Unveiling the influence of multi-layer snowpack in seasonal forecast system on model climatological bias

• DOI: 10.5194/egusphere-egu25-5714
• Publication Date: 2025-03-14T19:28:01Z
• Authors: Eunkyo Seo; Paul Dirmeyer

The representation of snow in land surface models is critical for accurate seasonal forecasting, yet traditional single-layer snow schemes fail to capture the full insulating properties of deep snowpacks. These limitations result in pronounced seasonal biases, including excessive winter cooling and springtime warming. This study explores the impact of introducing a multi-layer snow scheme within the Global Seasonal Forecast System (GloSea) to address these biases. Using 24 years of retrospective forecasts (1993–2016), we compare the latest version, GloSea6, incorporating the multi-layer scheme, with GloSea5, which relies on a single-layer approach. The multi-layer snow scheme in GloSea6 improves the onset of snowmelt, delaying it by approximately two weeks. This delay moderates spring soil moisture depletion, promoting greater latent heat flux and surface evaporative cooling. The wetter surface reduces the overestimation of water-limited processes and mitigates near-surface warming biases during summer. Additionally, the enhanced representation of snow improves the simulation of precipitation, particularly in snowmelt-driven regions such as the Great Plains, Europe, and South and East Asia, leading to substantial error reductions. These findings highlight the critical role of a multi-layer snow scheme in advancing seasonal forecast accuracy, not only for temperature and precipitation during snowmelt but also for subsequent summer climatic conditions through improved land-atmosphere feedback processes.