عنوان مقاله [English]
Climate changes and phenomena such as drought are effective in the yield of agricultural products. Replacing crisis management with risk management is one of the solutions for these phenomena. With risk assessment before crisis, the amount of damages will be reduced to the minimum amount. In this research, the risk assessment of drought by Monte Carlo method will be used in order to reduce the damages caused by drought as a natural and uncontrollable phenomenon on saffron product. The monthly Standardized Precipitation Index (SPI) of drought and the monthly average temperature are calculated as independent variables in the yield distribution function. The relationship between independent variables (temperature and SPI) and dependent variable (saffron yield) is established using Artificial Neural Network (ANN). After that, 2000 random data from independent variables are generated using MATLAB and 2000 simulated yields generated by a trained artificial neural network. Then, the cumulative distribution of the simulated yields are determined and these yields are standardized in order to unification of the yield data of each city. The risk factor is calculated by choosing a reference station and using the cumulative distribution. The relative risks of the stations are considered after drawing the diagram of Yield-Risk standard factor. The results of the research show that most of the studied years are in normal range and the drought frequency in the four stations of Khorasan Jonoobi province is twice the stations in Khorasan Razavi. Furthermore, the artificial neural network with a correlation coefficient of 0.85 could predict the yield of the product very well. The similarity of the cumulative distribution diagram of the real yield with the cumulative distribution of the yields simulated by Monte Carlo indicates that the results are correct. At the end The results of this research show that Ghayen has the highest relative risk compared to the reference station (Torbat-e- Heydariyeh) and Nehbandan has the lowest one.