A conjugate application of MODIS/Terra data and empirical method to assess reference evapotranspiration for the southwest region of Bangladesh

Abstract

Availability of water resources, especially irrigation water in the Southwest (SW) region of Bangladesh, is in threat due to climate change impacts, reduction of upstream flow, and excessive groundwater consumption. It is essential to assess the evapotranspiration ( $$\mathrm{ET}$$ ), especially reference evapotranspiration ( $${\mathrm{ET}}_{0}$$ ) to ensure efficient irrigation demand for this region. However, direct measurement of $${\mathrm{ET}}_{0}$$ is a difficult task as empirical equation-based assessment requires a complex set of input data. As remote sensing (RS) technology has evolved as a viable alternative to the ground-based estimation of ET0, this study investigated whether it is possible to attain a reliable assessment of $${\mathrm{ET}}_{0}$$ for the SW region of Bangladesh using satellite data. Land surface temperature (LST) data retrieved from the moderate-resolution imaging spectroradiometer (MODIS) satellite have been evaluated as an alternative of air temperature in three empirical $${\mathrm{ET}}_{0}$$ assessment methods, namely the Hargreaves–Samani (HS), Thornthwaite (TH), and Blaney–Criddle (BC) methods. $${\mathrm{ET}}_{0}$$ assessed from these methods have been calibrated and validated against the FAO-56 Penman–Monteith (PM) method. Several statistical indices, viz. root mean square error (RMSE), mean absolute error (MAE), mean relative error (MRE), and coefficient of determination ( $${R}^{2}$$ ) have been used to evaluate the performance of $${\mathrm{ET}}_{0}$$ assessment from the above-mentioned methods. MODIS-LST data have been found strongly correlated with air temperature ( $${R}^{2}$$ = 0.95). A good correlation in $${\mathrm{ET}}_{0}$$ assessment has been found between the PM method and HS ( $${R}^{2}$$ = 0.85), TH ( $${R}^{2}$$ = 0.81), and BC ( $${R}^{2}$$ = 0.79) methods. With the calibrated result, the calibrated HS method provides the most accurate results compared to the other methods, with an average RMSE of 0.44 mm/day, MAE of 0.39 mm/day, and MRE of 12.34%. Considering the consistency of the findings with the previous studies, this method has been recommended for assessing $${\mathrm{ET}}_{0}$$ under conditions of limited data availability.