Pseudo-proxy analysis of the pre-observational mean (POM) temperature model in borehole climatology

The thermal regime of the Earth’s shallow subsurface can be used to reconstruct past surface temperature on centennial timescales. The temperature versus depth data extracted from the subsurface via borehole temperature profiles are a direct record of past heat exchanges at the air-ground interface and can be transformed by inversion into temperature versus time data of a specified interval. Due to the nature of heat diffusion into the ground, temperature versus time data inverted from borehole logs contain remnants of prior thermal events superimposed on the interval as a reference temperature. In order to compare climatic inferences from geothermal data with other climatic indicators, it is necessary to identify a background or reference temperature for them. This background temperature is not known for meteorological and proxy data sources, and must be estimated. The pre-observational mean (POM) temperature model was created to estimate this pre-observational background temperature, thereby facilitating comparison between borehole temperature data and surface air temperature (SAT) or proxy data, and has since become commonplace in its application. A 1000-year paleotemperature dataset simulated by a general circulation model is used to perform a pseudo-proxy evaluation of the merits of the POM method. In a series of experiments we test the sensitivity of the POM method to such model parameters as SAT interval length, recent trend in the SAT interval, and borehole measurement depth. Results indicate that using short intervals of SAT (or proxy) data, or shallow borehole logs can result in ambiguous or erroneous conclusions regarding agreement between the two datasets.