Jsmetrics v0.2.0: a Python package for metrics and algorithms used to identify or characterise atmospheric jet streams

Abstract

Abstract. The underlying dynamics controlling jet streams are complex, but it is expected that they will have an observable response to changes in the larger climatic system. A growing divergence in regional surface warming trends across the planet, which has been both observed and projected since the start of the 20th century, has likely altered the thermodynamic relationships responsible for jet stream formation and control. Despite this, the exact movements and trends in the changes to the jet streams generally remain unclear and without consensus in the literature. The latest IPCC report highlighted that trends both within and between a variety of observational and modelling studies were inconsistent (Gulev et al., 2021; Lee et al., 2021). Trends in jet streams were associated with low to medium confidence, especially in the Northern Hemisphere. However, what is often overlooked in evaluating these trends is the confused message in the literature around how to first identify, and then characterise, the jet streams themselves. We classify the methods for characterising jet streams in the literature into three broad strategies: statistics that isolate individual values from the wind speed profile (jet statistics), methods for quantifying the sinuosity of the upper air (waviness metrics), and algorithms that identify a mask related to the coordinates of fast-flowing wind throughout the horizontal and/or vertical plane (jet core algorithms). While each approach can capture particular characteristics and changes, they are subject to the spatial and temporal specifications of their definition. There is therefore value in using them in combination to assess parametric and structural uncertainty and to carry out sensitivity analyses. Here, we describe jsmetrics version 0.2.0, a new open-source Python 3 module with standardised versions of 17 metrics that have been used for jet stream characterisation. We demonstrate the application of this library with two case studies derived from ERA5 climate reanalysis data.