Refining Surface Runoff Predictions in Missouri, United States Using Enhanced SCS Curve Number Method

Accurately modeling surface runoff is essential for effective water resource management, flood forecasting, and urban planning. This study refines surface runoff predictions in the Missouri Hydrological Area (MHA) using an enhanced Soil Conservation Service Curve Number (SCS-CN) method. Land use and land cover (LULC) data from 2001 to 2021 were analyzed to calculate weighted curve numbers, accounting for regional variability. Adjustments to the SCS-CN method, including improved formulations for initial abstraction, enhanced the predictive accuracy. The outputs were compared with observed and simulated surface runoff from the United States Geological Survey (USGS) and the North American Land Data Assimilation System (NLDAS), respectively.

To calculate the surface runoff, one of the major inputs is Curve Numbers (CN) which is predominantly based on land cover. The LULC analysis revealed that agricultural lands dominate the region, covering approximately 51% of the total area, followed by forests (30%), and developed or built-up areas (7%). Shrublands, grasslands, wetlands, and barren lands collectively account for the remaining area, with wetlands showing significant fluctuations due to environmental changes and restoration efforts. Weighted CNs were calculated for the study area, with values ranging from 30 to 98, depending on land use, soil type, and hydrological conditions. Agricultural lands and developed areas exhibited higher CN values, reflecting higher runoff potential, while forested and wetland areas had lower CNs, indicating greater infiltration capacity.

In addition to CN, precipitation is another input. Hourly precipitation data (0.125° × 0.125° lat/lon grids) are obtained from the NLDAS for the period January 2001 to December 2021. This dataset captures the region’s substantial variability, with annual precipitation ranging from 950 mm to 1,540 mm, reflecting distinct seasonal patterns and spatial heterogeneity within the study region.

The surface runoff estimated using the updated SCS-CN method was validated against the USGS Quick-Flow runoff estimates. Statistical metrics, including Nash-Sutcliffe Efficiency (NSE) and Kling-Gupta Efficiency (KGE), highlight the improved reliability of the enhanced method, with NSE and KGE values of 0.5608 and 0.4475, respectively, for the updated CN-formulation. In contrast, the original equation with the constant initial abstraction ratio of 0.2 yielded lower NSE and KGE values of 0.4055 and 0.1763. These results emphasize the importance of refining CN estimates, which explains more variance, and aligns more closely with observations. This adaptability to regional hydrological conditions makes the enhanced method a choice for accurate surface runoff predictions.