Abstract. May-to-September mean temperatures over the larger Pyrenees area (Northern Spain and Southern France) are reconstructed for the last Millennium from 22 maximum density (MXD) tree-ring chronologies. For the standardization of the tree-ring series, two detrending methods (Regional Curve Standardization (RCS) and 300-yr spline) were combined with and without an adaptive power transform (PT) for variance stabilization in the individual series. Thus, four different standardization procedures were applied to the data. Additionally, different regional chronologies were generated by computing a mean composite, averaging the local chronologies, or by applying Principal Components Analysis (PCA) to extract common variance from the subsets of individual MXD chronologies. Calibration-verification trials were performed using the product of the three regional aggregation methods in split periods: 1900–1952 and 1953–2006. Two methods were used to calibrate the regional chronology: regression and a simple variance-matching, sometimes also known as composite-plus-scaling. The resulting set of temperature reconstructions was compared with climate simulations performed with global (ECHO-G over the last Millennium for the target region) and regional (MM5) climate models. The reconstructions reveal inter-annual to multi-centennial temperature variations at the Pyrenees region for the last 750 yr. Generally, variations at inter-decadal timescales, including the cold periods associated with the solar minima, are common to all reconstruction variants although some discrepancies are found at longer timescales. The simulations of the global circulation model ECHO-G and the regional model MM5 agree with the tree-ring based reconstructions at decadal to multi-decadal time-scales. However, the comparison also highlights differences that need to be understood, such as the amplitude of the temperature variations and the discrepancies regarding the 20th century trends.
Tree-ring proxy based temperature reconstructions and climate model simulations: cross-comparison at the Pyrenees
