Analysis of spatial structure of some soil properties using geostatistical methods (Case study: west rangelands of Taftan- Khash)

The present study aimed to investigate the spatial variability of soil properties in the Taftan rangelands of Khash city. Homogeneous units were determined by overlyaing slope, aspect, elevation, and geology maps (1: 25000 scales) of the study area. Random systematic method was used to collect data by plots placed along three transects with 150 m length. Soil samples from 0 to 15 and 15 to 30 cm depths were collected by drilling 24 holes. Soil samples were analyzed and their characteristics including gravel percent, texture, saturation moisture, available water, lime, gypsum, organic matter, acidity (pH) and electrical conductivity (EC) were measured by the standard methods. Kriging analysis was used to examine and describing spatial structure by GS software version 9. Model were fitted to the variogram and model parameters were then determined. Accuracy assessment of the interpolation results was performed through cross-validation, and their Mean Bias Error (MBE) and Mean absolute Error (MAE) were calculated. According to the results, most of soil characteristics followed a spherical and exponential shape models in both surface and deep soil layers. Although some soil properties such as lime content of soil surface (with 4.70 nugget effect and 80% spatial structure), available water in the both soil layers (with 0.29, 0.1 nugget effect and 97%, 99% spatial structure respectively) and soil texture (sand, clay and silt percentage) had strong spatial structure. Some charactristics such as gypsum, saturation moisture and gravel did not show strong spatial structure in the both layers. According to the fitted model, kriging could not obtain a proper estimation of gravel amount in the first soil layer (MAE=11.93 and MBE= -0.708) and soil sand content in the second layer (MAE=10.51 and MBE= 0.15) because their high values of nugget effects and weak spatial structures resulted to the high error rate and estimation bias of these variables.