Sparse Gash model applied to seasonal dry tropical forest

Abstract

Rainfall interception represents a significant component of the water balance and its modelling and simulation are essential to understanding the hydrologic cycles in different ecosystems. Caatinga is a seasonal dry tropical forest, characterized by a deciduous and xerophilous vegetation that covers large areas in Brazil. This domain appears as a highly vulnerable natural water resource system, and is important for studies about droughts, impacts on soil erosion, as well as the adaptation to altered rainfall patterns and intensities. Specifically, knowledge about rainfall partitioning in such ecosystem can improve hydrologic modelling and efficiency. This work was carried out for parametrizing and validating the sparse Gash model, simulating the rainfall interception from five Caatinga species individually as well as aggregated together as a mixed-species forest. Proportions of gross rainfall into throughfall varied from 59.8 to 78.9%, and into interception were between 20.4 and 39.5%. Results also showed low scattering of predictions, as well as absence of constant and systematic errors during simulations. Rainfall interceptions predicted by the sparse Gash model resulted in mean absolute errors (MAE) ranging from 0.23 to 0.41 mm, while agreement indices (d) varied between 0.94 and 0.97 for the studied Caatinga species and the mixed vegetation. The sparse Gash model was reliable enough to apply to Caatinga ecosystem, appearing as a valuable tool for studying rainfall interception in this domain.