modeling of yield estimation for the main crops in iran based on mechanization index (hp ha-1)

agricultural mechanization is a method for transiting from traditional agriculture towards industrial and sustainable one. due to the limitation of natural resources and increasing population we need to have economical production of agricultural crops. for reaching this destination; agricultural mechanization has a remarkable role. so it is necessary to have an extensive view for mechanization, because with the help of mechanization the agricultural inputs such as seeds, fertilizer and even water and soil can effectively be managed for an economical and sustainable production. this study has been carried out in many provinces of iran. the data of agricultural tractors and cereal combine harvesters were firstly gathered by means of questionnaire. the tractors were categorized in four power levels of less than 45, 45 to 80, 80 to 110, and more than 110 hp. in addition, it was also carried out for cereal combine harvesters; it was in three power levels, i.e. between 100 to 110, 110 to 155 and 155 to 210 horse-power in 3 ages, i.e. less than 13, between 13 to 20, and more than 20 years. information regarding to cultivation areas, production volume, and yield of main crops gathered from statistics of ministry of jihad-e-agriculture. then agriculture mechanization level index (hp ha-1) in each province was calculated. four main crops including irrigated and rain-fed wheat and irrigated and rain-fed barley, which met the required criteria to be used in the model, were statistically analyzed. correlation analysis was carried out in order to get an effective model between yield of the four main crops in iran and agriculture mechanization level index. pearson correlation index showed that there is a direct and significant correlation between these variables. subsequently, outliers were identified in order to get a model with necessary efficiency to predict the yield through mechanization level index, by scatter diagram and estimating regression lines in 1% probability level. the effective model was estimated with acceptable coefficient of determination 0.851, after removing outliers.