Application of North American Multi-Model Ensemble for Iran's Seasonal Precipitation Forecasts

The assessment of Atmosphere-Ocean General Circulation climate model outputs have been become of much global interest. If the skill of the dynamic seasonal climate forecasts prove appropriate skill over the long-term (hindcast period), real-time seasonal forecast systems are able to support decision makers in managing monthly and seasonal time scales. The skill of dynamic models for target seasons depends on both the initialization and lead time. In this research, the output of 8 models enrolling in the North American Multi-Model Ensemble (NMME) with more than 102 ensemble members is evaluated over Iran. Analysis of the research is for the first 6 months of the water-year when DJF precipitation has the highest proportion within the study area. The anomaly correlation is calculated using PERSIANN-CDR for evaluation of three seasons (OND, DJF, JFM) in different initializations. The bias of NMME models is also analyzed over the period of 1983-2013. The results show that the anomaly correlation of NMME individual models is high over many parts of Iran. The bias between -2 to mm/day is identified over grid points within the study area. The results of the this research addressed the need to develop multi-model ensembles to treat systematic errors, downscaling and providing probabilistic forecasts in 30 river basins in Iran which suit the real-time application of NMME models during a water year.