The Examination of an Improved Analogue Method for Gridded Temperature Variation Reconstruction

Abstract

The analogue method (AM) is an approach for climate field reconstruction through combining proxy data and modelling data. This study improved the quantitative method of analogue pattern through considering the spatial distribution non-uniformity of proxy data (i.e., k _p term) and temperature changes capturing ability of proxy data (v _p term). Meanwhile, this study carries out pseudo-proxy experiments on temperature variations in the eastern Central Asia-East Asia region from 1902 to 1992 to examine the feasibility of the improved AM. The reconstruction results derived from improved AM match well with the instrumental data in terms of the temporal–spatial characteristics of temperature variation, with a correlations coefficient (r) of 0.5 (p < 0.05) for the mean annual temperature (MAT) series, which is higher than that from original AM. The accuracy of the reconstruction results derived from improved AM is primarily depending on the ability of proxy data to capture temperature variations, and is secondly depending on the quantity of available proxy data. In the case of 75 pseudo-proxy maintaining the exactly same distribution with available proxy data but prescribed explaining variances of 100%, 66%, and 33%, there are respectively correlations of r = 0.54, 0.51, and 0.44 (p < 0.05) between reconstructed and instrumental MAT series. When dealing with real proxies, whose explaining variances range from 4% to 24%, the correlation decreases to r = 0.28 (p < 0.05). For a prescribed explaining variance of 100%, corresponding to 75, 37, and 5 proxies, the correlations are r = 0.54, 0.50, and 0.35 (p < 0.05) respectively. These findings demonstrate the potential value of improved AM on the gridded temperature variation reconstruction and highlight the importance of proxy data quality.