Modelling 3D subsoil temperature evolution by combining deterministic-probabilistic approaches based on geostatistical conditional simulation

Modelling the underground temperature variability is a key element in geothermal energy designs which is affected by many parameters such as seasonal changing temperature of the surface, depth of bedrock, population density, etc. The most used approach is deterministic experimental models which can be modified and adapted based on a specific case study. Urban settlements dissipate heat in the ground, leading to uncertainties in defining temperature along depth. Regularly, the deterministic models of calculating subsoil temperature are valid for local areas and defined conditions, but fail to estimate the temperature evolution on the large-scale. In this research instead, the probabilistic approach - geostatistical multivariate conditional simulation - is tested to model the underground temperature variability using turning bands algorithm. A mixed urban-rural area (17 km2), with availability of various log temperature measurements, as well as surface information on natural and anthropic factors, has been considered as the case study. Results highlighted the potential of using probabilistic models to recreate the 3D- temperature underground maps in space and time. In contrary of the deterministic models with a sharp discontinuity at bedrock, simulation results demonstrated the more consistent gradient change at the transitional zone between the sediments and the bedrock leading to improve the suitability maps quality and sustainable use of underground space specifically at large-scale levels. The main objective of this study is to test the proposed methodology. Due to the lack of available data under the specific conditions required for the analysis, synthetic boreholes were generated based on four real boreholes with temperature measurements. These synthetic datasets were then used as input for the methodology.