Estimativa da evapotranspiração real via sensoriamento remoto

Abstract

Actual evapotranspiration (ETa) is one of the main hydrological cycle processes and the main cause of surface water loss. The precise knowledge of ETa rates along time and space is necessary for modeling water balance in watersheds and identifying agricultural net production, among other applications. ETa is determined in situ by flux towers and by lysimeters, which monitoring network is unable to represent its large spatial variability. Remote sensing is unable to directly measure ETa, but it makes possible the estimation of the evaporative fraction that, combined with meteorological data, is used to derive ETa. Several models use information on surface temperature and vegetation index, estimated by different remote sensors, such as MODIS and Landsat, to derive ETa. In this study, the SSEBop surface energy balance model was evaluated. The model was applied with MODIS data, via 8 different parameterizations, in the region of the Urucuia Aquifer System (SAU) and compared with monthly water balance data, estimated by the SMAP model, and annual water balance, both calculated in 4 basins. The model was also implemented with Landsat 7 and Landsat 8 images in Rio Grande do Sul State and validated with ETa data measured in two flux towers installed in irrigated rice fields in Paraíso de Sul and Cachoeira do Sul towns. Finally, the model was used to derive ETa with Landsat 7 and Landsat 8 images in a tomato plantation irrigated by a central pivot and compared to irrigation consumption data. Comparison of the model 8 parameterizations revealed that the last version of SSEBop results are closer to those of the water balance, however with unexpressive differences between the use of meteorological or climatological data, or one or more areas for computing c factor. SSEBop annual ETa was close to that calculated by annual water balance, with errors ranging from 10 to 20%, but with a low linear correlation. ETa was overestimated by the SSEBop in dry season and underestimated in rainy season, when compared to the SMAP model. Compared to flux tower ETa, the SSEBop model presented errors between 0.8 and 1.6 mm/day (17% and 34%), with an overestimation of intermediate ETa values. The application of SSEBop in the tomato plantation showed a good approximation between the values of ETa and irrigation depth. This study demonstrated the potential of remote sensing, especially the SSEBop model, for regional and local ETa estimation, as well as its use for estimating irrigation water consumption.