Performance of a distributed semi-conceptual hydrological model under tropical watershed conditions

Conceptual rainfall-runoff models at the watershed scale are useful tools for assisting in managing and planning water resources, making it possible to estimate streamflow and to predict hydrologic impacts due to land-use changes. The objective of this study was to calibrate and to validate the LAvras Simulation of Hydrology (LASH Model) (Beskow, 2009) in a Brazilian Tropical Watershed for daily streamflow. LASH is a continuous, distributed, semi-physically based model for simulation of different hydrologic components on a daily basis. The Shuffled Complex Evolution (SCE-UA) global search method was used with the LASH model in order to optimize model parameters that were found to be the most sensitive or not directly measurable. The LASH model was calibrated over a 2-year period, thereafter, the parameters obtained through the calibration were kept constant for the validation step using a different period of time from that analyzed during the calibration. The Nash-Sutcliffe coefficient (C-NS) values found were 0.820 and 0.764 during calibration and validation, respectively, whereas, the C-NS (log Q) values equal to 0.821 and 0.770 were obtained for the same periods. The SCE-UA method was found to be an efficient algorithm for searching 'optimal' model parameter values. It was possible to conclude that the model has a great potential for being applied in generating the long-term streamflow as well as flow-duration curves. Therefore, the model can reliably be applied under tropical conditions of this medium-sized watershed or other similar watersheds, thus making it useful to plan the sustainable development of similar tropical and subtropical watersheds. (C) 2011 Elsevier B.V. All rights reserved.