 24.
    Overview of datascience and sustainability analysis 7 designing membranes that exhibit less fouling in water treatment applications thereby reducing energy and chemicals used in wastewater treatment to exploring next genera- tion lithium-ion battery chemistries. Additionally, as the Industrial Internet of Things continues to expand, analysts will apply data science techniques to identify opportuni- ties to improve the energy, water, and material efficiency of industrial processes. Finally, evaluating the progress of consumers’ adoption of technology that will be more energy or water efficient, for example, is another important type of sustainability analysis.This type of analysis could be based on earth observation data in the case of adoption of large infrastructure or based on social media posts that indicate shifts in technology use in the home, on the road, or in the workplace. Two mainstays of sustainable systems analysis are life cycle assessment (LCA) and materials flow analysis (MFA). Whereas LCA evaluates the environmental effects of a product or process – from fuels to electronics to foods – MFA tracks the flows of com- modities within a system boundary, which could be a city, a region, or a nation. LCA and MFA are at the very beginning of applying data science techniques, in general because datasets are often insufficiently large to allow data science approaches to offer value.As the data revolution continues,these two analysis types have many opportunities to leverage data science techniques. Finally, evaluations of social well-being are another important pillar of sustainability analyses because sustainability is often described as having three pillars – economic, social, and environmental. One expanding enabler of using data science approaches in social well-being evaluations is satellite imagery, which provides a bird’s eye view of living conditions for Earth’s inhabitants.While these data can show us these conditions, they cannot identify what has caused them.This second and critical step will require the linkage of image interpretation and causal analysis. Regardless of analysis type, data availability is a cornerstone of all of these analyses. In some instances that remain data sparse, the use of data science techniques in these areas is anticipatory rather than widespread. Furthermore, the examples we provide here are not all-encompassing and the list of types of sustainability analyses that benefit from data science approaches today and into the future will evolve and grow. Data science tools Broadly speaking, data science is an inter-disciplinary field that adopts data collection, pre-processing, meaning/useful feature extraction methods, data exploration meth- ods, and predictive models to extract knowledge from a wide range of structured and unstructured data.Given the structure,size,heterogeneity,and complexity of the data sets, a wide range of data science tools and techniques have been developed. Among them, statistical machine learning is a prominent class of methods that are used and adopted for many data science task. Next, we will review three widely used subclass of ML methods.
  • 25.
    Data science appliedto sustainability analysis 8 Supervised learning It is used to model the functional relationship between the output variables and one or more independent input variables. Typically, the original function relationship is unknown and/or hard to derive in an analytical form.The approach starts with a set of training data given as a large set of input-output pair.The goal is to find a surrogate function for original function relationship such that the difference between prediction from the surrogate function and the observed value is minimal for all input-output pair in the training data and the unseen testing data. Several supervised learning algorithms exist in the ML literature. Based on the functionality, one can group them as follows: regularization, instance-based methods, recursive partitioning, kernel-based methods, artificial neural networks, bagging, and boosting methods. Often, the best method depends on the data and type of the modeling task, such as volume of data, variety of data, and speed required for training and inference. Here, we cover several widely adopted algorithms to cover different groups. We will review them from regression perspective (predicting a scalar value).Without loss of generality, most of these methods also handle classification (predicting a class). Multivariate linear regression (Bishop, 2006) is one of the most simple methods for modeling the functional relationship between inputs and output. It models the functional relationship using a linear equation.This is given by the sum of product of each input with a scaling factor. A bias factor is also added to the equation.The mul- tivariate linear regression involves finding the scaling factors and the bias. It is one of the well understood method and often preferred for interpretability and simplicity. It is important that data science practitioners try and adopt this method as a baseline and comparison to other methods. Ridge regression (Hoerl and Kennard, 1970) is a regularization algorithm that is designed to reduce the model complexity so that the model does not overfit the train- ing data. This overfitting occurs in supervised learning when the model learns small variations and/or noise in the training set and consequently loses prediction accuracy on the testing data.To do so, in addition to minimizing the error between predicted and actual observations, the method penalizes the training objective with respect to input coefficients and achieves tradeoff between minimizing the error and minimizing the sum of the square of the coefficients. k-nearest-neighbor regression (Bishop, 2006) belongs to the class of instance- based methods, where the training data is stored in memory and the model is built only during testing. Given a testing point, the method first finds k nearest input points in the training data and returns the prediction as the average of k outputs.Typically, k and the nearest distance metric are user defined hyperparameters. Support vector machine (Drucker et al., 1996) is a widely-used kernel-based method. It uses a kernel function to project the input space onto a higher-dimensional feature space; a linear regression is performed in the transformed space.The training is
  • 26.
    Overview of datascience and sustainability analysis 9 formulated as a convex quadratic optimization problem, for which efficient optimiza- tion algorithms are utilized.The effectiveness of this method depends on a good choice of kernel type and their hyperparameters. Decision tree regression (Breiman, 1984) belongs to the class of recursive parti- tioning methods. It recursively splits the multidimensional input space of training points into regions such that inputs with similar outputs fall within the same region.The splits give rise to a set of if-else rules. For each region, an average over the output values is computed and stored at the end of each rule. Given a new testing point, the decision tree employs the if-else rule to return the stored value as the predicted value. Random forest (Breiman, 2001) is a bagging approach that considers random subsamples of the training dataset and builds a decision tree on each subsample. Given a new test data point, the prediction from each tree is averaged to obtain the predicted value. Gradient boosting regression (Friedman, 2002) is similar to random forest but the trees are constructed sequentially on each random subsample. The key idea is to build each tree to minimize the error of the previous tree. Deep neural networks (Goodfellow et al., 2016) belong to the class of artificial neural networks.They are characterized by stacked layers, where each layer is composed of a number of units. Each unit receives inputs from units from previous layers, which are combined in a weighted linear fashion and passed through a nonlinear function. The first layer receives the training points and the predictions are obtained from the last layer of the stack.The training phase consists of modifying the weights of the stacked layers to minimize the prediction error on the training data set.This is typically done by stochastic gradient descent optimization method that computes the gradients of the objective function with respect to all the weights in the network and uses them to update the weights. Unsupervised learning Traditionally, unsupervised learning methods were used for exploratory analysis. (Bishop,2006) Notably,clustering and dimension reduction methods were adopted for a wide range of data science tasks.The former computes the distances between the points in the given data using a distance metric, which is then used to group similar points. The latter is often employed to project the high dimensional data into low dimensional embedding space for visualization. In recent years, auto encoders, a class of deep neu- ral networks, have received significant attention for dimension reduction method due to their ability to perform effective nonlinear dimension reduction and handle large amount of data.Another key advancement in the area of unsupervised learning is gener- ative modeling, which has potential to understand and explain the underlying structure of the input data when there are few–or even no–labels.A promising generative model- ing approach that has received much recent attention is generative adversarial networks
  • 27.
    Data science appliedto sustainability analysis 10 (GAN). (Goodfellow et al., 2014) The basic idea in GAN is to train two deep neural networks simultaneously and capture the domain-specific features and representations from the unlabeled data and deploy them as labeled data becomes available.For example, GANs can produce high quality synthetic images of real-world objects without having any explicit labels of what those objects are. By automatically extracting the underlying structure of the inputs without labels, GANs can empower supervised learning methods to understand the context of the domain in which they operate. Reinforcement learning It is an approach that is concerned with is concerned with training agents for autono- mous design and control. (Sutton and Barto, 2018) The agents interact within an envi- ronment, receive rewards, and use them to improve the actions iteratively using training settings.The agents once trained can be deployed for control in test settings. Tools The data science software ecosystem is quite vibrant has a wide range of software tools and many of them are open-source. Scikit-learn (Pedregosa et al., 2018) is one of the widely used package for numerous data science tasks. It has implementation of prepro- cessing, unsupervised, and supervised learning methods that are integral part of many data science pipelines. Similarly, R project for statistical computing (R Core Team, 2021) provides a number of libraries to build data science pipelines with minimal effort. Jupyter notebook (Kluyver et al., 2016) and R studio (Allaire, 2012) are productivity centric integrated development editors for interactive data science code development. Tensorflow (Abadi et al., 2016) and Pytorch (Paszke et al., 2019) are packages for dif- ferentiable computing and are widely used for the design and development of deep neural network models.Python and R ecosystem provides a number of libraries for data visualization (for example, Matplotlib (Hunter, 2007) and ggplot2 (Wickham, 2011)). RapidMiner, (Mierswa and Klinkenberg, 2018) Weka (Hall et al., 2009), and KNIME (Berthold et al., 2009) software tools designed for users with minimal programming experience.They provide easy to use interfaces to build data science pipelines but do not provide flexibility and configurability as programming-intensive software stack. Overview of case studies in data science in sustainability Data science techniques have been applied to numerous domains within the sustainabil- ity field.For example,social media data have been analyzed with data science techniques to inform an understanding of urban sustainability including aspects like mobility and economic development (Ilieva and McPhearson, 2018) and even waste minimization in beef supply chain. (Mishra and Singh, 2018) In general, the agricultural sector holds much promise for applications of data science to improve farming sustainability such as
  • 28.
    Overview of datascience and sustainability analysis 11 reducing use of fertilizer and irrigation. (Kamilaris et al., 2017) Considering the social side of sustainability, predictive analytics and data visualization have been used to study and improve the humanitarian supply chain. (Gupta et al., 2019) With such an expansive space of intersection between data science and sustainability, this book covers only a subset of an ever-growing field.We have focused on the broad top- ics of environmental quality and sustainability,energy and water,sustainable systems analy- sis, and society and policy. Fig. 1.7 places each chapter in this book in one of these topics. Environmental Quality and Sustainability focuses on how we can better under- stand natural ecosystems and design strategies to protect them and improve air, water, and soil quality. Swami (Chapter 2) examines the many ways artificial intelligence can contribute to conservation efforts. Bui (Chapter 3) describe the application of machine learning techniques including supervised pattern recognition, random forests, support vector machines, and deep learning to investigate spatial patterns such as species distri- bution, streamflow, and land use within Australian Critical Zones. For this application, machine learning techniques have proven helpful to predict spatial patterns, identify regions vulnerable to factors such as erosion or soil organic carbon loss, and to find the drivers of spatial patterns. Pauwels et al. (Chapter 4) describe several methods that have been used to improve hydrologic modeling to inform water resources for Australia. The methods described include Bayesian techniques and Monte Carlo methods that demonstrate improvements in parameter estimation over other methods and can better predict flooding. As described in Section 1.1, energy use and water consumption continue to rise. Kapousouz et al. (Chapter 5) use clustering to explore spatial and temporal use patterns Fig. 1.7 Organization of case studies in this book.
  • 29.
    Data science appliedto sustainability analysis 12 in energy and water consumption at the state level. Identifying patterns can lead to technology and policy development to reduce resource consumption. Developing and deploying technology to harness renewable energy is one important approach to mini- mizing the influence of energy production. Devereux and Cole (Chapter 7) explore avenues for using machine learning to develop solar photovoltaic cells that move this technology to optimal performance. One example is to use machine learning to predict property based on structure.Another relevant application of machine learning is to carry out high-throughput computational screening.A final option is to automatically gener- ate property databases. One helpful tool in this case is using natural language processing to mine the technical literature for property information that can be include in such a database.As new technology develops, it is helpful to evaluate how it is being deployed and used in the real world. One reality of solar photovoltaic deployment is that it tends to be limited in areas of lower socio-economic status. Castellanos et al. (Chapter 6) integrated Google Project Sunroof and United States Census data and carried out regression analysis and bootstrapping to explore racial and ethnic disparities of rooftop solar voltaic technology. With this information in hand, interventions can be better designed to increase solar PV deployment. In quantifying sustainability, it is important to remember that no one technology, policy, region, or other subdivision of a full system acts in isolation. For example, rede- signing a product with new materials to achieve a lighter weight could reduce the fuel consumed in shipping it, but the new materials themselves could be more energy inten- sive to produce than the original materials. For this reason, it is important to consider full systems when developing new technologies, considering changes to existing tech- nologies, or considering policies that could reduce environmental burdens in one por- tion of the supply chain only to increase them elsewhere. LCA is one analysis approach, standardized by the International Standards Organization, (ISO 2006; ISO 2006) that has been used to evaluate how changes in one step of a life cycle (e.g., manufactur- ing) alter the overall environmental effects of a product or process. Algren and Landis (Chapter 8) consider how machine learning can be best applied in LCA and provide several examples.Chakraborty and Helling (Chapter 9) describe how LCA in the indus- trial sector in particular has the power to guide supply chains to enhanced sustainability especially as companies gain access to data required to build comprehensive LCAs.They provide a vision for what this future state could look like. In addition to understanding the natural world’s health and how it functions, how technology influences sustainability through energy, water consumption and supply chain effects, focusing on societal behavior and policy development can also yield insights into enhancing overall societal sustainability. Some policies designed to limit pollution can be challenging to enforce. Handan-Nader et al. (Chapter 10) describe the use of machine learning with high-resolution satellite imagery to direct environmental regulation enforce- ment.They provide examples in detecting oil spills, deforestation, and air pollutant emis- sions.Subsequently,they describe their work with identifying concentrated animal feeding
  • 30.
    Overview of datascience and sustainability analysis 13 operations that are of interest to environmental regulators but may be difficult to locate because there is no public database of their locations. Earth observation (EO) data such as satellite imagery is in general showing increasing promise for supporting decision making and policy development in the sustainability space, including when relationships can be drawn between information from EO and data from other sources such as air pollutant monitors and economic databases, for example.Watmough and Marcinko (Chapter 11) explore how EO data can be used to target, develop, and evaluate policy to reduce pov- erty. Finally, Parish et al. (Chapter 12) describe how multiple field measurements and EO data can be integrated to inform sustainability assessments through the use of indicators, including developing techniques to monetize ecosystem services. The final chapter of this book outlines research needs that must be addressed to con- tinue to expand the application of data science techniques to sustainability challenges. As the intersection between data science and sustainability continues to develop and mature, the science of understanding earth system, technologies that are energy- and water-efficient, and policies that encourage the ongoing reduction of pollution, includ- ing greenhouse gas emissions, and poverty will benefit. References Abadi, M., Barham, P., Chen, J., Davis,A., Dean, J., Devin, M., Ghemawat, S., Irving, G., Isard, M., Kudlur, M., Levenberg, J., Monga, R., Moore, S., Murray, D.G., Steinter, B.,Tucker, P.,Vasudevan,V.,Warden, P., Wicke, M.,Yu,Y., Zheng, X., 2016.Tensorflow: a system for large-scale machine learning, Proceedings of the 12th USENIX Symposium on Operating Systems Design and Implementation, 265–283 https://www.usenix.org/conference/osdi16/technical-sessions/presentation/abadi. Allaire, J., 2012. RStudio: integrated development environment for R. in vol. 770, 394. Ashford, N.A., Hall, R.P., 2011.The Importance of regulation-induced innovation for sustainable develop- ment. Sustainability 3, 270–292. Balvanera, P., 2019. Societal burdens of nature loss. Science 366, 184–185. Berthold, M., Cebron, N., Dill, F., Gabriel,T.R., Kotter,T., Meinl,T., Ohl, P.,Thiel, K.,Wiswedel, B., 2009. KNIME-the Konstanz information miner: version 2.0 and beyond. AcM SIGKDD explorations Newsletter. https://doi.org/10.1145/1656274.1656280. Bishop, C.M., 2006. Pattern Recognition and Machine Learning. Springer, NewYork, NY. Boretti,A., Rosa, L., 2019. Reassessing the projections of theWorldWater Development Report. npj Clean Water 2, 1–6. Breiman, L., 1984. Classification and Regression Trees. Chapman Hall/CRC, NewYork, NY. Breiman, L., 2001. Random Forests. Mach. Learn. 45, 5–32. Drucker, H., Burges, C.J.C., Kaufman, L., Smola, A.,Vapnik,V, 1996. Support vector regression machines, Proceedings of the 9th International Conference on Neural Information Processing Systems. MIT Press, pp. 155–161. Food and Agriculture Organization of the United Nations. Status of the world’s soil resources. (2015). Friedman, J.H., 2002. Stochastic gradient boosting. Comput. Stat. Data Anal. 38, 367–378. Goodfellow, I., Bengio,Y., Courvile,A., 2016. Deep Learning. MIT Press. http://www.deeplearningbook. org/. Goodfellow, I.J. et al. Generative adversarial networks. arXiv:1406.2661[cs, stat] (2014). Gupta, S.,Altay, N., Luo, Z., 2019. Big data in humanitarian supply chain management: a review and further research directions.Ann Oper Res 283, 1153–1173. Hall, M., Eibe, F., Holmes, G., Pfahringer, B., Reutemann, P.,Witten,A.H., 2009.The WEKA data mining software: an update.ACM SIGKDD explorations newsletter 11, 10–18 https://www.kdd.org/explora- tion_files/p2V11n1.pdf.
  • 31.
    Data science appliedto sustainability analysis 14 Health Effects Institute, 2019. State of Global Air 2019. Special Report. https://www.stateofglobalair.org/ sites/default/files/soga_2019_report.pdf. Hoerl, A.E., Kennard, R.W., 1970. Ridge regression: biased estimation for nonorthogonal problems. Technometrics 12, 55–67. Hunter, J.D., 2007. Matplotlib: a 2D graphics environment. Comput. Sci. Eng. 9, 90–95. I. Mierswa R. Klinkenberg. RapidMiner Studio (9.1) [Data science, machine learning, predictive analyt- ics]. (2018). Ilieva, R.T., McPhearson,T., 2018. Social-media data for urban sustainability. Nat. Sustain. 1, 553–565. International Energy Agency. IEA, 2020.World total energy supply by source, 1971-2018. https://www.iea. org/data-and-statistics/charts/world-total-energy-supply-by-source-1971-2018. International EnergyAgency.Energy efficiency.2019.https://www.iea.org/reports/energy-efficiency-2019 (2019). ISO. Environmental management: life cycle assessment; Principles and framework. (2006). ISO. Environmental management: life cycle assessment; Requirements and guidelines. (2006). Kamilaris,A., Kartakoullis,A., Prenafeta-Boldú, F.X., 2017.A review on the practice of big data analysis in agriculture. Comput. Electron.Agric. 143, 23–37. Kluyver, T., et al., 2016. Jupyter Notebooks – a Publishing Format for Reproducible Computational Workflows, 87–90. Mishra, N., Singh, A., 2018. Use of Twitter data for waste minimisation in beef supply chain. Ann. Oper. Res. 270, 337–359. Paszke, A., et al., 2019. PyTorch: an imperative style, high-performance deep learning library. In:Wallach, H. et al (Ed.), Advances in Neural Information Processing Systems 32. Curran Associates, Inc., pp. 8026–8037. Pedregosa, F. et al. Scikit-learn: machine learning in python. arXiv:1201.0490[cs] (2018). R Core Team. http://www.R-project.org/. Sovacool, B.K, et al., 2020. Sustainable minerals and metals for a low-carbon future. Science 367, 30–33. Sutton, R.S., Barto,A.G., 2018. Reinforcement Learning:An Introduction.The MIT Press. UN Environment. Progress on ambient water quality, piloting the monitoring methodology and intitial findings for SDG indicator 6.3.2, 2018. (2018). Wickham, H., 2011. ggplot2.WIREs Computational Statistics 3, 180–185. World Bank. Poverty and shared prosperity 2018. Piecing together the poverty puzzle. https://www.world- bank.org/en/publication/poverty-and-shared-prosperity (2018). World Commission on Environment and Development. Our common future. (1987).
  • 32.
    15 PART 1 Environmental Healthand Sustainability 2. Applying AI to conservation challenges 17 3. Water balance estimation in Australia through the merging of satellite observations with models 29 4. Machine learning in the Australian critical zone 43
  • 34.
    17 Data Science Appliedto Sustainability Analysis © 2021 Elsevier Inc. DOI: 10.1016/C2018-0-02415-9 All rights reserved. CHAPTER 2 Applying AI to conservation challenges Niraj Swami Senior Director, Conservation Technology Strategy Enablement at The Nature Conservancy, Founding Partner, SCADVentures, Chicago, IL, USA Chapter Outlines Introduction 17 What is adaptive management? 19 Creating value with AI 21 Planning and implementing with AI 21 Monitoring evaluating with AI 22 Learning adapting with AI 23 The impact technology canvas: synchronizing AI initiatives 24 Summary 26 Introduction We live in a world where we’re surrounded by experiences,insights and tasks that are enriched by Artificial Intelligence (AI). From smart speakers in our living room that can distinctively understand commands from different voices to emails that basically write themselves based on your context. Conservation is no stranger to the breadth of capabilities AI enriches. A quick Google search on “AI in Conservation” will reveal numerous examples of how AI has been used in the conservation space.We did a quick survey of innovative AI-based initiatives at The Nature Conservancy and the greater sustainability technol- ogy community to highlight some intriguing areas of work: • Leveraging image-classification technologies to decipher land types from satellite imagery to further inform flood insurance policy schemes (Fig. 2.1). • Leveraging AI to extract patterns from user-generated social media posts to under- stand impact and economic significance of eco-tourism (near coral reefs) (Mapping Ocean Wealth na). • Using machine learning to forecast future water runoffs by modeling the physics of snowfall and snowmelt in a given region (Fig. 2.2). • Detecting and tagging specific development attributes from satellite imagery, such as plotting solar and wind energy plants in India and identify off-highway vehicle activ- ity in Mojave desert (Fig. 2.3). • Unmanned marine drones (Fig. 2.4) applying AI to self-navigate harsh ocean condi- tions using an array of on-device sensors, 360° cameras and GPS data, while alerting targets of interest remotely and also collecting hydrological data for on-shore analysis.
  • 35.
    Data science appliedto sustainability analysis 18 With data being collected on-device,these drones perform advanced image and time series data analysis, machine learning and other post-processing tasks on the edge. The use cases are aplenty. However, a notable pattern in these examples is the reli- ance on one particular data source - remote sensing data (images and other layers of insights) from satellites orbiting our planet. In 2014, the Wildlife Conservation Society shared “Ten ways remote sensing can contribute to conservation” (Rose et al., 2014) - primarily identifying use cases in monitoring conditions, understanding and predicting environmental changes. A quick Google search on “remote sensing datasets” will give you dozens of imagery datasets that can be leveraged for building AI solutions. While the readily available satellite imagery dataset is a great candidate for data sci- ence, what other scenarios could AI practitioners and data scientists pursue? How can we strategically look for conservation-related challenges best positioned for a boost Fig. 2.1 Community Rating System by Federal Emergency Management Agency (Open Space na). Fig. 2.2 Rapid progression of snowmelt and vegetation growth in the South Payette River Basin at three selected times in the spring (South Payette River Case Study na).
  • 36.
    Applying AI toconservation challenges 19 from cognitive technologies? In this chapter, we explore how AI-based solutions may be applied at various stages of a conservation project by looking through an innovator’s lens at Adaptive Management. What is adaptive management? Conservation is science guiding action, with the goal to prevent wasteful use of a natu- ral resource. Let’s explore the funnel of information in a typical conservation project. Fig. 2.3 Google Earth images of Off-Highway Vehicle (OHV) route proliferation, courtesy of John Zablocki and Michael Clifford at The Nature Conservancy. Fig. 2.4 Wind-Powered Ocean Drones created by SailDrone (Redefining Ocean Data Collection na).
  • 37.
    Data science appliedto sustainability analysis 20 A theory of change starts from a fundamental hypothesis. Formulated by research, a claim is then reviewed by peers, revised and published for a potential practitioner to implement.We then scope out a conservation project, detailing the business case, stake- holders and intended impact. Once implemented, we collect insights and analyze it to understand if our theory is actually working.Based on those insights,we learn and adapt accordingly.As work moves along this spectrum, the actors shift and so do stakeholders. This information funnel leans on the principles of adaptive management. Grounded in systems theory, experimental science and business, adaptive manage- ment is a systemic approach for improving natural resource management by learn- ing from management outcomes (as per the US Department of the Interior https:// www.doi.gov/sites/doi.gov/files/migrated/ppa/upload/Chapter1.pdf). Conservation programs often lean on this approach to drive strategic decision-making via a cyclical “always learning” mindset (Fig. 2.5). However, conservation is a team sport - one where scientists, conservationists, field workers, management professionals, engineers, analysts, policy makers, business leaders and citizens must play together to drive value for our planet.As we look at the broader scope of information needs, knowledge tasks and gaps in day-to-day workflows of these various stakeholders through the lens of the Adaptive Management approach, we start exposing new AI-ready potential. Fig. 2.5 The Adaptive Management Cycle (West, na).
  • 38.
    Applying AI toconservation challenges 21 Each stage of the plan-do-evaluate-learn-adjust cycle sparks new contexts to explore AI solutions for and the specific stakeholder relationships to solve for - how scientists and conservationists engage with management leaders, how can policy makers learn from field surveyors, how can we allocate enterprise value, private capital and citizen engagement to meet our sustainability goals. The following sections dive deeper into each of the stages of the cycle. Creating value with AI As we traverse the various stages of the Adaptive Management Cycle, we can expose AI-ready challenges and opportunities by looking for avenues to accelerate, augment, assist or automate sub tasks. Planning and implementing with AI Planning conservation projects often begins with science and outcome measures.A key task in this stage involves researching and collecting raw information from surveys, local reports,census data,stakeholder data,geospatial imagery and other unstructured content representing ground truth. Some potential areas for AI to aid are: • Natural language tools: extracting structured content from unstructured data sources can catalyze how scientists and policy makers can model and plan conserva- tion interventions by leaning on technologies like OCR (optical character recogni- tion), translation and natural language processing (to extract keywords and conser- vation-centric named entities). AI enthusiasts have also explored sentiment analysis of Twitter content (Fig. 2.6) to capture near real-time insights on how people and nature intersect (floods in Chennai India). Fig. 2.6 Sample social media content (Blog, 2017).
  • 39.
    Data science appliedto sustainability analysis 22 • Search and knowledge technologies: organizing and sharing information with science and management peers relies on contextually relevant knowledge - an area of search that has gained increased AI-based enrichment by services like Microsoft’s Azure Cognitive Search and Allen Institute’s Semantic Scholar (a free, AI-powered research tool for scientific literature, semanticscholar.org (AI-Powered Research Tool. Semantic Scholar na)).A growing area of opportunity is around analysis and extraction of structured data from multilingual content (e.g. handwritten survey and historical research reports in local languages). A majority of data collection in the field,for instance in regions like Africa and South Asia,has historically been done us- ing paper-based surveys and images. Scientists spend hours gleaning scanned images for insights they can use in their research and analysis.AI technologies like Optical Character Recognition (Matei et al., 2013) can help automate and speed up this digitization task, freeing up precious time for science teams. In addition, structured content from these handwritten corpora of knowledge can be used for more in- depth analysis (such as similarity analysis, entity and conceptual relationships). • Digital engagement: enterprises, government initiatives and citizens are increas- ingly accessing information and alerts via a wide spectrum digital channels - from social networks to smart assistants in their living rooms.AI-powered innovations such as Amazon Alexa Skills and notifications on smart watches enable personalized and engaging ways to connect real world data (e.g. weather forecasts, local water quality) to individual contexts (e.g. impact of drought to their businesses, water footprint of their purchases and behaviors).The City ofVirginia Beach has one such Alexa Skill (water sense na) that provides information on water levels and road closures for citizen users. • Data governance needs:downstream questions around data quality,availability and governance allow for experts in AI and data science to help guide what outcome measures are incorporated in the planning and implementation stages. Monitoring evaluating with AI Monitoring evaluating conservation projects entails engaging with field staff, remote technologies,analysis tools on the edge and a wide spectrum of monitoring methodolo- gies. Some potential areas for AI to aid are: • Edge assistance for field work:iNaturalist is a great example of how field data collec- tion can be combined with the image processing AI to give instant insights (iNaturalist na).Innovations in machine learning deployment,such asTensorflow Lite (Dokic et al., 2020) allow practitioners to expose prediction tools (e.g. to assess soil conditions from images) on observation data from mobile and embedded devices. SailDrone, a wind- powered unmanned sailing device (saildrone.com/technology), leverages edge analysis of solar-powered meteorological and oceanographic sensors and on-device cameras to autonomously navigate harsh ocean conditions and report data.
  • 40.
    Applying AI toconservation challenges 23 • Anomaly detection of time series data: in projects where monitoring of condi- tion is performed using Internet of Things (IoT) devices and sensors, there is op- portunity to tap into cognitive technologies to monitor data quality,sensor status and detecting patterns on the fly.Data science approaches used in other industries,such as fraud detection (banking transactions),fault monitoring (utilities and manufacturing) and intelligent alerts (healthcare) may transfer to conservation projects with stream- ing or time series data (e.g. monitoring soil nutrient levels, water quality, pH, etc.). • Developing new remote sensing datasets: access to high quality labeled data- sets is key for supervised learning. Data science tools allow us to generate broader geospatial representations from highly specific conservation projects with on-the- ground sensors. For instance, a biometrician at The Nature Conservancy developed novel AI-based approaches to analyze sound recordings from bioacoustics sensors in a forest and produced a model that represented characteristics of the ecological soundscape across measured forest conditions (Geospatial Conservation Atlas na). • Data science for evaluation: in cases where remote sensing imagery and monitoring datasets are readily available, data science can help assess condition and generate in- sights.The U.S.Geological Survey’s Land Cover Monitoring,Assessment,and Project dataset (Land Change Monitoring,Assessment, and Projection na) is one such can- didate dataset that may be leveraged for land change monitoring,land parcel tracking and building monitoring and alert systems by applying machine learning models and image analysis algorithms. Learning adapting with AI The final two stages of Adaptive Management Cycle demand synthesis of the conser- vation project’s decision-making process (from planning and implementation) and the evidence of its effects (from monitoring and evaluating) for two primary reasons. First, to report and share our approach and outcomes with stakeholders that can help scale impact to other geographies and contexts. Second, to feed actionable insights back into the next iteration of planning, (re)prioritization and adjustments of strategies. Cognitive technologies give us tools to understand, represent and model for these tasks in a variety of ways: • Unraveling connections from impact: knowledge representation is a key step in capturing the nuances of an AI problem’s domain. Representing decision-models, field knowledge, evidence and evaluation of outcomes in a knowledge graph al- low advanced data analyses, such as clustering and similarity detection. Social net- works and digital media platforms (like Netflix) use knowledge graphs to understand patterns of human behavior with the outcome to drive action (post something or consume a clip). In conservation projects, a knowledge graph can help us identify similarity in evidence sets and evaluation methodologies between different projects to assess, audit and adjust the underlying decision-making process.
  • 41.
    Data science appliedto sustainability analysis 24 • Selection and prioritization: another application of conservation-related knowl- edge graphs is for leveraging our learning’s, constraints (costs, labor and time) and contextual knowledge (like policy and other enabling factors) to select and prioritize work. Fig. 2.7 highlights an example scenario. Imagine you have limited funds to deploy for a conservation project - one in India and another in Argentina.What if you didn’t have to pick one project? What if you could pick a component, or two, from each project that advances the overall learning and insights on work in India and Argentina? Breaking down the knowledge about each region into enabling fac- tors,tools,policy support and other attributes allows us to compare common themes, focal areas and subsequently draw insights from graph-based analyses, such as com- monalities, linkages and cluster detection. • Creating digital twins:according to an article published on Nature.com (Tao and Qi, 2019), there is an increasing number of scenarios where a “digital twin” of a complex system has been used to help detect risks, optimize choices and model outcomes. For instance, NASA uses a digital copy of its spacecraft to monitor status (Glaessgen and Stargel,2012).Singapore uses a digital twin to monitor and improve utilities (Tomorrow. Mag.2019).What if we were to create a digital twin of a conservation project? The impact technology canvas: synchronizing AI initiatives In the entrepreneurial community, we have seen two key mindsets that have cata- lyzed powerful products - a ‘build-learn-iterate’ mindset (e.g. the Lean Canvas (1-Page Fig. 2.7 Using a Knowledge Graph to represent and analyze relationships between key project com- ponents.
  • 42.
    Applying AI toconservation challenges 25 Business Plan | LEANSTACK na)) and the user-centric mindset (e.g.Jobs-to-be-done by Clayton Christensen (JobsTo Be Done - Christensen Institute na)).In conservation, Adaptive Management offers a similar “continuous learning” mindset and a common lens for AI practitioners, researchers and innovators to understand and communicate their work. As we see contributions from all over the globe, we recommend a tool to capture, organize, share and collaborate on their AI-enabled efforts - the Impact Technology Canvas (Fig. 2.8). The Impact Technology Canvas systematically breaks down climate and sustainabil- ity work into utility (AI “jobs to be done”) and leverage (people, assets, resources, risks and capital) - regardless of the scale/scope of the solution (moonshot or roofshot). It enriches Adaptive Management’s continuous learning cycle with shared context: target personas, stakeholders, assumptions, partners, datasets, tools and other enabling contex- tual parameters (like location, outcome measures, focus areas, socio-economic factors). The canvas gives us a foundational structure and framing to capture problems, people, science, local knowhow, constraints and solutions behind specific climate or sustainability challenges.As the community of AI practitioners maps out the various AI initiatives using the canvas, we can more efficiently and effectively (i) identify reusable Fig. 2.8 The Impact Technology Canvas.
  • 43.
    Data science appliedto sustainability analysis 26 patterns and learning in technology, data and methods (sensor AI, machine learning models and economies of scope), and (ii) optimally analyze upstream/downstream risks gaps (and source new ideas, low-hanging fruit, research gaps and challenges), as illustrated in Fig. 2.9. Summary Our planet is a unique social network - the natural resources on which all life depends in turn are dependent on life itself. As we look to science, data, AI and technology to effectively mitigate climate change and drive sustainability,we can’t ignore the intercon- nectedness of this social network.Embedded in this interconnectedness is a key nuance - a constantly evolving, complex peoplenature relationship, which takes datasets beyond ecological domains and into human and societal factors.Take, for instance, the factors (human, ecological and socio-economic) you’d have to consider for a successful climate change intervention (say, a solar farm) in rural Northern India.These factors might be radically different for rural California! An electric vehicle might be a better fit for the same challenge. The interconnectedness of this complex peoplenature system and the noise (the uncertainties complexities of our actions and their effects) is a ripe environment for AI-shaped problem sets and innovation. Fig. 2.9 Unraveling scalability and transferability in the AI community using the canvas.
  • 44.
    Applying AI toconservation challenges 27 With increased computing capacity (cloud infrastructure), advanced data manage- ment (sensor networks and edge technologies) and our ability to collaborate without borders,AI practitioners can catalyze how fast we understand, learn, apply and traverse the dynamic people and nature relationship. References 1-Page Business Plan | LEANSTACK. na https://leanstack.com/leancanvas/. AI-Powered Research Tool. Semantic Scholar. na http://semanticscholar.org/. Blog, G., 2017.Tapping Twitter Sentiments: a Complete Case-Study on 2015 Chennai Floods. Analytics Vidhya https://www.analyticsvidhya.com/blog/2016/07/capstone-project/ Published May 29. Dokic, K., Martinovic, M., Mandusic, D., 2020. Inference speed and quantisation of neural networks with TensorFlow Lite for Microcontrollers framework, 2020 5th South-East Europe Design Automation, Computer Engineering, Computer Networks and Social Media Conference. SEEDA-CECNSM. doi:10.1109/seeda-cecnsm49515.2020.9221846. Glaessgen, E., Stargel, D., 2012. The Digital Twin Paradigm for Future NASA and U.S. Air Force Vehicles, 53rd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics and Materials ConferenceBR20th AIAA/ASME/AHS Adaptive Structures ConferenceBR14th. AIAA. doi:10.2514/6.2012-1818. iNaturalist. na https://iNaturalist.org/. Jobs To Be Done - Christensen Institute. na https://www.christenseninstitute.org/jobs-to-be-done/. Land Change Monitoring, Assessment, and Projection. na https://www.usgs.gov/core-science-systems/ eros/lcmap. Mapping Ocean Wealth. na http://oceanwealth.org/. Matei, O., Pop, P.C.,Vălean, H., 2013. Optical character recognition in real environments using neural net- works and k-nearest neighbor.Applied Intelligence 39 (4), 739–748. doi:10.1007/s10489-013-0456-2. Open Space. Coastal resilience. na https://coastalresilience.org/project/open-space/. “Redefining Ocean Data Collection.” Saildrone, saildrone.com/. na. Rose, R.A., Byler, D., Eastman, J.R., et al., 2014.Ten ways remote sensing can contribute to conservation. Conserv. Biol. 29 (2), 350–359. doi:10.1111/cobi.12397. Singapore experiments with its digital twin to improve city life Tomorrow.Mag. https://www.smartcitylab. com/blog/digital-transformation/singapore-experiments-with-its-digital-twin-to-improve-city-life/. Published May 30. South Payette River Case Study. na https://hydroforecast.com/case-study-south-payette/. Tao, F., Qi, Q., 2019. Make more digital twins. Nature News. https://www.nature.com/articles/d41586- 019-02849-1. Published September 25. The science of sound: acoustic soundscapes of mature forests in the temperate northern triangle of Myanmar. Geospatial Conservation Atlas. na https://geospatial.tnc.org/datasets/4899be47d4d34068 ac53035ac32bc7b3. “water sense”.naAlexa Skill.https://www.amazon.com/City-of-Virginia-Beach-storm/dp/B078K9F953. West, S. Meaning and action in sustainability science: interpretive approaches for social-ecological systems research. na 10.13140/RG.2.2.32127.10406.
  • 46.
    29 Data Science Appliedto Sustainability Analysis © 2021 Elsevier Inc. DOI: 10.1016/C2018-0-02415-9 All rights reserved. CHAPTER 3 Water balance estimation in Australia through the merging of satellite observations with models Chapter Outlines Introduction 29 Importance of water balance estimation in Australia 29 Problems with current methods for water balance estimation 30 Case studies 32 Inversion of rainfall rates from streamflow records 32 Estimation of flood forecast-effective river bathymetry through inverse modeling 34 Estimation of terrestrial water storage through the merging of Gravity Recovery and Climate Experiment (GRA CE) data with a hydrologic model 37 Conclusion 38 Introduction Importance of water balance estimation in Australia Due to its geographical properties,Australia faces a number of unique challenges related to its water management. It is the driest of all inhabited continents, while at the same time it is prone to devastating floods. According to the Queensland government, the average annual cost of floods is $377 Million (measured in 2008 Australian dollars). Especially disastrous were the 2010–11 floods, which costed the Australian economy an estimated $30 Billion. An estimate from Geoscience Australia resulted in a cost of $6.64 Billion (2013 Australian dollars) for the floods in the LockyerValley and Brisbane in January 2011 alone.This calculation included deaths and injuries but excluded most indirect losses. 35 deaths were confirmed, and 20,000 people were made homeless.The most important and cost-effective tools for the limitation and mitigation of the impact of floods are improved flood warning systems and community awareness, which are topics that are currently studied extensively in Australia. At the same time, droughts continue to damage Australia’s economy. Between 2006 and 2009 the drought reduced national GDP by roughly 0.75 percent. Between 1997 and 2002, the contribution of agriculture to the Australian economy averaged 2.9 per- cent. Between 2003 and 2009, the years of the millennium drought, this percentage Valentijn R.N. Pauwels, Ashley Wright, Ashkan Shokri, Stefania Grimaldi Monash University, Department of Civil Engineering, Clayton,VIC,Australia
  • 47.
    Data science appliedto sustainability analysis 30 reduced to 2.4 percent. During the two worst years of the drought, 2003 and 2007, this contribution further reduced to 2.1 percent [ABS, 2011, van Dijk et al., 2013]. Further, climate change is likely to worsen the impact of droughts in the South West and South East of Australia [Climate Council of Australia, 2018a]. It is already evident that droughts cause difficulties in the water security of Australia [Climate Council of Australia, 2018b]. For this reason, groundwater is an important contributor to the Australian economy.According to Deloitte Access Economics [2013], the economic contribution of groundwater use to the Australian Gross Domestic Product (GDP) is between $3.0 and $11.1 Billion,with households using approximately $410 Million worth of groundwater each year. These issues clarify the need for a reliable estimation of the water resources of Australia.A number of methods are currently available for this purpose, but each suffers from a number of issues.The remainder of this section provides an overview of these methods and their drawbacks, after which a number of studies are described that focus on these issues. Problems with current methods for water balance estimation Soil moisture is a key variable in the hydrologic cycle, as it determines the partitioning of the available energy into latent, sensible, and ground heat fluxes, and the net pre- cipitation into infiltration and surface runoff. Consequently, numerous satellite missions have been devoted to estimating this variable at large spatial scales through remote sens- ing. Relatively recent examples are the Sentinel-1 [Attema et al., 2007], the Advanced Scatterometer (ASCAT) [Figa-Saldana et al., 2002], the Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E) [Njoku et al., 2003], the Soil Moisture Ocean Salinity (SMOS) [Kerr et al., 2010], the Soil Moisture Active Passive (SMAP) [Entekhabi et al.,2010],and the Global Change Observation Mission (GCOM) [Imaoka et al., 2010] missions.Throughout the last decades, one major improvement in satellite remote sensing-derived soil moisture products is their temporal resolution.This has improved from 35 days, for the Synthetic Aperture Radar on European Remote Sensing satellites (SAR/ERS), to 3 days, for the ASCAT AMSR/E, SMOS, and SMAP missions [Li et al., 2018]. However, remaining issues with large-scale remote sensing of soil moisture contents are the potential to only provide soil moisture estimates of the upper centimeters of the soil, difficulties in the validation of the products, and errors and uncertainties in the inversion of the raw signal into soil moisture contents [Peng and Loew, 2017]. The only mission that attempts to estimate the water storage of the entire soil profile is the Gravity Recovery and Climate Experiment (GRACE), which provides infor- mation on the change of the terrestrial water storage (TWS) over time. Even though this mission has been proven extremely useful for a number of different objectives, the coarse temporal (monthly) and spatial (∼150,000 km2) resolutions, and problems with
  • 48.
    Water balance estimationin Australia through the merging of satellite observations with models 31 the conversion of the storage changes into volumes continue to complicate the use of the GRACE TWS retrievals [Girotto et al., 2016]. Remote sensing can also provide information on inundation extent and level. As water surfaces tend to be very smooth,a radar signal reaching open water will be reflect- ed away from the satellite sensor and a low backscatter will be observed. Conversely, rough dry areas will return part of the energy to the sensor resulting in higher back- scatter [Henderson and Lewis, 2008]. This is the principle of radar-based flood map- ping [Chini et al., 2017; Shen et al., 2019]. However, the interpretation of backscatter signal can be very difficult in complex environments, such as urban areas and areas with emerging vegetation [Pierdicca et al., 2018]. For this reason, increasing research efforts are being dedicated to the definition of accurate algorithms for the detection of floods using radar data.An overlay of the flooded area onto a Digital Elevation Model (DEM) can then provide an estimate of the flood water level [Hostache et al., 2009]. Both remote sensing-derived inundation extent and level can be used to evaluate the predic- tive skill of hydraulic flood forecasting models, and the selection of the most effective dataset is being debated in the literature. [Shen et al., 2019]. A common method to estimate the surface water balance at large spatial scales is through land surface modeling. These models simulate all processes related to the partitioning of precipitation into streamflow, evapotranspiration and change of stor- age, and aim to provide an answer to the fundamental question “What happens to the rain” [Penman, 1961]. Despite strong advances in computational power, the quality of the meteorological forcing and input data, model parameter estimation algorithms, and the representation of the physical processes, errors and uncertainties in each of these fields continue to lead to erroneous model predictions [Boughton et al., 2016]. Errors in the model prediction are propagated at each time step, which can severely affect the accuracy of the final model water balance estimates [Ellett et al., 2006].With respect to accurate flood modeling at large spatial scales, the most important obstacles are the lack of an accurate three-dimensional representation of the morphology of the floodplain and of the rivers [Garcia-Pintado et al., 2015;Wood et al., 2016]. Increasing efforts have been dedicated to the delivery of corrected satellite-derived DEMs for the implementa- tion of hydraulic models [Yamazaki et al., 2017, 2019]. However, DEMs usually provide information on floodplain morphology only; conversely, information on the river chan- nel depth and shape (i.e.river bathymetry) is rarely or not available for many catchments in the world [Domeneghetti, 2016].This problem can be overcome by measuring the river bathymetry, or depth, in-situ, but this is not practical even at the catchment scale. Furthermore, river bathymetry can be prone to significant changes over short time scales [Soar et al., 2017]. This chapter describes three studies tackling the above problems using large data sets. The methodologies were developed and tested using Australian catchments as test sites. The left hand side of Fig. 3.1 shows the location of the test sites.The right hand side of
  • 49.
    Data science appliedto sustainability analysis 32 Fig. 3.1 shows and overview of the required data, methods, and modeling techniques of these studies. Case studies Inversion of rainfall rates from streamflow records It is widely known that the estimation of spatially averaged rainfall rates is extremely challenging. An especially common method to achieve this goal is the interpolation of point scale rainfall observations, which suffers from the lack of spatial coverage of the stations.Weather radars and satellite observations can also be used for this purpose, but these suffer from uncertainties in the inversion algorithms and inconsistencies between the rainfall generated at the cloud level and the amounts reaching the ground [El Kenawy et al., 2019]. Inversion algorithms are used in environmental modeling to estimate independent variables, in non-linear systems, such as rainfall rates and model parameters from the measured dependent variables such as reflectivity and streamflow.A more recent approach to estimate spatially average rainfall rates makes use of the mobile phone network [Overeem et al., 2013], by relating the attenuation of the electromag- netic signal from the transmitting to the receiving antenna to the rain rate.However,the required commercial data are frequently quite difficult to acquire. Based on these issues and on the fact that discharge observations represent the catchment-integrated response to rainfall events, a different approach is to invert observed streamflow data into rainfall time series [Kirchner et al., 2009]. Because of the non-linear response of streamflow to rainfall advanced inverse modeling techniques may be required for this purpose. A methodology to estimate rainfall rates from streamflow records using inverse modeling techniques was developed by Wright et al. [2017b].To do so it is necessary to parameterize the current best estimate of areal rainfall which is derived from imperfect observations. The algorithm started by noting that most rainfall values at the hourly Fig. 3.1 Left hand side: Overview of the test sites for this chapter. Right hand side: Data, methods, and modelling techniques used in the three studies.
  • 50.
    Other documents randomlyhave different content
  • 51.
    devout demeanour atthe stake, 161, 164, 170, 175, 199; when in the flames begged the Cross to be held before her, 161, 175, 195; Jesus her dying utterance, 161, 176, 273, 301, 305; pity excited by her execution, 191, 192, 255; contrition of her executioner, 161, 163, 194; exact place of execution, 170, 175; death desired by the English, 186; her ashes cast into the Seine, 193, 207, 301, 302, 305; her appearance in June, 1429, 30; no authentic portrait known, 49; her abstemious diet, 237, 243, 296; prison diet, 15, 16; pious and simple life, xiii; physical hardihood, xiii; her presence controlled vice and raised tone of French army, xii- xiii, 243, 245, 249, 250, 251, 264, 268, 270; hospitable to poor, 221, 224, 272; problem as to her knowledge of logic and theology, xix; testimony to virtue and courage, xxvi, 319; eloquent and forensic, yet prudent and simple in answers, xxvii, 177, 179; Charles VII. orders posthumous enquiry nearly twenty years later, v, xxi, 371; abortive, 372; enquiry ordered by Pope Nicholas V., xxii, 372; no definite result, 373; Pope Calixtus, on petition of Jeanne’s mother, Isabella, causes solemn enquiry at Paris, xxii, 373–376; sworn information of events in the last days of Jeanne’s life, 147– 8, 150; official Latin text of trial and rehabilitation, xxv; sentence of rehabilitation xxiii, 321–328, 376 Jeanne d’Arc family, see d’Arc Jhesus Maria on banner, 31, 91, 361; on letters, 35, 36, 349, 350, 352, 369 Josephine, Empress, 249
  • 52.
    Joyart, Mengette, 222 Jumièges,Abbot of, 127 La Basque, standard-bearer, 316, 317 La Charité sur Loire, 53, 73, 317, 352, 361 Lacloppe, Bertrand, 218 Ladies’ Tree, see Tree Ladvenu, Br. Martin, 148, 150, 168, 170, 175, 191, 193–5, 328, 338, 372 Lagny, 29, 52, 78 La Hire, Maréchal, 233, 235, 250, 263, 264, 277, 279, 293, 308, 311, 312, 314 La Macée, Lady, 305 Lambert or Lombart, Jean, 306 Lancaster, House of, xvii Lapse, The, 121–134, 326 Lapau, Mme., 260 La Rose, Philippe, 373 La Rousse, woman, 9, 217, 219, 344 La Saussaye in diocese of Evreux, 209 Laval, Guy and André de, 30 Laverdy, M., 331 Laxart, Durand, 218, 225, 230 Laxart, Jean de, 6, 215 Lebouchier, Maître Guillaume, 100, 106 Lebouchier, Messire Pierre, 198 Lebuin, Michael, 225 Lecamus, Canon Jacques, 151
  • 53.
    Lecomte, Denis, 332,374 Ledoux, Maître Jean, 100, 118 Le Drapier, Perrin, 218 Lefevre, Maître Jean, Bishop of Démétriade, 101, 205, 210, 338 Lefumeux, Messire Jean, 231 Leguise, Bishop Jean, 240 Le Marie, Guillaume, 244, 306 Lemaître, Maître, Jean, Vice-Inquisitor of Beauvais, 8, 66, 67, 94, 95, 98, 99, 105, 129, 132, 133, 140, 141, 142, 168, 182, 193, 194, 322 Lenozolles, Maître Jean de, 290, 340 Leparmentier, Maugier, 300 Le Renard, see Thérouanne Leroyer, Catharine, 21, 227 Leroyer, Henry, 223, 227, 228 Le Royer, Thévenin, 218 L’Esbahy, Jacques, 248 Letters to English, 36–8, 235, 246–7, 286–7 Letter of Duke of Burgundy, 335 Letter of Count d’Armagnac, 34–5, 351; Jeanne’s reply, 35 Ligny, Count de, 178, 294, see Luxembourg, Jean Limoges, ix Lingué, Jean, 6 Lisle, 177 Loches, 268, 270 Lohier, Maître Jean, xix, xx, 166–7, 254, 257, 339 Loire, camps on the, 237
  • 54.
    Longueville, Lord Priorof, 100 Loré, Sieur Ambroise de, 275, 279 Lorraine, its traditions, viii, ix, 9 Lorraine, Charles, Duke of, 11, 30, 214, 218, 226, 272, 230 Lorraine, Réné of Anjou, Duke of, 11 Louis XI., vii, 245, 275, 289 Louviers, siege proposed, 157, 158 Loyseleur, Nicolas, 56, 117, 118, 130, 134, 148, 152, 165, 166, 167, 169, 179, 182, 183, 202, 258, 298–9, 340, 341 Lude, Sieur de, 278 Luxembourg, Jean de, 58, 334, 335, 336; his wife, Jeanne de Bethune, 46; see also Ligny, Count de Luxembourg, Count Waleran de, 46, 334 Luxembourg, Messire Louis de, Archbp. of Rouen, 163, 294; see Thérouanne Machet, Gerard, Bishop of Castres, 238 Maçon, Maître Jean, 248 Maçon, Robert le, 238 Mailly, Jean de, Bishop of Noyon, 255–6 Manchon, Guillaume, 56, 68, 77, 95, 126, 136, 146, 165, 172, 178–187, 188, 197, 212, 298, 331, 340, 372, 374 Mandrakes, 42 Manuel, Pierre, 304 Margaret of Anjou, 11, 275 Margaret of Bavaria, 272 Margaret of Scotland, 223
  • 55.
    Marguerie, André, 101,117, 118, 121, 192, 202, 208–9 Marie, Messire Thomas, 211 Marigny, 335 Marriage, action against Jeanne, 62, 64, 344 Martel, Charles, 27 Martin V., Pope, 34 Mary of Anjou, Queen, 46 Massieu, Maître Jean, 56, 68, 85, 117, 135, 141, 170, 171–176, 198, 339, 372 Maugier, Pierre, 373 Maurice, Maître Pierre, 56, 89, 95, 96, 101, 103, 121, 122, 148, 149, 166, 171, 180, 209, 302 Maxey-sur-Vays, 19, 225 Meaux, Bishop of, see Versailles Meaux, Viscountess de, 46 Mehun, 237 Melun, 57, 73 Melville, Lord, xi Merlin, prophecy, 21, 188, 241 Message, Mathieu, 244 Metz, Jean de Novelemport, called Jean de, x, 12, 136, 218, 223–5, 226, 228, 230, 265, 291, 301 Meung, 249, 263 Meung-sur-Loire, 237, 278, 289 Meung-sur-Yèvre, 245, 317 Midi, Maître Nicolas, 56, 61, 64, 67, 69, 74, 77, 79, 85, 89, 94, 95, 96, 100, 103,
  • 56.
    106, 109, 119,121, 134, 142, 166, 171, 176, 177, 207, 255, 258, 295, 300 Milan, Duke of, 6 Milet, Colette, 295 Milet, Pierre, 295, 296 Minet, Messire Jean, 6 Minier, Pierre, 209 Moen, Jean, 216 Monnet, Maître Jean, 258–9, 340 Montargis, battle, 232, 242 Moreau, Jean, 303 Morel, Maître Aubert, 101, 117, 118, 119 Morel, Jean, 6, 215 Morin, Maître Jourdin, 274, 282 Mortemer, Abbot of, 127 Mugot, see Contes, L. de Musée de Trocadéro, Paris, 49 Musnier, Simonin, 221 Muton, Guillaume, 103 Naples, vii Napoleon I., v Neufchâteau, 9, 10, 212, 214, 216, 218, 220, 344 Newman, Cardinal, xxii Nibat, Jean de, 100 Nicholas V., Pope, xxi, 372 Normandy, 371
  • 57.
    Norwich, Bishop of,127 Novelomport, Jean de, 12; see Metz Noyon, Bishop and Diocese of, 4, 121, 127, 142 Oath, administration of, 5, 6, 7 Olivier, Alain, 328 Orient, Pierre, 97 Orleans family, xv Orleans, Charles, Duke of, 72, 280 Orleans, Duke of, 12, 65, 307, 353 Orleans held by patriots, ix; its siege, vii, 31, 32, 35; relief ordered, x; siege raised by Jeanne d’Arc, 233–237, 242, 245, 246–7, 249, 260–263, 266–270, 275–6, 284–289, 292, 293, 296, 297, 307, 300–317 Ourches, Albert d’, 228 Paris, vii, xv, 181, 352; its assault, 14, 28, 73, 78, 353; Church of Notre Dame, 373; National Library, 331, 332; Parliament of, 336; University of, xxi, 119, 120, 122, 138, 177, 336 Paris, Guillaume Chartier, Bishop of, 321, 373 Partada, Alphonse de, 314 Pasquerel, Brother Jean, 32, 281, 282 Patay, Battle of, xii, 266, 280, 289, 293 Peter of Pomfret, ix Petit, Gerard, 229
  • 58.
    Philip II, 273 Philipthe Fair, xxii Picard ravages, 9 Pigache, Maître Jean, 209 Pinchon, Jean, 99 Poitiers, vii, ix, xi, xviii, xix, 24, 136, 265, 305; book at, 25; Church of, 116; Clergy of, 201, 244 Pole, William de la, Earl of Suffolk, 36, 241, 248, 263, 265, 277, 278 Pollichon, see Poulengey Pont l’Evêque, 73, 361 “Pontiffs, Three,” 34 Pope and Empire, xvi Pope Calixtus, v, xxii, 178, 373 Pope Nicholas V., 372, 373 Pope of Rome, 33, 36, 91, 114, 116, 128, 131, 145, 159, 160, 189, 210 Poulengey, Bertrand de, 12, 136, 218, 224, 226, 228–231, 265; see Pollichon, B. Poulnois, Hauves, 283 Pouthon, the Burgundian, 335 Preaux, Abbot of, 127 Pressy, Sieur Jean de, 47 Preston Manor, Sussex, 42 Prévosteau, Guillaume, 374 Process or transcript of trial, 180, 181, 187, 188, 196, 197, 200, 210, 325, 326, 327 Prophecy—France lost by a woman, saved by a maid, 226, 227
  • 59.
    Prophecy of d’Avignon,269 Queen of Charles VII, 270, 271 Quesnay, Maurice de, 106 Quicherat, xxv, 242 Rabateau, Jean, 24, 243, 265, 269, 306 Raguier, Hemon, Treasurer, 215, 275, 283 Raiguesson, Jean, 6, 215 Rais, Gilles de Laval, Seigneur de, 233 Raymond, 260 Regnault de Chartres, xiv; see Rheims Rehabilitation Sentence, 321–328, 371 Relapse, 135–146, 326 Reynel, Maître Jean de, 257 Rheims xii, 215, 237, 239, 244, 245, 264, 292, 304, 361; Jeanne’s house, 51 Rheims Cathedral xii, xiv, 51 Rheims, Jean Jouvenal des Ursins, Archbishop of, 321, 373 Rheims Reginald, Archbp. of 324 Rheims, Regnault de Chartres, Archbp. of, xi, xiv, xix, 24, 115, 118, 201, 233, 240, 305 Ricarville, Guillaume de, 245 Richard II, ix Richard, the Archer, 224, 226, 228, 230 Richard, Brother, 42, 48, 53 Richelieu, 289
  • 60.
    Ricquier, Jean, 301 Rose,Philippe de, 178 Roncessey-sous-Neufchâteau, 216 Rouel, Jean de, 257 Rouen, v, xix, 119, 372; Castle, 3, 110, 117, 119, 121, 290; the Ornament Room, 8, 340, 351; Archiepiscopal Chapel, 120, 138, 139, 328; Trial of Jeanne d’Arc, 3, 99, 121, 132, 181, 252; Jeanne’s Imprisonment, 96, 106, 192, 195, 199, 205, 255, 294, 299, 303, 305; Cemetery of Saint Ouen, 127, 132; Market Place, 170, 327 Roussel, Guillaume, 328 Roussel, Jean, 328 Roussel, Raoul, 118 Rouvray, Battle, 11 Royal Letters of Jeanne’s Surrender, 3 Saint Bernard, 119 St. Catherine, x, 23, 24, 26, 39, 40, 47, 60, 62, 65, 69, 71, 74, 75, 77, 79, 84, 88, 90, 92, 109, 115, 131, 137, 345, 352, 355, 357, 358, 363, 366, 368, 370 St. Catherine de Fierbois, village, 12, 27, 28, 89, 349 St. Charles, 234, 235 St. Denis, 13, 14, 29, 62, 88, 349, 353, 354, 361; war cry, 89 St. Eusebius, Cardinal of England, 127, 161, 187, 209 St. Gabriel, 118, 357, 366, 371 St. Lo of Rouen, Prior of, 127 St. Louis, 234, 235
  • 61.
    St. Margaret, x,23, 24, 26, 39, 40, 47, 60, 62, 65, 71, 74, 77, 84, 88, 90, 92, 109, 115, 131, 137, 345, 352, 354, 357, 358, 363, 366, 368, 370 Saint Mesmin, Amian de, 248 St. Michael, 25, 39, 42, 44, 45, 64, 84, 85, 199, 255, 344, 355, 357, 358, 366, 368, 370 St. Michel au Peril-de-la-Mer, Abbot of, 127 St. Nicholas du Port, 226, 277–8, 229 St. Ouen of Rouen, Abbot of, 127 St. Ouen, Jeanne preached to at, 172, 187, 191, 255, 259, 295, 376 St. Peter and his Successors, 124, 131 St. Pierre-le-Moustier, 246, 318 St. Quentin, Burgundy’s letter to people, 335 St. Thomas, 160 St. Urbain, town of, 12 Scales, Lord, 37, 248, 278 Scotland, King of, vii Séguin, Pierre, 244 Selles-en-Berry, 30, 271 Senlis, Bishop of, 51, 78 Shakespeare, ix Sicily, Yolande, Queen of, 275, 309 Soissons, 4, 54 Sologne, the, 260, 284 Stafford, Earl of, 183, 294 Suffolk, Earl of, 21; see Pole Surname, girls take mother’s, 95
  • 62.
    Surreau, Laurent, receiver-general,257, 328 Sword obtained at St. Catherine de Fierbois, 28, 89, 349, 353 Talbot, John, Earl of Shrewsbury, 37, 234, 235, 248, 266, 279, 280 Talbot, William, 7, 338 Taquel, Nicolas, 68, 146, 185, 195–198, 298, 340 Teresa, viii Thermes, Simon de, 229 Thérouanne, Bishop of (Cardinal de Luxembourg), 121, 127, 142, 208, 334 Thévenin, Jeannette, 6, 213, 215, 216 Thierry, Maître Reginald, 245 Thierry, Wautrin, 375 Thiesselin, Jeannette, 6, 213, 215, 217 Thou, Jacques de, 248 Tilly, Janet de, 232 Torcenay, Jean de, Bailly of Chaumont, 225, 229 Toul, 344 Touraine, Jacques de, 56, 89, 103, 106, 119, 166, 171, 183, 252, 257 Touroulde, Dame Marguerite la, 270–272 Tours, 28, 268, 283, 284 Torture, 117, 119, 126, 339 Toutmouillé, Brother Jean, 148, 150, 157, 372 Tree at Domremy, Ladies’ or Fairies’, 20, 214, 217, 219, 221, 229, 343, 344, 366 Tremouille, Seigneur de la, 60, 71, 78, 115 Tressart, Maître Jean, 192 Trèves, Sieur de, 238
  • 63.
    Trèves, Lady de,282 Trial, see Process Troyes, xii, 48, 49, 292; treaty of, viii, 226 Turquetil, Maître Eustace, 172 Turrelure, Pierre, 306 Vallée, Maître Guillaume, 190 Vatican, 331 Vaucouleurs, x, 9, 214, 225, 226, 227, 230; remains of castle, 11, 12, 65 Vaucouleurs, Alain de, 226 Vaux, Pasquier, de, 61, 64, 66 Venderès, Nicolas de, 99, 117, 118, 121, 135, 139, 147 Vendôme, Count de, 74 Verneuil, battle of, 277 Versailles, Pierre de, Abbot of Talmont, 243, 265, 269, 270, 274, 282 Vienne, Colet de, 12, 218, 224, 226, 228, 230 Villars, Sieur de, 232, 314 Viole, Maître Aignan, 297 Wandonne, Lionel Bastard de, 335 Ward, Jean Dieu-le-, 218 Warwick, Earl of, 106, 162, 164, 169, 174, 179, 183, 186, 189, 190, 197, 200, 212, 232, 254, 294, 299, 335, 338 Washington, George, xiv Waterin, Jean, 220 Well Sunday, 20, 214, 217, 219 William of Worcester, 336
  • 64.
    Woman’s dress, Jeanneand, 21, 46, 47, 95, 114–115, 134, 136, 138, 159, 169, 174, 186, 204, 228, 347, 348, 367, 368 Ysambard de la Pierre, 66, 67, 69, 77, 79, 85, 89, 118, 134, 135, 159– 162, 164, 168, 188, 190, 194, 340
  • 65.
    1. Had therebeen any desire on the part of the French King to rescue Jeanne from captivity, a ‘King’s ransom,’ which was later paid for her by Cauchon, could scarcely have been refused in those days for a prisoner of war, however renowned. Unhappily for the memory of Charles, she was left to the tender mercies of the English without any offer being made for her release, or any attempt at rescue. There existed a bitter feeling of jealousy towards Jeanne in consequence of her great successes in the field. This was notably shown during her attack upon Paris, where she was thwarted in every direction, and all possibility of victory was taken from her by the conduct of the King. Whether or not Flavy, the Governor of Compiègne, who was completely under the control of the King, betrayed Jeanne at Compiègne, by shutting the gates and closing the drawbridge at her approach, will never be known, but suspicion has always pointed to his betrayal of the Maid. Alain Bouchard states that, in the year 1488, he heard from two aged men of Compiègne, who had themselves been present, that a few days before her capture, the Maid was attending Mass in the Church of St. Jacques. After communicating and spending some time in devotion, she turned to the assembled congregation, and, leaning against a pillar, uttered this prediction: “My good friends, my dear little children, I am sold and betrayed. Soon I shall be given up to death. Pray to God for me, for I can no longer serve the King and the Kingdom of France.”—Grandes Annales de Bretagne, also Miroir des Femmes Vertueuses. 2. The House of Lancaster was fervidly orthodox. Persecution of heretics begins with Henry IV. The “Cardinal of England” (Beaufort Bishop of Winchester) was the malleus hereticorum at home and abroad. He spoke against the Hussites at the Council of Basle, and he planned Crusades against both heretics and “Saracens.” 3. The court before which Jeanne was brought to trial at Rouen was not a court of the Holy Office or Inquisition, neither was it,
  • 66.
    as the Englishcourts for the trial of heresy were in Lancastrian times, a statutable court of ecclesiastical jurisdiction on whose decision, certified by the bishop, the sheriff was bound to act. It was a composite tribunal. The Bishop of Beauvais claimed and exercised jurisdiction as Ordinary. But the Deputy Inquisitor was joined with him as co-ordinate judge with officers of his own. The Inquisition arose out of the troubles in Spain and South France, where heresy was to some extent necessarily a kind of treason to the polity of Christian Europe. Men were punished for heretical opinions, but these heretical opinions were in most cases lapses from allegiance at a time of national peril. The later Inquisition has no such excuse. 4. The Great Schism arose out of the Babylonian captivity at Avignon (1306–1376). Popes and anti-Popes contended for 40 years (1378–1418). France was on the side of the Avignon Popes, while the Empire and England supported the Popes in Rome. Philip the Fair, by arrangement with the Pope, changed the Papal chair to Avignon. During the seventy years of the captivity, when the Church was ruled by French Popes, France underwent the disasters of Crecy and Poitiers, and became almost a province of England. 5. It is agreed by all authorities that Jeanne was not captured in the Diocese of Beauvais, which ended at the Bridge of Compiègne. Jeanne was taken north of the Bridge, on the right bank of the river, and either in the Diocese of Noyon or Soissons, which of the two has not been determined. The Bishop’s assertion is distinctly untrue. 6. On January 6th, 1412. “In nocte Epiphiniarum Domini.” (Letter from Boulainvilliers to the Duke of Milan. Quicherat, vol. v., 116.) 7. The Font and Holy water stoup in the old Church at Domremy are said to be those in use in the 15th century.
  • 67.
    8. Jeanne appearsto have had a great many godparents. In the Enquiry made at Domremy in 1455, eight are mentioned, viz.: Jean Morel, Jean Barrey, Jean de Laxart, and Jean Raiguesson, as godfathers; and Jeannette Thévenin, Jeannette Thiesselin, Beatrix Estellin, and Edith Barrey, as godmothers. 9. John Gris, or Grey, a gentleman in the Household of the Duke of Bedford, afterwards knighted. He was appointed chief guardian to the Maid, with two assistants, all members of the King’s Body Guard. They appear to have left her entirely in the hands of the common soldiers five of whom kept constant watch over her. 10. There is no certain date for this event. By some it is placed between the first and second visits to Vaucouleurs, in 1428; by others, earlier, at the time of the Picard ravages of the neighbourhood in the September of 1426. 11. Robert de Baudricourt, Squire, Captain of Vaucouleurs in 1428; afterwards knighted and made Councillor and Chamberlain to the King and Bailly of Chaumont, 1454. 12. Of the ancient château the “Porte de France” alone survives. From this gate Jeanne rode out with her escort to visit the King at Chinon. The crypt of the chapel remains, where Jeanne constantly prayed. 13. This is said to have been on account of the impression produced on him by Jeanne’s prediction, on February 12th: “To-day the gentle Dauphin hath had great hurt near the town of Orleans, and yet greater will he have if you do not soon send me to him.” This “great hurt” proved to be the Battle of Rouvray, in which the French and Scottish troops were defeated by the English under Sir John Fastolf. 14. Charles I., the reigning Duke de Lorraine in 1428, was in very bad health, and, having no son, the succession was a matter of some anxiety. He died in 1431, and was succeeded by his son- in-law, Réné of Anjou, who had married his only daughter,
  • 68.
    Isabella. This Rénéwas a brother of Queen Mary, wife of Charles VII., and father of our own Queen Margaret, married in 1441 to Henry VI. 15. Jean de Novelomport, called de Metz, Bertrand de Poulengey, Colet de Vienne, the King’s Messenger, and three servants. 16. March 22nd, 1428. 17. This letter appears later, p. 36. Jeanne may have forgotten its contents, as both these expressions occur; or the Clerics who acted as her amanuenses may have inserted them without her knowledge. 18. Jeanne was entertained by command of the King in a small room on the first floor of the Tour de Coudray, within the Castle walls. Her room was approached by a staircase outside the tower. The vaulted roof of the room has fallen in and the fireplace is in ruins, but the room could easily be restored. Jeanne stayed here from March 8th to April 20th, 1429. She was two days at Chinon before she obtained access to the King. 19. Charles de Bourbon, Count de Clermont, Governor of the Duchy of the Bourbonnais and the Comté of Auvergne, during the captivity of his father in England. 20. On September 8th, 1429. 21. Up to the end of her life, Jeanne spoke of the Bishop as the person responsible for her trial and death. “Bishop, I die through you,” was her last speech to him, on May 30th, the day of her martyrdom. 22. This, and a subsequent enquiry, on February 27th, as to Jeanne’s habit of fasting, would seem to suggest a desire on the part of the questioner to prove that her visions had a more or less physical cause in a weak bodily state resulting from abstinence. As Jeanne’s usual food consisted of a little bread dipped in wine and water, and as she is reported to have had when at home (not in war) but one meal a day, it need hardly
  • 69.
    be supposed thatshe suffered much from the results of a Lenten Fast. 23. The fifteen days’ respite would coincide with the first Examination held in the Prison, May 10th, the first day on which the Allegory of the Sign was given. 24. Gérardin of Epinal, to whose child Jeanne was godmother, is probably the person alluded to; he gave witness in 1455 that Jeanne had called him “Burgundian.” 25. A small fortress in an island formed by two arms of the Meuse, nearly opposite the village of Domremy. 26. According to local tradition, this tree stood to within the last 50 years, and was struck by lightning; another has been planted in its place. The house, in which Jeanne was born, remained in the possession of the De Lys family till the 16th Century, when it passed into the hands of the Count de Salm, Seigneur of Domremy. In the 18th Century it became the property of Jean Gerardin, whose grandson, Nicolas, gave it up in 1818 to the Department of Vosges; so that it is now preserved as National property. 27. This is probably a survival of the Fontinalia, an old Latin festival. The custom of decorating the wells and springs was kept up in England until the last century, and still exists in a few remote villages. The name ‘Well Sunday’ survives, though the processions of youths and maidens have long passed away. The ‘fontaine aux Groseilliers’ is still in existence. It is an oblong tank of water, with the original spring flowing through it. The great beech tree stood close by. 28. Pierre de Bourlement, Head of the ancient house of Bassigny, and Lord of the Manor of Bourlement. He was the last of his race. 29. Merlin had foretold the coming of a maiden out of an Oak- wood from Lorraine; and a paper containing a prophecy to this
  • 70.
    effect had beensent, at the beginning of Jeanne’s career, to the English Commander, the Earl of Suffolk. There was also an old prophecy (quoted by Jeanne herself to Catharine Leroyer) that France, which had been “lost by a woman, should be saved by a Maid.” The conduct of Isabeau of Bavaria, wife of Charles VI., might certainly be said to have fulfilled the first half of this prophecy; and a tradition in the eastern counties that “deliverance should come from a maid of the Marches of Lorraine” must have directed many hopes to the mission of the Maiden from Domremy, though she herself does not seem to have known of the last prediction until some time later. The Oak-wood covers the hills above Domremy to this day. 30. This is the first identification of the “revelations” with any name; Jeanne had always spoken of her “Voices” or her “Counsel.” 31. This Examination at Poitiers had taken place in the Chapel attached to the Palace of the Counts of Poitou, which still exists and adjoins the ‘Salle des Pas Perdus,’ now the Great Hall of the Palais de Justice. It was conducted under the direction of the Archbishop of Rheims during the months of March and April, 1429, and extended over three weeks. At the conclusion, the assembly sent, as the result of their inquiries, a resolution to the King to the effect that he should follow the Maid’s guidance, and seek for the sign she promised him in the relief of Orleans, as a proof of the Divine origin of her mission, “for,” they added, “to doubt or forsake her without any appearance of evil would be to vex the Holy Spirit, and to make himself unworthy of the help of God: so saith Gamaliel in the Council of the Jews with regard to the Apostles.” Unfortunately, no trace of this Examination has been found: the ‘Book of Poitiers’ is referred to several times in the Trial; but it was not forthcoming at the time of the Rehabilitation. It was probably lost or destroyed by Jeanne’s enemies among her own party. The Archbishop of Rheims would have had it in his charge: and he was consistently opposed to Jeanne throughout.
  • 71.
    During her stayat Poitiers the Maid lodged in the house of Jean Rabatier. 32. According to local tradition, this Church was originally founded by Charles Martel in 732, after his victory over the Saracens, whom he here ceased to pursue, and deposited his sword as an offering. This is by some supposed to have been the sword which later Jeanne sent for; but the legend is not of an early date, and there is no suggestion of the kind in contemporary writings. According to one authority, the Greffier de la Rochelle, the sword was found in a reliquary, which had not been opened for twenty years or more. The Chronique de la Pucelle and the Journal of the Siege of Orleans state that it was one of many votive offerings, and was recognized by Jeanne’s description of the five crosses on the blade, possibly a Jerusalem Cross. Some of the old Chronicles say that Jeanne told the King she had never been at Fierbois: but this statement is disproved by her own words in this answer. The suggestion that, having been to three Masses in the Church, she might easily have seen the sword, is to some extent answered by the alleged difficulty of the Priests to find, among the many swords there, the one she had specially described. Of the ultimate fate of this sword there are many versions, and no two agree exactly as to date. It was certainly broken in striking a camp-follower, one of a class the Maid had forbidden to enter the Camp; but whether this was just after the retreat from Paris or earlier, it does not seem possible to decide. Jeanne herself says she “had it up to Saint-Denis” and “Lagny,” both of which dates would imply the autumn of 1429: but most witnesses tell the story of its being broken in the July preceding, though several different places are mentioned as the scene of the incident. 33. On September 13th, 1429. 34. A small town near Auxerre. In this neighbourhood some of the chronicles place the incident referred to of the breaking of the
  • 72.
    sword. The questionmay, therefore, have been intended to elicit the story. 35. The armour offered at Saint-Denis was the “blanc harnois” she wore during the earlier part of her career. When the church was pillaged by the English troops shortly after, this armour was sent to the King of England; but no further trace of it is known to exist. 36. Jeanne appears to have been a good horse-woman; she rode “horses so ill-tempered that no one would dare to ride them.” The Duke de Lorraine, on her first visit to him, and the Duke d’Alençon, after seeing her skill in riding a course, each gave her a horse; and we read also of a gift of a war-horse from the town of Orleans, and “many horses of value” sent from the Duke of Brittany. She had entered Orleans on a white horse, according to the Journal du Siège d’Orléans; but seems to have been in the habit of riding black chargers in war; and mention is also made by Châtelain of a “lyart” or grey. A story, repeated in a letter from Guy de Laval, relates that, on one occasion (June 6th, 1428), when her horse, “a fine black war-horse” was brought to the door, he was so restive that he would not stand still. “Take him to the Cross,” she said; and there he stood, “as though he were tied,” while she mounted. This was at Selles; and local tradition says that, from her lodging (a Dominican Monastery now the Lion d’Or hotel) the old iron town-cross was visible. It stood until about a century ago some fifteen paces in front of the north door of the Church, and was removed when the cemetery was converted into a market place. The Monastery was the property of the monks of Glatigny. The writers of the letter referred to above, Guy and André de Laval, were grandsons of Bertrand du Guesclin: the letter was dated Selles, June, 1429. The following are extracts: “... On Monday (June 6th) I left the King to go to Selles en Berry, four leagues from Saint Aignan. The King had summoned the Maid to come before him from Selles, where she then was, and many said this was much in my favour, so that I might see her. The said Maid treated my brother and me with
  • 73.
    great kindness: shewas armed at all points, save the head, and bore lance in hand. After we had arrived at Selles, I went to her lodging to see her, and she called for wine for me and said she would soon have me drink it in Paris. She seemed to me a thing divine, in all she did and all I saw and heard. “On Monday evening she left Selles to go to Romorantin.... I saw her mounting her horse armed all in white, save the head, a little axe in her hand.... And then, turning to the door of the Church, which was quite near, she said in a gentle woman’s voice, ‘You priests and clergy, make processions and prayers to God.’ Then she turned again on her way saying, ‘Draw on, draw on!’ her standard flying, borne by a gracious page, and her little axe in her hand. One of her brothers who arrived eight days since, left also with her, armed all in white.” 37. The banner was painted at Tours, while Jeanne was staying there, before her march to the relief of Orleans. The account for payment, in the “Comptes” of the Treasurer of War, gives: “À Hauvres Poulnoir, paintre, demourant à Tours, pour avoir paint et baillé estoffes pour une grand estandart et ung petit pour la Pucelle ... 25 livres tournois.” The description of this banner varies in different authors. The following account is compiled from them. “A white banner, sprinkled with fleur-de-lys; on the one side, the figure of Our Lord in Glory, holding the world, and giving His benediction to a lily, held by one of two Angels who are kneeling on each side: the words ‘Jhesus Maria’ at the side; on the other side the figure of Our Lady and a shield with the arms of France supported by two Angels” (de Cagny). This banner was blessed at the Church of Saint-Sauveur at Tours (Chronique de la Pucelle and de Cagny). The small banner or pennon had a representation of the Annunciation. There was also a third banner round which the priests assembled daily for service, and on this was depicted the Crucifixion (Pasquerel).
  • 74.
    Another banner ismentioned by the Greffier de la Rochelle, which Jeanne is said to have adopted as her own private pennon. It was made at Poitiers; and represented on a blue ground a white dove, holding in its beak a scroll, with the words, “De par le Roy du Ciel.” 38. May 7th, 1429. 39. This prophecy is recorded in a letter written, April 22nd, 1429, a fortnight before the event, by a Flemish diplomatist, De Rotslaer, then at Lyons. Her chaplain, Pasquerel, also states, in his evidence given in 1455, that she had told him of the coming injury on the previous day. 40. June 11th, 1429. 41. Gallicè: “en leur petite cotte,” i.e., with only the light clothing worn under their armour. 42. The “three Pontiffs” referred to are Martin V. (Colonna), the real and acknowledged Pope; the schismatic, Clement VIII.; and a mere pretender, Benedict XIV., who was supported only by one Cardinal. The Schism was practically at an end at the time of this letter, as Clement had abdicated a month earlier (July 26th). Clement VIII. is the true title, though called Clement VII. in Count d’Armagnac’s letter. 43. The English lost Paris in 1436. 44. Compiègne was relieved early in November; Saint Martin’s Day is November 11th. 45. The mandrake was a part of the accepted paraphernalia of a sorcerer. It was kept wrapped in a silk or linen cloth, and was supposed to preserve its owner from poverty. Brother Richard had recently preached a sermon against them (April, 1429); and many had been burned in consequence. 46. The balance was a frequent accessory to Saint Michael in the French stained glass windows of the 13th and 14th centuries. A
  • 75.
    noted example inthe Cathedral at Arles represents him weighing the souls of the departed in a balance as big as himself. One of the earliest examples in England is that in a fresco-painting at Preston Manor, Sussex, said to be of the reign of Edward I., in which Saint Michael appears weighing the souls of the faithful, accompanied by Jeanne’s saints, Saint Catherine and Saint Margaret. 47. Mary of Anjou, wife of Charles VII., daughter of Louis, Duke of Anjou and Yolande of Arragon. 48. Jeanne was taken from Beaurevoir early in August, and removed from there, when the negotiations for selling her were complete, about the middle of November. 49. Jeanne, Countess de Saint-Pol et Ligny, sister to Count Waleran de Luxembourg and aunt to Jean de Luxembourg. 50. Jeanne de Bethune, Viscountess de Meaux, wife of Jean de Luxembourg. Both these ladies were at Beaurevoir during Jeanne’s captivity, and shewed her great kindness, even interceding for her that she should not be sold to the English. 51. The Sieur de Pressy, in Artois. Present in the Burgundian camp when Jeanne was taken prisoner, and afterwards at Arras, where she was imprisoned on her way from Beaurevoir to Rouen. The questions seem to suggest that Beaupère had before him some information which has not come down to us. 52. This may perhaps refer to a popular belief in a halo, as of a Saint, surrounding the Maid’s head. 53. Brother Richard, a Mendicant Friar; some say, Augustan; some, Cordelier. He was preaching in Paris and the neighbourhood in 1428–9; and said, amongst other things, in a sermon at Sainte Géneviève, April 16th, 1419, that “strange things would happen in 1430.” He professed to have been in Jerusalem; and his sermons were so popular that congregations were found to listen to him for 10 or 11 hours,
  • 76.
    from 5 o’clockin the morning! He was driven out of Paris by the English and went to Troyes, where he joined the Maid. 54. No absolutely authentic portraits of Jeanne are known. A head of fine work, the portrait of a young girl wearing a casque and of Jeanne’s time, is at the Musée Historique at Orleans. Tradition asserts that when Jeanne entered Orleans in triumph with the relieving force a sculptor modelled the head of his statue of St. Maurice from Jeanne herself. This head is a portion of the statue which formerly stood in the church at Orleans dedicated to St. Maurice. The church was demolished in 1850. A photograph from the head is given as the frontispiece to this book, and an admirable copy maybe seen at the Musée du Trocadéro in Paris. It should have been stated on the frontispiece that the original is at Orleans, the copy in Paris. 55. Latin text adds: “dum rex suus consecraretur.” Tradition asserts that at the Coronation Jeanne stood on the left and slightly in front of the altar, coming direct from the sacristy of the cathedral. The coronation throne stood in front of the high altar. The cathedral and its painted glass exist as at the Coronation, with the exception of some comparatively recent stone work surrounding the choir. The Coronation of the Kings of France has taken place at Rheims Cathedral since the twelfth century. The King was not to all intents King of France until he had been anointed by the Holy Oil, brought in great state to the cathedral from the more ancient church of St. Remy. An inscription on the front of the Hotel Maison Rouge, situated near the west entrance of the cathedral, states that the town entertained Jeanne’s father and mother in that house during the Coronation. 56. About £200. 57. November 9th, 1429. 58. The Minute adds: “and I should be cured.”
  • 77.
    59. Surrendered July22nd. 60. In spite of this assertion, the Bishop was present at four out of the nine Examinations. 61. On May 23rd, 1430. 62. In the Minute only. 63. Not in the Minute. Latin text reads: “quod dedit regi suo dum venit ad eum.” 64. The “sign,” i.e. the appearance of “the White Lady.” 65. Jean, Duke d’Alençon: son of the Duke killed at Agincourt. He was of the blood-royal of France, and had married a daughter of the Duke d’Orléans. Jeanne was on very friendly terms with him, and always called him her “Beau Duc.” 66. The allegory of the Angel sent with a crown, here first given to avoid “perjury,” i.e., breaking her promise to preserve the King’s secret, is explained by Jeanne herself, on the last day of her life, to mean her own mission from Heaven to lead Charles to his crowning. 67. In the Minute: “et l’admener en trois ans”: not in the Latin Text. 68. The Minute reads: “la laissant faire de prisonniers.” 69. March 8th, 1428; it was before Easter, which in that year fell on March 7th. 70. The house in which Jeanne lodged at Chinon is said to have belonged to a certain Regnier de la Barrier, whose widow or daughter is the “worthy woman” referred to. Jeanne was afterwards lodged in the Tower of Coudray, where her room may still be seen. It is approached by a staircase outside the tower. The vaulted roof has fallen in, and the fireplace is damaged, but the walls are intact, and the room could easily be
  • 78.
    Welcome to ourwebsite – the perfect destination for book lovers and knowledge seekers. We believe that every book holds a new world, offering opportunities for learning, discovery, and personal growth. That’s why we are dedicated to bringing you a diverse collection of books, ranging from classic literature and specialized publications to self-development guides and children's books. More than just a book-buying platform, we strive to be a bridge connecting you with timeless cultural and intellectual values. With an elegant, user-friendly interface and a smart search system, you can quickly find the books that best suit your interests. Additionally, our special promotions and home delivery services help you save time and fully enjoy the joy of reading. Join us on a journey of knowledge exploration, passion nurturing, and personal growth every day! ebookmasss.com