Developing a national surface water flood map, Scotland
The Scottish Environment Protection Agency (SEPA) has used Flood Modeller’s 2D FAST solver to produce, for the first time, a national surface water flood map to cover the whole of Scotland (total area: 78,000 sq.km). This is referred to as the national pluvial flood map for Scotland.
Flood Modeller’s 2D FAST solver used to produce a national surface water flood map (total area: 78,000 sq.km)
Flood spreading was carried out using a set of approximately 4,000 2D FAST models per rainfall scenario
The total simulation run time for the whole of Scotland is under 48 hours per rainfall scenario
Allows SEPA to determine areas at significant flood risk and inform the requirements for future detailed flood risk mapping
Flood Modeller’s 2D FAST solver is designed to enable rapid assessments of flooding. It provides results up to 1,000 times quicker than traditional 2D solvers. Providing results in seconds or minutes, as opposed to hours or days, different types of flooding can be rapidly screened, considerably reducing the cost and time spent modelling.
The national pluvial flood map complements the fluvial and coastal national maps already available to SEPA to inform the preliminary Flood Risk Assessment, which will in turn enable determination of the areas at significant flood risk and inform the requirements for future detailed flood risk mapping.
The project was delivered in two phases. The first phase involved data collection and method
development together with trial application to a pilot area. The second phase involved application of the method to the whole of Scotland and reporting on this.
During the first phase of the project, a range of calculation methods and data options were assessed for a pilot area which covered Glasgow and the Clyde and Loch Lomond catchments in western and central Scotland.
The NextMap Synthetic Aperture Radar (SAR) digital terrain model (DTM) was the most accurate readily available national terrain data set. The data provides a representation of the ‘bare earth’ (i.e. without buildings and vegetation canopy) ground levels on a 5m grid resolution. During the DTM preparation the dataset was modified based on Ordnance Survey Vector Map to include an approximate representation of building thresholds by increasing ground levels by 0.3m within building footprints.
Representation of rainfall is based on the Depth-Duration-Frequency (DDF) model of the Flood Estimation Handbook (FEH). Rainfall was processed using a 100m resolution urban/rural classification of the country to generate a 100m grid of effective rainfall. Based on trials on the pilot region and literature survey, rainfall data was prepared for three rainfall scenarios: 30-year rainfall return period, a 200-year rainfall return period, and a 200-year rainfall return period with 20% increase for climate change. Under each of these scenarios, a further distinction was made for urban and rural areas taking into account representative percentage runoff, storm duration and sewer capacity for each land use.
Flood Modeller’s 2D FAST solver was used to simulate pluvial flood spreading. The 2D FAST solver uses simplified hydraulics to spread water over a DTM. Its high computational efficiency, robustness (numerical stability), and volume conservative approach (which focuses on identifying ponded areas) makes it suitable to identify risk of flooding from pluvial sources.
Flood spreading was carried out using a set of approximately 4,000 2D FAST models per rainfall scenario (each model representing a hydrological catchment). The total simulation run time for the whole of Scotland is under 48 hours per rainfall scenario.
The model run for each rainfall scenario produced a GIS file (raster grid) of pluvial flood depths for each catchment at a 5m grid resolution. These results were then processed to remove small ponded areas (<200 square meters), apply flood depth thresholds of 0.1m and 0.3m, and generate polygon GIS files of flooded areas for the two depth thresholds aggregated for the whole of Scotland.
The project deliverables consist of six sets of GIS polygon ‘shapefiles’ identifying the areas predicted to be susceptible to pluvial flooding for the three rainfall scenarios with depth thresholds set at 0.1m and 0.3m. In addition, GIS raster grids of simulated pluvial flood depth at a resolution of the 5m x 5m cells (consistent with the DTM data) were delivered thresholded at 0.1m and 0.3m.
Validation of the simulated pluvial flood outputs was undertaken using a comparison with historical observed pluvial flooding data and with simulated results generated by other, more detailed, methods. ‘Sensibility’ checks to identify non-physical behaviour was also carried out.
The comparison data sets identified during the project that were relevant, reliable and available for use included the following information: (a) post flood event report on Glasgow East End flooding of 30 July 2002, (b) selected interim results (March 2011) from the modelling for the Glasgow Surface Water Management Strategy which used a detailed integrated surface-drainage model and LiDAR data.
The qualitative assessment undertaken confirmed that the method provided satisfactory results for identification of areas susceptible to pluvial flooding influenced by topography (as represented in the DTM) at national and regional scales.