Evapotranspiration over a processing cassava field: a comparative analysis of micrometeorological methods and remote sensing

The need for reliable evapotranspiration (ET) estimates has prompted the introduction of new methodologies. This study aimed at studying the ET of a processing cassava field (13°6′39" S, 39°16′46" W, 154 m asl) in a tropical climate in Bahia, Brazil, under rainfed conditions from April to August 2019 by means of micrometeorological and remote sensing methods. The combination of surface renewal analysis and energy balance (SREB) demonstrated its potential to accurately determine the crop ET once calibrated against eddy covariance measurements of H. Remote sensing techniques were also applied with the METRIC and SAFER algorithms. Due to frequent cloud cover in the area, only three Landsat images from overpasses in May and June could be used. High agreement in terms of crop ET was found between the surface and the remote sensing methods. For the three images processed, METRIC and SAFER were 8.6% and 26.4% higher than SREB, on average. Among the proposed regression models (M1, M2, and M3) for estimation of processing cassava ET, M3 showed a better adjustment with the highest coefficient of determination (r² = 0.952) and lowest error (RMSE = 0.205 mm day⁻¹). In the M3 model, ET/Rn was expressed as a function of the NDVI/LAI ratio. These three biophysical parameters, Rn, NDVI, and LAI, can routinely be determined from image processing for field applications in water management at the studied region. Therefore, with a limited set of variables, this approach can be satisfactorily applied using data collection methodologies that provide enhanced temporal and spatial resolution.