Mapping Global Croplands for Food Security

U.S. Geological Survey
U.S. Department of Interior
Global croplands for food and water security in the twenty-first century using
Earth Observing (EO) big-data analytics, machine learning, and cloud computing
May 8-10, 2023, New Techniques for Water Management II @
2023 Water for Food Global Conference, Nebraska , USA
Dr. Prasad S.Thenkabail, Dr. Pardhasaradhi G. Teluguntla
(pthenkabail@usgs.gov; pteluguntla@usgs.gov)
Mr. Adam Oliphant, Dr. Itiya Aneece, and Mr. Daniel Foley
United States Geological Survey (USGS)

Background
https://www.usgs.gov/apps/croplands/app/map (www.croplands.org)
https://lpdaac.usgs.gov/news/release-of-gfsad-30-meter-cropland-extent-products/
https://lpdaac.usgs.gov/news/release-of-lgrip30-data-product/
www.usgs.gov/wgsc/gfsad30

Global Croplands for Food and Water Security in the Twenty-first Century
Importance of Global Croplands for Life and Livelihoods
Requires
Energy
Availability of
Land
for Food Security
Availability of
Water
for Food Security
Contributes to
Nutritional needs
Needed for
Feeding animals
Adapting to
Climate Change
for Food Security
Global
Croplands
Affects
Economy,
Population,
Peace
Leveraging
Virtual Land &
Water
for Food Security

Background: Global Population Dynamics
Source: Data from the History Database of the Global Environment, HYDE V3.2

1540 Mha
240 Mha
Sources:
Cao et al., 2021, Goldewijk et al., 2010, Ramankutty and Foley,
1999, HYDE 2.0, Houghton et al., 1983, Thenkabail et al., 2015
140 Mha
36 Mha
84 Mha
52 Mha
145 Mha
200 Mha
293 Mha
840 Mha
260 Mha
429 Mha
Background: Global Cropland Dynamics
1800 Mha in 2020

Current State-of-Art

….typically, in 250m-10 km & as a result precision and accuracies of
croplands location as well as cropland areas are lacking
Existing Irrigated and Rainfed Cropland Products @ Coarse Resolution
Sno Name
*
Institution
**
Sensors
***
Resolution, Nominal Time, Nominal Classes Reference
(meters) (year)
A. Global cropland products
1 GCEP30 USGS Landsat 7 & 8 30 2021 Croplands Thenkabail et al. (2021)
2 UCL UCDL Multiple 250 2010 Croplands, irrigated, rainfed Waldner et al. (2016)
3 MODIS Cropland SDSU & UMD MODIS 250 2000 Croplands Pitman et al. (2010)
4 GRIPC BU MODIS 500 2005 Croplands, irrigated, rainfed Salmon et al. (2015)
5 GFSADCD1KM USGS AVHRR 1000 2010 Crop dominance, irrigated, rainfed Thenkabail et al. (2012)
6 GFSADCM1KM USGS AVHRR, MODIS, Landsat 1000 2010 Croplands, irrigated, rainfed Teluguntla et al. (2015)
7 GIAM IWMI AVHRR, SPOT VEG 1000-10,000 2000 Croplands, irrigated Thenkabail et al. (2009)
8 GMRCA IWMI AVHRR, SPOT VEG 1000-10,000 2000 Croplands, rainfed Biradar et al. (2009)
9 MIRCA2000 UB 10,000 2000 Croplands. irrigated Portman, Seibert & Doll (2009)
10 SAGE-Crop UW 100,000 2000 Croplands, crop dominance Monfreda, Ramankutty & Foley (2008)
11 SAGE-Agri UW 100,000 2000 Croplands, crop dominance Ramankutty et al.(2008)
B. Global LULC products in which cropland classes exist
12 WorldCover10V2 ESA Sentnel-2 10 2021 LULC Zanaga et al. (2022)
13 WorldCover10V1 ESA Sentnel-2 10 2020 LULC Zanaga et al. (2021), Karra et al. (2021)
14 Globeland30 NGCC Landsat 7 30 2010 LULC Chen et al. (2015)
15 FROM-GLC CAS Landsat 7 30 2000 Croplands Yu et al. (2013)
16 FROMGC CESS Landsat 7 30 Circa 2010 LULC Gong et al. (2013)
17 CGLS-LC100 Copernicus PROBA-V 100 2015 LULC Buchhorn et al.(2020)
18 MODIS-JRC JRC/MARS MODIS, Landsat 250 2009 LULC Vancutsem et al. (2012)
19 Globcover ESA MERIS 300 2005, 2009 LULC Defourny et al. (2009)
20 MCD12Q1 NASA MODIS 500 2004 - now LULC Leroux et al. (2014)
21 GLC BU MODIS 500 LULC Friedal et al. (2010)
22 DISCover USGS AVHRR 1000 1992-93 LULC Loveland et al.(2000)
23 LULC 2000 USGS AVHRR 2000 2000 LULC Soulard et al. (2014)
24 GLC 2000 JRC SPOT 1/112° 2000 LULC Fritz et al. (2010)
Note:
* = GCEP30= Global cropland extent product@30m; UCL = Unified Cropland Layer; GFSADCM1KM= Global Food Security-Support Analysis Data Cropland Mask @ 1-km;
GRIPC= Global Rainfed, Irrigated, and Paddy Croplands;GFSADCD1KM= Global Food Security-Support Analysis Data Crop Dominance @ 1-km;
MIRCA2000= Global monthly irrigated and rainfed crop areas around the year 2000; GIAM- Global Irrigated Area Map; GMRCA= Global Map of Rainfed Cropland Areas;
SAGE= Center for Sustainability and the Global Environment;LULC = land use and land cover: MCD12Q1= MODIS Land Cover Type product;
FROM-GLC = Fine Resolution Observation and Monitoring of Global Land Cover;FROMGC = Finer Resolution Observation and Monitoring of Global Land Cover
CGLS= Copernicus Global Land Cover Layers; MODIS-JRC = MODIS Joint Research center; GLC = Global Land Cover.
** = UCDL = Université Catholique de Louvain, Belgium; USGS = United States Geological Survey; BU = Boston University SDSU = South Dakota State University.
UMD = University of Maryland; UB = University of Bonn; IWMI= International Water Management Institute; UW = University of Wisconsin.
NGCC = National Geomatics Center of China; NASA = National Aeronautics and Space Administration; CAS = Chinese Academy of Sciences.
MARS = Monitoring Agricultural Resources; ESA = The European Space Agency.
*** = AVHRR = Advanced Very High Resolution radiometer; MODIS = Moderate Resolution Imaging Spectroradiometer; SPOT = Satellite Pour l’Observation de la Terre
MERIS = Medium-Spectral Resolution, Imaging Spectrometer; Proba-V sensor is a European follow-on to the SPOT VGT mission

GRIP1000-2class map. An aggregated 2-class SPOT-Vegetation, AVHRR and other ancillary data
derived global rainfed and irrigated-area product @ 1000m (GRIP1000) developed using multiple-
satellite sensor time-series data and ancillary data as reported in Teluguntla et al. (2016, 2015).
GFSAD @ 1KM (GRIP1000): 2 Class Map

Thenkabail et al., 2012. PE&RS, Vol. 78, No.8 using data from Glieck, 2011
How Countries consume water for Agriculture to grow food
Predicting and Preventing Crisis in Cropland Water Use in a Changing Climate through
Agricultural Crop Water Use

Foley, D.J., Thenkabail, P.S., Aneece, I., and Teluguntla, P. 2019. A meta-analysis of global crop water productivity of three
leading world crops (wheat, corn, and rice) in the irrigated areas over three decades, International Journal of Digital Earth,
DOI: 10.1080/17538947.2019.1651912 Link to this article: https://doi.org/10.1080/17538947.2019.1651912 IP-105160
Predicting and Preventing Crisis in Cropland Water Use in a Changing Climate through
Global Food productivity Increases through Crop Water Productivity: Wheat Crop (kgm3)

Goal and Products

Global Food Security Support Analysis Data (GFSAD) Project
Mapping Global Irrigated and Rainfed Croplands
The overarching goal of GFSAD project is to make
effective use of USGS and NASA multiple satellite sensor
Petabyte-volume Planet-scale data on the cloud using
automated machine learning algorithms to derive a
series of nominal 30-m cropland products to help us
understand, assess, and manage global agriculture,
water, and food security.

1. Global Cropland Extent Product
2. Global Irrigated versus
Rainfed Croplands
4. Global Major Crop Types
3. Global Cropping Intensities
Baseline Cropland Products for Food and Water Security
Four
Baseline
Cropland
Products

Within and Across Field Variability
1. Croplands versus non-croplands
2. Irrigated versus Rainfed croplands
3. Cropping intensities
4. Crop types
5. Crop growth stages
6. Cropland fallows versus croplands
7. Crop productivity
8. Crop water use
9. Crop water productivity
10. Species composition (e.g., Oryza sativa, O. glaberrima);
11. Vegetation or crop type (e.g., soybeans vs. corn);
12. Biophysical properties (e.g., LAI, biomass, yield, density);
13. Biochemical properties (e.g, Carotenoids, Chlorophyll);
14. Disease and stress (e.g., insect infestation, drought),
15. Nutrients (e.g., Nitrogen),
16. Moisture (e.g., leaf moisture),
17. Light use efficiency,
18. Net primary productivity and so on.
Global Cropland Products from Satellites Allow us to
Model, Map, Monitor Every Farm in the World for Every Parameter

Mapping the World’s Croplands
from Earth Observing (EO) Satellites

Landsat-derived Global Cropland Extent Product at 30m (LGCEP30)
Understanding Cropland Extent
Definitions: Net Cropland Extent Includes
Croplands + cropland fallows + Plantations
All croplands cultivated for food, feed, and fiber, including plantations (e.g., orchards,
vineyards, coffee, tea, rubber) or croplands that are left fallow

Landsat-derived Global Rainfed and Irrigated-Cropland Product at 30m (LGRIP30)
Definitions: Irrigated Areas
• Irrigation is defined as artificial application of any amount of water to
overcome crop water stress.
• Irrigated areas are those areas which are irrigated one or more times
during crop growing season. This definition is key to mapping. Even if
croplands get artificial watering one or more times during the growing
season, it is defined as irrigated areas for purpose of this mapping.
• Artificial irrigation can come from various sources such as canals
delivering water from reservoirs (e.g., from major, medium, or minor
dams or tanks), and ground water pumping. Water is, typically,
delivered by gravity (e.g., river diversions or canals delivering water
directly by flooding) or drip or sprinklers or any other means
including hand carried and delivered.

Irrigated Areas Sample Illustrations
Center Pivots (Texas, USA) Run of the river (Ganges, India) Major Reservoirs (Mysore, India)
River deltas (Tamil Nadu, India)
Canal systems (California, USA) Deltas, Reservoirs Riverine systems (Egypt) Flood plains (Pakistan & India)
Minor reservoirs (Central India) Canal systems (Rajasthan, India)

Yellow river basin, China Xin Jiang, China Yangon, Myanmar
Bali, Indonesia N of Cape Town, South Africa
Mekong River Delta, Vietnam Chao Phraya Basin near Bangkok, Thailand Java, Indonesia
South of Casablanca, Morocco
Irrigated Areas Sample Illustrations

Definitions: Rainfed Areas
Cropland areas that have no irrigation whatsoever and
are fully dependent on precipitation (rainfall or moisture
retained from snowfall).
Thereby, rainfed areas are those croplands that are not
watered by artificial means but purely dependent on
nature.

Mid West, Ohio, USA Winnipeg, Canada Pampas, Argentina
Sahel, Niger
NE Dublin, Ireland
Wales, UK West of Perth, Australia
Sudan Savanna, Nigeria Derzhavinskiy, Kazakhstan
Rainfed Areas Sample Illustrations

North of Birmingham, UK North of Nantes, France N&NW Hamburg, Germany
Rajasthan, India
East of Nairobi, Kenya
North of Dakar, Senegal East of Bulawayo, Zimbabwe
South of Moscow, Russia NE China
Rainfed Areas Sample Illustrations

Methods
Global Food Security-support Analysis Data (GFSAD) Products

3. Machine LearningDeep Learning Algorithms
Pixel-based supervised MLAs) along with
Object-based hierarchical segmentation (HSEG)
A. Random forest algorithms
B. Support vector machines
C. Automated cropland classification algorithms
D. Spectral matching techniques
E. Decision Tree algorithms
F. Linear discriminant analysis
G. Principal component analysis, change detection analysis
H. kMeans, Isoclass clustering
I. Classification and Regression Tree (CART)
J. Tree-based regression algorithm
K. Phenology based methods
L. Fourier harmonic analysis
M. Hierarchical segmentation
4. Cloud Computing (e.g., Google Earth Engine)
1. Multiple Satellite Image Acquisitions
2. Analysis Ready Data (ARD)
Credit: NASAEarthdataCEOS
Mega-file Data-cube: Harmonized;Standardized
USGS Landsat Collection 2 release for the World: 49 years of data
• https://www.usgs.gov/news/usgs-releases-most-advanced-landsat-archive-date
Landsat-derived Global Cropland Mapping Methods
A “paradigm-shift” in how we Produce Global Cropland Maps
5. Automation and Artificial Intelligence

Global Mapping Methods: Croplands, Irrigated Vs. Rainfed
Machine Learning Algorithms (MLAs)

74 Agroecological Zones (AEZs) of the World
Derived off: https://gaez.fao.org/

Analysis Ready Data (ARD) Data Cubes
Also see: USGS Landsat Collection 2 release for the World: 49 years of data
• https://www.usgs.gov/news/usgs-releases-most-advanced-landsat-archive-date
• Landsat-8, 9, Sentinel-2, and NASA’s Harmonized Landsat Sentinel-2 (HLS) Landsat product (HLSL30) for 2013-
present and HLS S2 product (HLSS30) for 2015-present (Masek et al., 2022)

72 AEZ’s
Landsat 2013-2016
Gorelick, N., et al., 2017. Google Earth Engine: Planetary-
scale geospatial analysis for everyone, Remote Sensing of
Environment, http://dx.doi.org/10.1016/j.rse.2017.06.031
Analysis Ready Data, Machine Learning, Cloud computing

Table. Characteristics of some of the twenty-first century
satellites1.
PlanetScope Doves RapidEye SkySat Landsat 8 Sentinel-2
Number of
satellites 200+ 5 13 1 2
Spectral
resolution 4-8 bands 5 bands 5 bands (R, G, B, NIR, pan) 11 bands (VNIR, SWIR, TIR, pan) 13 bands (VNIR, SWIR)
Spatial
resolution 3-4 m 6.5 m 0.5-1 m 30 m multispectral, 15 m pan 10, 20, or 60 m depending on band
Temporal
resolution Daily at nadir
Daily off-nadir, 5.5
days at nadir 4-5 days 16 days 5 days
Image
dimension 24 by 7 km
77 km by up to
300 km 3.2 by 1.3 km 185 by 180 km 290 by 300 km
Radiometric
resolution 8 to 16 bit 8 to 16 bit 8 to 16 bit 12 to 16 bit 12 bit
Note:
1 = This Table was compiled by Dr. Itiya Aneece, Mendenhall Post-doctoral researcher @
Timelapse credit:
Will Marshall,
Planet Labs Inc.
200+ satellites gather 0.5 to 3
meter data daily for the entire
Planet. Realistically, weekly of
bi-weekly composites @ 3-m to
overcome cloud issues
Timelapse credit:Planet Labs Inc.
Planetscope 200+ satellites ; 350 M+ km2/day
Rapideye 5 satellites ; 6.5 M km2/day
Skysat 15 satellites ; 0.5 M km2/day
Source: Planet Labs Inc.
Note: 7 Petabytes currently; adding ~ 7+ terabytes daily
Sourcing Reference data from Hyperspatial (sub-meter to 5-m) Data of the World

Reference samples of irrigated areas from sub-meter to 5-m very high-resolution imagery (VHRI). These
data are used in :
1. Training machine learning and other algorithms,
2. Identifying and labeling classes, and
3. Validation of cropland products.
Sourcing Reference Data from Very High-Resolution Imagery: Irrigated Areas

Sourcing Reference Data for Training and Validation: Croplands
Illustrations for Australia (~2000 samples) and China (~4000 samples)
Australia: Ground-data (top); VHRI (bottom) China: Ground-data (top); VHRI (bottom)

Ground data samples of irrigated areas. These data were gathered from various sources and compiled into
database. These data are used in :
3. Validation of cropland products.
Sourcing Reference Data from the Ground: Irrigated Areas

Sourcing Reference Data from Very High-Resolution Imagery: Rainfed Areas
Reference samples of rainfed areas from sub-meter to 5-m very high-resolution imagery (VHRI). These
data are used in :

Ground data samples of rainfed areas. These data were gathered from various sources and compiled into
database. These data are used in :
Sourcing Reference Data from the Ground: : Rainfed Areas

Distribution of Reference training, testing, and validation data gathered from multiple sources either through ground data or
through sub-meter to 5-m very high-resolution imagery data (Teluguntla et al., 2023, Thenkabail et al., 2021, Congalton et al., 2017).
These data are used in :
Reference Data Synthesis: Croplands Vs Non-croplands for the World

Reference Data Synthesis: Irrigated cropland samples for the World

Reference Data Synthesis: Rainfed cropland samples for the World

Sampling for Knowledge generation for Croplands vs. Non-croplands

Knowledge generation for Machine Learning Algorithms to Map Croplands vs. Non-croplands

Croplands vs. Non-croplands Map for an area in Africa using Multiple Machine Learning Algorithms
(a) the pixel-based classification from random forest classifier; (b) the object-based RHSeg image segmentation
result; (c) the merged results with RHSeg segmentation result with pixel-based Random Forest classification; and
(d) a true color Google Earth Imagery for reference.

Knowledge generation for Machine Learning Algorithms to Map Irrigated vs. Rainfed

Unsupervised Machine Learning Algorithm for Irrigated vs. Rainfed croplands
Irrigated Classes of Africa: 40 to 21 Classes
Grouping classes from the irrigated area mask. All similar classes
were grouped together and labeled. Here we illustrate 40 classes
reduced to 21 similar classes.
2. Irrigated area classes: Preliminary 40 class signatures
3. Grouping distinct Irrigated area classes together: 40 to 21 classes
1. Irrigated area classes: Preliminary 40 classes

Irrigated Class Identifying and Labelling: Spectral matching Techniques (SMTs)
1. Ideal spectra from reference data
2. Class spectra of similar classes from Isoclass
3. Match Class spectra with ideal spectra
4. Group class spectra and match with ideal spectra
Quantitative spectral
matching techniques
(SMTs):
1. Spectral correlation
similarity (SCR) R2:
Shape measure.
2. Spectral similarity
value (SSV): Shape and
Magnitude measure.
3. Modified spectral
angle similarity (MSAS):
Angle measure
(Thenkabail et al., 2007,
Teluguntla et al, 2017)

Irrigated Class Identifying and Labelling: Spectral matching Techniques
(SMTs)
)
1. Ideal spectra from reference data 2. Class spectra of similar classes from Isoclass
3. Match Class spectra with ideal spectra 4. Group class spectra and match with ideal spectra
Classes are
grouped,
identified, labeled
Quantitative
spectral matching
techniques (SMTs):
1. Spectral
correlation
similarity (SCR) R2:
Shape measure.
2. Spectral
similarity value
(SSV): Shape and
Magnitude
measure.
3. Modified
spectral angle
similarity (MSAS):
Angle measure
(Thenkabail et al.,
2007, Teluguntla et

Space-time spectral curves (ST SCs): See how the irrigated, rainfed, and
other classes move around in 2D space during a calendar year.
Irrigated versus Rainfed: Space-Time Spiral Curves (ST SCs)
Dots indicate Julian dates of a calendar year and
how they move around in 2-D space

Irrigated Area Classes for Africa (Level I): 24 Classes
Irrigated,
single crop
is dominant
in Africa
……………
………..
But,
there is
significant
irrigated
double crop

Rainfed Area Classes for Africa (Level I): 23 Classes
Rainfed,
single
crop is
dominant
in Africa

Irrigated 5 classes and Rainfed 6 Classes (Level II)
Africa rainfed cropland area: 263.5 Mha

Irrigated and Rainfed combined 11 Classes (Level II)
rainfed
rainfed
irrigated
irrigated

Results

Landsat-derived Global Cropland Mapping: Results
Landsat-derived Global Cropland Extent at 30m (LGCEP30) for Africa

Landsat-derived Global Rainfed and Irrigated-Cropland Product at 30m (LGRIP30) for Africa
rainfed
rainfed
irrigated
irrigated

Landsat-derived Global Cropland Extent at 30m (LGCEP30) for CONUS

Landsat-derived Global Rainfed and Irrigated-Cropland Product at 30m (LGRIP30) for CONUS

Results: Global

Landsat-derived Global Cropland Extent at 30m (LGCEP30)
Total cropland areas = 1.8 billion hectares

Landsat-derived Global rainfed & irrigated product @ 30m (LGRIP30)
for the nominal year 2015
39%
61% Total cropland areas = 1.8 billion hectares

Landsat-derived Global Rainfed and Irrigated-Cropland Product at 30m (LGRIP30) :
State and County-wise Irrigated and Rainfed Area
25%
75%
Total cropland Area 168 Mha
Irrigated croplands (~25%)
Rainfed croplands (~75%)

for the nominal year 2015. Zoom in views of Nebraska, USA.
County-wise Irrigated and Rainfed Area: Nebraska

for the nominal year 2015. Zoom in views of California, USA.
County-wise Irrigated and Rainfed Area: California

Causes for Increase in Irrigated Areas of the World
1.Definition of irrigated area: Supplemental irrigation has become a norm. So, these irrigated
areas are swiftly increasing when we consider supplemental irrigation.
2. Technological advances in water delivery to agriculture: In much of the world technological
advances has created greater and greater irrigation facility. For example, most farmers these
days use some form of irrigation either by pumping water directly from rivers or from open
wells or from deep acquirer wells. In India, for example, between 1950 and 2010, the number of
drilled tube wells increased from 1 million to nearly 30 million (World Bank, 2020). The number
of deep tube wells in the country saw surge from 1.46 million to 2.6 million in just between 2006-
07 and 2013-14 (Business Standard, 2017).
3. Equipped area for irrigation: our irrigated areas consider all irrigated areas equipped for
irrigation as irrigated areas even when part of those are not irrigated at times (or left fallows).
For example, significant pasturehay are included in irrigated areas in few countries.
4. Climate factor: In an increasingly changing climate, purely rainfall dependent agriculture is
reducing swiftly. Even when rainfall amount is normal, it’s timeliness in delivery varies. Crops
need water in timely manner and when rainfall fails to follow a pattern during the growing
season, plants will need water from irrigation and hence supplemental irrigation is greatly
increasing, primarily through pumping from open wells, tube-wells, and directly from rivers.

Causes for Increase in Irrigated Areas of the World
The global net irrigated areas (GNIA) reported in our study (715 Mha) is substantially higher
compared to previous studies which report GNIA that varied between 307-450 Mha (Puy et al.,
2020, Salmon et al., 2015, Siebert et al., 2013, 2001, FAO, 2012b, FAO, 2015, Thenkabail et al.
2012, 2011, 2009a). This is not surprising. Indeed, in a recent study by Puy et al., (2020) of
Princeton University in a paper in the Nature Communications suggested that the amount of
irrigated land could in fact increase to as high as 1.8 billion hectares by 2050. They suggested in
the paper that “Policymakers should acknowledge that irrigated areas can grow much more than
previously thought in order to avoid underestimating potential environmental costs”. Puy et al.
(2020) clearly demonstrate that the current models underestimate irrigated areas. Rosa et al.
(2020) showed that up to 35% of the current rainfed areas have potential for irrigation
expansion in a climate change scenario of temperature warming between 1.5-3 degree Celsius.
Our results show that this irrigation expansion is already happening at a rapid phase. In an
increasingly changing climate, complete dependance on precipitation which has wide variability
is becoming unsustainable and some supplemental irrigation in the least is necessary to ensure
optimal productivity andor to sustain any productivity at all in rainfed croplands. Further,
most of lower irrigated areas reported in the range around 307-450 Mha estimates were for the
2000-2005 period (Salmon et al., 2015, Thenkabail et al., 2011, 2009a). These estimates were 307-
311 Mha for 2000-2008 period (Siebert et al., 2015, FAO, 2012b, Siebert and Döll, 2001, 2010,
Siebert at al., 2013), 314 Mha for 2005 by Salmon et al. (2015), and 399 Mha by Thenkabail et
al. (2011, 2009a) for nominal 2000.

Accuracies

Landsat-derived Global Cropland Mapping: Accuracies
Zone-by-Zone Error Matrices
Reference Data
Crop No-Crop Total User Accuracy
Map
Data Crop 255 48 303 84.16%
No-Crop 51 830 881 94.21%
Total 306 878 1,184
Producer Accuracy 83.33% 94.53% 91.64%
Reference Data
Map
Data
Crop 140 18 158 88.61%
No-Crop 24 67 91 73.63%
Total 164 85 249
Reference Data
Map
Data
Crop 43 5 48 89.58%
No-Crop 4 198 202 98.02%
Total 47 203 250

Accuracy Error Matrices: Global
For the entire world, the global cropland extent product had an overall accuracy of 91.7%.
For the cropland class, the producer’s accuracy was 83.4% (errors of omission of 16.6%)
and user’s accuracies of 78.3% (errors of commissions of 21.7%).

Distribution of Reference validation data used accuracy assessments of Landsat-
derived Global Rainfed and Irrigated-Cropland Product @ 30-m (LGRIP30).
There were 10,477 samples of which 5667 were rainfed and 4810 were irrigated.
Reference Validation Data: Irrigated versus Rainfed

Accuracy error matrix of the world.:
Global accuracy assessment error matrix showing the overall accuracies, user’s
accuracies, and producer’s accuracies. Number of validation samples, N = 10477.
Accuracy Error Matrices: Global
ReferenceData
Class Irrigated Rainfed Row total
Commission
error
Irrigated 4171 774 4945 15.7%
Rainfed 639 4893 5532 11.6%
Column total 4810 5667 10477
Omission error 13.3% 13.7%
Producer accuracy 86.7% 86.3%
User accuracy 84.3% 88.4%
Overall accuracy 86.5%

Area Statistics
/

What you would like to get out of the summit
●
●
Landsat-derived Global Cropland Mapping: Area Results
Cropland Area by Country as % of Total Net Global Cropland Areas

Irrigated Areas as percentage of the Global Net Irrigated Areas

Rainfed Areas as percentage of the Global Net Rainfed Areas

Irrigated Area Map of CONUS: County-wise
Fresno
Merced,
Tulare,
San Joaquin
& Kenn
Grant
Mississippi

Rainfed Area of CONUS: County-wise Rainfed Area
Chouteau & Hill
Cass,
Stutsman &
Ward
Whitman county

Total (Irrigated and Rainfed) cropland area Map of CONUS: County-wise

Way Forward
/

Landsat-derived Global Cropland Mapping: Level I, Detailed Mapping
Telangana State, India: Irrigated Areas, Level I 37 classes

Landsat-derived Global Cropland Mapping: Level I, Class Signatures
Telangana State, India: Irrigated, Level I, 37 Class Signatures

Landsat-derived Global Cropland Mapping: Level I, Detailed Mapping
Telangana State, India: Rainfed Areas, Level I, 22 classes

Landsat-derived Global Cropland Mapping: Level II, Aggregated Classes
Telangana State, India: Rainfed and Irrigated (LGRIP30-Telangana), Level II, 6 classes

Landsat-derived Global Cropland Mapping: Level III, Aggregated Classes
Telangana State, India: Rainfed and Irrigated (LGRIP30-Telangana), Level III, 2 classes

Landsat-derived Global Cropland Mapping: Level III, Aggregated Classes
Telangana State, India: Rainfed and Irrigated (LGRIP30-Telangana), Level III, 2 classes
Accuracies:
Irrigated area class
Producer accuracy:
92.9%
(Errors of
Omissions: 7.1%)
User accuracy:
98.5%
(Errors of
Commissions:
1.5%)

Impact

Landsat-derived Global Cropland Mapping: Impact
Monthly Data Users and Downloads: 126 Countries Have used GFSAD Data
Oliphant, A., Thenkabail, P.S., and Teluguntla, P., 2022. Global Food Security-support
Analysis Data at 30 meters (GFSAD30) Cropland Product: Impact Analysis November
2017 through December 2019. Open File Report, United States Geological Survey. IP-
119165.
GFSAD data from
NASA’s
LP DAAC.
2018-2021,
126 countries,
2022
72 countries
2021
96 countries
2023
Every month about 20
countries use our data

DATA AccessDissemination
Peer-reviewed Manuscripts
USGS Professional Paper:
Thenkabail, P.S., Teluguntla, P.G., Xiong, J., Oliphant, A., Congalton,
R.G., Ozdogan, M., Gumma, M.K., Tilton, J.C., Giri, C., Milesi, C., Phalke,
A., Massey, R., Yadav, K., Sankey, T., Zhong, Y., Aneece, I., and Foley, D.,
2021, Global cropland-extent product at 30-m resolution (GCEP30)
derived from Landsat satellite time-series data for the year 2015 using
multiple machine-learning algorithms on Google Earth Engine cloud:
U.S. Geological Survey Professional Paper 1868, 63 p.,
https://doi.org/10.3133/pp1868. https://lpdaac.usgs.gov/news/release-of-
gfsad-30-meter-cropland-extent-products/ IP-119164.
For 12 other related publications, go to “Publications”
section in:
Note: From 2017-present: these papers have over 2000 citations

Dissemination

What you would like to get out of the summit
●
●
1. GFSAD Web Sites:
2. Data Download from LP DAAC:
30m Products:
https://lpdaac.usgs.gov/products/lgrip30v001/
1Km products:
https://lpdaac.usgs.gov/products/gfsad1kcdv001/
https://lpdaac.usgs.gov/products/gfsad1kcmv001/
3. Browse and View
www.croplands.org
Landsat-derived Global Cropland Mapping

Mapping Global Croplands for Food Security

Recommended

Recommended

More Related Content

Similar to Mapping Global Croplands for Food Security

Similar to Mapping Global Croplands for Food Security (20)

More from Daugherty Water for Food Global Institute

More from Daugherty Water for Food Global Institute (20)

Recently uploaded

Recently uploaded (20)

Mapping Global Croplands for Food Security