[A-HW30] 水循環・水環境

[AHW30-21] Estimation of Evapotranspiration Fluxes using METRIC Model over Sub-tropical Moist Deciduous Forest in Doon Valley, India

*Saurabh Purohit^1,2、N R Patel¹、Bhaskar R Nikam¹、S P Aggarwal¹、Triparna Sett¹、Prakash Chauhan¹ (1.Indian Institute of Remote Sensing, Dehradun, India、2.Forest Research Institute Deemed to be University, Dehradun, India)

キーワード：Evapotranspiration, METRIC, Eddy Covariance

Knowledge of Forest Evapotranspiration (ET) provides insights on the role of forests in the hydrological budget. Higher spatial resolution (30 m) ET maps can be helpful to assess its variability associated with the forests at the finer scale. METRIC (Mapping Evapotranspiration at high Resolution with Internalized Calibration) model estimates the amount of actual evapotranspiration of each pixel of the satellite image. Sal (Shorea robusta Gaertn.f) forest is the important and the dominant forest type covering Doon valley and there is little investigation to estimate ET using the presented model. There are fewer studies on the forest ET in this region using the remote sensing-based surface energy-based models. The present study was done to evaluate the performance of the METRIC (Mapping Evapotranspiration at high Resolution with Internalized Calibration) model for ET estimation at the Landsat scale (30 m) over the sub-tropical Shorea robusta Gaertn.f forest in Doon Valley located in foothills of Himalaya in India. Total of 12 cloud-free Landsat 8 scenes over the study area for the year 2018 were processed using the water package in R Environment for the estimation of ET. Results for daily ET of the cloud-free images were compared with the Eddy Covariance (EC) Flux Tower installed in the Barkot Reserve Forest for 2018. Overall results showed good agreement between the daily ET derived from METRIC model and estimated from Flux Tower data with value of coefficient of determination (r²) 0.76, Mean Absolute Error (MAE) 0.68, Root Mean Square Error (RMSE) 0.72 mm, Nash Sutcliffe Efficiency (E) 0.85 and Wilmott Index (d) 0.70. Comparison of monthly ET derived from METRIC model and Flux Tower data for January to June 2018 showed similar trends however, METRIC model overestimated ET for months of March to May. The present study is one of the first attempts to prepare high-resolution ET maps at the Landsat scale for the Sal Forests of Doon valley using METRIC model and its validation with the Flux Tower data. It will enhance our knowledge in understanding the energy balance dynamics in the forest.