Combination of conventional and geostatistical mapping methods to estimate some soil ‎physical and chemical properties ‎(A case study: Sarduieh region, Kerman province)‎

Background and objectivesþ:þý Because of wide variety of soil properties (including physical, chemical ýand biological), the type and rate of change of these properties in relation to the cultivation of ýagricultural products is great importance. For optimal management of soil, it is need to know these ýproperties. Since it is very expensive to measure soil characteristics in the laboratory and for a broad ýarea, estimation of their spatial distribution is required. This study has estimated some soil physical and ýchemical properties in geoform map units of some parts of Sarduieh region using the conventional soil ýmapping method, kriging and their combination as single estimator and has compared the results with ýeach other.ý
150 observation points (with a mean distance of 200 m) as a random classified ýsampling pattern were taken from the topsoil (0-30 cm) of the area. Besides, undisturbed samples were ýtaken from each observation point by cylinder. After air drying the samples and passing them through ýthe 2 mm sieve, required laboratory analyses were done on the samples. After calculation of estimated ývalues and error variance of the studied variables using the mentioned estimators, their validation was ýconducted using coefficient of determination (R2) and relative root mean square error (RMSE%). ý
All the studied soil properties, except bulk density and soil reaction, had moderate variability. ýThe spatial structure of all variables (except volume percentage of coarse fragments, organic matter ýpercentage, and electrical conductivity, which had strong spatial structure) was moderate and spherical ýand exponential models were well able to model the spatial structure of soil properties. The range of ýmodels for various properties was variable between 450 m for field capacity and 1945 m for soil ýreaction. ý
A distance of approximately 880 meters, which represents an average range of studied properties, can be ýconsidered as an optimum sampling distance for future studies in this region. ý
Results showed when the spatial correlation class of a variable was strong, the kriging ýestimator and combined method had a more successful estimation of that property. Besides, in such ýsituation, using combined estimator leading to a significant reduction in amount of estimations’ bias. ýOn the other hand, using the combined method for estimating other soil properties in the study area ýthat had moderate spatial correlation classes did not have a significant preference than two other ýestimators, i.e, geoform and kriging maps.