This study uses historical weather observations rescued from across Europe, as well as machine learning techniques to improve our understanding of a series of heatwaves in 1806. The results indicate that while machine learning models can do a good job of reproducing historical weather observations, they can't capture the physics behind why a heatwave would occur.

The paper shows that historical records of wine yield and quality from Central Europe (1416–1988) can be combined into a Yield Quality Index (YQI). This index serves as a reliable indicator of climate conditions from May to July. Using long-term tax, tithe, and estate records from the Moselle region in Germany and the Swiss Plateau, the study shows that high YQI values correspond to warm summers, early harvests, and supportive tree-ring evidence. Conversely, low values often reflect cold summers or frost damage. These results confirm that wine production data can significantly improve climate reconstructions and help explain the social and economic impacts of past climate extremes.