Development of soils data layers is an important task for LTER sites, How to determine the optimum spatial resolution for these layers for a given objective and how to gauge their accuracy are cutting- edge problems in ecosystem science. Spatially explicit soils data are needed to drive predictions of essentially all aspects of ecosystem function and pattern, e.g., hydrologic response, primary production, trace-gas exchange, and biodivcrsity. Major soil variables needed are:
- Bulk density and stone content by depth or genetic horizon uu capacity to hold plant-available water uu hydraulic conductivity and gas diffusivity
- Water infiltrability
- Pore-size distribution (which determines the previous factors, for soil organisms)
- Texture (content of sand, silt and clay)
- Clay mineralogy (this accounts for much of the spatial variation in soil chemical processes) carbon content
- Percent base saturation
Several important variables have been omitted but it is expected that they can be calculated from this subset. Also, the variables listed above are ones that change only slowly through time and are thus ap’propriate for development as spatially explicit data layers. Nutrient availability is needed also but changes rapidly in response to meteorological variables and biological processes and is thus more appropriate for development as variables in dynamic simulatioh models, not as soils data layers.
How far along are the LTER goal? A first step is a traditional soils map, which divides the area of interest into map units keyed to causal factors such as geologic substrate, topography, and (sometimes) climate. In practice, especially in mountainous terrain, the resulting map units are tremendously heterogenous internally.
Nonetheless, they offer a place to start. Without an initial soils map, there is no way to even begin to design a sampling and analysis scheme that will eventually produce needed data layers.
Phil Sollins (H.J. Andrews, AND) and Rich Boone (Harvard Forest, HFR) have polled the Sites as to the status of soil survey. Of 14 sites responding, all but Niwot (NWT) have Soil Conservation Service (SCS, or equivalent) county soil surveys, but only Coweeta CWT), HFR, Konza (KNZ), and Luquillo (LUOJ have high-intensity maps (scale larger than 1:2500) of all or part of a site. Not surprisingly, neither of the most mountainous sites (AND and NWT), where short-range variability (on the order of meters) is often larger than even the coarsest-scale differences at the other sites, considers its survey to be very useable. All classification is based on old versions of Soil Taxonomy (1975 or earlier). GIS availability is excellent -- all but KNZ, NWT, and Virginia Coast have SCS county surveys or equivalent available as a GIS data layer. Availability of lab results is generally not good, and samples collected during the soil survey have been archived by only CWT, HFR, Kellogg, and LUQ.
The full results of the survey are available at LTERnet.edu through Gopher (reports/soils), and an initial SOILS mailing list has been set up. John Kimball of SCS has again offered his agency’s help in designing and implementing intensive surveys.
The initial process of soil survey is well along across the LTER Network. But, before soil pattern and variability can be taken into account when scaling up from plot-based results, methods are needed for determining:
- Spatial resolution optimal for the questions to be asked and influences turnover time sites toward reaching this
- Accuracy of the resulting data layers. Soil samples must also be archived if long-term trends are to be assessed.
—Phil Sollins and Rich Boone