Improvement of SCS-CN model based on antecedent precipitation and rainfall duration

Abstract: An accurate prediction of the runoff has been one of the important steps in the water supply in recent years. However, there is a sudden jump in the runoff prediction under the general configuration of Antecedent Moisture Condition (AMC) by Soil Conservation Service Curve Number (SCS-CN). At the same time, the rainfall duration cannot be considered as an important component of the surface runoff. Therefore, it is necessary to modify the prediction model of rainfall runoff for ecological and economic significance. Taking the Miyun reservoir in Beijing of China as the research object, this study aims to propose an improved SCS-CN model using antecedent precipitation and rainfall duration. A partial correlation analysis was first made between the rainfall factors (rainfall, rainfall duration, and average rainfall intensity) and slope runoff. The rainfall and rainfall duration were then selected as the important factors affecting the slope runoff in the study area. Subsequently, an updated SCS-CN model was established to combine with the early rainfall and rainfall duration. The Antecedent Prediction Index (API) was also used to simulate the soil's early water conditions. A static infiltration equation was considered the soil early rainfall and infiltration in a rainfall event. Among them, the potential maximum water storage was equal to the sum of the previous soil moisture and the potential soil water storage. The maximum static infiltration rate was the static infiltration rate when the watershed soil moisture reached the full storage, and the static infiltration coefficient was dimensionless related to the soil structure and land use. The monitoring data was collected from the 200 rainfall runoff events in the runoff community of the Shixia basin from 2006 to 2010 and 2014 to 2020. The newly improved model was finally verified to compare with the original and two improved SCS-CN models. The results showed that the improved model performed best among the four runoff models, where the Nash efficiency coefficient was 0.77, the determination coefficient was 0.79, and the root mean square error was 3.21 mm. The Nash efficiency coefficient, the determination coefficient and increased by 319%, 97.5%, and root mean square error was reduced by 107.5% respectively, compared with the original SCS-CN model. Furthermore, the improved model was much better than the rest, where the four SCS runoff models had underestimated the runoff. It was found that the improved SCS-CN model was positively correlated with the yield. Nevertheless, the improved model was not suitable for the rainfall runoff events with the normal soil moisture, grassland land use type, and rainfall type I (short rainfall duration, small rainfall, high rainfall intensity, and low frequency). The parameter sensitivity analysis showed that the most sensitive parameters were the potential maximum water storage and static infiltration coefficient. Specifically, there was a general parameter sensitivity of the maximum static infiltration velocity, whereas, the initial loss rate was the worst. Consequently, the improved model presented strong applicability for the rainfall runoff of Chaobai River Basin in the upper reaches of Miyun reservoir. This finding can provide a strong reference basis for the calculation of runoff yield.