Impact of input uncertainty within large-scale shallow geothermal assessment

The presence of underground infrastructure has been shown to affect subsurface temperatures beneath dense urban areas, in a phenomenon known as the subsurface urban heat island (SUHI). The increase in temperature can impact the subsurface in several ways, including groundwater quality and ecosystems, goods production and storage, and infrastructure maintenance. Importantly, and most relevant to this work, this phenomenon can have considerable impact on the shallow geothermal potential of the ground under cities, and accounting for this is an important aspect of estimating and planning the comprehensive provision of heating and cooling using the ground.

 

A barrier to the accurate assessment of city-scale shallow geothermal potential is scarcity of data on ground conditions, within both the natural and the built environment. The cost to acquire these subsurface data is prohibitively high, and uncertainties in the parameter values to use in numerical modelling remain, giving rise to propagated uncertainty in the results calculated for the potential, which is seldom accounted for. Quantifying the uncertainty in the determined geothermal potential given uncertain input parameters is an important step towards establishing meaningful potential estimates as well as understanding which parameters require more and/or more precise data measurements.

 

In a step towards this, this work builds on a previously published methodology for large-scale thermal and geothermal potential mapping, based on the identification of ground thermal archetypes. The methodology is expanded through the propagation of sources of input uncertainty, such as ground thermal parameters and temperature of subsurface infrastructure, to determine the variability in the ground temperature and, by extension, the large-scale geothermal potential within two boroughs of London, United Kingdom. Critical parameters are identified via an archetype-level sensitivity analysis and surrogate models are generated for each of the archetypes identified within the modelled domain. Uncertainty in the input parameters is propagated through to the volume-averaged temperature, using Monte Carlo simulations. The results show the effect of uncertainty from individual inputs as well as combined effects from multiple sources of uncertainty, contributing to an improved understanding of the reliability of shallow geothermal for space heating and cooling.