Evaluation of CFSv2 Enssmble Precipitation Forecasts with Water Resources Management Application Perspective ( Case Study: Dez Dam River Basin)

Recent advances in numerical weather and seasonal forecasts ,make the water resources manager and operators very interested to involve the forecasts in real operation.Generally in the studies, the skill assessment of the forecasts would be done by comparing the forecasts with station gauge or gridded data which maight not reflect the real skill of the forecasts in basin scale. In this paper , the forecasts generated by the Noaa climate forecast system version 2( CFSv2) were evaluated in the Dez dam basin. A combination of rain gauge data with satelite base precipitation estimates( Persiann-CDR) were used as the basis of evaluation. Result showed monthly precipitation forecasts with lead 1 has the best performance based on R and ROCauc although the underestimation is higher than other leads. Evaluation of the forecasts in 3 different percentile including below normal, normal and above normal, showed well performance in percentiles above 0.6 and below 0.4, also best monthly forecast performance with lead1 was 0.88 for Nov, 0.66 for Dec and 0.8 for Mar respectively for percentile of category 1 to 3 based on ROCauc. Winter precipitation evaluation based on the SPI with 5 category, suggested that 50 to 60 percent of the times, the exact category was truly predicted while with on category difference the accuracy imrove up to 80 percent of the times.