Anticipation of different phenological phases of apple using artificial neural network

Background/Question/Methods Apple is one of the oldest trees in the world which is widely cultivated because of it’s highly compatibility with various climatic conditions. In this study, we applied phenological statistics of agricultural meteorology of Golmakan and also used observation which are performed in Gardening station of Karaj to anticipate different phonologic phases of apple using intelligent neural network. At first the matrix of input data which is consisting of climatic parameters such as minimum temperature, maximum temperature, the mean of daily temperature, absolute minimum temperature, and absolute maximum temperature has been established. The range of temperature changes, growing days & timed chilling unit (in silver tip phase) had been prepared for different phenological stages during 1999-2004; then the matrix of out coming data which, in fact, were the occurrence dates of different phenological stages of apple was prepared and the modeling of different phenological stages was performed by using neural network.

Results/Conclusions .

In this study the accuracy of model was examined by using RMSE index and by contrasting real & anticipation dates during 2 years. For this purpose observed climatic & phenological data was also used in similar figure at out of investigating zone. i.e. Karaj zone using paired t- test, we specified there isn't meaningful different between the rate of estimated error in the model of Karaj station in the model of Golmakan station. It’s also specified we could anticipate phenological stages of apple with acceptable accuracy using atmospheric parameters.