Spatial modeling of rainfed wheat yield using agroclimaticmicrozonation in Kurdistan

This study investigated, the role of variables and agroclimatic indices on rainfed wheat yield in Kurdistan province. The database was formed by collecting 31-year circuit data on planting area, production, losses and yield of wheat in 10 regions of Kurdistan province, and data of climatic elements on the hourly, daily, decade, monthly, seasonal and yearly levels related to 22 nearby synoptic stations. Data base completed after data preparation. 113 climatic and agroclimatic indicators were calculated using 15 primary meteorology elements of meteorological stations. Finally based on thecrop calendar and phonological stages of wheat, to investigate their effects on yield, 128 independent variables were selected. In order to calibrate the dates of the calculated phonological stages, the estimated agronomic calendar was compared with the recorded calendar results of neighboring rainfed wheat research stations of International Center for Agricultural Research in Dry Areas (ICARDA) in Kurdistan and other adjoining cool provinces. To identify and isolate variables that independently of each other affect yield, most correlated (Significantly) variables were eliminated. The correlation of the variables determined with wheat yield was measured. The effect of variables on wheat yield rate (kg/ha.year) was evaluated using multivariate regression “Enter” and “Stepwise” methods. Identified effective variables and indicators were interpolated based on geographical and height characteristics. According to the spatial variations of variables, the spatial model of wheat yield was introduced for province and areas of the province.
The results of this study showed that with a 99 percent confidence, climatic elements (variables) vary in different regions. Most of the independent variables have a significant effect on wheat yield in simple linear regression, but in Stepwise method, due to the internal correlation between variables, just variables entered that have insignificant correlation with each other and have more effects than other variables. The variables affecting the performance are differentin various regions, and from the point of view of effectiveness, the arrangement of the variables in different areas vary too. In other words, even in two regions with a climatic type (based on the Modified De Martonne method), both agro-climatic indices and wheat yield are significantly different. The impact of effective variables on yield at any time and place depends on the time of year and the phonological stage of wheat. At one time the environmental conditions of different regions in terms of temperature, humidity and precipitation differ, based on phonological stages of the regions. The time of the vulnerability of wheat varies from place to place. Wheat vulnerability at flowering stage is more than other stages. The effect of independent variables on yield at different times of year is proportional to the phonological stage in years Different and different in different regions. In Kurdistan province, the number of rainy days of the year, total degree hours the temperature reaches below -11 °C (sum of hours with below -11 °C temperature) from germination to tillering stage, the annual precipitation, the rainfall in the fifth decade of the water year (the precipitation of 11-20 of November), annual relative humidity and total degree hours the temperature reaches above 30°Ctemperature (sum of hours with above 30 °C temperature) in milky and dough stage is the determinants of the production of rainfed wheat. In Baneh and Marivan areas, the coefficient of variation (CV) is lower and in Bijar and Divandareh regions CV is more than other regions. Kamyaran region has the highest yield, Baneh and Marivan were ranked secondjointly. Sanandaj and then Bijarhave the lowest yield. Each region has a model for wheat yield and determinant factors vary from region to region. Although the annual production of Bijar is higher than other areas, wheat production in the Bijar region has a higher risk than other areas.