New baselines, new extremes? Assessing the impact of reference periods on climate extremes in Central Europe

Climatological normal periods are used to describe the typical climate conditions that can be expected in a given location. These climate normals – statistical values such as means or totals for specific climate elements – are calculated for non-overlapping 30-year periods. The 1961–1990 period was probably the most widely used reference baseline in modern climatology, as it was recommended by the World Meteorological Organization for long-term climate change assessments. Now that 2020 has passed, a new climatological normal period (1991–2020) can be adopted. This raises an important question: how will this updated baseline affect the way we define anomalies and extremes, given that climate warming has already shifted what is now considered “normal” in many regions? With ongoing climate change, the statistical definition of “normal” itself is evolving – potentially influencing climate assessments and regional comparisons.

The main aim of this study is to assess how the selection of the reference period (1961–1990 vs. 1991–2020) influences how extremes are defined and distributed across Central Europe, based on a new high-resolution dataset. First, we compared how the choice of baseline period affects spatial and seasonal patterns of air temperature and precipitation indices. These indices include mean values and counts of extreme days defined by absolute thresholds. Next, we assessed how standard percentile thresholds used to define extremes (such as the 90th percentile) vary depending on the baseline period, and how these differences impact the spatial distribution of corresponding extremes. The analyses are based on a newly developed, high-resolution gridded daily dataset for Central Europe (1 × 1 km), covering the years 1961–2020, which includes daily maximum and minimum temperatures and precipitation totals. The dataset is based on in-situ measurements covering Poland, Czechia, Slovakia, and the border regions of Germany and Austria.

The results show large differences in temperature and precipitation patterns between indices computed using the two baseline periods, demonstrating that the 1991–2020 reference is strongly affected by recent climate change. The selection of a more recent baseline can significantly understate the extent of ongoing warming, especially when extremes are defined using percentile thresholds. Precipitation indices appear less sensitive to baseline changes, although differences still exist. The 1961–1990 period may offer a more stable reference for climate studies, and careful consideration is needed when communicating changing baselines to decision-makers.