Download to read offline
![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.](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-29-2048.jpg)
![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).](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-30-2048.jpg)
![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](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-46-2048.jpg)
![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](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-47-2048.jpg)
![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.](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-48-2048.jpg)
![What Temple saw
in Dublin.
An amended
proclamation, Oct.
29.
The Very Rev.
Henry Jones.
of Scotch fishing boats in the bay, and five hundred men volunteered
to land and be armed for the service of the State. The offer was
accepted, but never acted on, for the fishermen were seized with a
panic, put to sea, and never reappeared until the next year. The
fugitives from Ulster soon began to pour into Dublin. Temple is open
to criticism for his account of what happened in the northern
province, but this is what he saw himself:
‘Many persons of good rank and quality, covered
over with old rags, and some without any other
covering than a little to hide their nakedness, some
reverend ministers and others that had escaped with their lives
sorely wounded. Wives came bitterly lamenting the murders of their
husbands; mothers of their children, barbarously destroyed before
their faces; poor infants ready to perish and pour out their souls in
their mothers’ bosoms; some over-wearied with long travel, and so
surbated, as they came creeping on their knees; others frozen up
with cold, ready to give up the ghost in the streets; others
overwhelmed with grief, distracted with their losses, lost also their
senses.... But those of better quality, who could not frame
themselves to be common beggars, crept into private places; and
some of them, that had not private friends to relieve them, even
wasted silently away, and so died without noise.... The greatest part
of the women and children thus barbarously expelled out of their
habitations perished in the city of Dublin; and so great numbers of
them were brought to their graves, as all the churchyards within the
whole town were of too narrow a compass to contain them.’ Two
large additional burial grounds were set apart.[277]
On October 29 the Lords Justices issued a second
proclamation. The words ‘Irish Papists’ in the first
had been misunderstood, and they now desired to
confine it to the ‘old mere Irish in the province of
Ulster’; and they straitly charged both Papists and
Protestants on their allegiance to ‘forbear
upbraiding matters of religion one against the](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-50-2048.jpg)
![State of the Pale.
Lord Gormanston.
Sir N. Barnewall.
Sir T. Nugent.
Sir C. Bellew.
The Earl of
Kildare.
professed to have such orders. If, however, the inhabitants of the
county Cavan would peaceably return to their own dwellings, restore
every possible article to its plundered owner, and abstain from all
hostile acts in future, then the Lords Justices would forward their
petition to his Majesty and ‘humbly seek his royal pleasure therein.’
The O’Reillys were in the meantime preparing to attack Dublin in
force.[278]
As regards the gentry of the Pale, Roman Catholics
for the most part, the Lords Justices were in a
difficult position. By mistrusting them they ran the
risk of driving them into rebellion; by trusting them they increased
their power for mischief, should they be already tainted. For the
moment the first danger seemed the greater of the two, and
commissions as governors of counties with plenary powers were
accordingly issued to several of them, by which they were authorised
to proceed by martial law against the rebels, ‘hanging them till they
be dead as hath been accustomed in time of open rebellion,’
destroying or sparing their houses and territories according to their
discretion. They were also empowered to grant protections.
Viscount Gormanston was thus made governor of
Meath, and arms were given him for 500 men. He
was in open rebellion a few weeks later. Sir
Nicholas Barnewall of Turvey, afterwards created
Viscount Kingsland by Charles I., became governor
of the county of Dublin, and had arms for 300
men. Barnewall was a good deal involved in
political intrigues, but soon fled to England to avoid
taking arms against the Government. A commission
as governor of Westmeath and arms for 300 men
were given to Sir Thomas Nugent, who afterwards
tried to fill the difficult part of neutral. Sir
Christopher Bellew was governor of Louth, with arms for 300, but he
very soon joined the Irish. To George Earl of Kildare, Cork’s son-in-
law, his own county was entrusted and arms for 300; but he was a](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-52-2048.jpg)
![Ormonde made
general.
Sir H. Tichborne.
Protestant and suffered severely for his loyalty, while he was quite
unable to curb his neighbours. Finding after a time that the arms
given out would, if used at all, be used against them, the Lords
Justices endeavoured to get them back, but they recovered only 950
out of 1700, and the enemy had the rest.[279]
Ormonde was at his own house at Carrick-on-Suir
when the rebellion broke out. The Lords Justices
sent for him at once, and the first letter being
delayed in transmission, a second was sent with a
commission to him and Mountgarret to govern the
county of Kilkenny and to take such other precautions as were
possible. The gentry met at Kilkenny and offered to raise 240 foot
and 50 horse, while Callan and other towns made similar promises.
There were, however, no arms, and the Lords Justices would give
none out of the stores. Before purchases could be made in England
the situation was greatly changed. Ormonde arrived at Dublin with
his troop early at the end of the first week in November, and on the
10th Sir Patrick Wemyss returned from Edinburgh with his
nomination as Lieutenant-General, to command the army as he had
done in Strafford’s time. The Lords Justices made out his commission
next day, with warrant to execute martial law, but without prejudice
to Leicester’s authority as Lord Lieutenant. It was not till six months
later that the King gave him power to appoint subordinate officers
according to the ‘constant practice and custom of former times,’ it
having by then become evident that Leicester would not reside in
Ireland. The defence of Drogheda had already been provided for by
Sir Henry Tichborne, who was living at Dunshaughly, near Finglas,
and who had brought his family into Dublin on the first day, having
already ‘scattered a parcel of rogues’ that threatened his country
house. Having received a commission from the Lords Justices, he
raised and armed 1000 men in nine days among the Protestants who
had left their homes, and with this regiment he entered Drogheda
on November 4. Three additional companies were sent to him a few
days later.[280]](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-53-2048.jpg)
![Ormonde
disagrees with the
Lords Justices.
The Irish
Parliament after
the outbreak.
Both Houses
protest against
the rising.
Vain hopes of
peace.
Prorogation, Nov.
17, 1641.
One of Ormonde’s first acts as general was to
commission Lord Lambert, Sir Charles Coote, and
Sir Piers Crosbie to raise regiments of 1000 men
each, and thirteen others to raise independent
companies of 100 each. The ranks were filled in a few days, for all
business was at a standstill, and Protestant fugitives poured in in
great numbers. There were 1500 disciplined men of the old army
about Dublin. Strafford had left a fine train of field artillery with
arms, tents, and all necessaries for 10,000 men. Under these
circumstances Ormonde was for pushing on, and putting down the
northern rebellion at once. To this the Lords Justices would not
consent, and it may be that they were jealous of their general; but it
must be confessed that there was also something to be said for a
cautious policy. With the Pale evidently disaffected Dublin could not
be considered as very safe.[281]
When the rebellion broke out the Lords Justices by
their own authority prorogued Parliament till
February 24, fearing a concourse of people to
Dublin, and also because the state of Ulster made
it almost certain that there would not be a
Protestant majority. The gentry of the Pale, and the
Roman Catholic party generally, protested strongly,
and there were doubts about the legality of the
prorogation. Some lawyers held that Parliament
would be dissolved by the mere fact of not meeting
on the appointed day. To get over the difficulty the
Lords Justices agreed that Parliament should meet
as originally announced, but that it should sit only
for one day, and should then be prorogued to a date earlier than
February 24. Ormonde and some others were in favour of a regular
session, but they were overruled by the official members of the
Council. Parliament met accordingly on November 9, and
immediately adjourned till the 16th, so as to give time for private
negotiations. The attendance was thin in both Houses, partly on
account of the state of the country and partly because many thought](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-54-2048.jpg)
![Leicester Lord
Lieutenant.
He never came to
Ireland.
The rebellion
reported to the
English
Parliament.
that the prorogation till February was still in force. Mr. Cadowgan
significantly remarked that ‘many members of the House are traitors,
and whether they come or not it is not material.’ There was a great
military display about the Castle gates, according to the precedent
created by Strafford, and offence was taken at this; but the two
Houses agreed to a protestation against those who, ‘contrary to their
duty and loyalty to his Majesty, and against the laws of God, and the
fundamental laws of the realm, have traitorously and rebelliously
raised arms, have seized on some of his Majesty’s forts and castles,
and dispossessed many of his Majesty’s faithful subjects of their
houses, lands, and goods, and have slain many of them, and
committed other cruel and inhumane outrages and acts of hostility
within the realm.’ And the Lords and Commons pledged themselves
to ‘take up arms and with their lives and fortunes suppress them and
their attempts.’ There was some grumbling about the words
‘traitorously and rebelliously’ on the principle that birds are not to be
caught by throwing stones at them, but the majority thought the
Ulster rebels past praying for, and the protest was agreed to without
a division. There was also unanimity in appointing a joint committee,
fairly representing different sections, with power, subject to royal or
viceregal consent, to confer with the Ulster people. Two days were
occupied in these discussions, and on the evening of the 17th the
Lords Justices prorogued Parliament till January 11. When that day
came things had gone far beyond the parliamentary stage.[282]
The Earl of Leicester was appointed Lord
Lieutenant early in June 1641, and the Lords
Justices were directed by the King to furnish him
with copies of all their instructions. He remained in
England, and to him the Irish Government
addressed their account of the outbreak. This was
brought over by Owen O’Connolly, received on or
before October 31, and at once communicated to
the Privy Council, who had a Sunday sitting. On
Monday, November 1, the Upper House did not sit
in the morning, ‘for,’ says Clarendon, ‘it was All Saints’ Day, which the](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-55-2048.jpg)
![The news reaches
the King, Oct. 27.
Letter from the
O’Farrells.
Catholic
grievances
represented to
the King.
Lords yet kept holy, though the Commons had
reformed it.’ To the House of Commons accordingly
the Privy Council proceeded in a body, headed by
the Lord Keeper. There was no precedent for such a visitation, but
after a short discussion chairs were placed in the body of the House
and Leicester, with his hat off, read the Lords Justices’ letter of
October 25. Clarendon testifies from personal knowledge that the
rebellion was odious to the King, and confidently asserts that none
of the parliamentary leaders ‘originally and intentionally contributed
thereunto,’ though he believes that their conduct afterwards added
fuel to the flame. When the Privy Councillors had withdrawn the
House went into committee, Mr. Whitelock in the chair, and drew up
heads for a conference with the Peers. As to money they resolved to
borrow 50,000l., giving full security, and to pay O’Connolly 500l.
down with a pension of 200l. until an estate of greater value could
be provided. Resolutions were passed against Papists, and
particularly for the banishment of the Queen’s Capuchins. The Lords
met in the afternoon, and after this the two Houses acted together.
Three days later the estimate for Ireland was raised to 200,000l.,
and Leicester was authorised to raise 3,500 foot and 600 horse,
while arms were provided for a further levy. News of the outbreak
came to the King at Edinburgh direct from Ulster four days before it
reached the English Parliament. Tradition says that he was playing
golf, and that he finished his game.[283]
Lord Dillon of Costello, who was a professing
Protestant, produced at the Council on November
10 a letter signed by twenty-six O’Farrells in county
Longford. This paper is well written, and contains
the usual pleas for religious equality, which modern
readers will readily admit, though they were not
according to the ideas of that day either at home
or abroad. The O’Farrells had taken an oath of
allegiance, but their sincerity is open to doubt, for they demanded
‘an act of oblivion and general pardon without restitution on account
of goods taken in the times of this commotion.’ No government could](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-56-2048.jpg)
![Weakness of the
Irish Government.
Relief comes but
slowly.
Monck, Grenville
and Harcourt.
possibly grant any such amnesty, and the suggestion came at a time
when Ulster was in a blaze and when Dublin was crowded with
Protestants who had escaped with their bare lives. Dillon and Taaffe
were commissioned by the Roman Catholic lords to carry their
grievances to the King. When returning with instructions they were
stopped at Ware and their papers overhauled, the Lords Justices
having warned their parliamentary friends.[284]
The influence of Carte has led historians generally
to think that the Lords Justices were either too
desperately frightened to think of anything but
their own safety, or that they let the rebellion
gather head to suit the views of the English
parliamentary party. There is not much evidence
for either supposition. Just at the moment when
the Pale was declaring against them they reported
their destitute condition to Leicester. The troops
were unpaid. At Dublin they had but 3000 foot and 200 horse, and
the capital as well as Drogheda was surrounded by armed bands
who had already made food scarce, and who threatened to cut off
the water. A large extent had to be defended, and many of the
inhabitants were not to be trusted. A crusade was being preached all
over the country, and at Longford, notwithstanding the oath of the
O’Farrells, a priest was reported to have given the signal for a
massacre by ripping up the parson with his own hand. The mischief
was spreading daily, and agitators industriously declared that no help
would be sent from England. Ireland was not, however, forgotten,
but Parliament, to whom the King had specially entrusted it, had its
own business to do, and a popular assembly has no administrative
energy. It was not till the last day of December that Sir Simon
Harcourt landed with 1100 men. Three hundred more followed
quickly, and George Monck with Leicester’s own regiment was not far
behind. Grenville brought 400 horse about the same time. Harcourt
had long military experience in the Low Countries, and had lately
commanded a regiment in Scotland. He had a commission as
Governor of Dublin, but Coote was in possession and was not](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-57-2048.jpg)
![Sir Charles Coote.
disturbed. Harcourt was very angry with the Lords Justices, but he
got on well with Ormonde and did good service until his death.[285]
The number of troops available in Dublin was
small, but they were much better armed than the
insurgents. It was thus a matter of policy to act on
the offensive and clear the surrounding country, demolishing houses
and castles where troublesome posts might be established. This
work, cruel in itself, was performed in a very ruthless manner, and
particular blame has always fallen upon Sir Charles Coote, whose
ferocity seems to have been as conspicuous as his courage. One
story told both by Bellings and Leyburn is that he called upon a
countryman to blow into the mouth of his pistol, that the simple
fellow obeyed, and that Coote shot him in that position. He never
went to bed during a campaign, but kept himself ready for any
alarm, and lost his life in a sally from Trim during a night attack at
the head of only seventeen men, the place being beset by
thousands.[286]
FOOTNOTES:
[268] Alice Thornton’s Autobiography; Irish Lords Journals,
February 22, 1640-1; Petition of the Irish Committee to the King,
Cal. State Papers, Ireland, 1640, addendum; Radcliffe’s answer to
the Committee, ib. January 9, 1641, and their rejoinder, ib.
February 12.
[269] Irish Commons Journals, February 16, 1640-1. The queries,
with the answers and declaration of the Commons, are in Nalson,
ii. 572-589.
[270] Irish Commons Journals, 1641, p. 211; Irish Lords Journals,
February 27, March 4.
[271] Irish Commons Journals, June 7, July 10. The story about
the powder is from Borlase’s Rebellion, ed. 1680, p. 12; he is not
a very good authority, but on this occasion is speaking of his
father’s action.](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-58-2048.jpg)
![[272] Examination of Henry Macartan, quartermaster to Owen
Roe O’Neill, February 12, 1641-2, Contemp. Hist. i. 396; Vane to
the Lords Justices, March 16, 1640-1, Cox’s Hibernia Anglicana, ii.
65; Cole to the Lords Justices, October 11, 1641, printed in
Nalson and elsewhere; Lords Justices and Council to Vane, June
30, 1641, State Papers, Ireland; Deposition as to the
Multifarnham meeting, May 3, 1642 (misprinted 1641), in
Hickson’s Seventeenth Century, ii. 355. Temple produces evidence
as to the rebellion being threatened long before it actually
happened, O’More himself having admitted as much, p. 103.
Patrick O’Bryan of Fermanagh swore on January 29, 1641-2 ‘that
he heard Colonel Plunket say that he knew of this plot eight years
ago, but within these three years hath been more fully acquainted
with it’—Somers Tracts, v. 586. Lieutenant Craven, who had been
a prisoner with the Ulster Irish, was prepared to swear that on
March 3, 1641-2, he had heard Bishop Heber Macmahon tell his
friends that he had planned the rebellion years before, and knew
from personal knowledge that all Catholic nations would help;
urging them to persevere and extirpate heresy. Macmahon
repeated this at Monaghan in January 1643-4—Carte MSS. vol.
lxiii. f. 132.
[273] Lord Maguire’s Relation, written by him in the Tower (after
August 1642) printed from the Carte Papers in Contemp. Hist. i.
501. Parsons to Vane, August 3, State Papers, Ireland. Temple’s
History is valuable here, for he was present in Dublin and signed
the proclamation on October 23, Bellings, i. 7-11.
[274] O’Connolly’s Deposition, October 22, in Temple’s History,
with the author’s remarks, and his further Relation printed from a
manuscript in Trinity College in Contemp. Hist., i. 357.
[275] Chiefly from Temple’s History, where O’Connolly’s evidence,
and the proclamation of October 23, are given in full. There is an
independent account by Alice Thornton, Wandesford’s daughter,
who was in Dublin at the time, aged fifteen. According to her
O’Connolly swam the Liffey. ‘What shall I do for my wife?’ he
asked the conspirators, and they answered ‘Hang her, for she was
but an English dog; he might get better of his own country.’—
Autobiography, Surtees Society, 1875.
[276] Sir F. Willoughby’s narrative among the Trinity College
MSS., 809-841, vol. xxxii. f. 178.
[277] Temple, pp. 93-4. Macmahon’s Deposition, October 23,
Contemp. Hist. i. Appx. xix. Lords Justices and Council to](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-59-2048.jpg)
![Leicester, October 25, printed in Temple’s History and elsewhere.
Macmahon’s latter evidence, ‘taken at the rack’ on March 22,
1641-2, gives further details regarding the Ulster conspirators,
but he knew nothing about the Pale, and does not even mention
O’More’s name. Reports of Maguire’s trial have been often
printed.
[278] Proclamation of October 29, 1641, in Temple and
elsewhere. Dean Jones’s ‘Relation of the beginning and
proceedings of the rebellion in Cavan, c.,’ was printed in London
by order of the House of Commons in the spring of 1642, and
reproduced in vol. v. of the Somers Tracts as well as in Gilbert’s
Contemporary History, where the Cavan Remonstrance, received
November 6, and the Lords Justices’ answer dated November 10,
are also printed. Rosetti at Cologne heard that many Protestants
had joined the rebels, which was certainly not true, though some
pretended to do so. Roman Transcripts, R.O., December 10, 1641.
Another paper from Cologne speaks of the rebels ‘quali vanno
decapitando et appiccando li Protestanti che non gli vogliono
assistere,’ ib. December 22.
[279] Temple prints the commission to Gormanston as a
specimen. Lords Justices and Council to Leicester, December 14,
in Nalson, ii. 911.
[280] Sir Henry Tichborne’s letter to his wife, printed with
Temple’s History, Cork, 1766. Carte’s Ormonde, i. 193, and the
King’s letters in vol. iii. Nos. 31 and 82.
[281] Carte’s Ormonde, i. 192-5; Lords Justices to Ormonde,
October 24, 1641, printed in Confederation and War, i. 227.
[282] Bellings gives the two documents referred to. He was a
member of this Parliament, and one of the Joint Committee. Irish
Commons Journals.
[283] Rushworth, iv. 398-406; Nicholas to the King, November 1,
1641, in Evelyn’s Correspondence; Macray’s edition of Clarendon’s
History, i. 408; May’s Long Parliament, p. 127. May is a good
authority for what happened in London, but for events in Ireland
he depends chiefly on Temple. Lords Journals, November 1;
Lang’s Hist. of Scotland, iii. 100; Vane to Nicholas, October 27,
Nicholas Papers, i. 58.
[284] Nalson, ii. 898; Rushworth, iv. 413; Diurnal Occurrences,
December 20-25, 1641.](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-60-2048.jpg)
![[285] Despatch of December 14, in Nalson, ut sup. Monck’s letter
from Chester, ib. 919, shows how little money Parliament had to
spare. In clerical circles abroad it was rumoured a little later that
Dublin would soon fall, and that five hundred Protestants who
objected to the cross in baptism had been marked with it on the
forehead and sent back to England—Roman Transcripts, R.O.,
February 2, 1642. Four letters from Sir Simon Harcourt, January
3, 1641-42 to March 21, in vol. i. of Harcourt Papers (private
circulation). As late as September 16, 1642, Sir N. Loftus wrote
from Dublin that the enfeebled garrison could not hold out for six
weeks if seriously attacked. Food and ammunition were wanting,
and the surviving soldiers sick or starving—Portland Papers, i.
700.
[286] Bellings, i. xxxii. 35; George Leyburn’s Memoirs, Preface;
Borlase’s Irish Rebellion, p. 104, ed. 1743. Coote was killed May
7, 1642; when the name occurs later the reference is to his son,
also Sir Charles.](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-61-2048.jpg)
![The massacre in
Island Magee.
The rising in
Tyrone, Oct. 23,
1641.
English tenants
plundered.
Murder of
Protestants.
several thousand Protestants were massacred, that the murders
were not confined to one province or county, but occurred in almost
every part of the island, that the retaliation was very savage,
innocent persons often suffering for the guilty, and that great
atrocities were committed on both sides. ‘The cause of the war,’ says
Petty, ‘was a desire of the Romanists to recover the Church revenue,
worth about 110,000l. per annum and of the common Irish to get all
the Englishmen’s estates, and of the ten or twelve grandees of
Ireland to get the empire of the whole.... But as for the bloodshed in
the contest, God best knows who did occasion it.’ He thought the
population of Ireland in 1641 was about 1,400,000, out of which
only 210,000 were British.[287]
One of the worst cases of retaliation was the
massacre by Scots of many Roman Catholic
inhabitants of Island Magee in Antrim, but it is
necessary to point out that this took place in January 1642, because
it has been asserted that it was the first act of violence and the real
cause of the whole rebellion. Some of those who took part in the
outrage were alive in 1653, and were then prosecuted by the
Cromwellian Government.[288]
Dublin was saved, but the rebellion broke out in
Ulster upon the appointed day. According to
Captain John Creichton, his grandfather’s house
near Caledon in Tyrone was the first attacked. The
rebellion certainly began upon Sir Phelim O’Neill’s
property at Caledon or Kinard during the night of
October 22, when O’Connolly was telling the Lords
Justices what he had heard. William Skelton, who
lived as a servant in Sir Phelim’s house, was
ploughing in the afternoon when an Irish fellow
servant came to him with about twenty companions and said that
they had risen about religion. Armed only with cudgels, they
attacked several of Sir Phelim’s English tenants, who were well-to-do
and apparently well-beloved by their Irish neighbours, ‘and differed](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-64-2048.jpg)
![Sir Phelim O’Neill
at Charlemont.
The Caulfield
family.
Dungannon,
Mountjoy,
Tanderagee and
Newry taken
Bishop Henry
Leslie.
not in anything, save only that the Irish went to mass, and the
English to the Protestant church in Tinane, a mile from Kinard.’
Taken by surprise, the Protestants were easily disarmed, and robbed
in the first instance only of such horses as would make troopers. All
the English and Scots neighbours were thus plundered in detail,
cattle, corn, furniture, and clothes being taken in succession. In
about a fortnight the Irish began to murder the Protestants. Among
those whom Skelton knew of his own knowledge to be killed in cold
blood before the end of the year was ‘one Edward Boswell, who was
come over but a year before from England, upon the invitation of
the said Sir Phelim, his wife having nursed a child of the said Sir
Phelim’s in London.’ Boswell’s wife and child were murdered at the
same time, and seventeen others in Kinard itself, men, women, and
children. Skelton and some others were saved by the intercession of
Daniel Bawn, whose wife was an Englishman’s daughter.[289]
While his English servant was ploughing at Kinard,
Sir Phelim O’Neill was on his way to Charlemont
with an armed party. He had invited himself to
dinner and was hospitably received by Lady
Caulfield and her son, who had not long succeeded
to the peerage. In after days there was a family
tradition that the butler, an old and trusty servant,
was alarmed by the attitude of Sir Phelim’s
followers and imparted his fears to his mistress. His
advice was neglected, and when the meal was over
he left the house and made the best of his way to
Dublin. The Caulfields and the unsuspecting men
who ought to have defended the fort were
surprised and captured, and O’Neill occupied Dungannon the same
night. Next day the O’Quins took Mountjoy, the O’Hanlons
Tanderagee, and the Magennises Newry. All were surprised, and
there was practically no resistance. In the course of the day a
fugitive trooper came to Lisburn, where Henry Leslie, Bishop of
Down, was living, with news of the disasters at Charlemont and
Dungannon, and four hours later another runaway announced that](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-65-2048.jpg)
![Fermanagh. Rory
Maguire.
Murders at
Lisgoole and
elsewhere.
Treatment of the
English Bible.
Newry was taken. Leslie at once sent the news on to Lord
Montgomery, who was at or near Newtownards, and to Lord
Chichester at Belfast; and they both wrote to the King.
Chichester said only one man had been slain, which has been
adduced as a proof that there was no massacre, but he knew only
what Leslie had told him, and there were no tidings from any point
beyond Dungannon. Other districts could tell a very different tale.
[290]
Lord Maguire was a prisoner, but his brother Rory
raised Fermanagh before any account of the doings
in Dublin had come so far. The robbing and
murdering began on October 23, and very soon the
whole county was at the mercy of the rebels.
Enniskillen was never taken, and it will be seen
that walled towns, if well defended, were generally
maintained. Alice Champion, whose husband was killed in her
presence on the first day, heard the murderers say that ‘they had
special orders from Lord Maguire not to spare him or any of the
Crosses that were his followers and tenants.’ About twenty-four
others were murdered at the same time, and Mrs. Champion
afterwards heard them boast that they had ‘killed so many
Englishmen that the grease or fat that remained on their swords
might have made an Irish candle,’ ninety being despatched at
Lisgoole alone. The latter massacre is also sworn to by an eye-
witness. Anne Ogden’s husband was murdered in the same way. She
was allowed to fly to Dublin with her two children, but all were
stripped on the way, and the children afterwards died ‘through the
torments of hunger and cold they endured on that journey.’
Edward Flack, a clergyman, was plundered and
wounded on the 23rd, and his house burned. The
rebels in this case vented some of their fury on his
Bible, which they stamped upon in a puddle, saying ‘A plague on this
book, it has bred all this quarrel,’ and hoping that all Bibles would
have this or worse treatment within three weeks. Much more of the](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-66-2048.jpg)
![Cavan. The
O’Reillys.
Pretended orders
from the King.
Colonel Richard
Plunkett.
same kind might be said, and the events sworn to in Fermanagh
alone fully dispel the idea that there were no murders at the first
outbreak.[291]
In Cavan, where the O’Reillys were supreme, there
were no murders at the very beginning. Here, as in
other places, the first idea seems to have been to
spare the Scots and not to kill the English unless
they resisted their spoilers. On the night of October
23, the Rev. George Crichton, vicar of Lurgan, who
lived at Virginia, was roused out of his first sleep by
two neighbours, who told him of the rising further
north. Many of the Protestant inhabitants fled into
the fields, but Crichton thought it better to stand his ground, and
very soon a messenger came from Captain Tirlogh McShane McPhilip
O’Reilly, to say that the Irish would harm no Scot. Crichton perhaps
profited also by the fact that ‘no man ever lost a penny by him in the
Bishop’s Court, and none ever paid to him what he did owe,’ which
may have been a result of Bedell’s influence. He went out and met
this chief at Parta wood, about a mile to the east of the town.
O’Reilly, who had some twenty-four men with him, announced that
Dublin and all other strong places were taken, and that they ‘had
directions from his Majesty to do all these things to curb the
Parliament of England; for all the Catholics in England should have
been compelled to go to Church, or else they should be all hanged
before their own doors on Tuesday next.’ Crichton said he did not
believe such a thing had been ever dreamed of, whereupon O’Reilly
declared his intention of seizing all Protestant property and of killing
anyone who resisted. Next morning Virginia was sacked accordingly,
but no lives were taken, for no one made any defence. The canny
Scots clergyman managed to keep the Irish in pretty good humour,
lodged nine families in his own house, and provided food for the
fugitives from Fermanagh who began to arrive in a few days. Many
thousands from Ballyhaise, Belturbet and Cavan passed through
Virginia on their way towards the Pale. Crichton obtained help from
Colonel Richard Plunkett, who wept and blamed Rory Maguire for all.](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-67-2048.jpg)
![Cavan and
Belturbet.
Philip MacHugh
O’Reilly.
Horrors of a
winter flight.
On being asked whether the Irish had made a covenant he said,
‘Yea, the Scots have taught us our A B C; in the meantime he so
trembled that he could scarce carry a cup of drink to his head.’
Nevertheless he boasted that Dublin was the only place not taken,
that Geneva had fallen, and that there was war in England. Many of
the wretched Fermanagh Protestants were wounded, and the state
of their children was pitiable. The wounded were tended and milk
provided for the children, Crichton telling his wife and family that it
was their plain duty to stay, and that ‘in this trouble God had called
them to do him that service.’ All this happened within the first week
of the outbreak, and when the long stream of refugees seemed to
have passed, Crichton and his family prepared to go; but they were
detained, lest what they had to tell might be inconvenient.
Protestants from the north continued to drop in for some time, and
Crichton was allowed to relieve them until after the overthrow at
Julianstown at the end of November. The O’Reillys took part in the
affair, and their followers became bolder and less lenient.[292]
Another clergyman, Henry Jones, Dean of Kilmore,
was living at Bellananagh Castle, near Cavan, at
the time of the outbreak. Philip MacHugh
MacShane O’Reilly, member for the county, was the
chosen leader of the Irish. The actual chief of the
clan was Edmund O’Reilly, but the most active part
was taken by his son, Miles O’Reilly, the high
sheriff, a desperate ‘young man,’ who at once
assumed his native name of Mulmore Mac Edmond.
Under the pretence of raising the posse comitatus he sent bailiffs to
the scattered houses of Protestants and collected their arms. He
himself seized the arms at Farnham Castle, and took possession of
Cloghoughter, with whose governor, Arthur Culme, he had been on
terms of friendship. Next day, October 24, the sheriff proceeded to
Belturbet, which was the principal English settlement and contained
some 1500 Protestants. Sir Stephen Butler was dead, but his widow
had married Mr. Edward Philpot and was living there with her five
children. Sir Francis Hamilton, who was at Keilagh Castle, tried to](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-68-2048.jpg)
![The O’Reillys were
not unanimous.
Doctor Henry
Jones.
Weakness of the
Irish Government.
Divisions among
the Irish.
organise some resistance, but Philip MacHugh O’Reilly took the
settlers under his protection, and they gave up their arms. Yet
Captain Ryves with some thirty horse had no difficulty in reaching
the Pale by O’Daly’s Bridge on the Blackwater, and in occupying
Ardbraccan for the Lords Justices. Cavan surrendered, and on the
29th Bellananagh, which was indefensible, surrendered to the
sheriff’s uncle, Philip MacMulmore O’Reilly. It had been determined
to clear all the English out of the county, and though Lady Butler
with 1500 others were escorted as far as Cavan they were attacked
just beyond the town, and stripped of everything. Those who did not
die of exposure reached Dublin, to starve and shiver among the
other fugitives there. Those who remained at Belturbet had a still
worse fate.[293]
The O’Reillys had always been more civilised than
other natives of Ulster, and they almost seem to
have felt that the Government must win in the end.
Rose O’Neill, the wife of Philip MacHugh, wished to
kill all the English and Scotch at Ballyhaise, but he
would not allow it. ‘The day,’ he said, ‘may come
when thou mayest be beholding to the poorest
among them.’ With a view no doubt to that distant
day, they resolved to petition the Lords Justices
and to send an Englishman with the message.
Bedell refused to go on account of his age and
because his plundered flock could not spare him,
but Jones, who in his time played many parts, thought it safer to do
as he was asked. He left his wife and children as hostages and went
to Dublin, with a memorial signed by seven O’Reillys which spoke of
former misgovernment, and rumours that worse was to come. They
protested their loyalty and desired the Lords Justices ‘to make
remonstrance to his Majesty for us ... so that the liberties of our
conscience may be secured unto us, and we eased of our other
burdens in the civil government.’ The Lords Justices and eight Privy
Councillors, of whom Ormonde was one, sent an answer, dealing in
generalities ‘suitable to the weak condition of affairs in Dublin.’ The](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-69-2048.jpg)
![Rising in
Monaghan.
Murder of Richard
Blayney.
A sham royal
commission.
most they could promise was that if they would restore all the Cavan
Protestants to their homes and properties and cease from further
hostilities, that then their memorial should be forwarded to the King.
On his return Jones found the O’Reillys preparing to invade the Pale.
He managed to keep the Dublin Government well informed, at the
same time dissuading the Irish from attacking the capital, whose
means of defence he exaggerated. Drogheda, he said, was more
assailable, and to Drogheda they determined to go. They mustered
first at Virginia, where Mr. Crichton made friends with Philip
MacHugh’s mother on the ground of common kinship with Argyle, ‘of
which house it seemeth that she was well pleased that she was
descended. This kindred stood me in great stead afterwards, for
although it was far off and old, yet it bound the hands of the ruder
sort from shedding my blood.’ Many lives, says Crichton, were also
saved by the quarrels of the Irish among themselves. Philip
MacHugh not only shielded his far away cousin, and others for his
sake, but was evidently disinclined to the task in hand, regretted
that he had not kept the Protestants safe at Belturbet, ‘blamed Rory
Maguire for threatening to kill and burn them, and cursed those
among the English that gave them counsel to leave their
habitations.’ Crichton thought O’Reilly a deep dissembler, but he
should have the credit for comparative humanity. He and others
seem to have thought that the war was on the point of breaking out
in England, and that it would be impossible to send any troops to
Ireland for years to come.[294]
In Monaghan there was a general rising on October
23, but a number of murders were committed
during the first few days, and the Macmahons
behaved worse than the O’Reillys. Richard Blayney,
member for the county, and commissioner of
subsidies, was hanged by Sir Phelim O’Neill’s direct
orders, and his dead body barbarously treated. At
Carrickmacross Essex’s bailiff, Patrick McLoughlin
Macmahon, took the lead among the local rebels,
and about 600l. of the great absentee’s rents came into their hands.](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-70-2048.jpg)
![The Portadown
massacre, about
Nov. 1, 1641.
The church at
Blackwater.
Alleged
apparitions.
Investigation by
Owen Roe O’Neill.
In Monaghan, as elsewhere, the Irish professed to do everything by
the King’s orders, but at Armagh Sir Phelim O’Neill professed to
show the actual commission with a broad seal to it, adding that he
would be a traitor if he acted of his own accord. ‘We are a sold
people,’ said an Englishman who witnessed the scene. A number of
Protestants took refuge in the cathedral, but they had to surrender,
and being stripped and robbed were sent to keep the Caulfields
company at Charlemont. A miscellaneous collection of Protestants,
including many children and poor people, from whom no ransom
could be expected, were driven to the bridge of Portadown and
there murdered.[295]
The Portadown massacre has been more discussed
perhaps than any episode in the Irish rebellion, and
it has left behind it an ineffaceable impression of
horror. The victims were only a part of those
murdered in the county of Armagh, but more than
100—one account says 160—were killed at one
time—and the affair was carefully planned
beforehand. The chief actor was Captain Manus
O’Cahan, but many of the sufferers had received
passes from Sir Phelim himself. O’Cahan and his
men, Mrs. Price deposed, forced and drove all
those prisoners, and amongst them the deponent’s
five children, by name Adam, John, Anne, Mary,
and Jane Price, off the bridge into the water. Those that could swim
were shot or forced back into the river. When Owen Roe O’Neill
came to the country he asked in Mrs. Price’s hearing how many
Protestants the rebels had drowned at Portadown, and they said
400. If this is correct the cruel work on the Bann must have
continued for some time. They also said that those drowned in the
Blackwater were too many to count, and that the number thrust into
lakes and bog-holes could not even be guessed at. On November 17
they burned the church at Blackwaterstown with a crowd of
Protestants in it, ‘whose cries being exceeding loud and fearful, the
rebels used to delight much in a scornful manner to imitate them,](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-71-2048.jpg)
![Bedell at Kilmore.
He is allowed to
relieve many
Protestants.
He refuses to
leave his post.
and brag of their acts.’ Attempts have been made to discredit the
evidence on the ground that Mrs. Price and others refer to
apparitions at the scene of the Portadown massacre. Screams and
cries are easily explained, for wolves and dogs fed undisturbed upon
the unburied dead. But Mrs. Price says she actually saw a ghost
when she visited the spot where her five children had been
slaughtered, and that Owen Roe O’Neill, who came expressly to
inform himself as to the alleged apparitions, was present with his
men, who saw it also. It was twilight, and ‘upon a sudden, there
appeared unto them a vision, or spirit assuming the shape of a
woman, waist high, upright in the water, naked, her hair dishevelled,
very white, and her eyes seeming to twinkle in her head, and her
skin as white as snow; which spirit or vision, seeming to stand
upright in the water, divulged, and often repeated the word
“Revenge! Revenge! Revenge!”’ O’Neill sent a priest and a friar to
question the figure both in English and Latin, but it answered
nothing. He afterwards sent a trumpet to the nearest English force
for a Protestant clergyman, by whom the same figure was seen and
the cries of ‘Revenge!’ heard, but Mrs. Price does not say she was
present on this occasion. The evidence of this lady shows no marks
of a wandering mind, and yet it is evident that she believed in an
apparition. It is quite possible that some crazed woman who had lost
all that was dear to her may have haunted the spot and cried for
vengeance, but in any case a belief in ghosts was still general in
those days, and especially in Ireland. The evidence as to the
massacre is overwhelming.[296]
Bedell was at Kilmore when the rebellion broke out.
The Protestants were surprised, but it was
remembered afterwards that there had been an
invasion or migration of rats, and that caterpillars
had appeared in unusual numbers. It was more to
the purpose that a crack-brained Irish scholar who
wandered from house to house was heard
frequently to exclaim, ‘Where is King Charles now?’
and that he wrote in an old almanac ‘We doubt not](https://image.slidesharecdn.com/34730-250502192952-c4c2e0fb/75/Data-Science-Applied-to-Sustainability-Analysis-1st-edition-eBook-PDF-72-2048.jpg)
Data Science Applied to Sustainability Analysis 1st edition - eBook PDF
- 1. Data Science Appliedto Sustainability Analysis 1st edition - eBook PDF download https://ebooksecure.com/download/data-science-applied-to- sustainability-analysis-ebook-pdf/ Download full version ebook from https://ebooksecure.com
- 2. We believe theseproducts will be a great fit for you. Click the link to download now, or visit ebooksecure.com to discover even more! Innovation in Nano-polysaccharides for Eco- sustainability: From Science to Industrial Applications 1st Edition - eBook PDF https://ebooksecure.com/download/innovation-in-nano- polysaccharides-for-eco-sustainability-from-science-to- industrial-applications-ebook-pdf/ (eBook PDF) Intro to Python for Computer Science and Data Science: Learning to Program with AI, Big Data and The Cloud http://ebooksecure.com/product/ebook-pdf-intro-to-python-for- computer-science-and-data-science-learning-to-program-with-ai- big-data-and-the-cloud/ (eBook PDF) Environmental Science and Sustainability 1st Edition by Daniel J. Sherman http://ebooksecure.com/product/ebook-pdf-environmental-science- and-sustainability-1st-edition-by-daniel-j-sherman/ (eBook PDF) Introduction to Statistics and Data Analysis 6th Edition http://ebooksecure.com/product/ebook-pdf-introduction-to- statistics-and-data-analysis-6th-edition/
- 3. An Introduction toCategorical Data Analysis 3rd Edition (eBook PDF) http://ebooksecure.com/product/an-introduction-to-categorical- data-analysis-3rd-edition-ebook-pdf/ Data, Statistics, and Useful Numbers for Environmental Sustainability: Bringing the Numbers to Life 1st Edition Benoit Cushman-Roisin - eBook PDF https://ebooksecure.com/download/data-statistics-and-useful- numbers-for-environmental-sustainability-bringing-the-numbers-to- life-ebook-pdf/ (eBook PDF) An Introduction to Statistical Methods & Data Analysis 7th http://ebooksecure.com/product/ebook-pdf-an-introduction-to- statistical-methods-data-analysis-7th/ Numerical Methods in Environmental Data Analysis 1st Edition- eBook PDF https://ebooksecure.com/download/numerical-methods-in- environmental-data-analysis-ebook-pdf/ Biomass, Biofuels, Biochemicals: Green-Economy: Systems Analysis for Sustainability 1st edition - eBook PDF https://ebooksecure.com/download/biomass-biofuels-biochemicals- green-economy-systems-analysis-for-sustainability-ebook-pdf/
- 5.
- 7. 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
- 8. 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
- 9. 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
- 10. 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
- 11. 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
- 12. 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
- 13. 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
- 14. 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
- 15. 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
- 17. 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
- 18. 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).
- 19. 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
- 20. 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
- 21. 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
- 22. 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.
- 23. 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.
- 24. 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
- 25. 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
- 26. 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
- 27. 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.
- 28. 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
- 29. 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.
- 30. 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).
- 31. 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
- 33. 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.
- 34. 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).
- 35. 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).
- 36. 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).
- 37. 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).
- 38. 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.
- 39. 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.
- 40. 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.
- 41. 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.
- 42. 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.
- 43. 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.
- 45. 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
- 46. 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
- 47. 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
- 48. 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.
- 49. Another Random ScribdDocument with Unrelated Content
- 50. What Temple saw inDublin. An amended proclamation, Oct. 29. The Very Rev. Henry Jones. of Scotch fishing boats in the bay, and five hundred men volunteered to land and be armed for the service of the State. The offer was accepted, but never acted on, for the fishermen were seized with a panic, put to sea, and never reappeared until the next year. The fugitives from Ulster soon began to pour into Dublin. Temple is open to criticism for his account of what happened in the northern province, but this is what he saw himself: ‘Many persons of good rank and quality, covered over with old rags, and some without any other covering than a little to hide their nakedness, some reverend ministers and others that had escaped with their lives sorely wounded. Wives came bitterly lamenting the murders of their husbands; mothers of their children, barbarously destroyed before their faces; poor infants ready to perish and pour out their souls in their mothers’ bosoms; some over-wearied with long travel, and so surbated, as they came creeping on their knees; others frozen up with cold, ready to give up the ghost in the streets; others overwhelmed with grief, distracted with their losses, lost also their senses.... But those of better quality, who could not frame themselves to be common beggars, crept into private places; and some of them, that had not private friends to relieve them, even wasted silently away, and so died without noise.... The greatest part of the women and children thus barbarously expelled out of their habitations perished in the city of Dublin; and so great numbers of them were brought to their graves, as all the churchyards within the whole town were of too narrow a compass to contain them.’ Two large additional burial grounds were set apart.[277] On October 29 the Lords Justices issued a second proclamation. The words ‘Irish Papists’ in the first had been misunderstood, and they now desired to confine it to the ‘old mere Irish in the province of Ulster’; and they straitly charged both Papists and Protestants on their allegiance to ‘forbear upbraiding matters of religion one against the
- 51. The Protestants at Belturbet. TheLords Justices mark time. other.’ They soon had authentic evidence of how the old mere Irish were behaving in one Ulster county. Dean Jones came to Dublin at the beginning of November with the Remonstrance of the O’Reillys, which Bedell had excused himself from carrying. ‘I must confess,’ says Jones, ‘the task was such as was in every respect improper for me to undergo ... but chiefly considering that thereby I might gain the opportunity of laying open to the Lords what I had observed ... which by letters could not so safely be delivered, I did therefore accept.’ The O’Reillys declared that the outbreak was caused by oppression and by the fear of worse oppression; that there was no intention to rebel against the King; and that the people had attacked the English settlers without their orders and against their will. To prevent greater disorders they had seized strong places for the King’s use, and they demanded liberty of conscience and security for their property. Jones saw clearly that the rising was general and that the native gentry had no wish to restrain it, and he could tell what had happened to the English inhabitants of Belturbet. Philip Mac Hugh O’Reilly and the rest had promised these people a safe passage, and had allowed them to carry away some of their property, which they were thus induced not to hide. In the town of Cavan they were attacked, the guard given by the O’Reillys joining in the treachery, and robbed of everything. ‘Some were killed, all stripped, some almost, others altogether naked, not respecting women and sucking infants, the Lady Butler faring herein as did others. Of these miserable creatures many perished by famine and cold, travelling naked through frost and snow, the rest recovering Dublin, where now many of them are among others, in the same distress for bread and clothes.’ After a week’s hesitation, the Lords Justices sent back an answer by Jones, whose wife and children remained as hostages. This he describes as ‘fair, but general and dilatory, suitable to the weak condition of affairs in Dublin, the safety whereof wholly depending upon the gain of time.’ The Government yielded no point of importance. They reminded the remonstrants that fortresses could not be legally seized without orders from the King, and that the rebels had falsely
- 52. State of thePale. Lord Gormanston. Sir N. Barnewall. Sir T. Nugent. Sir C. Bellew. The Earl of Kildare. professed to have such orders. If, however, the inhabitants of the county Cavan would peaceably return to their own dwellings, restore every possible article to its plundered owner, and abstain from all hostile acts in future, then the Lords Justices would forward their petition to his Majesty and ‘humbly seek his royal pleasure therein.’ The O’Reillys were in the meantime preparing to attack Dublin in force.[278] As regards the gentry of the Pale, Roman Catholics for the most part, the Lords Justices were in a difficult position. By mistrusting them they ran the risk of driving them into rebellion; by trusting them they increased their power for mischief, should they be already tainted. For the moment the first danger seemed the greater of the two, and commissions as governors of counties with plenary powers were accordingly issued to several of them, by which they were authorised to proceed by martial law against the rebels, ‘hanging them till they be dead as hath been accustomed in time of open rebellion,’ destroying or sparing their houses and territories according to their discretion. They were also empowered to grant protections. Viscount Gormanston was thus made governor of Meath, and arms were given him for 500 men. He was in open rebellion a few weeks later. Sir Nicholas Barnewall of Turvey, afterwards created Viscount Kingsland by Charles I., became governor of the county of Dublin, and had arms for 300 men. Barnewall was a good deal involved in political intrigues, but soon fled to England to avoid taking arms against the Government. A commission as governor of Westmeath and arms for 300 men were given to Sir Thomas Nugent, who afterwards tried to fill the difficult part of neutral. Sir Christopher Bellew was governor of Louth, with arms for 300, but he very soon joined the Irish. To George Earl of Kildare, Cork’s son-in- law, his own county was entrusted and arms for 300; but he was a
- 53. Ormonde made general. Sir H.Tichborne. Protestant and suffered severely for his loyalty, while he was quite unable to curb his neighbours. Finding after a time that the arms given out would, if used at all, be used against them, the Lords Justices endeavoured to get them back, but they recovered only 950 out of 1700, and the enemy had the rest.[279] Ormonde was at his own house at Carrick-on-Suir when the rebellion broke out. The Lords Justices sent for him at once, and the first letter being delayed in transmission, a second was sent with a commission to him and Mountgarret to govern the county of Kilkenny and to take such other precautions as were possible. The gentry met at Kilkenny and offered to raise 240 foot and 50 horse, while Callan and other towns made similar promises. There were, however, no arms, and the Lords Justices would give none out of the stores. Before purchases could be made in England the situation was greatly changed. Ormonde arrived at Dublin with his troop early at the end of the first week in November, and on the 10th Sir Patrick Wemyss returned from Edinburgh with his nomination as Lieutenant-General, to command the army as he had done in Strafford’s time. The Lords Justices made out his commission next day, with warrant to execute martial law, but without prejudice to Leicester’s authority as Lord Lieutenant. It was not till six months later that the King gave him power to appoint subordinate officers according to the ‘constant practice and custom of former times,’ it having by then become evident that Leicester would not reside in Ireland. The defence of Drogheda had already been provided for by Sir Henry Tichborne, who was living at Dunshaughly, near Finglas, and who had brought his family into Dublin on the first day, having already ‘scattered a parcel of rogues’ that threatened his country house. Having received a commission from the Lords Justices, he raised and armed 1000 men in nine days among the Protestants who had left their homes, and with this regiment he entered Drogheda on November 4. Three additional companies were sent to him a few days later.[280]
- 54. Ormonde disagrees with the LordsJustices. The Irish Parliament after the outbreak. Both Houses protest against the rising. Vain hopes of peace. Prorogation, Nov. 17, 1641. One of Ormonde’s first acts as general was to commission Lord Lambert, Sir Charles Coote, and Sir Piers Crosbie to raise regiments of 1000 men each, and thirteen others to raise independent companies of 100 each. The ranks were filled in a few days, for all business was at a standstill, and Protestant fugitives poured in in great numbers. There were 1500 disciplined men of the old army about Dublin. Strafford had left a fine train of field artillery with arms, tents, and all necessaries for 10,000 men. Under these circumstances Ormonde was for pushing on, and putting down the northern rebellion at once. To this the Lords Justices would not consent, and it may be that they were jealous of their general; but it must be confessed that there was also something to be said for a cautious policy. With the Pale evidently disaffected Dublin could not be considered as very safe.[281] When the rebellion broke out the Lords Justices by their own authority prorogued Parliament till February 24, fearing a concourse of people to Dublin, and also because the state of Ulster made it almost certain that there would not be a Protestant majority. The gentry of the Pale, and the Roman Catholic party generally, protested strongly, and there were doubts about the legality of the prorogation. Some lawyers held that Parliament would be dissolved by the mere fact of not meeting on the appointed day. To get over the difficulty the Lords Justices agreed that Parliament should meet as originally announced, but that it should sit only for one day, and should then be prorogued to a date earlier than February 24. Ormonde and some others were in favour of a regular session, but they were overruled by the official members of the Council. Parliament met accordingly on November 9, and immediately adjourned till the 16th, so as to give time for private negotiations. The attendance was thin in both Houses, partly on account of the state of the country and partly because many thought
- 55. Leicester Lord Lieutenant. He nevercame to Ireland. The rebellion reported to the English Parliament. that the prorogation till February was still in force. Mr. Cadowgan significantly remarked that ‘many members of the House are traitors, and whether they come or not it is not material.’ There was a great military display about the Castle gates, according to the precedent created by Strafford, and offence was taken at this; but the two Houses agreed to a protestation against those who, ‘contrary to their duty and loyalty to his Majesty, and against the laws of God, and the fundamental laws of the realm, have traitorously and rebelliously raised arms, have seized on some of his Majesty’s forts and castles, and dispossessed many of his Majesty’s faithful subjects of their houses, lands, and goods, and have slain many of them, and committed other cruel and inhumane outrages and acts of hostility within the realm.’ And the Lords and Commons pledged themselves to ‘take up arms and with their lives and fortunes suppress them and their attempts.’ There was some grumbling about the words ‘traitorously and rebelliously’ on the principle that birds are not to be caught by throwing stones at them, but the majority thought the Ulster rebels past praying for, and the protest was agreed to without a division. There was also unanimity in appointing a joint committee, fairly representing different sections, with power, subject to royal or viceregal consent, to confer with the Ulster people. Two days were occupied in these discussions, and on the evening of the 17th the Lords Justices prorogued Parliament till January 11. When that day came things had gone far beyond the parliamentary stage.[282] The Earl of Leicester was appointed Lord Lieutenant early in June 1641, and the Lords Justices were directed by the King to furnish him with copies of all their instructions. He remained in England, and to him the Irish Government addressed their account of the outbreak. This was brought over by Owen O’Connolly, received on or before October 31, and at once communicated to the Privy Council, who had a Sunday sitting. On Monday, November 1, the Upper House did not sit in the morning, ‘for,’ says Clarendon, ‘it was All Saints’ Day, which the
- 56. The news reaches theKing, Oct. 27. Letter from the O’Farrells. Catholic grievances represented to the King. Lords yet kept holy, though the Commons had reformed it.’ To the House of Commons accordingly the Privy Council proceeded in a body, headed by the Lord Keeper. There was no precedent for such a visitation, but after a short discussion chairs were placed in the body of the House and Leicester, with his hat off, read the Lords Justices’ letter of October 25. Clarendon testifies from personal knowledge that the rebellion was odious to the King, and confidently asserts that none of the parliamentary leaders ‘originally and intentionally contributed thereunto,’ though he believes that their conduct afterwards added fuel to the flame. When the Privy Councillors had withdrawn the House went into committee, Mr. Whitelock in the chair, and drew up heads for a conference with the Peers. As to money they resolved to borrow 50,000l., giving full security, and to pay O’Connolly 500l. down with a pension of 200l. until an estate of greater value could be provided. Resolutions were passed against Papists, and particularly for the banishment of the Queen’s Capuchins. The Lords met in the afternoon, and after this the two Houses acted together. Three days later the estimate for Ireland was raised to 200,000l., and Leicester was authorised to raise 3,500 foot and 600 horse, while arms were provided for a further levy. News of the outbreak came to the King at Edinburgh direct from Ulster four days before it reached the English Parliament. Tradition says that he was playing golf, and that he finished his game.[283] Lord Dillon of Costello, who was a professing Protestant, produced at the Council on November 10 a letter signed by twenty-six O’Farrells in county Longford. This paper is well written, and contains the usual pleas for religious equality, which modern readers will readily admit, though they were not according to the ideas of that day either at home or abroad. The O’Farrells had taken an oath of allegiance, but their sincerity is open to doubt, for they demanded ‘an act of oblivion and general pardon without restitution on account of goods taken in the times of this commotion.’ No government could
- 57. Weakness of the IrishGovernment. Relief comes but slowly. Monck, Grenville and Harcourt. possibly grant any such amnesty, and the suggestion came at a time when Ulster was in a blaze and when Dublin was crowded with Protestants who had escaped with their bare lives. Dillon and Taaffe were commissioned by the Roman Catholic lords to carry their grievances to the King. When returning with instructions they were stopped at Ware and their papers overhauled, the Lords Justices having warned their parliamentary friends.[284] The influence of Carte has led historians generally to think that the Lords Justices were either too desperately frightened to think of anything but their own safety, or that they let the rebellion gather head to suit the views of the English parliamentary party. There is not much evidence for either supposition. Just at the moment when the Pale was declaring against them they reported their destitute condition to Leicester. The troops were unpaid. At Dublin they had but 3000 foot and 200 horse, and the capital as well as Drogheda was surrounded by armed bands who had already made food scarce, and who threatened to cut off the water. A large extent had to be defended, and many of the inhabitants were not to be trusted. A crusade was being preached all over the country, and at Longford, notwithstanding the oath of the O’Farrells, a priest was reported to have given the signal for a massacre by ripping up the parson with his own hand. The mischief was spreading daily, and agitators industriously declared that no help would be sent from England. Ireland was not, however, forgotten, but Parliament, to whom the King had specially entrusted it, had its own business to do, and a popular assembly has no administrative energy. It was not till the last day of December that Sir Simon Harcourt landed with 1100 men. Three hundred more followed quickly, and George Monck with Leicester’s own regiment was not far behind. Grenville brought 400 horse about the same time. Harcourt had long military experience in the Low Countries, and had lately commanded a regiment in Scotland. He had a commission as Governor of Dublin, but Coote was in possession and was not
- 58. Sir Charles Coote. disturbed.Harcourt was very angry with the Lords Justices, but he got on well with Ormonde and did good service until his death.[285] The number of troops available in Dublin was small, but they were much better armed than the insurgents. It was thus a matter of policy to act on the offensive and clear the surrounding country, demolishing houses and castles where troublesome posts might be established. This work, cruel in itself, was performed in a very ruthless manner, and particular blame has always fallen upon Sir Charles Coote, whose ferocity seems to have been as conspicuous as his courage. One story told both by Bellings and Leyburn is that he called upon a countryman to blow into the mouth of his pistol, that the simple fellow obeyed, and that Coote shot him in that position. He never went to bed during a campaign, but kept himself ready for any alarm, and lost his life in a sally from Trim during a night attack at the head of only seventeen men, the place being beset by thousands.[286] FOOTNOTES: [268] Alice Thornton’s Autobiography; Irish Lords Journals, February 22, 1640-1; Petition of the Irish Committee to the King, Cal. State Papers, Ireland, 1640, addendum; Radcliffe’s answer to the Committee, ib. January 9, 1641, and their rejoinder, ib. February 12. [269] Irish Commons Journals, February 16, 1640-1. The queries, with the answers and declaration of the Commons, are in Nalson, ii. 572-589. [270] Irish Commons Journals, 1641, p. 211; Irish Lords Journals, February 27, March 4. [271] Irish Commons Journals, June 7, July 10. The story about the powder is from Borlase’s Rebellion, ed. 1680, p. 12; he is not a very good authority, but on this occasion is speaking of his father’s action.
- 59. [272] Examination ofHenry Macartan, quartermaster to Owen Roe O’Neill, February 12, 1641-2, Contemp. Hist. i. 396; Vane to the Lords Justices, March 16, 1640-1, Cox’s Hibernia Anglicana, ii. 65; Cole to the Lords Justices, October 11, 1641, printed in Nalson and elsewhere; Lords Justices and Council to Vane, June 30, 1641, State Papers, Ireland; Deposition as to the Multifarnham meeting, May 3, 1642 (misprinted 1641), in Hickson’s Seventeenth Century, ii. 355. Temple produces evidence as to the rebellion being threatened long before it actually happened, O’More himself having admitted as much, p. 103. Patrick O’Bryan of Fermanagh swore on January 29, 1641-2 ‘that he heard Colonel Plunket say that he knew of this plot eight years ago, but within these three years hath been more fully acquainted with it’—Somers Tracts, v. 586. Lieutenant Craven, who had been a prisoner with the Ulster Irish, was prepared to swear that on March 3, 1641-2, he had heard Bishop Heber Macmahon tell his friends that he had planned the rebellion years before, and knew from personal knowledge that all Catholic nations would help; urging them to persevere and extirpate heresy. Macmahon repeated this at Monaghan in January 1643-4—Carte MSS. vol. lxiii. f. 132. [273] Lord Maguire’s Relation, written by him in the Tower (after August 1642) printed from the Carte Papers in Contemp. Hist. i. 501. Parsons to Vane, August 3, State Papers, Ireland. Temple’s History is valuable here, for he was present in Dublin and signed the proclamation on October 23, Bellings, i. 7-11. [274] O’Connolly’s Deposition, October 22, in Temple’s History, with the author’s remarks, and his further Relation printed from a manuscript in Trinity College in Contemp. Hist., i. 357. [275] Chiefly from Temple’s History, where O’Connolly’s evidence, and the proclamation of October 23, are given in full. There is an independent account by Alice Thornton, Wandesford’s daughter, who was in Dublin at the time, aged fifteen. According to her O’Connolly swam the Liffey. ‘What shall I do for my wife?’ he asked the conspirators, and they answered ‘Hang her, for she was but an English dog; he might get better of his own country.’— Autobiography, Surtees Society, 1875. [276] Sir F. Willoughby’s narrative among the Trinity College MSS., 809-841, vol. xxxii. f. 178. [277] Temple, pp. 93-4. Macmahon’s Deposition, October 23, Contemp. Hist. i. Appx. xix. Lords Justices and Council to
- 60. Leicester, October 25,printed in Temple’s History and elsewhere. Macmahon’s latter evidence, ‘taken at the rack’ on March 22, 1641-2, gives further details regarding the Ulster conspirators, but he knew nothing about the Pale, and does not even mention O’More’s name. Reports of Maguire’s trial have been often printed. [278] Proclamation of October 29, 1641, in Temple and elsewhere. Dean Jones’s ‘Relation of the beginning and proceedings of the rebellion in Cavan, c.,’ was printed in London by order of the House of Commons in the spring of 1642, and reproduced in vol. v. of the Somers Tracts as well as in Gilbert’s Contemporary History, where the Cavan Remonstrance, received November 6, and the Lords Justices’ answer dated November 10, are also printed. Rosetti at Cologne heard that many Protestants had joined the rebels, which was certainly not true, though some pretended to do so. Roman Transcripts, R.O., December 10, 1641. Another paper from Cologne speaks of the rebels ‘quali vanno decapitando et appiccando li Protestanti che non gli vogliono assistere,’ ib. December 22. [279] Temple prints the commission to Gormanston as a specimen. Lords Justices and Council to Leicester, December 14, in Nalson, ii. 911. [280] Sir Henry Tichborne’s letter to his wife, printed with Temple’s History, Cork, 1766. Carte’s Ormonde, i. 193, and the King’s letters in vol. iii. Nos. 31 and 82. [281] Carte’s Ormonde, i. 192-5; Lords Justices to Ormonde, October 24, 1641, printed in Confederation and War, i. 227. [282] Bellings gives the two documents referred to. He was a member of this Parliament, and one of the Joint Committee. Irish Commons Journals. [283] Rushworth, iv. 398-406; Nicholas to the King, November 1, 1641, in Evelyn’s Correspondence; Macray’s edition of Clarendon’s History, i. 408; May’s Long Parliament, p. 127. May is a good authority for what happened in London, but for events in Ireland he depends chiefly on Temple. Lords Journals, November 1; Lang’s Hist. of Scotland, iii. 100; Vane to Nicholas, October 27, Nicholas Papers, i. 58. [284] Nalson, ii. 898; Rushworth, iv. 413; Diurnal Occurrences, December 20-25, 1641.
- 61. [285] Despatch ofDecember 14, in Nalson, ut sup. Monck’s letter from Chester, ib. 919, shows how little money Parliament had to spare. In clerical circles abroad it was rumoured a little later that Dublin would soon fall, and that five hundred Protestants who objected to the cross in baptism had been marked with it on the forehead and sent back to England—Roman Transcripts, R.O., February 2, 1642. Four letters from Sir Simon Harcourt, January 3, 1641-42 to March 21, in vol. i. of Harcourt Papers (private circulation). As late as September 16, 1642, Sir N. Loftus wrote from Dublin that the enfeebled garrison could not hold out for six weeks if seriously attacked. Food and ammunition were wanting, and the surviving soldiers sick or starving—Portland Papers, i. 700. [286] Bellings, i. xxxii. 35; George Leyburn’s Memoirs, Preface; Borlase’s Irish Rebellion, p. 104, ed. 1743. Coote was killed May 7, 1642; when the name occurs later the reference is to his son, also Sir Charles.
- 62. Outbreak in Ulster. Savage character ofthe contest. Contemporary accounts of the massacre. Later estimates. The number of victims cannot be ascertained. CHAPTER XX PROGRESS OF THE REBELLION ‘There are,’ says Hume, ‘three events in our history which may be regarded as touchstones of party men: an English Whig who asserts the reality of the popish plot, an Irish Catholic who denies the massacre in 1641, and a Scotch Jacobite who maintains the innocence of Queen Mary, must be considered as men beyond the reach of argument or reason, and must be left to their prejudices.’ The fact of a massacre cannot be denied, but its extent is quite another matter. There is no evidence of any general conspiracy of the Irish to destroy all the Protestants, but so far as Ulster was concerned there was no doubt one to regain the land and in so doing to expel the settlers. Rinuccini admitted that the northern Irish, though good Catholics, were often great savages; and it is not surprising that there should have been many murders, sometimes of the most atrocious character, and that a much larger number of lives should have been lost through starvation and exposure. It is also true that many acts of kindness were done by the successful insurgents, and that the retaliation of the English was cruel and indiscriminating. As to the number killed during the early part of the rebellion and before it assumed the dignity of civil war, it is impossible to form anything like a satisfactory estimate. Temple, whose book was published in 1646,
- 63. says that inthe first two years after the outbreak ‘300,000 British and Protestants were cruelly murdered in cold blood, destroyed some other way, or expelled out of their habitations according to the strict conjecture and computation of those who seemed best to understand the numbers of English planted in Ireland, besides those few that perished in the heat of fight during the war.’ The great exaggeration of this has been dwelt on by writers who wish to disparage Temple’s authority, but these enormous figures were generally believed in at the time. May, who depended partly on Temple, says ‘the innocent Protestants were upon a sudden disseized of their estates, and the persons of above 200,000 men, women, and children, murdered, many of them with exquisite and unheard of tortures, within the space of one month.’ Dr. Maxwell learned from the Irish themselves that their priests counted 154,000 killed during the first five months. The Jesuit Cornelius O’Mahony, writing in 1645, says it was admitted on all sides that 150,000 heretics had been killed up to that time; he exults in the fact, and thinks the number was really greater. Clarendon says 40,000 or 50,000 English Protestants were murdered at the very beginning of the rebellion. Petty was the first writer of repute who attempted anything like a critical estimate. He had a genius for statistics and he knew a great deal, but owing to the want of trustworthy data, even he can do little more than guess that ‘37,000 were massacred in the first year of tumults.’ So much for those who lived at or near the time; modern writers can scarcely be better informed, but may perhaps be more impartial. Froude, who was not inclined to minimise, thinks even Petty’s estimate too high, and quotes the account of an eye-witness who says 20,000 were killed or starved to death in about the first two months. Warner, who wrote in 1767, was inclined to adopt Peter Walsh’s estimate of 8000. Reid rejected the higher figures, but without venturing on any decided opinion, Lecky very truly said that certainty was unattainable, but was inclined to agree with Warner. Miss Hickson, who examined the depositions more closely than any other writer, said the same, but thought the number killed in the first three or four years of the war could hardly fall short of 25,000. The conclusion of the whole matter is that
- 64. The massacre in IslandMagee. The rising in Tyrone, Oct. 23, 1641. English tenants plundered. Murder of Protestants. several thousand Protestants were massacred, that the murders were not confined to one province or county, but occurred in almost every part of the island, that the retaliation was very savage, innocent persons often suffering for the guilty, and that great atrocities were committed on both sides. ‘The cause of the war,’ says Petty, ‘was a desire of the Romanists to recover the Church revenue, worth about 110,000l. per annum and of the common Irish to get all the Englishmen’s estates, and of the ten or twelve grandees of Ireland to get the empire of the whole.... But as for the bloodshed in the contest, God best knows who did occasion it.’ He thought the population of Ireland in 1641 was about 1,400,000, out of which only 210,000 were British.[287] One of the worst cases of retaliation was the massacre by Scots of many Roman Catholic inhabitants of Island Magee in Antrim, but it is necessary to point out that this took place in January 1642, because it has been asserted that it was the first act of violence and the real cause of the whole rebellion. Some of those who took part in the outrage were alive in 1653, and were then prosecuted by the Cromwellian Government.[288] Dublin was saved, but the rebellion broke out in Ulster upon the appointed day. According to Captain John Creichton, his grandfather’s house near Caledon in Tyrone was the first attacked. The rebellion certainly began upon Sir Phelim O’Neill’s property at Caledon or Kinard during the night of October 22, when O’Connolly was telling the Lords Justices what he had heard. William Skelton, who lived as a servant in Sir Phelim’s house, was ploughing in the afternoon when an Irish fellow servant came to him with about twenty companions and said that they had risen about religion. Armed only with cudgels, they attacked several of Sir Phelim’s English tenants, who were well-to-do and apparently well-beloved by their Irish neighbours, ‘and differed
- 65. Sir Phelim O’Neill atCharlemont. The Caulfield family. Dungannon, Mountjoy, Tanderagee and Newry taken Bishop Henry Leslie. not in anything, save only that the Irish went to mass, and the English to the Protestant church in Tinane, a mile from Kinard.’ Taken by surprise, the Protestants were easily disarmed, and robbed in the first instance only of such horses as would make troopers. All the English and Scots neighbours were thus plundered in detail, cattle, corn, furniture, and clothes being taken in succession. In about a fortnight the Irish began to murder the Protestants. Among those whom Skelton knew of his own knowledge to be killed in cold blood before the end of the year was ‘one Edward Boswell, who was come over but a year before from England, upon the invitation of the said Sir Phelim, his wife having nursed a child of the said Sir Phelim’s in London.’ Boswell’s wife and child were murdered at the same time, and seventeen others in Kinard itself, men, women, and children. Skelton and some others were saved by the intercession of Daniel Bawn, whose wife was an Englishman’s daughter.[289] While his English servant was ploughing at Kinard, Sir Phelim O’Neill was on his way to Charlemont with an armed party. He had invited himself to dinner and was hospitably received by Lady Caulfield and her son, who had not long succeeded to the peerage. In after days there was a family tradition that the butler, an old and trusty servant, was alarmed by the attitude of Sir Phelim’s followers and imparted his fears to his mistress. His advice was neglected, and when the meal was over he left the house and made the best of his way to Dublin. The Caulfields and the unsuspecting men who ought to have defended the fort were surprised and captured, and O’Neill occupied Dungannon the same night. Next day the O’Quins took Mountjoy, the O’Hanlons Tanderagee, and the Magennises Newry. All were surprised, and there was practically no resistance. In the course of the day a fugitive trooper came to Lisburn, where Henry Leslie, Bishop of Down, was living, with news of the disasters at Charlemont and Dungannon, and four hours later another runaway announced that
- 66. Fermanagh. Rory Maguire. Murders at Lisgooleand elsewhere. Treatment of the English Bible. Newry was taken. Leslie at once sent the news on to Lord Montgomery, who was at or near Newtownards, and to Lord Chichester at Belfast; and they both wrote to the King. Chichester said only one man had been slain, which has been adduced as a proof that there was no massacre, but he knew only what Leslie had told him, and there were no tidings from any point beyond Dungannon. Other districts could tell a very different tale. [290] Lord Maguire was a prisoner, but his brother Rory raised Fermanagh before any account of the doings in Dublin had come so far. The robbing and murdering began on October 23, and very soon the whole county was at the mercy of the rebels. Enniskillen was never taken, and it will be seen that walled towns, if well defended, were generally maintained. Alice Champion, whose husband was killed in her presence on the first day, heard the murderers say that ‘they had special orders from Lord Maguire not to spare him or any of the Crosses that were his followers and tenants.’ About twenty-four others were murdered at the same time, and Mrs. Champion afterwards heard them boast that they had ‘killed so many Englishmen that the grease or fat that remained on their swords might have made an Irish candle,’ ninety being despatched at Lisgoole alone. The latter massacre is also sworn to by an eye- witness. Anne Ogden’s husband was murdered in the same way. She was allowed to fly to Dublin with her two children, but all were stripped on the way, and the children afterwards died ‘through the torments of hunger and cold they endured on that journey.’ Edward Flack, a clergyman, was plundered and wounded on the 23rd, and his house burned. The rebels in this case vented some of their fury on his Bible, which they stamped upon in a puddle, saying ‘A plague on this book, it has bred all this quarrel,’ and hoping that all Bibles would have this or worse treatment within three weeks. Much more of the
- 67. Cavan. The O’Reillys. Pretended orders fromthe King. Colonel Richard Plunkett. same kind might be said, and the events sworn to in Fermanagh alone fully dispel the idea that there were no murders at the first outbreak.[291] In Cavan, where the O’Reillys were supreme, there were no murders at the very beginning. Here, as in other places, the first idea seems to have been to spare the Scots and not to kill the English unless they resisted their spoilers. On the night of October 23, the Rev. George Crichton, vicar of Lurgan, who lived at Virginia, was roused out of his first sleep by two neighbours, who told him of the rising further north. Many of the Protestant inhabitants fled into the fields, but Crichton thought it better to stand his ground, and very soon a messenger came from Captain Tirlogh McShane McPhilip O’Reilly, to say that the Irish would harm no Scot. Crichton perhaps profited also by the fact that ‘no man ever lost a penny by him in the Bishop’s Court, and none ever paid to him what he did owe,’ which may have been a result of Bedell’s influence. He went out and met this chief at Parta wood, about a mile to the east of the town. O’Reilly, who had some twenty-four men with him, announced that Dublin and all other strong places were taken, and that they ‘had directions from his Majesty to do all these things to curb the Parliament of England; for all the Catholics in England should have been compelled to go to Church, or else they should be all hanged before their own doors on Tuesday next.’ Crichton said he did not believe such a thing had been ever dreamed of, whereupon O’Reilly declared his intention of seizing all Protestant property and of killing anyone who resisted. Next morning Virginia was sacked accordingly, but no lives were taken, for no one made any defence. The canny Scots clergyman managed to keep the Irish in pretty good humour, lodged nine families in his own house, and provided food for the fugitives from Fermanagh who began to arrive in a few days. Many thousands from Ballyhaise, Belturbet and Cavan passed through Virginia on their way towards the Pale. Crichton obtained help from Colonel Richard Plunkett, who wept and blamed Rory Maguire for all.
- 68. Cavan and Belturbet. Philip MacHugh O’Reilly. Horrorsof a winter flight. On being asked whether the Irish had made a covenant he said, ‘Yea, the Scots have taught us our A B C; in the meantime he so trembled that he could scarce carry a cup of drink to his head.’ Nevertheless he boasted that Dublin was the only place not taken, that Geneva had fallen, and that there was war in England. Many of the wretched Fermanagh Protestants were wounded, and the state of their children was pitiable. The wounded were tended and milk provided for the children, Crichton telling his wife and family that it was their plain duty to stay, and that ‘in this trouble God had called them to do him that service.’ All this happened within the first week of the outbreak, and when the long stream of refugees seemed to have passed, Crichton and his family prepared to go; but they were detained, lest what they had to tell might be inconvenient. Protestants from the north continued to drop in for some time, and Crichton was allowed to relieve them until after the overthrow at Julianstown at the end of November. The O’Reillys took part in the affair, and their followers became bolder and less lenient.[292] Another clergyman, Henry Jones, Dean of Kilmore, was living at Bellananagh Castle, near Cavan, at the time of the outbreak. Philip MacHugh MacShane O’Reilly, member for the county, was the chosen leader of the Irish. The actual chief of the clan was Edmund O’Reilly, but the most active part was taken by his son, Miles O’Reilly, the high sheriff, a desperate ‘young man,’ who at once assumed his native name of Mulmore Mac Edmond. Under the pretence of raising the posse comitatus he sent bailiffs to the scattered houses of Protestants and collected their arms. He himself seized the arms at Farnham Castle, and took possession of Cloghoughter, with whose governor, Arthur Culme, he had been on terms of friendship. Next day, October 24, the sheriff proceeded to Belturbet, which was the principal English settlement and contained some 1500 Protestants. Sir Stephen Butler was dead, but his widow had married Mr. Edward Philpot and was living there with her five children. Sir Francis Hamilton, who was at Keilagh Castle, tried to
- 69. The O’Reillys were notunanimous. Doctor Henry Jones. Weakness of the Irish Government. Divisions among the Irish. organise some resistance, but Philip MacHugh O’Reilly took the settlers under his protection, and they gave up their arms. Yet Captain Ryves with some thirty horse had no difficulty in reaching the Pale by O’Daly’s Bridge on the Blackwater, and in occupying Ardbraccan for the Lords Justices. Cavan surrendered, and on the 29th Bellananagh, which was indefensible, surrendered to the sheriff’s uncle, Philip MacMulmore O’Reilly. It had been determined to clear all the English out of the county, and though Lady Butler with 1500 others were escorted as far as Cavan they were attacked just beyond the town, and stripped of everything. Those who did not die of exposure reached Dublin, to starve and shiver among the other fugitives there. Those who remained at Belturbet had a still worse fate.[293] The O’Reillys had always been more civilised than other natives of Ulster, and they almost seem to have felt that the Government must win in the end. Rose O’Neill, the wife of Philip MacHugh, wished to kill all the English and Scotch at Ballyhaise, but he would not allow it. ‘The day,’ he said, ‘may come when thou mayest be beholding to the poorest among them.’ With a view no doubt to that distant day, they resolved to petition the Lords Justices and to send an Englishman with the message. Bedell refused to go on account of his age and because his plundered flock could not spare him, but Jones, who in his time played many parts, thought it safer to do as he was asked. He left his wife and children as hostages and went to Dublin, with a memorial signed by seven O’Reillys which spoke of former misgovernment, and rumours that worse was to come. They protested their loyalty and desired the Lords Justices ‘to make remonstrance to his Majesty for us ... so that the liberties of our conscience may be secured unto us, and we eased of our other burdens in the civil government.’ The Lords Justices and eight Privy Councillors, of whom Ormonde was one, sent an answer, dealing in generalities ‘suitable to the weak condition of affairs in Dublin.’ The
- 70. Rising in Monaghan. Murder ofRichard Blayney. A sham royal commission. most they could promise was that if they would restore all the Cavan Protestants to their homes and properties and cease from further hostilities, that then their memorial should be forwarded to the King. On his return Jones found the O’Reillys preparing to invade the Pale. He managed to keep the Dublin Government well informed, at the same time dissuading the Irish from attacking the capital, whose means of defence he exaggerated. Drogheda, he said, was more assailable, and to Drogheda they determined to go. They mustered first at Virginia, where Mr. Crichton made friends with Philip MacHugh’s mother on the ground of common kinship with Argyle, ‘of which house it seemeth that she was well pleased that she was descended. This kindred stood me in great stead afterwards, for although it was far off and old, yet it bound the hands of the ruder sort from shedding my blood.’ Many lives, says Crichton, were also saved by the quarrels of the Irish among themselves. Philip MacHugh not only shielded his far away cousin, and others for his sake, but was evidently disinclined to the task in hand, regretted that he had not kept the Protestants safe at Belturbet, ‘blamed Rory Maguire for threatening to kill and burn them, and cursed those among the English that gave them counsel to leave their habitations.’ Crichton thought O’Reilly a deep dissembler, but he should have the credit for comparative humanity. He and others seem to have thought that the war was on the point of breaking out in England, and that it would be impossible to send any troops to Ireland for years to come.[294] In Monaghan there was a general rising on October 23, but a number of murders were committed during the first few days, and the Macmahons behaved worse than the O’Reillys. Richard Blayney, member for the county, and commissioner of subsidies, was hanged by Sir Phelim O’Neill’s direct orders, and his dead body barbarously treated. At Carrickmacross Essex’s bailiff, Patrick McLoughlin Macmahon, took the lead among the local rebels, and about 600l. of the great absentee’s rents came into their hands.
- 71. The Portadown massacre, about Nov.1, 1641. The church at Blackwater. Alleged apparitions. Investigation by Owen Roe O’Neill. In Monaghan, as elsewhere, the Irish professed to do everything by the King’s orders, but at Armagh Sir Phelim O’Neill professed to show the actual commission with a broad seal to it, adding that he would be a traitor if he acted of his own accord. ‘We are a sold people,’ said an Englishman who witnessed the scene. A number of Protestants took refuge in the cathedral, but they had to surrender, and being stripped and robbed were sent to keep the Caulfields company at Charlemont. A miscellaneous collection of Protestants, including many children and poor people, from whom no ransom could be expected, were driven to the bridge of Portadown and there murdered.[295] The Portadown massacre has been more discussed perhaps than any episode in the Irish rebellion, and it has left behind it an ineffaceable impression of horror. The victims were only a part of those murdered in the county of Armagh, but more than 100—one account says 160—were killed at one time—and the affair was carefully planned beforehand. The chief actor was Captain Manus O’Cahan, but many of the sufferers had received passes from Sir Phelim himself. O’Cahan and his men, Mrs. Price deposed, forced and drove all those prisoners, and amongst them the deponent’s five children, by name Adam, John, Anne, Mary, and Jane Price, off the bridge into the water. Those that could swim were shot or forced back into the river. When Owen Roe O’Neill came to the country he asked in Mrs. Price’s hearing how many Protestants the rebels had drowned at Portadown, and they said 400. If this is correct the cruel work on the Bann must have continued for some time. They also said that those drowned in the Blackwater were too many to count, and that the number thrust into lakes and bog-holes could not even be guessed at. On November 17 they burned the church at Blackwaterstown with a crowd of Protestants in it, ‘whose cries being exceeding loud and fearful, the rebels used to delight much in a scornful manner to imitate them,
- 72. Bedell at Kilmore. Heis allowed to relieve many Protestants. He refuses to leave his post. and brag of their acts.’ Attempts have been made to discredit the evidence on the ground that Mrs. Price and others refer to apparitions at the scene of the Portadown massacre. Screams and cries are easily explained, for wolves and dogs fed undisturbed upon the unburied dead. But Mrs. Price says she actually saw a ghost when she visited the spot where her five children had been slaughtered, and that Owen Roe O’Neill, who came expressly to inform himself as to the alleged apparitions, was present with his men, who saw it also. It was twilight, and ‘upon a sudden, there appeared unto them a vision, or spirit assuming the shape of a woman, waist high, upright in the water, naked, her hair dishevelled, very white, and her eyes seeming to twinkle in her head, and her skin as white as snow; which spirit or vision, seeming to stand upright in the water, divulged, and often repeated the word “Revenge! Revenge! Revenge!”’ O’Neill sent a priest and a friar to question the figure both in English and Latin, but it answered nothing. He afterwards sent a trumpet to the nearest English force for a Protestant clergyman, by whom the same figure was seen and the cries of ‘Revenge!’ heard, but Mrs. Price does not say she was present on this occasion. The evidence of this lady shows no marks of a wandering mind, and yet it is evident that she believed in an apparition. It is quite possible that some crazed woman who had lost all that was dear to her may have haunted the spot and cried for vengeance, but in any case a belief in ghosts was still general in those days, and especially in Ireland. The evidence as to the massacre is overwhelming.[296] Bedell was at Kilmore when the rebellion broke out. The Protestants were surprised, but it was remembered afterwards that there had been an invasion or migration of rats, and that caterpillars had appeared in unusual numbers. It was more to the purpose that a crack-brained Irish scholar who wandered from house to house was heard frequently to exclaim, ‘Where is King Charles now?’ and that he wrote in an old almanac ‘We doubt not
- 73. He is imprisoned atLough Oughter. He is released. Fate of his library. of France and Spain in this action’—words which he may have heard in some conventicle of the Irish. The fugitive Protestants crowded to Kilmore, where they were all sheltered and fed, the better sort in the palace and the rest in out-buildings. The bishop’s son, who was there, likens the stream of poor stripped people to ‘Job’s messengers bringing one sad report after another without intermission.’ After a few days, Edmund O’Reilly, the sheriff’s father, ordered Bedell to dismiss his guests, who were about 200, chiefly old people, women and children. On his refusal those in the detached buildings were attacked at night and driven out almost naked into the cold and darkness. The bishop’s cattle were seized, but he had stored some grain in the house, and was still able in an irregular way to relieve many stray Protestants. On one occasion he sallied forth to rescue some of them, and two muskets were placed against his breast. He bade them fire, but they went away, and still for some time the palace walls were allowed to shelter those within. One of these was John Parker, afterwards Bishop of Elphin, who had fled from his living at Belturbet. ‘For the space of three weeks,’ says Parker, ‘we enjoyed a heaven upon earth, much of our time spent in prayer, reading God’s word, and in good conference; inasmuch as I have since oft professed my willingness to undergo (if my heart did not deceive me) another Irish stripping to enjoy a conversation with so learned and holy a man.’ Church service was regularly continued, but the investment of the house became closer, Bedell resolutely refusing to quit his post, although the Irish urged him to leave the country and promised all his company safe convoy to Dublin. His own children wished him to accept this offer, and it is probable that the Bishop himself and possible that his guests might have reached the capital in safety, but the experience of others had not been encouraging. Some prisoners having been taken by the Scottish garrisons at Keilagh and Croghan, and Eugene Swiney, the rival Bishop of Kilmore, pressing for restoration to his palace, Bedell and his family were at last expelled. ‘I arrest you,’ said Edmund O’Reilly, laying his hand on the Bishop’s shoulder, ‘in the King’s name.’ Having
- 74. Bedell’s death, first arrangedthat the Church plate provided by himself should be handed over to the other Bishop, Bedell was conveyed to a castle upon an island in Lough Oughter. He was allowed to take his money with him, and his two sons with their wives accompanied him. They were well treated on the whole, but the castle had neither glass nor shutters to the windows, and they spent a cold Christmas. Some of the prisoners were in irons, and Bedell earnestly desired to share their fate, but this was refused. The party were dependent on the Irish for news, and at first they heard much of the disaster at Julianstown and of the certain fall of Drogheda. But an English prisoner who knew Irish listened one night through a chink in the floor, and heard a soldier fresh from Drogheda tell the guard that the siege was raised. ‘The bullets,’ he said, ‘poured down as thick from the walls as if one should take a fire-pan full of coals and pour them down upon the hearth, which he acted before them, sitting altogether at the fire. And for his own part he said he would be hanged before he would go forth again upon such a piece of service.’ At last Bedell and his sons were exchanged for some of those in the hands of the Scots, and released from the castle. The Bishop’s remaining days were spent in the houses of Dennis Sheridan, a clergyman ordained and beneficed by him, whose vicarage was near at hand. Sheridan, though a Protestant, was a Celt, and respect for his clan secured him a certain toleration. He was instrumental in saving some of Bedell’s books, among them a Hebrew Bible, now at Emmanuel College, Cambridge, and the Irish version of the Old Testament which had cost so much trouble, and which was not destined to be printed for yet another generation. Most of the books and manuscripts were taken away first by friars and afterwards by English soldiers, who sold them. ‘Certain of the Bishop’s sermons,’ says his son, ‘were preached in Dublin, and heard there by some of his near relations, that had formerly heard them from his own mouth: some even of the episcopal order were not innocent in this case.’ Bedell remained for some weeks with Sheridan, preaching often and praying with those that were
- 75. Welcome to OurBookstore - The Ultimate Destination for Book Lovers Are you passionate about testbank and eager to explore new worlds of knowledge? At our website, we offer a vast collection of books that cater to every interest and age group. From classic literature to specialized publications, self-help books, and children’s stories, we have it all! Each book is a gateway to new adventures, helping you expand your knowledge and nourish your soul Experience Convenient and Enjoyable Book Shopping Our website is more than just an online bookstore—it’s a bridge connecting readers to the timeless values of culture and wisdom. With a sleek and user-friendly interface and a smart search system, you can find your favorite books quickly and easily. Enjoy special promotions, fast home delivery, and a seamless shopping experience that saves you time and enhances your love for reading. Let us accompany you on the journey of exploring knowledge and personal growth! ebooksecure.com