Comparison of runoff generation methods for land use impact assessment using the SWAT model in humid tropics

Abstract

Hydrological responses due to deforestation in a humid tropical catchment were analyzed using two runoff generation methods available in the Soil Water Assessment Tool (SWAT) model: the Curve Number (CN) and the Green-Ampt (GA) methods. The calibrated model, which performed well in simulating runoff under present land use condition in the Batanghari River Basin, Indonesia (42,960 km²), was then used to simulate runoff using past and future land use scenarios. Simulations showed similar changes in the annual water budget: decreasing evaporation and increasing total discharge. However, the two methods showed opposite changes in flow regimes: high flow increased (13%) under the CN while low flow increased (27%) under the GA. These results are associated with differences in runoff generation mechanisms, where surface runoff contributes to total discharge to a much larger extent under the CN (43%) than the GA (4%). Land use changes caused a reduction in infiltration rate, leading to higher high flow under the CN, while high flow did not change under the GA. Instead, lower evapotranspiration increased groundwater flow under the GA, and thus the steady low flow increased. This study suggests that the runoff generation method should be selected carefully based on the dominant flow pathway of a catchment, particularly for land use impact studies in the humid tropics.