Location: Dale Bumpers Small Farms Research CenterTitle: Selection of terrain attributes and its scale dependency on soil organic carbon prediction
|GUO, ZHIXING - Guangdong Academy Of Agricultural Sciences|
|CHELLASAMY, MENAKA - Aarhus University|
|GREVE, M - Aarhus University|
|GREVE, M - Aarhus University|
Submitted to: Geoderma
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/8/2019
Publication Date: 2/1/2019
Citation: Guo, Z., Adhikari, K., Chellasamy, M., Greve, M.B., Owens, P.R., Greve, M.H. 2019. Selection of terrain attributes and its scale dependency on soil organic carbon prediction. Geoderma. 340;303-312. https://doi.org/10.1016/j.geoderma.2019.01.023.
Interpretive Summary: Topography is among the key environmental variables affecting the distribution of soil carbon across landscapes. However, the scale dependency of terrain attributes on SOC prediciton performance has not been reported. The aim of this paper was to evaluate topography/terrain attributes at varying grid resolutions or scales to select the most influencing tool for prediction for soil organic carbon at a regional scale in Denmark. We found that SOC distribution in a recently glaciated landscape, the three terrain attributes relative slope position, channel altitudei, and standard height were the most important at all resolutions and scales. This research can aid management decisions for improving soil health by understanding natural soil variability to tailor management practices to provide the most improvement.
Technical Abstract: Soil organic carbon (SOC) is an important component of the global carbon pool and therefore makes a significant contribution to the global carbon cycle. Terrain attributes are often used as predictors of SOC in digital soil mapping, but there are no fixed rules, only a few empirical, as well as, pedological guidelines on which terrain attributes and their scale dependency have been used. The aim of this paper was to evaluate 22 terrain attributes at varying grid resolutions and select the most influencing attributes for SOC prediction. A typical 7,500 km2 area located in Denmark was selected; a total of 2,514,820 prediction models based on data-mining were constructed for 71 different grid sizes ranging from 12.8 m to 2,304 m, and 22 terrain attributes derived from digital elevation model. Relative importance and usage of each attribute in each model were computed. Major indicators for the SOC prediction model were determined by weighting, grouping, summation and normalization. The results showed there were statistically significant differences in the SOC model depending on terrain attributes used. Relative Slope Position (RSP), Channel Altitude (Chnl_alti), and Standard Height (StandH) were the three most important terrain attributes in the five-attribute-model at all resolutions and the remaining two models being Normalized Height (NormalH) and Valley Depth (Vall_depth) at resolutions finer than 40 m, and Elevation and Channel Base (Chnl_base) at coarser than 40 m. The models at 88 m and 92.8 m pixel size (nearest the 90 m of the Globalsoilmap or SRTM projects) and 30.4 m (nearest the 30 m TM satellite image) were validated. We found that SOC distribution in a recently glaciated landscape, the three terrain attributes RSP, Chnl_alti, and StandH were the most important at all resolutions.