A paper entitled ‘Combining the SWAT model with sequentialuncertainty fitting algorithm for streamflow prediction and uncertaintyanalysis for the Lake Dianchi Basin, China’ was recently published onlinein Hydrological Processes (DOI: 10.1002/hyp.9605).
Streams play an importantrole in linking the land with lakes. Nutrients released from agricultural orurban sources flow via streams to lakes, causing water quality deteriorationand eutrophication. Therefore, accurate simulation of streamflow is helpful towater-quality improvement in lake basins. In China, the degradation of waterquality in Lake Dianchi has been of great concern since the 1980s. To assistenvironmental managers in decision making, it is important to assess andpredict hydrological processes at the basin scale. This study evaluated theperformance of the soil and water assessment tool (SWAT) for predictingstreamflow in the Lake Dianchi Basin. The model was calibrated and validatedusing monthly observed streamflow values for three flow stations within theLake Dianchi Basin through application of the sequential uncertainty fittingalgorithm (SUFI-2). The results of the autocalibration method for calibratingand prediction uncertainty from different sources were also examined. Together,the p-factor (the percentage of measured data bracketed by 95% prediction ofuncertainty, or 95PPU) and the r-factor (the average thickness of the 95PPUband divided by the standard deviation of the measured data) indicated the strengthof the calibration and uncertainty analysis. The results showed that the SUFI-2algorithm performed better than the autocalibration method. Comparison of theSUFI-2 algorithm and autocalibration results showed that some snowmelt factorswere sensitive to model output upstream of the Panlongjiang flow station. The95PPU captured more than 70% of the observed streamflow at the three flowstations. The corresponding p-factors and r-factors suggested that some flowstations had relatively larger uncertainty, especially in prediction of somepeak flows. Although uncertainty existed, statistical criteria including R2 andNash-Sutcliffe efficiency were reasonably determined. The model produced auseful result and can be used for further applications.