Development and Evaluation of a Time-Lagged Ensemble of WRF-Downscaled CFSv2 Forecasts for Extended-Range Forecasts in South Korea

While weather forecasts up to 10 days are widely available and readily accessible to the public, forecasts beyond this period often lack necessary details, yet are crucial for informed decision-making across diverse weather-dependent sectors. Although GCM-driven seasonal forecasts offer valuable long-range meteorological information, their practical application is often hindered by their coarse resolution, inherent model errors, and uncertainties. In this regard, this study focuses on improving extended-range predictions for South Korea, specifically up to 1 month ahead, through dynamic downscaling of the NCEP Climate Forecast System version 2 (CFSv2) seasonal forecasts and development of a time-lagged ensemble using the downscaled data. Our preliminary analysis shows that the ECMWF seasonal forecast system 5 (SEAS5) generally outperforms CFSv2 for South Korea, but SEAS5's 12-hourly pressure-level data make it unsuitable as input for dynamic downscaling. Therefore, CFSv2, which provides 6-hourly output at both surface and pressure levels, is utilized for high-resolution RCM simulations, with SEAS5 incorporated into the selection of CFSv2 forecast members to be downscaled out of a large pool of available members with varying lead times for each target period. The state-of-the-art Weather Research and Forecasting (WRF) model is employed to dynamically downscale the selected CFSv2 forecast members to capture finer-scale atmospheric processes with more detailed geographical features, and the downscaled forecasts are then integrated into a time-lagged ensemble to better reflect uncertainty and enhance forecast accuracy. The added value of the high-resolution time-lagged ensembles is assessed by comparing the ensemble outputs with observations. By combining the two advanced approaches, dynamical downscaling and a time-lagged ensemble, this study aims to bridge the gap between low-resolution GCM forecasts and informational needs for high-resolution forecasts and provide more reliable and actionable meteorological information up to 1 month ahead, thereby supporting effective weather assessments for optimal management and operations in various sectors. [Acknowledgements: This study was supported by the “Research Program for Agricultural Science & Technology Development (Project No. RS-2024-00399847)”, National Institute of Agricultural Sciences, Rural Development Administration, Republic of Korea.]