Following the curve? Reviewing the physical basis of the SCS curve number method for estimating storm runoff

Abstract Less attention has been paid to runoff generation from semi‐arid than from humid‐temperate catchments. The SCS curve number approach is simple to apply and widely used, but lacks physical underpinning. Here output from a runoff generation models is compared with data from field measurements, making use of 11 years data from rainfall and runoff events at the Sierra de Enguera Soil Erosion Experimental Station in Eastern Spain. Runoff from natural rainfall events was monitored for 10 years on bare plots of 1–16 m length. The largest storm event was of 142 mm, generating runoff of up to 115 mm on the smallest plots. The model presented simulates overland storm flow on a sloping rough and unvegetated surface, representing an area of 320 × 320 m. Green‐Ampt infiltration constants are randomly assigned to each cell in a 128 × 128 grid, and rectangular storms applied at a range of total amounts and intensities to simulate runoff at each transect across the area. A simple algebraic expression is developed to estimate total runoff and storage in terms of storm size and duration, and plot length, with parameters that reflect infiltration behaviour, and this expression is compared with the SCS curve number approach. For the very largest storms, both expressions converge asymptotically towards 100% runoff, but the revised expression greatly improves estimates of runoff from smaller events. Output of these simulations is compared with measured storm runoff data on bare runoff plots at the Sierra de Enguera experimental Station in SE Spain and gives further support to the proposed expression for storm runoff.