Mesoscale high-wind features within European winter storms, which include the warm jet (WJ), cold jet (CJ), sting jet (SJ), cold-frontal convection (CFC) and cold-sector winds (CS), can cause immense damage. RAMEFI is a novel approach to objectively identify these features using a probabilistic random forest based on surface observations, namely pressure, temperature, wind, precipitation parameters. Note that the SJ is included in the CJ category here. High winds not associated with any of these features are detected as ‘no feature’ (NF). At each point, RAMEFI generates probabilities of occurrence for each of the features.
Here, RAMEFI is applied to the deterministic ICON-EU forecast for grid points where the wind speed exceeds 80% of its 98th percentile. For details, we refer to Eisenstein et al. (2022). You can switch between the unfiltered RAMEFI output and output filtered by cyclones, i.e. local mean-sea level pressure minima. Filtered output includes an area of +/- 15°E and +5°N to -15°N around a cyclone. You can further switch to surface gusts.