Mapeamento da evapotranspiração real por imagens orbitais em bacia representativa no estado de Pernambuco

Abstract

The study, assessment and monitoring of hydrological processes in a river basin are fundamental for an efficient management of water use. In this context, quantifying evapotranspiration is of great relevance considering that it is one of the most important components of a hydrological cycle, because it represents the loss of surface water to the atmosphere and its use by the crops. The objective of this research is to improve the actual evapotranspiration (Etr) provided by the Mu et al. (2007) algorithm. This enhancement will be achieved by the use of meteorological data with higher resolution in modelling, and through a better estimation of the parameters involved in the computation of aerodynamic resistance and the crop. The research was carried out in the Tapacurá Basin in the state of Pernambuco, Northeast Brazil, where an experimental meteorological tower was installed, which served as the basis for the input meteorological data provision, as well as for ETr determination by the energy balance method by the Bowen ratio (BERB). The model was processed from eight-day products (MOD09Q1) between 2011 and 2012. Due to the atmospheric conditions of the Tapacura Basin, which are very cloudy, the selection of cloud-free pixels was performed resulting in a more cleaned image every thirty days. Thus forming a monthly composition of albedo and vegetation indexes (IAF and EVI 2). It was then admitted that these variables remain constant during the month, enabling the determination of daily net radiation (Rn24h) and daily actual evapotranspiration. For the Rn24h, a local calibration of the model proposed by De Bruim (1987) was performed through the CNR4 sensor installed at the site in 2015 and 2016. The calibrated model was used to determine the Rn24h in 2011 and 2012. Through the composition of albedo, solar radiation data, daily atmospheric transmittance and calibration of the Rn24h model, it was possible to obtain results for all of 2011 and 2012. The Rn24h results estimated by remote sensing showed a good correlation with the ones measured at the station (R² = 0.9692, for the year 2011 and R² = 0.9373, for the year 2012). The validation of the ETr by remote sensing by the BERB presented the Mean Absolute Error (EAM), Mean Relative Error (ERM) and Root Mean Square Error (REQM), which were respectively, 0.73 mm day -1, 17, 1% and 0.93 mm day -1 for 2011 and 0.58 mm day -1, 26.8% and 0.70 mm day -1 for 2012. The results were consistent with the literature and showed that the proposed methodology is efficient in studies of spatial and temporal distribution of ETr for the studied region.