Assessment of water demands for irrigation using energy balance and satellite data fusion models in cloud computing: A study in the Brazilian semiarid region

Assessment of irrigation in arid and semiarid regions is imperative to ensure the sustainable use of limited water resources and guarantee food production. Therefore, this study aimed to assess actual evapotranspiration – ETa derived from the Surface Energy Balance Algorithm for Land – SEBAL model with and without satellite image fusion as input of a soil water balance in a pilot area of sugarcane in the semiarid region of Brazil. A fusion of Landsat sensors’ and Moderate Resolution Imaging Spectroradiometer – MODIS’ images was completed through a Spatial and Temporal Adaptive Reflectance Fusion Model – STARFM script developed using cloud computing, and its performance in estimating key variables for the radiation balance was evaluated. ETa and irrigation were daily estimated between June, 2015 and May, 2016 by combining STARFM with SEBAL and evaluated according to the Bowen ratio technique and irrigation data. In addition, one-minute surface meteorological elements at the satellite overpass times were used. STARFM performed well with RMSE of 17.00 W m−2, 2.28 K, 0.07, and 0.01 for Rn, Ts, NDVI, and albedo, respectively. The metrics employed to evaluate ETa estimates indicated that the SEBAL+STARFM has low mean errors (PBIAS = −2.75% and RMSE = 0.97 mm d−1 and 16.66 mm month−1) and high coefficient of determination (0.87 for daily ETa–ET24, and 0.91 for monthly ETa), in comparison with SEBAL using Landsat-only images (PBIAS = −5.25%, RMSE = 0.97 mm d−1 and 17.66 mm month−1, r² = 0.92). Adding fused images resulted in a better fit of the estimated cumulative ET24 curve to the measured ET24. The water balance indicated that the cultivated cane suffered water stress, which was better represented by estimates using the ET24 curve with the addition of fused images than Landsat images alone. Although this increase in temporal resolution of the estimated ET24 data indicated a greater water consumption, it informs a quantity that would be sufficient to meet the water demand of the crops. •Spatial and Temporal Adaptive Reflectance Fusion model (STARFM) was integrated with SEBAL.•The STARFM applied to key-net radiation variables produces good estimates.•STARFM+SEBAL was validated in an irrigated area in Brazilian semiarid region.•STARFM+SEBAL fitted the cumulative actual ET curve better than Landsat-only images.•The python code used can aid water demand evaluation in other sites.