Efficiency of some geostatistical estimators for interpolation and mapping some soil quality properties
Subject Areas : Farm water management with the aim of improving irrigation management indicators
Seyedeh Mahbube
Mousavifard
1
(M.Sc. of Soil Science, College of Agriculture, University of Urmia, Iran)
Hamid Reza
Momtaz
2
(Assistant Professor, Soil Science Department, College of Agriculture, University of Urmia, Iran)
Habib
Khodaverdiloo
3
(Assistant Professor, Soil Science Department, College of Agriculture, University of Urmia, Iran)
Keywords: geostatistics, interpolation techniques, Kriging, soil quality properties,
Abstract :
For design and implementation of soil conservation programs, accurate and optimized mapping of soil properties is required. The type of interpolation method is one of the most important issue involved in mapping soil properties. Six ordinary methods of spatial interpolation were compared to determine their suitability for estimating some quality properties of topsoil including clay content, electrical conductivity, equivalent calcium carbonate and organic carbon in the Naqade region farmlands located in west Azerbaijan. The area of study region was about 8300 ha. For this purpose, 282 topsoil samples were collected randomly and tested for laboratory analysis. Deterministic methods (radial basis function and global polynomial interpolation) as well as geostatistical estimators (ordinary Kriging) were utilized to map related soil properties. The highest and lowest CV was obtained for EC and clay content variables, respectively. For clay and electrical conductivity, the exponential model, for equivalent calcium carbonate the Gaussian model, and for organic carbon the spherical were the best fitting models to calculate the experimental variogram. According to the results, the ordinary Kriging approach characterized with the minimum error indicator of bias-variance trade-off (BVTO) for evaluating clay, electrical conductivity, equivalent calcium carbonate and organic carbon. The thin plat spline model of radial basis function showed the maximum error in assessment of studied attributes. Eventually, the interpolation maps were generated regard to the most appropriate interpolators.
