Abstract:

This study evaluates the forecast performance of the CMA_GFS model in the Arctic using surface observational data and GPM precipitation observations for January, April, July, and October 2023. The model performance for 2 m air temperature, 2 m dew point temperature, surface pressure, 10 m wind speed, and precipitation was systematically assessed. The spatiotemporal characteristics of forecast errors and the model performance under extreme warming events were analyzed, and comparisons were conducted with the GFS and IFS models. The results indicate that the CMA_GFS model exhibits systematic error structures in the prediction of continuous meteorological variables over the Arctic. Specifically, air temperature, surface pressure, and wind speed are generally overestimated, while the dew point temperature shows pronounced seasonal-dependent biases. Forecast errors increase with forecast lead time, and winter errors are significantly larger than those in summer. High-error regions are mainly concentrated over areas with complex underlying surfaces and near coastal boundaries. The precipitation forecasts can reasonably reproduce the overall precipitation structure and show relatively good predictive skill for precipitation events within the 2~5 mm·d^–1 range. Under extreme warming events, the forecast errors of thermodynamic variables increase significantly. The comparison results further show that the overall forecast performance of the CMA_GFS model is comparable to that of the GFS model but remains inferior to that of the IFS model. However, the precipitation forecast skill of CMA_GFS is superior to that of both GFS and IFS. This study reveals the error structure characteristics of the CMA_GFS model in the Arctic and identifies their main influencing factors, providing a scientific basis for further model improvement and operational application in polar regions.

Key words: CMA_GFS, validation assessment, surface meteorological elements, Arctic region