Makumbura, Randika K ORCID: 0000-0001-6030-1567, Gunathilake, Miyuru B ORCID: 0000-0001-7052-1942, Samarasinghe, Jayanga T ORCID: 0000-0001-8491-0092, Confesor, Remegio ORCID: 0000-0002-7306-1432, Muttil, Nitin ORCID: 0000-0001-7758-8365 and Rathnayake, Upaka ORCID: 0000-0002-7341-9078 (2022) Comparison of calibration approaches of the Soil and Water Assessment Tool (SWAT) model in a tropical watershed. Hydrology, 9 (10). ISSN 2306-5338

Abstract

Hydrologic models are indispensable tools for water resource planning and management. Accurate model predictions are critical for better water resource development and management decisions. Single-site model calibration and calibrating a watershed model at the watershed outlet are commonly adopted strategies. In the present study, for the first time, a multi-site calibration for the Soil and Water Assessment Tool (SWAT) in the Kelani River Basin with a catchment area of about 2340 km2 was carried out. The SWAT model was calibrated at five streamflow gauging stations, Deraniyagala, Kithulgala, Holombuwa, Glencourse, and Hanwella, with drainage areas of 183, 383, 155, 1463, and 1782 km2, respectively, using three distinct calibration strategies. These strategies were, utilizing (1) data from downstream and (2) data from upstream, both categorized here as single-site calibration, and (3) data from downstream and upstream (multi-site calibration). Considering the performance of the model during the calibration period, which was examined using the statistical indices R2 and NSE, the model performance at Holombuwa was upgraded from “good” to “very good” with the multi-site calibration technique. Simultaneously, the PBIAS at Hanwella and Kithulgala improved from “unsatisfactory” to “satisfactory” and “satisfactory” to “good” model performance, while the RSR improved from “good” to “very good” model performance at Deraniyagala, indicating the innovative multi-site calibration approach demonstrated a significant improvement in the results. Hence, this study will provide valuable insights for hydrological modelers to determine the most appropriate calibration strategy for their large-scale watersheds, considering the spatial variation of the watershed characteristics, thereby reducing the uncertainty in hydrologic predictions.

Search Google Scholar

Repository staff login