| Abstrakt: |
Soils are natural bodies that exhibit intricate variations across diverse landscape, evolving dynamically over both space and time. When evaluating and modeling spatial variability, multiple variables are commonly sampled alongside the primary point of interest. Moreover, efficient land management can be supported by grouping spatial data to delineate homogeneous zones. As soil heterogeneity is characterized by both continuous variation and discontinuities, the aim of this research was the combination of the classic pedological approach, using only soil data, with a digital mapping approach that uses both soil and ancillary data. The study area, covering an area of 100,000 ha is located in the Bam region, southeast Iran. The major land use in the study area includes pasture; bare soil and farmland. Mean annual precipitation and temperature of the region are 59 mm and 23 °C, respectively. Soil samples were taken up to 0.20 m-depth at 116 locations and some physical and chemical properties, including sand, silt, clay, sodium adsorption ratio (SAR), electrical conductivity (EC), pH, calcium carbonate equivalent (CCE), soil organic matter (SOM) and CaSO4, were determined. First, the variables were interpolated using a combination of block kriging with irregular geographical units and simple kriging with varying local means, determined according to a previous soil stratification, and then submitted to a clustering approach including also terrain attributes from an existing DEM, remote sensing variables and spatial coordinates. Pseudo F statistic, R squared and Cubic Clustering Criterion were used to optimize the number of clusters that was set at four zones which differed significantly with respect to the selected properties. The proposed method described in this research could be efficiently used to delineate spatially homogeneous zones to optimize land-use planning. [ABSTRACT FROM AUTHOR] |
