Analysis of variations and controls of evapotranspiration over major Indian River Basins (1982-2014)

This study analyses long-term (1982-2014) estimates of evapotranspiration (ET) over four major river basins of India with the primary objective of understanding the factors controlling its space-time variability. Here we utilize terrestrial water storage change (TWSC) estimates, computed from WaterGAP Global Hydrology Model (WGHM) simulations, in monthly water balance computations to obtain the best available estimates of long-term ET for the study region. Trend analysis shows significant increase in annual ET over Ganga (2.72 mm/year) and Krishna (3.90 mm/year) River Basins, while in Godavari and Mahanadi River Basins the observed trends are insignificant. The relative contribution of potential factors (represented by precipitation, soil moisture, temperature and Normalized Difference Vegetation Index (NDVI)) that affect the variability of monthly ET is assessed using Hierarchical Partitioning Analysis (HPA). Results reveal that ET variance is largely controlled by the availability of water (represented by precipitation and soil moisture) in all the river basins. Precipitation (soil moisture) accounts for 65% (16%), 70% (20%), 77% (15%) and 67% (18%) of the variability in monthly ET over the Ganga, Godavari, Krishna and Mahanadi River Basins, respectively. Similarly, highest contributions from precipitation are also observed in annual scale variations of ET in all the river basins. Multiple regression analysis performed to assess the overall influence of controlling variables demonstrate that precipitation, soil moisture, temperature and NDVI explain 84% (Ganga), 86% (Godavari), 91% (Krishna) and 82% (Mahanadi) of variations observed in monthly ET over the respective basins. Results presented in this study have major implications for the understanding of ET variability, appropriateness and discrepancies in different ET products and compliment the contemporary efforts of extending GRACE-based ET estimates in space and time.