عنوان مقاله [English]
Background and objectives: Accuracy of spatial estimation of soil properties such as clay, organic matter and calcium carbonate equivalent is very important in plant nutrition and environmental planning. The most important step before statistical analysis and using geostatistical estimators is data check. In addition to investigation of outlier data and data distribution, another important action in geostatistical studies is trend surface analysis. Analyzing trend surface can evaluate the role of factors such as stone kin, climate, topography and generally regional anomalies. The objective of this study was trend-surface analysis and its effects on variogram modeling and mapping of clay, organic matter and calcium carbonate equivalent.
Materials and Methods: For this purpose, 100 surface soil samples in 0-15 cm depth were selected randomly based on different classes area slope from 41353 ha area in Selin plain farmland located in Kaleibar region, East-Azerbaijan. Soil properties such as clay, calcium carbonate equivalent and organic matter were measured by hydrometer, return titration and wet oxidation method, respectively. For analyzing trend surface used multiple regression models which its independent variables was geographical coordinate and dependent variable was a soil properties. For zoning clay, organic matter and calcium carbonate equivalent and residuals of removing trend used ordinary kriging. The effect of removing trend surface in variogram modeling and kriging estimating were evaluated by cross-validation method with indexes mean error (ME), root mean square error (RMSE) and determination coefficient (R2).
Results: Trend surface analysis showed that the best regression models for trend determination of clay, calcium carbonate equivalent and organic matter were first order, first order and quadratic, respectively. Removing the detected trend led to decrease in sill but the nugget effect did not changed. However, no significant difference was observed between accuracy of kriging estimator in presence and remove of trend. This can be attributed to the fact that both abnormal manner of environment and activations of human. So that, the regression models of the trends were 35, 18 and 21% of clay, organic matter and calcium carbonate equivalent variations, respectively. However, removing the detected trend led to increase 9.1, 2.7 and 6.6% of R2 for clay, organic matter and calcium carbonate equivalent, respectively.
Conclusion: generally, investigation of trend surface recommended in soil studies that is more deals to spatial data. Because the trend depends on the location and conditions of the study area as well, the source of the creating trend trend trend trend trend.