Calibration and validation of the SWMM hydrological model on steep terrain watersheds in Costa Rica

Abstract

The hydrologic model SWMM was applied in the experimental watersheds of the Toyogres and Zopilote rivers in Costa Rica. The suite PEST was used for model‘s parameter optimization and sensitivity analysis. Initial parameterization of the model was based on high-resolution spatial information.  In the calibration/validation process, 13 rainstorms, recorded between September and November 2011, were used. Results show an average NS efficiency of 0.85±0.07 in calibration and 0.87±0.04 in validation for the Toyogres River, whereas 0.80 ± 0.04 and 0.77 ± 0.01 were respectively found for the Zopilote River. The Manning‘s roughness coefficient for runoff generation over impervious area (nimp), runoff generation over pervious area (nper) and hydraulic transit (ncha) were the most sensitive parameters of them all.

Their behavior is partially dependent on total volume and rainfall intensity. Furthermore, there is a profound dependency with the antecedent soil moisture conditions and the state of the hydraulic flow in both, surface runoff generation and drainage network.

It is concluded that there is no unique set of optimum parameters capable to satisfy every possible condition. Even when the coefficients of variation of the sensitive parameters are relatively high, SWMM has shown sufficient flexibility to be used even in conditions where three is no calibration/validation data.