The basic soils information used in FLDAS was derived from the global soils dataset of Reynolds, Jackson, and Rawls [1999, WRR2000], which is available online here. That dataset includes fractions of sand, silt, and clay, and porosity, among other fields, and it is based on the FAO Soil Map of the World linked to a global database of over 1300 soil pedons (sample cores). The spatial resolution of the Reynolds maps is 5 minutes and there are two layers: 0-30 cm and 30-100 cm. These maps were spatially resampled to simulation resolution (1/4 or 1 degree) for FLDAS, and vertically inter/extrapolated to 0-2 cm, 2-150 cm, and 150-350 cm. (See GLDAS soils site for more info.)
Certain model parameters are indexed based on a soil texture class and different model uses different texture class. Therefore a routine is included in LDAS to compute texture class based on fractions of sand, silt, and clay in a given grid cell. (Same as GLDAS, outside CONUS) Noah land model uses soil texture map.
GeoWRSI uses a derived Water Holding Capacity (WHC) map available from the FAO World Harmonized Soil Database.