Determinação da evapotranspiração (ET) em citros por meio do algoritmo SAFER mediante o uso de sensoriamento remoto

Abstract

Brazil is the largest producer of oranges and the fifth largest producer of lemons, therefore, water management practices that improve the production of these crops in the country are essential. Within irrigation management, there is a need to estimate evapotranspiration (ET). Applying the SAFER algorithm on satellite images is a promising technique to determine ET. Thus, this work aimed to determine the evapotranspiration of orange and lemon crops in Capitão Poço, PA, using the SAFER algorithm. Images from Landsat 7 and 8 satellites were used, in addition to collecting meteorological information on the cultures available at the weather station in the study area, from July to December 2021. For the determination of ET by the SAFER method, the ET was determined with data collected from satellite images (reflectance, surface albedo, Normalized Difference Vegetation Index (NDVI), spectral radiance, and surface temperature) and the reference evapotranspiration (ET0), determined from data collected at a weather station (air temperature, air humidity, radiation balance, and wind speed), where ETSAFER corresponds to the quotient of ET and ET0 multiplied by ET0, was calculated. Subsequently, the algorithm was validated based on the Penman-Monteith method from the calculation of mean absolute error and mean relative error. The mean ETSAFER values for lemon and orange were 3.25 (±0.05) and 3.36 (±0.01), respectively. A maximum albedo value of 0.40 was observed between crops in December; due to the larger canopy volume, more significant densification in the lemon crop favored the higher NDVI values. The images from the Landsat 7 and 8 satellites can be used to calculate ET via the SAFER algorithm because they offered precious information that helped the algorithm estimate the variable accurately for the period studied. The values found via the algorithm (ETSAFER) presented conformity with the observed data (ETC) with an accuracy of 75%; the ET estimate for the lemon tree presented the best performance with 74%, while the orange-grown area obtained an accuracy of 60%.