Threshold values in spatially sliding Jenkinson–Collison–Lamb circulation classifications

The Jenkinson–Collison classification is among the most widely used methods for classifying synoptic-scale sea-level pressure fields and examining their links to surface weather. Its principal advantage, compared with approaches such as self-organizing maps or cluster analysis, is that it predefines circulation patterns based on simple and easily interpretable circulation indices (geostrophic wind speed, direction, and vorticity). Because the method was originally designed to objectively reproduce the manual Lamb weather types for the British Isles, the thresholds governing the assignment of fields to circulation types were chosen accordingly. When applied to markedly different climates with distinct circulation regimes and synoptic relationships, however, the default thresholds are likely to yield suboptimal or even misleading results.

One recent development of the Jenkinson–Collison methodology is its simultaneous application to an array of spatially overlapping (sliding) domains, producing “sliding classifications” that address a key limitation of circulation classifications—their restricted spatial validity. However, particularly at hemispheric or global scales, sliding classifications are highly sensitive to the choice of circulation indices and threshold values, and thus strongly affected by suboptimal parameter settings.

In this contribution, we test a wide range (~10,000) of threshold combinations across multiple Northern Hemisphere extratropical regions to (1) evaluate whether the default thresholds produce meaningful and useful classifications and (2) identify regions where alternative threshold sets improve classification utility. Because the quality and usefulness of a classification depend strongly on the evaluation criteria and research objectives, we employ several metrics and applications to assess performance. These include the tendency toward “snowballing” (i.e., the formation of one dominant class alongside many small ones) and the ability of classifications to explain variability in surface weather elements such as daily mean and maximum temperature, daily maximum wind gust, and daily precipitation totals.