Abstract:

 In Northeast China, snowmelt is an important source of spring runoff, and the reliable simulation and forecasting of snowmelt runoff can provide a reference for reasonable management and scheduling of water resources during the spring flood season. There are two runoff-generating models of snowmelt runoff and rainfall runoff in the cold Northeast region during the year. The differences in the physical and spatial properties of the basin have a significant impact on the snowmelt runoff generation. To solve the problem of spatial heterogeneity of snowmelt runoff, and to improve the simulation and forecasting accuracy of the SWAT model, we propose a single-site and multi-site combined parameter calibration for the SWAT model. Firstly, the initial parameter calibration of single station was carried out by using the inflow of Baishan Reservoir, and then the snowmelt runoff parameters were calibrated by multi-station, and the parameters were transplanted to the initial parameter set of single station. The correlation coefficients (R2) of monthly average flow in validation periods of the annual (January-December), flood season (June-September), and spring season (March-May) were 0.76, 0.74, and 0.58, the daily average flow R2 were 0.71, 0.75 and 0.51, after the initial parameter rate determination at a single station. The correlation coefficients R2 in validation periods of the annual (January-December), flood season (June-September), and spring season (March-May) were 0.80, 0.74 and 0.79, the daily average flow R2 were 0.74, 0.78 and 0.61, after using a combined single-site and multi-site parameter calibration strategy. The results showed that the simulation accuracy of snowmelt runoff was improved by 20% and 10% respectively.

Key words: Northeast cold region, snowmelt runoff, SWAT, parameter transplantation, spatial heterogeneity, Baishan basin