Importance of observational reliability for hydrological parameter optimization: a case study of the Upper Chao Phraya River in Thailand

Rainfall-runoff models associated with optimal parameters are essential for the effective planning and management of water resources. However, there may be difficulties in parameter optimization, particularly in a basin with incomplete hydrological data. Therefore, we investigated whether a parameter optimization tool called hydroPSO could outperform the existing manual tuning approach and whether a larger number of tuning parameters would yield better model results in the Upper Chao Phraya River Basin of Thailand. We applied the particle swarm optimization (PSO) algorithm to H08, a grid-based land surface model, to systematically search for two setups containing four and 12 parameters at hydrological gauges with both adequate and inadequate observational data. The overall H08-PSO simulations with 12 parameters associated with land use outperformed those of both the H08-Plain and the H08-PSO with four parameters at most hydrological gauges. However, the simulations with 12 parameters produced an unsatisfactory performance at the stations with inadequate observations. This finding suggests that the parameter optimization tool could replace the laborious manual tuning approach because it improves model performance; however, more observations and monitoring are needed in regions with poor simulation performance.