Hydrologic simulation (runoff) models incorporate numerical algorithms, which are individually subject to computational errors due to spatial and temporal discretization. Infiltration errors can greatly affect parameter calibration for runoff, groundwater recharge, and evapotranspiration, and can diminish the predictive capability of a model. Errors in infiltration attributed to temporal discretization were quantified for the Hydrological Simulation Program-FORTRAN (HSPF), a widely used hydrologic model. Analysis is provided to quantify the maximum rate and cumulative volume errors that result from time-step selection. Maximum errors occur at initial surface saturation and immediately thereafter. Maximum rate and cumulative volume errors for HSPF are found to be unacceptably high for some combinations of infiltration parameter values and time-step length. For example, the maximum rate and volume errors are 158 and 74%, respectively, for a 5 cm/h storm event using a 1 h model time step. Equations are provided to estimate infiltration errors for varying combinations of model parameter values and time-step lengths. Recommendations for time-step selections are provided for shallow depth-to-water-table conditions, which can have implications for spatial discretization. For sandy soils with 1-3 m depths-to-water table, the results suggest that a time-step length of less than 15 min may be necessary to manage numerical errors in infiltration.
