2012 AGU Chapman Conference on Remote Sensing of the ...

for various water resources studies with certain degree oferror and uncertainty.Thenkabail, Prasad S.Water Use and Water Productivity of Key WorldCrops using Hyperion-ASTER, and a LargeCollection of in-situ Field Biological and SpectralDataThenkabail, Prasad S. 1 ; Mariotto, Isabella 11. Geography, U.S. Geological Survey, Flagstaff, AZ, USAThe overarching goal of this study is to assess water useand water productivity of key world crops using Hyperion-ASTER and a large collection of in-situ field biological andspectral data. The study will be based on existing datasets,collected during the 2006 and 2007 crop growing seasons,over large-scale irrigated areas of the arid Syr Darya riverbasin (444,000 km2) in Central Asia where recent studiesshow snowmelt water supplies from Himalayas are on swiftdeclines. The irrigated cropland data acquired include: (a)Hyperion narrow-band data (5 images) from EarthObserving-1 (EO-1), (b) spectroradiometer data from 400-2500 nanometers, (c) broad-band data from ASTER as wellas ETM+, ALI, IKONOS, and Quickbird, and (d) field-plotbiological data. The field-plot data of 5 crops (wheat, cotton,maize, rice and alfalfa) were collected, every 15-20 days,throughout the summer crop growing seasons (April-October) of 2006 and 2007 for a total of 1003 samplelocations and consisted of: several thousand spectralmeasurements, crop variables (e.g. biomass, LAI, yield), soiltype and salinity, water variables (e.g., inflow, outflow), andmeteorological data (e.g., rainfall, ET). The study of 5 cropsusing Hyperion-ASTER-field spectral and biological datawill: (a) develop and test water productivity models (WPMs),(b) establish shifts in phenology depicting canopies’integrated response to environmental change and\orcontrolled-planted by humans, (c) highlight best performinghyperspectral water indices (HWIs); and (d) establish chiefcauses of water productivity variations and identifyhyperspectral wavebands and indices that are most sensitiveto them. The water productivity (WP; kg\m3) is obtained bydividing crop productivity (CP kg\m2; based on the bestHyperion models) by water use (WU; m3\m2; based on thebest ASTER models). Water use of irrigated crops will bedetermined by the Simplified Surface Energy Balance Model(SSEBM), which is derived by multiplying evaporativefraction obtained using ASTER thermal imagery withreference ET derived from the Penman-Monteith equation.Crop productivity is determined through the besthyperspectral models. Hyperspectral wavebands and indicesin studying water productivity, water use, and phenology willbe established and will involve: (a) two-band vegetation index(TBVIs) models, (b) Optimum multiple-band vegetationindex (OMBVI) models, (c) derivative index models, (d)broad-band models (NIR-red based, mid-infrared, soiladjusted,and atmospherically corrected), (e) principalcomponent analysis, (f) discriminant or separability analysis(e.g., stepwise discriminant analysis), and (g) crop waterstress indices using red-edge, NIR and SWIR water bands, aswell as thermal bands. The outcome of the research will leadto: 1. Determining proportion of irrigated areas in low,medium, or high water productivity (WP; kg\m3) and theirdrivers (management practice, soil type, salinity status, etc.);3. Pin-pointing areas of low and high WP, 3. Establishingwater use (m3\m2) of 5 irrigated crops, and 4.Recommending optimal Hyperion waveband centers andwidths, in 400 to 2500 nanometer range, required to beststudy irrigated cropland characteristics;Thenkabail, Prasad S.Global croplands and their water use assessmentsthrough advanced remote sensing and non-remotesensing approachesThenkabail, Prasad S. 11. Geography, U.S. Geological Survey, Flagstaff, AZ, USAThe global cropland area estimates amongst differentstudies are quite close and range between 1.47 to 1.53 billionhectares. The total water use by global croplands, estimatedbased on existing cropland maps, varies between 6,685 to7500 km3 yr-1; of this 4,586 km3 yr-1 is by rainfed croplands(green water use) and the rest by irrigated croplands (bluewater use). Irrigated areas use about 2,099 km3 yr-1 (1,180km3 yr-1 of blue water and the rest by rain that falls directlyover irrigated croplands). However, 1.6 to 2.5 times the bluewater required by irrigated croplands (1,180 km3 yr-1) isactually withdrawn (from reservoirs or pumping of groundwater) making irrigation efficiency of only between 40-62percent. However, there are major uncertainties indetermining: a) precise location of croplands; (b) whethercrops were irrigated or rainfed; (c) exact crop types and theirphenologies; and (d) cropping intensities. This in turn leadsto uncertainties in green water use (from rainfed croplands)and blue water use (from irrigated croplands) computations.Given that about 80% of all human water use goes towardsagriculture, precise estimates of global croplands and theirwater use is of major importance. The causes of differenceswere as a result of definitions in mapping, data types used,methodologies used, resolution of the imagery, uncertaintiesin sub-pixel area computations, inadequate accounting ofstatistics on minor irrigation (groundwater, small reservoirsand tanks), and data sharing issues. To overcome the abovelimitations this research proposes an implementationstrategy to produce products using advanced remote sensing142