Mathematical models have become increasingly popular in both research and management problems involving flow and transport processes in the subsurface. The unsaturated hydraulic functions are key input data in numerical models of vadose zone processes. These functions may be either measured directly or estimated indirectly through prediction from more easily measured data based using quasi-empirical models. Rosetta V1.0 is a Windows 95/98 program to estimate unsaturated hydraulic properties from surrogate soil data such as soil texture data and bulk density. Models of this type are called pedotransfer functions (PTFs) since they translate basic soil data into hydraulic properties. Rosetta can be used to estimate the following properties:

Water retention parameters according to van Genuchten (1980) Saturated hydraulic conductivity Unsaturated hydraulic conductivity parameters according to van Genuchten (1980) and Mualem (1976)

Detailed description of the hydraulic functions Rosetta offers five PTFs that allow prediction of the hydraulic properties with limited or more extended sets of input data. This hierarchical approach is of a great practical value because it permits optimal use of available input data. The models use the following hierarchical sequence of input data

Soil textural class Sand, silt and clay percentages Sand, silt and clay percentages and bulk density Sand, silt and clay percentages, bulk density and a water retention point at 330 cm (33 kPa). Sand, silt and clay percentages, bulk density and water retention points at 330 and 15000 cm (33 and 1500 kPa)

The first model is based on a lookup table that provides class average hydraulic parameters for each USDA soil textural class. The other four models are based on neural network analyses and provide more accurate predictions when more input variables are used. In addition to the hierarchical approach, we also offer a model that allows prediction of the unsaturated hydraulic conductivity parameters from fitted van Genuchten (1980) retention parameters (Schaap and Leij, 1999). This model is also used in the hierarchical approach such that it automatically uses the predicted retention parameters as input, instead of measured (fitted) retention parameters. All estimated hydraulic parameters are accompanied by uncertainty estimates that permit an assessment of the reliability of Rosetta's predictions. These uncertainty estimates were generated by combining the neural networks with the bootstrap method (see Schaap and Leij (1998) and Schaap et al. (1999) for more information).