The most important parameters of the LISEM model associated to surface runoff production (antecedent soil moisture (θap) and soil layer thickness), were calibrated by individual precipitation events in the Marcela Creek Watershed, Alto Rio Grande Region, MG. The hydrological effects of different land-uses in the watershed were also simulated. The model was highly sensitivity to soil moisture, which was the most important calibration paramete. The influence of the water balance layer was related to the soil moisture during calibration. It was observed that the deeper the soil layer, the greater the soil moisture required to calibrate the model. Results of simulation showed that the peak discharges were reduced under native forest, eucalyptus plantation and 35 months-old coffee plantation, respectively, by 37.6; 42.2 and 28.0 %, in the case of a precipitation event of 48 mm h-1. The results of the hydrological model LISEM in hydrological simulations of direct surface runoff were satisfactory, suggesting that it may be used for hydrological behavior prediction in tropical watersheds after adequately calibrated.
distributed modeling; land-use; direct surface runoff; soil moisture
