Spatial variability of digital soil maps and its impact on regional ecosystem modeling

This paper discusses some of the practical difficulties encountered in using generalized regional soils maps such as the U.S. Soil Conservation Service State Soil Geographic Data Base (STATSGO) for landscape to regional level ecosystem modeling. For highlighting regional trends, the STATSGO-derived maps are adequate. However, caution is warranted when using the same maps for modeling at a landscape level due to the generalized nature of the STATSGO data coupled with the inherent spatial variability of soils. The variability of soil water holding capacity (SWHC), within a STATSGO class was quite large with a median coefficient of variation of between 40 and 60%. Comparison of SWHC values derived from finer scale county level soil surveys with STATSGO data showed large discrepancies. Simulated outputs from the forest ecosystem model, PnET, differed by approximately 20% depending on the source of SWHC input data. To improve the utility of regional soils databases for modeling purposes, estimates of the spatial variability of soil properties need to be better quantified and communicated to the prospective data user.