|Slater, Brian - THE OHIO STATE UNIVERSITY|
Submitted to: Soil Science Society of America Annual Meeting
Publication Type: Abstract Only
Publication Acceptance Date: October 1, 2003
Publication Date: November 4, 2003
Citation: Venteris, E.R., McCarty, G.W., Slater, B.K. 2003. Geospatial techniques for soil organic carbon mapping across multiple scales. Soil Science Society of America Annual Meetings (S05-venteris3132000oral). CDROM. Technical Abstract: Discrete soil measurements can be converted to continuous soil maps through GIS-based soil-landscape modeling. The general approach (environmental correlation) is to use statistical, geostatistical or numerical models to link discrete sample data with continuous spatial data sets such as DEMs, soil maps, and remote imagery. Such techniques hold promise for measuring soil organic carbon inventory at scales ranging from the field to continents. Several key scaling issues must be addressed. Model error is expected to increase with scale. The choice of data and modeling techniques should be optimized to minimize this trend. This goal is challenging as the resolution and accuracy of GIS data generally decreases with the increase in spatial extent. In addition, the variance in soil properties increases with spatial extent as a wider range of geomorphological and ecological environments are encountered. Dominant physio-ecological processes can change as area of interest increases. Transitions between such processes across scale must be understood. The sample efficiency of spatial modeling techniques changes with scale. Geostatistical modeling methods are practical at field scales but can be more difficult to execute at regional scales due to intense sampling requirements for univariate modeling and the need for very strong correlation between cokriging parameters. Regression techniques are proposed as one alternative to geostatistics for meso and large area studies. Models should be optimized to maximize accuracy and minimize sampling requirements. Case studies from Iowa, Ohio and Maryland are presented to illustrate scaling concepts.