Post-wildfire runoff was investigated by combining field measurements and modelling of infiltration into fire-affected soils to predict time-to-start of runoff and peak runoff rate at the plot scale (1 m²). Time series of soil-water content, rainfall and runoff were measured on a hillslope burned by the 2010 Fourmile Canyon Fire west of Boulder, Colorado during cyclonic and convective rainstorms in the spring and summer of 2011. Some of the field measurements and measured soil physical properties were used to calibrate a one-dimensional post-wildfire numerical model, which was then used as a ‘virtual instrument’ to provide estimates of the saturated hydraulic conductivity and high-resolution (1 mm) estimates of the soil-water profile and water fluxes within the unsaturated zone.

Field and model estimates of the wetting-front depth indicated that post-wildfire infiltration was on average confined to shallow depths less than 30 mm. Model estimates of the effective saturated hydraulic conductivity, K_s, near the soil surface ranged from 0.1 to 5.2 mm h⁻¹. Because of the relatively small values of K_s, the time-to-start of runoff (measured from the start of rainfall),  t_p, was found to depend only on the initial soil-water saturation deficit (predicted by the model) and a measured characteristic of the rainfall profile (referred to as the average rainfall acceleration, equal to the initial rate of change in rainfall intensity). An analytical model was developed from the combined results and explained 92–97% of the variance of  t_p, and the numerical infiltration model explained 74–91% of the variance of the peak runoff rates. These results are from one burned site, but they strongly suggest that  t_p in fire-affected soils (which often have low values of K_s) is probably controlled more by the storm profile and the initial soil-water saturation deficit than by soil hydraulic properties.