In this study two commonly used automated methods to detect atmospheric fronts in the lower troposphere are compared in various synoptic situations. The first method is a thermal approach, relying on the gradient of equivalent potential temperature (TH), while the second method is based on temporal changes in the 10 m wind (WND). For a comprehensive objective comparison of the outputs of these methods of frontal identification, both schemes are firstly applied to an idealised strong baroclinic wave simulation in the absence of topography. Then, two case-studies (one in the Northern Hemisphere (NH) and one in the Southern Hemisphere (SH)) were conducted to contrast fronts detected by the methods. Finally, we obtain global winter and summer frontal occurrence climatologies (derived from ERA-Interim for 1979-2012) and compare the structure of these. TH is able to identify cold and warm fronts in strong baroclinic cases that are in good agreement with manual analyses. WND is particularly suited for the detection of strongly elongated, meridionally oriented moving fronts, but has very limited ability to identify zonally oriented warm fronts. We note that the areas of the main TH frontal activity are shifted equatorwards compared to the WND patterns and are located upstream of regions of main WND front activity. The number of WND fronts in the NH shows more interseasonal variations than TH fronts, decreasing by more than 50% from winter to summer. In the SH there is a weaker seasonal variation of the number of observed WND fronts, however TH front activity reduces from summer (DJF) to winter (JJA). The main motivation is to give an overview of the performance of these methods, such that researchers can choose the appropriate one for their particular interest.

This paper introduces a newly compiled set of feature-based climatologies identified from ERA-Interim (1979–2014). Two categories of flow features are considered: (i) Eulerian climatologies of jet streams, tropopause folds, surface fronts, cyclones and anticyclones, blocks, and potential vorticity streamers and cutoffs and (ii) Lagrangian climatologies, based on a large ensemble of air parcel trajectories, of stratosphere–troposphere exchange, warm conveyor belts, and tropical moisture exports. Monthly means of these feature climatologies are openly available at the ETH Zürich web page (http://eraiclim.ethz.ch) and are annually updated. Datasets at higher resolution can be obtained from the authors on request. These feature climatologies allow studying the frequency, variability, and trend of atmospheric phenomena and their interrelationships across temporal scales. To illustrate the potential of this dataset, boreal winter climatologies of selected features are presented and, as a first application, the very unusual Northern Hemispheric winter of 2009/10 is identified as the season when most of the considered features show maximum deviations from climatology. The second application considers dry winters in the western United States and reveals fairly localized anomalies in the eastern North Pacific of enhanced blocking and surface anticyclones and reduced cyclones.