From Heterogeneous Well Data to Probabilistic 3D Temperature Modelling of the Bavarian Molasse Basin for Geothermal Exploration

Reliable subsurface temperature models are a key prerequisite for geothermal exploration, reservoir assessment, and broader subsurface energy applications. Within the GeoChaNce research project, we present an integrated geological and thermal characterization of the Bavarian part of the North Alpine Foreland Basin (NAFB), combining petrophysical analyses of a large heterogeneous well dataset with advanced geostatistical modelling approaches.

The thermal analysis focuses on developing a fully volumetric 3D temperature model that covers depths ranging from 300 m to 5000 m true vertical depth. The temperature dataset comprises 196 bottom-hole temperature (BHT) values, which were corrected using Monte Carlo methods to account for uncertainty, and 19 high-quality continuous temperature logs, including wireline and fiber-optic measurements. To robustly account for data heterogeneity and measurement uncertainty, particularly in the error-prone BHT correction methods, Empirical Bayesian Kriging (EBK) was applied within a 3D framework. The model was computed on a 100 × 100 × 100 m voxel grid and provides probabilistic temperature distributions for P10, P50, and P90 scenarios. Cross-validation using a leave-one-out approach yields a mean standard error of 5.6 K, with more than 87% of predictions falling within the modelled 90% confidence interval.

The resulting temperature model reproduces well-known regional thermal anomalies of the Molasse Basin, including positive anomalies in the Munich and Landshut areas and a pronounced negative anomaly associated with the Wasserburg Trough. In addition, a 3D Empirical Bayesian Indicator Kriging approach was used to derive probability maps for reaching specific temperature thresholds (e.g., 80 °C and 100 °C), providing a robust probabilistic framework for geothermal assessment.

Ongoing work focuses on coupling the solely statistical EBK temperature model with lithology-specific thermal conductivity data derived from laboratory measurements, mixing-law models, and petrophysical interpretations of logging data. This will allow calibration of the temperature field, derivation of regional heat-flow densities, and calculation of horizon-based temperature gradients. The GeoChaNce results provide an improved, uncertainty-aware thermal framework for the Bavarian Molasse Basin, contributing to more reliable geothermal resource assessments and forming a key component for a future geothermal decision-support system for the reservoir.