Premium
Seasonal precipitation forecast skill over Iran
Author(s) -
Shirvani Amin,
Landman Willem A.
Publication year - 2015
Publication title -
international journal of climatology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.58
H-Index - 166
eISSN - 1097-0088
pISSN - 0899-8418
DOI - 10.1002/joc.4467
Subject(s) - climatology , downscaling , geopotential height , environmental science , sea surface temperature , atmospheric model , quantitative precipitation forecast , precipitation , forcing (mathematics) , predictability , forecast skill , climate forecast system , geopotential , climate model , meteorology , climate change , geology , statistics , mathematics , geography , oceanography
This paper examines the skill of seasonal precipitation forecasts over Iran using one two‐tiered model, three National Multi‐Model Ensemble ( NMME ) models, and two coupled ocean–atmosphere or one‐tiered models. These models are, respectively, the ECHAM4 .5 atmospheric model that is forced with sea surface temperature ( SST ) anomalies forecasted using constructed analogue SSTs ( ECHAM4 .5‐ SSTCA ); the IRI‐ECHAM4 .5‐DirectCoupled, the NASA‐GMAO ‐062012 and the NCEP‐CFSv2 ; and the ECHAM4 .5 Modular Ocean Model version 3 ( ECHAM4 .5‐ MOM3‐DC2 ) and the ECHAM4 .5‐ GML‐NCEP Coupled Forecast System ( CFSSST ). The precipitation and 850 hPa geopotential height fields of the forecast models are statistically downscaling to the 0.5° × 0.5° spatial resolution of the Global Precipitation Climatology Centre ( GPCC ) Version 6 gridded precipitation data, using model output statistics ( MOS ) developed through the canonical correlation analysis ( CCA ) option of the Climate Predictability Tool ( CPT ). Retroactive validations for lead times of up to 3 months are performed using the relative operating characteristic ( ROC ) and reliability diagrams, which are evaluated for above‐ and below‐normal categories and defined by the upper and lower 75th and 25th percentiles of the data record over the 15‐year test period of 1995/1996 to 2009/2010. The forecast models' skills are also compared with skills obtained by (a) downscaling simulations produced by forcing the ECHAM4 .5 with simultaneously observed SST , and (b) the 850 hPa geopotential height NCEP‐NCAR (National Centers for Environmental Prediction‐National Center for Atmospheric Research) reanalysis data. Downscaling forecasts from most models generally produce the highest skill forecast at lead times of up to 3 months for autumn precipitation – the October‐November‐December ( OND ) season. For most seasons, a high skill is obtained from ECHAM4 .5‐ MOM3‐DC2 forecasts at a 1‐month lead time when the models' 850 hPa geopotential height fields are used as the predictor fields. For this model and lead time, the Pearson correlation between the area‐averaged of the observed and forecasts over the study area for the OND , November‐December‐January ( NDJ ), December‐January‐February ( DJF ) and January‐February‐March ( JFM ) seasons were 0.68, 0.62, 0.42 and 0.43, respectively.