LAUSR

Abstract The importance of water resources forecasting is growing due to changes in precipitation caused by climate change and with hydrologic models are being used mainly for water resource management. Calibration and validation through adjustment of parameters are important processes for increasing the reliability of the application of hydrologic models. This is different from flow rates that can be estimated in real time through level-flow rating curve. The reliability of data depends on observation; therefore, using data with a short can increase the reliability of the results. In this study, daily loads of suspended solids (SS), total nitrogen (TN), and total phosphorous (TP) were generated using three statistical methods: general method, perfect prognosis method, and model output statistics (MOS). Fitness of calibration and validation of hydrologic models were then evaluated using correlation, R2, and anomaly analysis. When these three methods were applied to the Geumho and Hwang rivers, perfect prognosis method and the MOS method showed better results than the general method. For the Geumho River, correlation and R2 improved by up to 0.08% for SS, 0.04% for TN, and 0.12% for TP. For Hwang River, results improved by up to 27% for SS, 12% for TN, and 15% for TP. When the anomaly analysis results were represented by correlation and R2, they showed significant improvements of up to 0.86 in SS, 0.62 in TN, and 0.57 in TP for the Geumho River and up to 0.58 in SS, 0.49 in TN, and 0.49 in TP for the Hwang River. The statistical method suggested in this study can be applied when observation for the water quality data is long or data are insufficient.