• Atmosphere
• /
• v.34 no.4
• /
• pp.463-480
• /
• 2024

This study aims to assess the performance of climate prediction systems around the world, and understand objective seasonal prediction skill of KMA's GloSea6. Using the 2023 hindcast verification values provided by the WMO Lead Centre for Long-Range Forecast (LC-LRF), we analyzed the skill in the global, East Asia, and European regions. The differences in prediction skill and RMSE between GPC (Global Producing Centers) were very small in this challenging area. Overall, GloSea6 showed the best ACC across variables and periods. Operating this outstanding climate prediction system not only ensures the provision of the best forecasting services but also offers excellent research and development tools. This result also suggests that seasonal forecasting requires different strategies against short- to medium-range forecast to account for climate prediction sources and reduce uncertainties. The skill differences between GloSea6-Seoul and GloSea6-Exeter, especially in high latitude, could be due to differences in snow and soil temperature initialization. Understanding these differences is important for future prediction system development. GPCs that use atmospheric only models instead of coupled, showed the limitations for seasonal predictions. Systems developed a relatively long time ago tended to perform low, suggesting that continuous improvements and upgrades are important. Among variables, SST showed the best prediction skill with the lowest RMSE. Temperature and pressure variables showed practical skill levels, around 0.5. We aimed to quantitatively assess the skills of climate prediction systems, and this assessment can guide the improvement and development of future systems and serve as a reference.