Irrigated area mapping in heterogeneous landscapes with MODIS time series, ground truth and census data, Krishna Basin, India

T. W. Biggs, P. S. Thenkabail, M. K. Gumma, Christopher A Scott, G. R. Parthasaradhi, H. N. Turral

Research output: Contribution to journalArticle

100 Citations (Scopus)

Abstract

Diverse irrigated areas were mapped in the Krishna River Basin (258,912 km2), southern India, using an irrigated fraction approach and multiple ancillary data sources. Unsupervised classification of a monthly time series of net difference vegetation index (NDVI) images from the Moderate Resolution Imaging Spectrometer (MODIS) over January-December 2002 generated 40 classes. Nine generalized classes included five irrigated classes with distinct NDVI time signatures: continuous irrigation, double-cropped, irrigated with low biomass, minor irrigation, and groundwater irrigation. Areas irrigated by surface water began greening 45 days after groundwater-irrigated areas, which allowed separation of surface and groundwater irrigation in the classification. The fraction of each class area irrigated was determined using three different methods: ground truth data, a linear regression model calibrated to agricultural census data, and visual interpretation of Landsat TM imagery. Irrigated fractions determined by the three methods varied least for the double-cropped irrigated class (0.62-0.79) and rangeland (0.00-0.02), and most for the minor irrigated class (0.06-0.43). Small irrigated patches (<0.1 km2) accounted for more irrigated area than all major surface water irrigated areas combined. The irrigated fractions of the minor and groundwater-irrigated classes differed widely by method, suggesting that mapping patchy and small irrigated areas remains challenging, but comparison of multiple data sources improves confidence in the classification and highlights areas requiring more intensive fieldwork.

Original languageEnglish (US)
Pages (from-to)4245-4266
Number of pages22
JournalInternational Journal of Remote Sensing
Volume27
Issue number19
DOIs
StatePublished - Oct 2006

Fingerprint

Irrigation
MODIS
Spectrometers
Groundwater
Time series
census
irrigation
time series
Imaging techniques
groundwater
vegetation index
Surface waters
basin
surface water
unsupervised classification
rangeland
Linear regression
Landsat thematic mapper
Catchments
fieldwork

ASJC Scopus subject areas

  • Computers in Earth Sciences

Cite this

Irrigated area mapping in heterogeneous landscapes with MODIS time series, ground truth and census data, Krishna Basin, India. / Biggs, T. W.; Thenkabail, P. S.; Gumma, M. K.; Scott, Christopher A; Parthasaradhi, G. R.; Turral, H. N.

In: International Journal of Remote Sensing, Vol. 27, No. 19, 10.2006, p. 4245-4266.

Research output: Contribution to journalArticle

Biggs, T. W. ; Thenkabail, P. S. ; Gumma, M. K. ; Scott, Christopher A ; Parthasaradhi, G. R. ; Turral, H. N. / Irrigated area mapping in heterogeneous landscapes with MODIS time series, ground truth and census data, Krishna Basin, India. In: International Journal of Remote Sensing. 2006 ; Vol. 27, No. 19. pp. 4245-4266.
@article{41934d77cdeb48cb82263edb256be78c,
title = "Irrigated area mapping in heterogeneous landscapes with MODIS time series, ground truth and census data, Krishna Basin, India",
abstract = "Diverse irrigated areas were mapped in the Krishna River Basin (258,912 km2), southern India, using an irrigated fraction approach and multiple ancillary data sources. Unsupervised classification of a monthly time series of net difference vegetation index (NDVI) images from the Moderate Resolution Imaging Spectrometer (MODIS) over January-December 2002 generated 40 classes. Nine generalized classes included five irrigated classes with distinct NDVI time signatures: continuous irrigation, double-cropped, irrigated with low biomass, minor irrigation, and groundwater irrigation. Areas irrigated by surface water began greening 45 days after groundwater-irrigated areas, which allowed separation of surface and groundwater irrigation in the classification. The fraction of each class area irrigated was determined using three different methods: ground truth data, a linear regression model calibrated to agricultural census data, and visual interpretation of Landsat TM imagery. Irrigated fractions determined by the three methods varied least for the double-cropped irrigated class (0.62-0.79) and rangeland (0.00-0.02), and most for the minor irrigated class (0.06-0.43). Small irrigated patches (<0.1 km2) accounted for more irrigated area than all major surface water irrigated areas combined. The irrigated fractions of the minor and groundwater-irrigated classes differed widely by method, suggesting that mapping patchy and small irrigated areas remains challenging, but comparison of multiple data sources improves confidence in the classification and highlights areas requiring more intensive fieldwork.",
author = "Biggs, {T. W.} and Thenkabail, {P. S.} and Gumma, {M. K.} and Scott, {Christopher A} and Parthasaradhi, {G. R.} and Turral, {H. N.}",
year = "2006",
month = "10",
doi = "10.1080/01431160600851801",
language = "English (US)",
volume = "27",
pages = "4245--4266",
journal = "International Joural of Remote Sensing",
issn = "0143-1161",
publisher = "Taylor and Francis Ltd.",
number = "19",

}

TY - JOUR

T1 - Irrigated area mapping in heterogeneous landscapes with MODIS time series, ground truth and census data, Krishna Basin, India

AU - Biggs, T. W.

AU - Thenkabail, P. S.

AU - Gumma, M. K.

AU - Scott, Christopher A

AU - Parthasaradhi, G. R.

AU - Turral, H. N.

PY - 2006/10

Y1 - 2006/10

N2 - Diverse irrigated areas were mapped in the Krishna River Basin (258,912 km2), southern India, using an irrigated fraction approach and multiple ancillary data sources. Unsupervised classification of a monthly time series of net difference vegetation index (NDVI) images from the Moderate Resolution Imaging Spectrometer (MODIS) over January-December 2002 generated 40 classes. Nine generalized classes included five irrigated classes with distinct NDVI time signatures: continuous irrigation, double-cropped, irrigated with low biomass, minor irrigation, and groundwater irrigation. Areas irrigated by surface water began greening 45 days after groundwater-irrigated areas, which allowed separation of surface and groundwater irrigation in the classification. The fraction of each class area irrigated was determined using three different methods: ground truth data, a linear regression model calibrated to agricultural census data, and visual interpretation of Landsat TM imagery. Irrigated fractions determined by the three methods varied least for the double-cropped irrigated class (0.62-0.79) and rangeland (0.00-0.02), and most for the minor irrigated class (0.06-0.43). Small irrigated patches (<0.1 km2) accounted for more irrigated area than all major surface water irrigated areas combined. The irrigated fractions of the minor and groundwater-irrigated classes differed widely by method, suggesting that mapping patchy and small irrigated areas remains challenging, but comparison of multiple data sources improves confidence in the classification and highlights areas requiring more intensive fieldwork.

AB - Diverse irrigated areas were mapped in the Krishna River Basin (258,912 km2), southern India, using an irrigated fraction approach and multiple ancillary data sources. Unsupervised classification of a monthly time series of net difference vegetation index (NDVI) images from the Moderate Resolution Imaging Spectrometer (MODIS) over January-December 2002 generated 40 classes. Nine generalized classes included five irrigated classes with distinct NDVI time signatures: continuous irrigation, double-cropped, irrigated with low biomass, minor irrigation, and groundwater irrigation. Areas irrigated by surface water began greening 45 days after groundwater-irrigated areas, which allowed separation of surface and groundwater irrigation in the classification. The fraction of each class area irrigated was determined using three different methods: ground truth data, a linear regression model calibrated to agricultural census data, and visual interpretation of Landsat TM imagery. Irrigated fractions determined by the three methods varied least for the double-cropped irrigated class (0.62-0.79) and rangeland (0.00-0.02), and most for the minor irrigated class (0.06-0.43). Small irrigated patches (<0.1 km2) accounted for more irrigated area than all major surface water irrigated areas combined. The irrigated fractions of the minor and groundwater-irrigated classes differed widely by method, suggesting that mapping patchy and small irrigated areas remains challenging, but comparison of multiple data sources improves confidence in the classification and highlights areas requiring more intensive fieldwork.

UR - http://www.scopus.com/inward/record.url?scp=33749472628&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33749472628&partnerID=8YFLogxK

U2 - 10.1080/01431160600851801

DO - 10.1080/01431160600851801

M3 - Article

AN - SCOPUS:33749472628

VL - 27

SP - 4245

EP - 4266

JO - International Joural of Remote Sensing

JF - International Joural of Remote Sensing

SN - 0143-1161

IS - 19

ER -