A Temporal Consistency Analysis of Long-Term Precipitation Trends in ERA5 and other ECMWF reanalyses over the Alps, Italy, and Fennoscandia

Reanalyses are commonly employed for the analysis of climatological trends due to their emphasis on temporal consistency. The ERA5 reanalysis family, comprising the deterministic ERA5-HRES and the ensemble-based ERA5-EDA, has proven to be a valuable resource for trend extraction. In this study, long-term trends (1941–2000) in total annual precipitation are examined across three regions: the Alps, Italy, and Fennoscandia. It is recognized that variations in the observational systems used for data assimilation impact water cycle components like precipitation. This fact raises the question of to what extent temporal variations in ERA5 precipitation amounts are solely a result of climate variations and the influence of changes in the observational system impacting simulation accuracy. Addressing this issue, this work compares ERA5-HRES and ERA5-EDA with three additional ECMWF reanalyses, CERA-20C, ERA-20C, and ERA-20CM, using homogenized, trend-focused observational datasets: LAPrec1901 (for the Alps), UniMi/ISAC-CNR (for Italy), and NGCD (for Fennoscandia). The results show that isolating the climatological signal in ERA5 from the effects of observational system changes presents a significant challenge. Distinct trend behavior is identified in ERA5 data for the period 1940–1970, especially over the Alps and, to a lesser extent, Italy and Norway, diverging from observed trends. An increasing, though non-linear, trend in the deviation from observational datasets is observed prior to 1970, with widespread differences over large mountain areas. In contrast, more localized increasing trends are detected after 1970. These findings emphasize the necessity of accounting for the influence of evolving data assimilation systems when interpreting precipitation trends from reanalyses. Enhancing future reanalysis interpretability could benefit from adopting a model-only integration for the same period, as implemented in ERA-20C and ERA-20CM, to help disentangle climate-driven signals from those introduced by the assimilation process. This assessment underlines the critical need for careful evaluation of long-term precipitation trends in reanalyses and supports the continued development of auxiliary configurations to improve the robustness of climate trend analyses.