Parameterization and calibration of the SWAT hydrological model using SUFI-2 and GLUE algorithms in Bay of Plenty, New Zealand
Vicky Anand, Shailesh Kumar Singh, Bakimchandra Oinam
Abstract
ABSTRACT Planning and managing water resources necessitate the utilization of hydrologic models. However, the calibration of these models poses a significant challenge due to the high degree of parameter uncertainty, which has global implications. Moreover, regions with high seasonal variation in precipitation exhibit strong heteroskedasticity, further complicating the calibration process. This study aims to determine an optimized calibration approach and parameter optimizations by integrating two sensitivity methods with two optimization techniques. The methodology was demonstrated in the sub-basins of the Bay of Plenty, New Zealand, using the SWAT hydrological model. The performance indices of the calibrated models were analyzed, revealing striking similarities between GLUE and SUFI-2 approaches. Although GLUE marginally outperformed SUFI-2 in terms of the fit indices, indicating better model accuracy, SUFI-2 demonstrated less uncertainty, making it a more stable option. Among the parameters, the initial SCS curve number for moisture condition II, baseflow factor, groundwater delay, groundwater revap coefficient, soil evaporation compensation factor, Manning's coefficient value for the main channel, and effective hydraulic conductivity in the main channel were identified as the most sensitive. These findings provide critical insights into hydrological modeling, supporting the development of policies not only in the Bay of Plenty but also in similar regions.