Around 4.7 million UK properties sit on land that can flood from below—groundwater rising through permeable rock and soil. Yet no national system exists to forecast where that water will actually emerge. Current groundwater flood warnings rely on water levels at individual boreholes. Forecasters must guess the flood’s spatial extent from scattered points. A handful of local studies have linked groundwater models to surface flood models, but nobody has scaled that up to the whole country. This project will couple the British Groundwater Model with a high-resolution surface flood code called HiPIMS. The student will identify where the groundwater model falls short—particularly in timing and resolution—and correct those errors, potentially using machine learning. The resulting framework will then be tested against historical flood observations and aerial imagery. If it works, the Met Office and Environment Agency’s Flood Forecasting Centre could issue spatially explicit groundwater flood warnings for the first time. That would shift flood response from guesswork to prediction, giving local authorities and infrastructure operators—water treatment plants, rail networks, power substations—actionable maps of where water will rise, not just that it will.
View original technical description
It has been estimated that 4.7 million properties are susceptible to flooding from groundwater within the UK; 1.5 million from clearwater flooding and 3.2 million from rising groundwater in permeable superficial deposits. Whilst we have estimates of susceptibility, we do not currently possess tools to simulate and forecast the spatial extent of groundwater flooding, nor groundwater flood risk, nationally. Current groundwater flood forecasts are based on models of groundwater levels at observation boreholes, which use 'trigger levels' to initiate groundwater flood warnings. Consequently, the spatial extent of a forecasted groundwater flood has to be inferred from point data. A small number of studies have developed approaches to integrate the modelling of groundwater discharge at the land surface, resulting surface water flows, and flood inundation, but these have all been applied over small areas. The study will investigate approaches to develop a national scale groundwater flood modelling system, which integrates distributed groundwater and surface water flood modelling codes, to simulate and forecast the spatio-temporal extent of flood events. The research will aim to support the flood hazard modelling and forecasting community e.g. the operational flood forecasting and warning service of the Met-Office-EA Flood Forecasting Centre, and wider FLOOD-CDT partners and stakeholders. The British Groundwater Model will provide the underpinning tool to simulate groundwater levels and emergence across Britain. We will assess the limitations of the BGWM and implement required changes (e.g. relating to spatial and temporal resolution, structure / parameterisation) considering performance metrics focusing on simulating peak groundwater levels. The student will investigate model errors and consider approaches to correct for these when coupling it to a surface water flood inundation model; potentially involving machine learning methods. Approaches to model surface water flood inundation, driven by groundwater, will be investigated and the project will seek to integrate the surface water flood modelling code HiPIMS with BGWM. Performance of the modelling framework in forecasting groundwater flooding will be quantified. This will make use of historical observations of groundwater floods, such as groundwater level time-series, and field observations and aerial imagery (e.g. as collected by the EA) of flood extents.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know