Integrating seasonal forecasting using K- nearest neighbor (k-NN) method and CERES-wheat model for management of rainfed wheat cultivation

Seasonal weather forecast for the upcoming season may serve as a usefool tool for making management decisions which may decrease the production costs and associated risks. In this study attemps haven been made to combine a seasonal weather forecast approach based on k-NN nearest neighbor and dynamic simulation model CERES-Wheat model as a decision support system of farm management practices (planting date and nitrogen applicatin level) for rainfed wheat (variety Dehdasht) using the data of a field experiment at Izeh research station, Khuzastan province, Iran during 2015-2016 growing season. The results showed that the k-NN approach and CERES-Wheat model have an acceptable performance in seasonal weather forecast and crop growth simulation, respectively. By combining k-NN and CERES-Wheat models, the appropriate sowing time of the selected variety in Izeh region was determined to be between November 5 and early December. The recommended amount of applied nitrogen fertilizer in dry and rainy seasons are 50 and 150 kg ha-1, respectively. The proposed combined approach can be used as a suitable decision support system of rainfed crops in other climatic regions.