Estimation of evapotranspiration from a suite of geostationary satellites

Abstract

ABSTRACT Multi-spectral and diurnal meteorological data from optical-thermal observations of Indian Geostationary satellites were used to estimate country-scale evapotranspiration (ET). In this study radiation and vegetation products from a suite of Indian geostationary satellites, Kalpana-1 (K-1) and Indian National SATellite (INSAT) 3A Charged Couple Detector (CCD), were used. ET was derived as a function of net available energy and evaporative fraction within a single-source energy balance framework. The scaling function for albedo, net radiation and soil heat flux were developed with ground measured data from twenty-three Agro-Met Station (AMS) installed in different agro-climatic zones of India. These scaling functions were applied on the satellite data to derive intermediate inputs for surface energy balance. Further, evaporative fraction was derived from Priestley-Taylor parameter (φ) and day-time net available energy. On regional scale, pixel-by-pixel dynamic φi was computed from upper (φ max) and lower (φ min) limits of warm and wet edges of thermal inertia-Normalized Difference Vegetation Index (NDVI) triangle. Here, thermal inertia was represented by morning rise in surface brightness temperature (ΔT) between 02:30 and 06:30 Greenwich Mean Time (GMT). The dynamic φi with upper and lower limits were computed using ΔT – NDVI triangle. The derived net radiation was validated with AMS measured net radiation at seven agro-climatic zone of India showed Root mean Square Error (RMSE) between 37 and 87 W m−2. The estimated ET was validated at six agricultural and five natural grassland and wetland vegetation sites with AMS measured ET for a year. The RMSE of 10 days ET sum were varying from 7.50 to 17.70 mm for agricultural and for natural vegetation sites it varies from 8.60 to 13.10 mm. The errors variation among all site is depending on the sub-pixel heterogeneity and difference in K-1 and AMS footprints. The ET sum over agricultural kharif (June to October) and rabi (November to April) season also able distinguish normal and drought year. This study provides a robust methodology for generating near real time ET at country scale from Indian Space Research Organisation (ISRO’s) present and future high-resolution geostationary satellites over India.