Bimodality in Ensemble Forecasts of 2-Meter Temperature: Event Aggregation

Abstract. A previous study has shown that a large portion of subseasonal-to-seasonal European Centre for Medium-Range Weather Forecasts (ECMWF) ensemble forecasts for 2-meter temperature exhibit properties of bimodality, in some locations reaching occurrence rates of over 30 %. This study introduces novel methodology to help identify `bimodal events', meteorological events which trigger the development of widespread bimodality in forecasts. Understanding such events not only provides insight into the dynamics of the meteorological phenomena causing bimodal events, but also has drastic implications for the skill of forecasts affected. The methodology that is developed allows one to systematically characterize the spatial and temporal scales of the derived bimodal events, and thus uncover the flow states that lead to them. Three distinct regions that exhibit high occurrence rates of bimodality are studied: one in South America, one in the Southern Ocean and one in the North Atlantic. It is found that each region's bimodality appears to be triggered by large-scale atmospheric processes interacting with geographically specific processes: In South America bimodality is related to an atmospheric wave interacting with the Andes, in the Southern Ocean bimodality is related to an atmospheric wave interacting with sea ice, and in the North Atlantic bimodality is connected to an atmospheric wave deforming near the Gulf Stream. This common pattern of large-scale circulation anomalies interacting with local boundary conditions suggests that any deeper dynamical understanding of these events should incorporate such interactions.