Effects of snowfall on lower atmosphere temperature in the Korean Integrated Model (KIM)

The Korea Meteorological Administration (KMA) began operating the Korean Integrated Model (KIM) in April 2020 and has been improving the model’s performance through continuous research and development since then. The horizontal resolution of the KIM has recently increased to 8 km from 12 km for more detailed predictions, and we have started the semi operation of the high-resolution KIM in November 2024, alongside the operational KIM with a resolution of 12 km.
The operational KIM has systematic biases in 2m temperature forecasts over land in the mid-to-high latitudes of Northern Hemisphere. The temperature biases vary seasonally, with a warm bias in summer and a cold bias in winter compared to the IFS analysis. The cold biases are reduced in the high resolution (8km) KIM but still remain. In this study, we focus on the cold bias, especially around the Ural Mountains during the winter season. 
To analyze the source of the cold bias in the lower atmosphere, we diagnosed the land surface process and found that KIM overestimates snowfall and, consequently, results in the surface cold bias.
To address this issue, we performed sensitivity experiments focusing on the precipitation partitioning method (PPM) that determines the rain-snow separation in precipitation processes. The results demonstrate that snowfall amounts are highly sensitive to the PPM configuration, which significantly influences both surface and atmospheric temperatures. Furthermore, we found that cloud-radiation interactions amplify this sensitivity through positive feedback mechanisms. In future work, we plan to evaluate the overall performance of the KIM by implementing an improved PPM scheme.