Are temperature reconstructions regionally biased?

Abstract

Are temperature reconstructions possibly biased due to regionally differing density of utilized proxy-networks? This question is assessed utilizing a simple process-based forward model of tree growth in the virtual reality of two simulations of the climate of the last millennium with different amplitude of solar forcing variations. The pseudo-tree ring series cluster in high latitudes of the northern hemisphere and east Asia. Only weak biases are found for the full network. However, for a strong solar forcing amplitude the high latitudes indicate a warmer first half of the last millennium while mid-latitudes and Asia were slightly colder than the extratropical hemispheric average. Reconstruction skill is weak or non-existent for two simple reconstruction schemes, and comparison of virtual reality target and reconstructions reveals strong deficiencies. The temporal resolution of the proxies has an influence on the reconstruction task and results are sensitive to the construction of the proxy-network. Existing regional temperature biases can be attenuated or accentuated by the skill of the reconstruction approach.