Global variability in multi-century ground warming inferred from geothermal data

Geothermal temperature‐depth measurements can be used to reconstruct past climate without the need for calibration against meteorological data. The global geothermal database is one of opportunity and is therefore subject to variations in measurement protocol, quality control and potential non‐climatic influences. As the climatic history recovery is sensitive to these factors, we developed a Bayesian hierarchical model that allows us to treat errors and uncertainty formally. This way we may better isolate the climate signal. For the Northern Hemisphere extra‐tropics our reconstruction shows a warming beginning in CE 1750 and it captures the observed two‐phase warming over the past century. We clearly identify the northern tropics as a region of greatest ground warming and hypothesize that this reflects land‐use change. For the Southern Hemisphere, the inclusion of newer data leads to a modest cooling until CE 1750. Outside the tropics, agreement with multi‐proxy reconstructions has improved relative to earlier studies.