Dryland wheat is highly dependent on climatic factors and therefore yield is fluctuating accordingly. Since wheat plays an important role in food security, beforehand wheat yield prediction can help government decision-making. Possibility of predicting dryland wheat yield by precipitation and edaphic data was studied using regression models. Dryland wheat yield data during 1362-83 were collected from Jahad-e-agriculture of Khorasan. Eight major wheat growing areas (Bojnourd, Shirvan, Farouj, Esfarayen, Dargaz, Quchan, Mane-Semelghan and Raz-Jargalan) which have maximum average yield and cultivated area were selected. The precipitation data was used in the forms of annual (beginning from Mehr), monthly during growing season (Aban to Khordad) and total growing season precipitation (8 months from Aban to Khordad). In order to evaluate edaphic data such as soil texture, moisture content at field capacity (FC) and permanent wilting point (PWP) and also soil available water, soil samples were taken. To calculate the moisture content at FC and PWP, standard transfer functions were adopted. To obtain the regression models, some Excel, Minitab and Sigma stat software
were used. The regression methods included simple regression, multiple linear regression followed forward stepwise regression. The results obtained from 10 different regression models showed that the most important parameters for dryland wheat prediction were precipitation in Farvardin, Khordad, Aban and Esfand, soil moisture content at FC and PWP and soil clay percentage. According to regression coefficient of models, the best equations were those models which used 1) precipitation in Farvardin, khordad and Aban and soil moisture content at FC and PWP (R2 = 0.78, RMSE= 27.3) and 2) total precipitation during growing season, soil clay percentage and soil moisture content at FC and PWP (R2 = 0.71, RMSE= 33.9). |
