250 years of daily weather: a reconstruction of temperature and precipitation in Switzerland since the late 18th century

Numerous historical sources report on hazardous past climate and weather events that had considerable impacts on society. Studying for example mechanisms of such events is however hampered by a lack of spatial weather information. Gridded high-resolution daily data sets mostly cover the past few decades. For Switzerland, Pfister et al. (2019) reconstructed daily fields of precipitation and temperature back to 1864, but the century before this year would be particularly relevant to study the transition from the Little Ice Age climate to the Anthropocene and to analyze the anomalous, volcanically-perturbed climate in the early 19^th century and late 18^th century. Here we present a daily high-resolution (1x1 km²) reconstruction of temperature and precipitation fields for Switzerland for the years 1763 to 1960 using the analog resampling method (ARM). Together with the present-day meteorological fields, this forms a 250-year data set. The ARM samples temperature and precipitation fields for the historical period from the most similar days in a reference period. These most similar days are selected based on the smallest distance calculated between the observational data in the historical period and the reference period. As observational data, we use temperature, pressure, precipitation, and precipitation occurrence. The resampled fields are then post-processed by assimilating historical temperature measurements and adjusting precipitation fields using isotonic distributional regression. Despite the much scarcer data availability in the period before 1864, evaluation results are promising for the temperature reconstruction with correlation values of on average 0.9 and root mean square errors of on average 1.8°C. For precipitation, the evaluation results are less promising with correlation values of on average 0.7 and root mean square errors of on average 5 mm. Due to its high spatial variability and the small number of records in the historical period, precipitation is more difficult to reconstruct. With the here presented data set, it is now possible to study historical weather and climate events in their spatial extent, such as the warm summer in 1807, and bring it into context leveraging other historical sources. The data set can further be used to calculate impact-relevant indices, for agricultural, phenological, and hydrological modeling.