Skill of Medium-Range Forecast Models Using the Same Initial Conditions

• DOI: 10.1175/bams-d-21-0234.1
• Publication Date: 2022-06-30T19:56:25Z
• Authors: L. Magnusson; D. Ackerley; Y. Bouteloup; J.-H. Chen; J. Doyle; P. Earnshaw; Y. C. Kwon; M. Köhler; S. T. K Lang; Y.-J. Lim; M. Matsueda; T. Matsunobu; R. McTaggart-Cowan; A. Reinecke; M. Yamaguchi; L. Zhou
• Keywords: Initialization; Quantitative precipitation forecast; Ensemble forecasting; Forecast verification; Consensus forecast

Abstract In the Different Models, Same Initial Conditions (DIMOSIC) project, forecasts from different global medium-range forecast models have been created based on same initial conditions. The dataset consists of 10-day deterministic seven and includes 122 dates spanning one calendar year. All are initialized ECMWF operational analyses to minimize differences due initialization. run at or near their respective resolutions explore similarities between models. main aims this study 1) evaluate skill how it depends model formulation, 2) assess systematic errors short lead times, 3) compare multimodel ensemble spread uncertainty schemes, 4) identify that generate similar solutions. Our results show all in capable producing high-quality given a analysis. But time, we find large variety biases, both terms temperature precipitation. We able whose more each other than they those systems, use physics packages. However, spread, our also demonstrate sensitivities formulations still substantial. therefore believe diversity design stems parallel development efforts modeling centers around world remains valuable for future progress numerical weather prediction community.