ArtificialINTELLIGENCE FOR GEOTHERMOMETRY – A CRITICAL REVIEW

The chemistry of a geothermal water sample encodes the temperature of the reservoir it came from, and reading that signal is the cheapest exploration tool geothermal has. Classical geothermometers do the reading through empirical equations that assume full fluid-rock equilibrium, an assumption routinely broken by mixing, boiling and re-equilibration on the way to surface. Geologic® prepared a critical review of every significant attempt over the last two decades to replace those equations with artificial intelligence: the first review dedicated specifically to AI in geothermometry.

Fourteen studies from 2001 to 2024 are dissected to a depth no prior review attempted, covering each model's architecture, hyperparameters, dataset size and pedigree, temperature range and validation discipline, across solute and gas-phase geothermometers, classification hybrids and unsupervised prospecting. The verdict is consistent: machine learning already outperforms classical equations where data exist, with test accuracies now rivalling measured temperatures, but the binding constraint is data, not algorithms. Models trained on tens of samples do not transfer between geologic settings, and the review closes with a concrete agenda for the datasets, architectures and explainability the field needs next.