Leveraging Wetland Soil Carbon Data Sets to Address Spatial Representation of Wetland Carbon Stocks

Understanding the spatial representation of wetland soil carbon stocks is important for overall carbon cycling projections because wetlands have the capability to sequester a large amount of organic carbon relative to their area. An in-depth analysis of wetland soil data was conducted to produce a map of wetland soil organic carbon stocks for the conterminous United States and leverage these databases to explore carbon related questions. This project includes CONUS wetland sample data from the EPA National Wetland Condition Assessment (NWCA) data set, wall-to-wall organic carbon percentage data from the USDA NRCS Soil Survey Geographic Database (SSURGO) data set, coastal wetland soil carbon data from Smithsonian Coastal Carbon Research Coordination Network (CCRCN), and additional soil data from the wetlands surveyed as a part of NSF National Ecological Observatory Network (NEON) data set. Information from SSURGO was extracted as a binned frequency table using a Python script. All further analyses were carried out in R programming language, allowing for the mean and several quantiles to be compared between data sets on a region-by-region and vegetation type-by-vegetation type basis. A broken stick linear regression technique was applied to the NWCA data on organic carbon content, to determine "mineral vs organic" model representation for soil carbon stocks to 1m depth. Both a linear least square and a quantile fit technique were explored for this analysis. Overall the results of these analyses support the position that SSURGO and NWCA datasets are cross validated at a regional scale, although there may be issues of representative sampling in some cases. This work serves as a foundational element for an ongoing USGS National Wetland Carbon Assessment and modeling project.