INFILTRATION SIMULATIONS AMONG 5 HYDRAULIC PROPERTY MODELS

Analytical functions are used within computer models to give continuous representation of discrete data obtained from measurement. Functionally different equations have been used for soil hydraulic properties without concern for their smoothing effect on the data or their relationship to the stability of the numerical algorithm. This study evaluates the effect of five different power-function representations of the soil water-retention function and corresponding pore-interaction hydraulic-conductivity representations on numerical solutions of Richards' equation. The functions were curve-fit to 46 data sets for Hecla loamy fine sand (sandy, mixed Aquic Haploboroll). Minimum, median, and maximum function parameters (Campbell's b coefficient and air entry) were used to identify data sets from which one-dimensional finite-difference infiltration simulations were conducted. The wetting-front location varied 6.0 to 8.0 cm (T = 6 min) among models, whereas variation within a model due to soil-series-imposed parameter variability was slightly larger (5.5-8.5 cm). Infiltration amounts varied -27 to 12% from the average among models. The results quantify sources of error due to curve-fitting soil water-retention data to analytical functions assuming pore-interaction theory.