Estimation of Pan Evaporation and Actual Evapotranspiration Using GIS and Remote Sensing

Abstract

This study demonstrates the estimation of actual evapotranspiration and pan evaporation based on the satellite images. LISS III and Landsat images are used in this study. LISS III and Landsat images are available in different bands. All bands are composited and mosaicked to form single image of the study area. Composited image is classified into five different classes namely Forestland, Agricultural Land, Fallow Land, Water Bodies and Built-up Area. Separate methodology for the estimation of actual evapotranspiration and pan evaporation has been proposed. To determine actual evapotranspiration, Normalized Difference Vegetation Index (NDVI) map is generated using ARCGIS. For the different years, average NDVI values and corresponding Kc values are obtained. Reference evapotranspiration (Eto) is obtained by Penman–Monteith model. Actual evapotranspiration is obtained by multiplying reference evapotranspiration to a crop coefficient (Kc). RMSE and NRMSE between observed and predicted evapotranspiration are found to be 0.27 mm and 0.7 mm, respectively. To determine pan evaporation, Normalized Difference Water Index (NDWI) map is generated using ARCGIS. Total 24 Landsat satellite images are used to obtain NDWI values on the Water Bodies. NDWI is correlated with the relative humidity-dependent Pan coefficient. Correlation coefficient between pan coefficient and NDWI was found to be 0.92. A new model is proposed relating to modified pan coefficient (Kwp) and NDWI. Pan evaporation is obtained by dividing Eto by NDWI-dependent modified pan coefficient (Kwp). Coefficient of correlation and RMSE between observed and predicted pan evaporation is found to be 0.70 and 0.5 mm, respectively.