Methodological and physical biases in global to subcontinental borehole temperature reconstructions: an assessment from a pseudo-proxy perspective

Abstract

Borehole-based reconstruction is a well-established technique to recover information of the past climate variability based on two main hypotheses: (1) past ground surface temperature (GST) histories can be recovered from borehole temperature profiles (BTPs); (2) the past GST evolution is coupled to surface air temperature (SAT) changes, and thus, past SAT changes can be recovered from BTPs. Compared to some of the last millennium (LM) proxy-based reconstructions, previous studies based on the borehole technique indicate a larger temperature increase during the last few centuries. The nature of these differences has fostered the assessment of this reconstruction technique in search of potential causes of bias. Here, we expand previous works to explore potential methodological and physical biases using pseudo-proxy experiments with the Community Earth System Model Last Millennium Ensemble (CESM-LME). A heat-conduction forward model driven by simulated surface temperature is used to generate synthetic BTPs that are then inverted using singular value decomposition. This procedure is applied to the set of simulations that incorporates all of the LM external forcing factors as well as those that consider the concentration of the green house gases (GHGs) and the land use land cover (LULC) changes forcings separately. The results indicate that methodological issues may impact the representation of the simulated GST at different spatial scales, with the temporal logging of the BTPs as the main sampling issue that may lead to an underestimation of the simulated GST 20th-century trends. Our analysis also shows that in the surrogate reality of the CESM-LME the GST does not fully capture the SAT warming during the industrial period, and thus, there may be a further underestimation of the past SAT changes due to physical processes. Globally, this effect is mainly influenced by the GHG forcing, whereas regionally, LULC changes and other forcings factors also contribute. These findings suggest that despite the larger temperature increase suggested by the borehole estimations during the last few centuries of the LM relative to some other proxy reconstructions, both the methodological and physical biases would result in a underestimation of the 20th-century warming.