Performance evaluation of swat and HEC-HMS mathematical models for Atrai-karatoa river basin

Abstract

The use of mathematical models for simulating the hydrological processes in a catchment scale is getting popular. As a result, numerous hydrological models have been developed by researchers over the years. These complex mathematical models have contributed significantly in understanding the hydrological processes in different scales. Therefore, those models have been providing logical supports to the strategists and policy makers for sustainable watershed management. However, those models were developed based on various theories and assumptions considering distinct goals, therefore, the performances of hydrological models are not similar.Therefore, a comparative study using different hydrological models required to evaluate the performances of different hydrological models applicable to a particular watershed. In this study, two popular hydrological modeling software SWAT and HEC-HMS have been used to assess the performance of those models in a data limited watershed of Bangladesh.Atrai-Karatoa River system is one of the prominent rivers in the Northwest Hydrological Region (NWHR)of Bangladesh.The Atrai-Karatoa River basin does not have year-round continuous discharge measurements, unlike other rivers in Bangladesh. As a result, utilizing historical climate, land use, and soil data, calibratedand validated hydrological model can be beneficial for estimating runoff for the watershed. Results from this study revealed that during the calibration and validation phase, the performances of two models were satisfactory and both models were able to reproduce the hydrological characteristics of the Atrai-Karatoa watershed during dryand wet seasons.During calibration (2017–2019) and validation (2013–2016), Coefficient of Determination (R2) and a Nash Sutcliffe Efficiency (NSE) were found 0.86, 0.75, and 0.84, 0.75, respectively for the SWAT model. During the same periods, the HEC-HMS model results showed the values were 0.70, 0.56, and 0.68, 0.55, respectively during the calibration and validation.These results inferred that high flows were captured well by the SWAT model, while medium flows were captured well by the HEC-HMS model. It is noteworthy that the simulated low flows were close to the observed values by both models. Furthermore, dry and wet seasonal flows were simulated reasonably well by the SWAT model with slight under predicting compared to the observed values. The HEC-HMS model slightly over predicted the dry flow during validation stage and slightly over predicted the wet seasonal peak flows in calibration stage and slightly under predicted the peak flow during validation stagecompared to observed flows. However, overall SWAT model performed better compared with HEC-HMS for the Atrai-Karatoa River Basin. In addition, in this study peak flows for different return periods have been estimated considering the return period of different rainfall events. Results show that the HEC-HMS estimated flow varied between 5% to 10% compared with the data using observed predicted flow, whereas, the SWAT estimated flow varied between 3% to 6%. These findings indicated that for the Atrai-Karatoa River basin both HEC-HMS and SWAT model performs satisfactorily but between the two models SWAT performed better than the HEC-HMS model.