Incorporação de produtos de sensoriamento remoto na calibração de um modelo hidrológico distribuído

Abstract

The calibration of hydrological model parameters only based on the reproduction of the observed streamflow (Q) regime does not guarantee that the hydrological processes are satisfactorily represented. Alternatively, calibration with data on variables such as actual evapotranspiration (AET) and soil moisture (SM) obtained by remote sensing (RS) is a promising solution to overcome the limitations of traditional calibration. This research explores the value of combining RS products that reflect different components of the water balance. Different multi-objective calibration schemes are developed and applied for two river basins with different physical, climatic and hydrological characteristics (Piancó River Basin, in the semiarid region of Paraíba, and the Grande River Basin tributary to Furnas, in the Southeast). Two spatial averaging and spatial distribution approaches were tested, and simple (flow only – Q or AET or SM) and multivariable (Q, AET and SM) calibration procedures were used in fourteen calibration procedures that incorporate streamflow (Q) and three satellite products (AET estimated by the MOD16 algorithm - MODerate-resolution Imaging Spectroradiometer product 16; and by GLEAM - Global Land Evaporation Amsterdam Model; and the SM estimated by the European Space Agency in Climate Change Initiative program - ESA CCI). The large-scale hydrological model MGB was adapted to work with such calibration schemes, being run in a daily time step from 2001 to 2011. The incorporation of the AET of MOD16 did not help to improve the performance of the hydrological model regardless of the configuration of calibration when applied to the semi-arid basin, with better performance for the Furnas basin. The introduction of AET estimates from GLEAM as a calibration variable led the hydrological model to reproduce these AET values more closely and, at the same time, better represent the observed Q for both study areas. The results of the multivariate calibration with Q, AET and SM show that gains in soil moisture and evapotranspiration performance resulted in decreases in flow performance. The findings of this research highlight the benefit of incorporating AET and SM as calibration variables to improve overall hydrological model performance, which can even outperform calibration based on flow alone. However, there is a need to test different calibration configurations and different RS data sources, which can result in very contrasting results.