عنوان مقاله [English]
نویسنده [English]چکیده [English]
The main objective of this study is finding a proper model for predicting the yield of dry farmed wheat with the use of climate parameters in the province. For predicting the yield of dry farmed wheat, neural networks have been used. Firstly, the yield data of wheat during statistical period of 1995-2003 from information bank of Ministry of Agriculture for each township was prepared separately and then meteorological statistics from the existing stations in these townships were extracted from information bank of Iran’s Meteorological organization for similar statistical period. 7 years of the existing statistic were considered for model training and two years were considered for the test file. To adopt the best model, it was required to determine the best input matrix of meteorology data. For this purpose, the first input matrix containing 9 initial meteorological parameters which finally through calculating the error amount of the model, the best composition was obtained when the parameters of precipitation, temperature, number of days with heat and cold stress, transpiration, evaporation and number of rainy days have been included in the input matrix.
The results showed that, the first factor has the highest role in determining the yield of dry farm wheat in East Azarbaijan province.
The second effective factor on the yield of dry wheat is the amount of evaporation and transpiration. To evaluate the accuracy of model based on the predicted yield, the index for compliance rate (d) was calculated, and the results showed that the accuracy rate is 0.82.