Comparison of Brooks-Corey and Rossi-Nimmo Retention Functions in Modeling Soil-Water Movement at a Site in the Amargosa Desert, Nevada

B J Andraski (US Geological Survey, 333 West Nye Lane, Carson City, NV 89706; 775-887-7636; e-mail: andraski@usgs.gov) and E A Jacobson (Dept. of Geoscience, University of Nevada-Las Vegas, Las Vegas, NV 89154; 702-895-1237)

Recent work has emphasized development of modified functions to improve the description of water retention across the entire range of soil moisture from saturation to oven dryness, but application and testing of such functions in computer simulation modeling has been limited. The objective of this work was to compare the Brooks-Corey (BC) function and a variant, the Rossi-Nimmo (RN) two-parameter function, in modeling soil-water movement at an arid site. Laboratory characterization data, field monitoring data, and 3.85-yr simulations of water and heat flow for a devegetated native-soil site were used in the analysis. The highly stratified soil profile served as a natural analog for a multi-layered, waste-cover system and consists of three layers: layer 1, loamy sand, 0-1 m; layer 2, gravelly coarse sand, 1-2 m; layer 3, gravelly coarse sandy loam, 2-5.5 m. Under dry conditions, use of the RN function improved the tracking of trends indicated by thermocouple psychrometer data (0 to -8 MPa) that were available for the 0.25-m depth; the maximum difference between simulated values was 25 MPa (BC, -43 MPa; RN, -68 MPa). Root-mean-square errors for the simulations ranged from 0.20 to 2.1 MPa for depths of 0.75, 1.05, 1.85, 2.75, and 3.85 m; within depths, the maximum difference between BC and RN values was <0.5 MPa. Field data and model results showed that the capillary break between the two uppermost soil layers impeded percolation. Neither simulation replicated the magnitude of seasonal water-potential variations measured in layer 2; an evaluation of the data suggested that field-measured variations were driven primarily by water flux changes as opposed to temperature changes. For layer 3, measured and simulated water potentials were in good agreement. For all depths evaluated, simulated net water fluxes were downward, and BC values were 6-15% greater than RN values. Results showed that use of the RN function improved simulation of drying trends in the surface-active zone.