Estimating regional carbon stocks and spatially covarying edaphic factors using soil maps at three scales

Abstract

Most estimates of regional and global soil carbon stocks are based on extrapolations of mean soil C contents for broad categories of soil or vegetation types. Uncertainties exist in both the estimates of mean soil C contents and the area over which each mean should be extrapolated. Geographic information systems now permit spatially referenced estimates of soil C at finer scales of resolution than were previously practical. We compared estimates of total soil C stocks of the state of Maine using three methods: (1) multiplying the area of the state by published means of soil C for temperate forests and for Spodosols; (2) calculating areas of inclusions of soil taxa in the 1:5,000,000 FAO/UNESCO Soils Map of the World and multiplying those areas by selected mean carbon contents; and (3) calculating soil C for each soil series and map unit in the 1:250,000 State Soil Geographic Data Base (STATSGO) and summing these estimates for the entire state. The STATSGO estimate of total soil C was between 23% and 49% higher than the common coarse scale extrapolations, primarily because STATSGO included data on Histosols, which cover less than 5% of the area of the state, but which constitute over one-third of the soil C. Spodosols cover about 65% of the state, but contribute less than 39% of the soil C. Estimates of total soil C in Maine based on the FAO map agreed within 8% of the STATSGO estimate for one possible matching of FAO soil taxa with data on soil C, but another plausible matching overestimated soil C stocks. We also compared estimates from the 1:250,000 STATSGO database and from the 1:20,000 Soil Survey Geographic Data Base (SSURGO) for a 7.5 minute quadrangle within the state. SSURGO indicated 13% less total soil C than did STATSGO, largely because the attribute data on depths of soil horizons in SSURGO are more specific for this locality. Despite localized differences, the STATSGO database offers promise of scaling up county soil survey data to regional scales because it includes attribute data and estimates of areal coverage of C-rich inclusions within map units. The spatially referenced data also permit examination of covariation of soil C stocks with soil properties thought to affect stabilization of soil C. Clay content was a poor predictor of soil C in Maine, but drainage class covaried significantly with soil C across the state.