摘要：An attempt is made to produce probabilistic precipitation forecasts from deterministic model output over Iran. Due to the chaotic nature of the atmosphere, probabilistic weather forecasts can be used to quantify the intrinsic uncertainty in the output of meteorological forecasting models. Probabilistic forecasts are more flexible to use and of greater economic value to data users when compared to deterministic forecasts. In this study, the results of the application of a simple and practical method for converting deterministic precipitation forecasts into probabilistic forecasts are presented. The model prediction for precipitation in the spatio-temporal neighborhood around each grid point is used to calculate the probability of precipitation at that point. The method was applied to the Pennsylvania State University-NCAR Mesoscale Model version 5 (MM5) precipitation forecasts for January 2005 over Iran. The quality and economic value of probabilistic forecasts was evaluated for 6 and 12 h accumulated precipitation forecasts. Results showed that the derived probabilistic forecasts were superior to the corresponding deterministic forecasts in quality, economic value and consistency. Also, since the method is easy to implement with minimal computational requirements, it is thus an affordable and appropriate method for operational implementation in a variety of settings.
