Comparing soil carbon estimates in glaciated soils at a farm scale using geospatial analysis of field and SSURGO data

Soil carbon is a key soil property related to ecosystem services and it is often used in soil carbon content estimates at various scales. Uncertainties in soil carbon estimates often arise from variability in field, laboratory, and/or geospatial data at a farm scale. The objectives of this study were to quantify and compare levels of soil organic carbon (SOC), soil inorganic carbon (SIC), and total soil carbon (TC) for a 147-hectare field site in upstate New York based on three alternative analysis procedures: a) using carbon concentrations reported by the Soil Survey Geographic (SSURGO) spatial databases for each soil map unit (SMU) present at the field site and applying that value across each SMU; b) averaging the carbon contents of soil cores collected within a specific SMU boundary and applying the averaged value across each SMU; and c) interpolating carbon contents across the field site based on the individual soil cores. Maps of SOC, SIC, and TC contents based on the interpolated core samples were different from maps created by applying averaged core results or SSURGO values across the SMUs. Differences in the magnitudes and spatial distributions of carbon can be attributed to several factors. For example, SSURGO soil carbon values are frequently measured for a selected pedon(s) from a “type location” and not from the actual study location. These “type locations” can be located far from study sites and even in different states. Also, SSURGO soil carbon values may overestimate the actual contents when compared to systematic field measurements because the SSURGO values at lower depths are often extrapolated from upper soil horizons. Such extrapolations affect inorganic carbon to a much greater extent than organic carbon, therefore better agreement is observed in the present study between SOC estimates from the SSURGO database and field measurements. Because regional and/or global carbon estimates are rarely made with detailed field data due to the high costs of field and laboratory measurements, additional field sampling is needed to constrain and improve these estimates and also to assess the potential variability present.