Trends in short-term precipitation variability and their connection to atmospheric circulation

Long-term changes in precipitation occurrence and short-term variability influence drought and flood risk and therefore represent an important aspect of climate change assessment. In this study, we analyse trends in wet-day probability and precipitation persistence across Europe and evaluate their consistency across three widely used datasets: station observations (ECA&D), gridded observations (E-OBS), and the ERA5 reanalysis. Wet-day probability is defined as the seasonal probability of daily precipitation exceeding 1 mm. Short-term variability is characterized by (i) transition probabilities between dry and wet days (dry-to-wet and wet-to-wet), assuming a two-state first-order Markov process, and (ii) persistence of precipitation quantified as the lag-1 autocorrelation of this process.

The analysis covers the period 1961–2020 and is conducted separately for seasons, with particular focus on winter and summer. Trends are estimated using non-parametric (Mann-Kendall test, Kendall statistic). To interpret spatial patterns and assess potential drivers, precipitation characteristics are related to atmospheric circulation types derived from sea-level pressure using the Jenkinson–Collison classification, which describes daily flow direction, strength, and vorticity. Circulation types are further categorized as conducive or non-conducive to wet days based on their relative wet-day frequency within each region (e.g., Central, Southern, and Northern Europe).

The results reveal a pronounced north–south gradient in winter, with increasing wet-day probability in northern Europe. Positive trends in both wet-to-wet and dry-to-wet transition probabilities also prevail in northern Europe. In contrast, southern regions show weaker or negative trends. In summer, trends are generally weaker and exhibit no clear large-scale spatial pattern. Differences among datasets highlight the sensitivity of trend estimates to data source, particularly in regions with sparse station coverage.