Impact of simplifications on numerical modelling of the shallow subsurface at city-scale and implications for shallow geothermal potential.

Anthropogenic infrastructures in the shallow subsurface, such as heated basements, tunnels or shallow geothermal systems, are known to increase ground temperatures, particularly in urban areas. Numerical modelling helps inform on the extent of thermal influence of such structures, and its potential uses. Realistic modelling of the subsurface is often computationally costly and requires large amounts of data which is often not readily available, necessitating the use of modelling simplifications. This work presents a case-study on the city centre of Cardiff, UK, for which high resolution data is available, and compares modelling results when three key modelling components (namely ground elevation, hydraulic gradient distribution and basement geometry) are implemented either 'realistically', i.e. with high resolution data, or 'simplified', utilising commonly accepted modelling assumptions. Results are presented at a point (local) scale and at a domain (aggregate) scale to investigate the impacts such simplifications have on model outputs for different purposes. Comparison to measured data at individual locations shows that the accuracy of temperature outputs from numerical models is largely insensitive to simplification of the hydraulic gradient distribution implemented, while changes in basement geometry affect accuracy of the mean temperature predicted at a point by as much as 3.5 °C. At the domain scale, ground temperatures within the first 20 m show a notable increase (approximately 1 °C volume-averaged and 0.5 °C surface-averaged), while the average heat flux over the domain is about 0.06 W/m at 20 m depth. These increased temperatures result in beneficial conditions for shallow geothermal utilisation, producing drilling cost savings of around £1700 per typical household system or about 9% increase in thermal energy potential. Simplifications of basement geometry and (to a lesser degree) the hydraulics can result in an overestimation of these temperatures and therefore over-predict geothermal potential, while the elevation simplification showed little impact.