Abstract Snow cover and cold compound extremes can generate multiple hazards. Subseasonal forecasts of these compound extremes hold considerable socioeconomic value; however, snow–atmosphere interactions introduce additional challenges that may limit forecast skill. This study evaluates the subseasonal forecast performance of these compound extremes in midlatitude East Asia using the Model for Prediction Across Scales–Atmosphere (MPAS‐A), with a focus on snow–atmosphere interactions. At the onset of these compound extremes, rapid snow cover expansion and abrupt decreases in surface air temperature occur regionally across midlatitude East Asia. MPAS‐A can forecast these extreme variations in the snow cover fraction and temperature up to two pentads in advance, with skill declining but remaining detectable at three pentads. Forecasts beyond four pentads exhibit limited reliability. Further analysis indicates that biases in MPAS‐A snow cover fraction predictions primarily originate from an underestimation of snowfall. These biases are not confined to snow cover; sensitivity experiments reveal that they propagate into surface air temperature forecasts. In addition, the forecast performance of MPAS‐A is sensitive to the snow cover fraction scheme. MPAS‐A with the Noah‐MP land surface scheme produces more snow cover and colder temperatures than those with Noah because of differences in the snow cover fraction formulation. These findings highlight the critical role of snow cover in subseasonal forecasts and suggest that improving snowfall representation and snow cover fraction parameterization can enhance subseasonal forecasting models.
Li et al. (Sat,) studied this question.