Optimizing the soil conservation service curve number model by accounting for rainfall characteristics: a case study of surface water sources in Beijing

• Published in: Environmental monitoring and assessment 2021-03, Vol.193 (3), p.115, Article 115
• Main Authors: Song, Wenlong, Jiao, Jian, Du, Pengfei, Liu, Hongjie
• Format: Article
• Language: English
• Subjects: Accuracy; Atmospheric Protection/Air Quality Control/Air Pollution; Beijing; Catchment area; Correlation analysis; Cultivated lands; Drinking water; Earth and Environmental Science; Ecology; Ecotoxicology; Environment; Environmental Management; Environmental Monitoring; Environmental science; Land use; Model accuracy; Monitoring/Environmental Analysis; Optimization; Rain; Rainfall; Rainfall amount; Rainfall processes; Runoff; Runoff forecasting; Simulation; Soil; Soil conservation; Surface runoff; Surface water; Water; Water Movements; Water sources; Watersheds
• ISSN: 0167-6369; 1573-2959
• DOI: 10.1007/s10661-021-08862-0

The upper catchment of the Miyun reservoir is an important drinking water source in Beijing. In recent years, researchers have used the soil conservation service curve number (SCS-CN) model to calculate surface runoff for the district. Although the runoff forecasting accuracy was unsatisfactory, the lack of understanding of rainfall processes and their influence on runoff may explain the observed deviations. Our study sought to optimize and assess the SCS-CN model simulation accuracy for the district by proposing an SCS-CN calculation method for each runoff event (CN t ) based on observation data for 253 rainfall and runoff events from 7 plots in the Miyun Shixia watershed. This study elucidated a significant positive correlation between the ratio of CN t and the average SCS-CN (CN 1 ), as well as the ratio of the maximum X-minute rainfall amount ( P X ) to the total rainfall amount for each rainfall event ( P ). Furthermore, a calculation method involving power function equations between CN t /CN 1 and P X / P was proposed for CN t . When X  = 5 min and the initial abstraction ratio ( λ ) = 0.01, the simulation performance of the optimized model was the highest, with a Nash-Sutcliffe efficiency coefficient of 0.791, which was significantly higher than that of the non-optimized SCS-CN model. The simulation performance for bare and cultivated land was higher than that of other land uses, with E f values of 0.831 and 0.828, respectively. Future research should focus on improving the prediction accuracy of runoff events resulting from high-intensity and short-duration rainfall events.