Skip to main content
ARS Home » Research » Publications at this Location » Publication #147709

Title: USING SOIL-LANDSCAPE MODELING TO DEVELOP TRUE SPATIO-TEMPORAL SOIL CARBON MODELS AT MULTIPLE-SCALES

Author
item Venteris, Erik
item McCarty, Gregory
item Ritchie, Jerry
item SLATER, BRIAN - OHIO STATE UNIV.

Submitted to: BARC Poster Day
Publication Type: Abstract Only
Publication Acceptance Date: 3/25/2003
Publication Date: 3/25/2003
Citation: Venturis, E.R., McCarty, G.W., Ritchie, J.C., Slater, B.K. 2003. Using soil-landscape modeling to develop true spatio-temporal soil carbon models at multiple-scales [abstract]. Abs 24. BARC Poster Day.

Interpretive Summary:

Technical Abstract: At the present, there are general stochastic spatial models which can be applied to SOC inventory and temporal models (CENTURY) which predict values at a single point. Temporal models that fully account for spatial landscape processes are needed. Our initial task in the development of such models is to use stochastic spatial modeling to identify which landscape processes are most important. An essential component to this task is to identify how these processes change in importance over a range of scales. Multi-scale soil sampling programs were launched at the National Soil Tilth Laboratory in Ames, IA; at the North Appalachian Experimental Watershed in Coshocton, OH; and at the OPE3 site in Beltsville, MD. The first step in the scaling study was to sample and model small (1,000-10,000 m^2) zero and first order watersheds at these sites. One-hundred to 200 samples were collected for each site. SOC, bulk density, and Cs137 inventory (to estimate soil redistribution since 1965) were measured at each site. At this scale, gradients in carbon were hypothesized to be due to differences in soil moisture and erosion/depositional processes. These processes are strongly related to topography, so precision digital elevation models and topographic parameters (slope, wetness index, etc.) were calculated for each site. Least-squares regression models were used to predict SOC from terrain parameters (multivariate regression) and Cs137 data (univariate regression). SOC in the Iowa mollisol was strongly correlated to both terrain (wetness index) and Cs137 (r^2 0.8). SOC variability in Ohio and in Maryland ultisols was not as predictable with r^2 ranging from 0.2 to 0.5 for both terrain and Cs137. The results indicate that the strength of interaction between SOC and topography driven processes varies widely between soils. The reasons for these differences needs to be better understood to enable effective spatio-temporal modeling.