guest ::
| Home > Publications database > Geostatistical co-regionalization of soil hydraulic properties in a micro-scale catchment using terrain attributes |
| Home > Publications database > Geostatistical co-regionalization of soil hydraulic properties in a micro-scale catchment using terrain attributes |
| Journal Article | PreJuSER-50874 |
; ;
2006
Elsevier Science
Amsterdam [u.a.]
This record in other databases: 
Please use a persistent id in citations: doi:10.1016/j.geoderma.2005.05.008
Abstract: Any effort of distributed hydrological modeling requires the spatially distributed input of soil hydraulic properties and soil thickness. Most of the hydrological models are sensitive concerning these soil properties, thus the use of point measurements and co-variables should be optimized for a most accurate spatial prediction. During this study, we focus on the use of terrain attributes as co-variables. In order to determine the dependencies between the soil properties and topography, we derived 17 terrain attributes for a small rural catchment (28.6 ha). Correlation statistics between these terrain attributes and soil hydraulic properties calculated from measured grain size distribution and organic carbon content with pedo-transfer functions were used to identify terrain attributes as co-variables for the spatial prediction of the soil properties. We detected in particular for the following terrain attributes a high prediction potential for soil properties: relative elevation, slope of the catchment area, radiation angle and morphometric units such as slope elements. We also compared the performance of multiple regression, ordinary kriging, external drift kriging and regression kriging model C to estimate the spatial distribution of topsoil and subsoil hydraulic properties and horizon thickness. The prediction errors for the spatial structure of soil hydraulic properties according to Mualem/Van Genuchten and horizon thickness were quantified by a cross validation procedure. We determined the regression kriging model C as the most appropriate method with, on average, the smallest prediction errors and because the resulting spatial structure corresponds to recent models of soil properties spatial structure. Compared to ordinary kriging without covariables, the spatial prediction of soil properties could be improved with up to 15% by using terrain attributes as co-variables. (c) 2005 Elsevier B.V. All rights reserved.
Keyword(s): J ; DEM (auto) ; kriging (auto) ; PTF (auto) ; interpolation (auto) ; soil physical properties (auto) ; spatial distribution (auto)
|
The record appears in these collections: |