Understanding Sampling Bias in the Global Heat Flow Compilation

Abstract

Geothermal heat flow is commonly inferred from the gradient of temperature values in boreholes. Such measurements are expensive and logistically challenging in remote locations and, therefore, often targeted to regions of economic interest. As a result, measurements are not distributed evenly. Some tectonic, geologic and even topographic settings are overrepresented in global heat flow compilations; other settings are underrepresented or completely missing. These limitations in representation have implications for empirical heat flow models that use catalogue data to assign heat flow by the similarity of observables. In this contribution, we analyse the sampling bias in the Global Heat Flow Database of the International Heat Flow Commission (IHFC), the most recent and extensive heat flow catalogue, and discuss the implications for accurate prediction and global appraisals. We also suggest correction weights to reduce the bias when the catalogue is used for empirical modelling.From comparison with auxiliary variables, we find that each of the following settings is highly overrepresented for heat flow measurements; continental crust, sedimentary rocks, volcanic rocks, and Phanerozoic regions with hydrocarbon exploration. Oceanic crust, cratons, and metamorphic rocks are underrepresented. The findings also suggest a general tendency to measure heat flow in areas where the values are elevated; however, this conclusion depends on which auxiliary variable is under consideration to determine the settings. We anticipate that the use of our correction weights to balance disproportional representation will improve empirical heat flow models for remote regions and assist in the ongoing assessment of the Global Heat Flow Database.