Spatial Variability of Wheat Grain Yield and Soil Properties in Semi-Arid Rainfed Croplands in South of Zanjan Province

For zoning of wheat grain yield, simple kriging, cokriging and regression- kriging methods were evaluated. For this purpose, in 2014, 298 rainfed wheat fields with distances of 0.59 km from each other were selected at the Khodabandeh county located in the south of Zanjan province. Wheat grain yield of the rainfed lands was measured using the quadrant plots in the field. Soil samples were collected from the surface soil (0-30 cm) of the fields at two replicates. Physical and chemical properties of soil including sand, silt, clay, moisture, organic matter, pH, EC, calcium carbonate equivalent, total nitrogen, available phosphorus and plant-available potassium were measured. Soil data were used to assess the geostatistical models and evaluate the soil characteristics roles on wheat grain yield over the fields. Out of total data, 238 data were used to develop geostatistical models and 60 data were applied to test the models. Results showed that cokriging estimated wheat grain yield in new locations more accurate (R2= 0.74, RMSE= 118.95) than the simple kriging (R2= 0.74, RMSE= 127.87) and regression- kriging (R2= 0.26, RMSE= 198.84). Soil total nitrogen was used as auxiliary variable in cokriging method to determine the spatial variation of wheat grain yield. In addition, according to the results by understanding the spatial variation of the soil total nitrogen spatial grain yield variation can be quantified and used as geostatistical map for management scopes.