On the reconstruction of temperature records from proxy data in mid Europe

Abstract

The problem of the reconstruction of mean summer temperatures from proxy data on a rigorous statistical basis is addressed: Cross spectral analysis is used for evaluating the suitability of linear transfer models between proxy data and temperature, the linear model is estimated from the data of a calibration period and it is verified using the data of a different period. Strict statistical tests are applied at every step of the analysis. Three examples using the proposed method are given: the reconstruction of mean April–July temperatures at Basel from grape harvest data in NE France (1484–1768), mean April–September temperatures at Basel from late wood density of trees from the Swiss Prealps (1270‐1768), and mean April–September temperatures at Great St. Bernhard from the late wood proxy too (1270–1850). The grape harvest data give a very good reconstruction, the resulting temperatures from 1500–1768 are documented. The late wood data fails to give a good reconstruction.