GWL-REA: An improved method for classifying Hess-Brezowksy Grosswetterlagen based on pattern correlations in combination with Lamb Weather Type statistics

Abstract

<p>The Hess-Brezowsky Grosswetterlagen (GWL) are a widely used set of 29 synoptic weather patterns, focussed on Central Europe. Previous algorithmic methods for classifying GWLs, based on multi-parameter correlations with respective climatological GWL patterns, yielded acceptable results in terms of the larger-scale synoptic character. However, synoptic patterns did not always correspond well to the specific nomenclature of the respective chosen GWL type in terms of central (anti)cyclonicity or location of the primary steering High (Low). To overcome this problem, the original method has been expanded by introducing statistics from the well-known Lamb Weather Types (LWT) system. LWTs centred over specific European locations allow a direct determination of e.g. local cyclonicity, allowing biases to be placed on the pattern correlations to improve the nomenclature correspondence of the results. The pattern correlations themselves have also been refined by introducing a many-to-one mapping, in which each GWL is represented by several different synoptic patterns, spanning a much wider range of the within-type variability in the multi-variable phase-space than was previously possible. This new method (GWL-REA), designed within the context of climate research at DWD, produces a much-improved GWL classification of reanalysis data with a potentially wide-range of applications. A version of the new method (GWL-EPS) is also deployed in operational forecasting at DWD for classifying medium-range ensemble forecasts (ECMWF-EPS15). The 51 ensemble runs are each classified and the likely alternative GWL developments are shown as a structure of GWL-branches, resulting in an efficient summary of the expected synoptic developments over the next two weeks.</p>