2D Hydrodynamic Modelling of Kendal with CityCAT

This lesson explains how CityCAT was used to model flooding in Kendal during Storm Desmond, based on Using the CityCAT Hydrodynamic Model to Assess Fluvial and Pluvial Flood Risk in Kendal (MSc Dissertation, Newcastle University, 2019). It covers the governing equations and numerical approach, the Kendal model set-up, validation using flood extent and depth evidence, and the model’s key limitations.

1. What CityCAT represents

The dissertation places CityCAT within the family of 2D hydrodynamic flood models used to simulate shallow flood flows across urban surfaces. In general, 2D models treat flow as varying across a horizontal field and assume water depth is shallow. The source notes that many 2D flood models use the full shallow water equations, which represent conservation of mass and momentum after depth-averaging the Navier–Stokes equations. These equations do not have analytical solutions, so numerical schemes are needed to approximate water depth and velocity through time.

In Using the CityCAT Hydrodynamic Model to Assess Fluvial and Pluvial Flood Risk in Kendal (MSc Dissertation, Newcastle University, 2019), CityCAT is used to represent both fluvial flooding from the River Kent system and pluvial flooding from intense rainfall and overland flow pathways in Kendal.

2. Governing equations and numerical scheme

The dissertation discusses CityCAT as a 2D hydrodynamic model designed for urban flooding, and it is framed against common shallow-water modelling approaches. The key physical basis is:

• Conservation of mass: water volume must balance inflows, outflows, and storage on the grid.
• Conservation of momentum: flow movement is controlled by water-surface slope, bed slope, and resistance.
• Shallow-flow assumption: horizontal scales are much larger than water depth, allowing depth-averaged treatment.

The source also refers to dynamic, diffusive, and kinematic wave approaches in the background discussion of 2D modelling. In this context, CityCAT is presented as a hydrodynamic model that solves the full 2D shallow water equations using a finite volume, shock-capturing numerical scheme, with Green-Ampt infiltration for permeable surfaces and buildings (from OS MasterMap) removed from the computational grid, rather than a simple static inundation method.

Numerically, the dissertation uses CityCAT on a raster grid over Kendal. The model updates flood depths across cells over time from the imposed hydraulic and hydrological inputs. This means that the flood pattern emerges from:

• the local elevation differences in the DEM,
• boundary inflows from rivers,
• spatially distributed rainfall or runoff contributions, and
• surface roughness controlling resistance.

So, for this lesson, the most important interpretation from Using the CityCAT Hydrodynamic Model to Assess Fluvial and Pluvial Flood Risk in Kendal (MSc Dissertation, Newcastle University, 2019) is that CityCAT solves 2D shallow-flow behaviour numerically on a grid to simulate time-varying inundation, rather than just mapping low-lying land.

3. Storm Desmond context in Kendal

The dissertation states that Storm Desmond caused major flooding in Kendal in December 2015. Across the catchment, an average of 183 mm of rainfall fell in 48 hours, around 10.1% of the average annual total. Kendal was heavily affected, with an estimated 2,150 residential and business properties impacted.

The catchment setting matters for the model because Kendal is influenced by several watercourses. The dissertation identifies the River Kent catchment as including important subcatchments such as the River Mint and Stock Beck. Their combined behaviour helped generate the flooding observed during the event.

4. Model set-up for Kendal

4.1 Digital elevation model (DEM)

The model was built on a high-resolution raster terrain representation of Kendal. The dissertation indicates a 2 m model resolution for the CityCAT simulation. Terrain is central in CityCAT because it controls where water ponds, spills, and forms flow pathways across the town.

The source also highlights an important issue related to buildings. Their representation in the DEM was necessary to simulate realistic urban flow pathways, but this complicated validation of flood extent. Because buildings interrupt raster flood patterns, direct comparison between simulated and observed flood polygons can show apparent under-representation at building locations and over-representation at edges after resampling.

4.2 Inflows and flood-generating mechanisms

The Kendal simulation was designed to recreate the combined effects of fluvial and pluvial flooding during Storm Desmond. Based on the dissertation, the important inflow mechanisms included:

• River inflows associated with the River Kent system, especially the Kent, Mint, and Stock Beck.
• Rainfall-driven overland flow, including runoff from surrounding land into urban streets.
• Local drainage and surcharge-related effects, although these were simplified because the model did not explicitly include the real storm-drain network.

The source provides examples of observed mechanisms that the model sought to reproduce, such as:

• right-bank overtopping of the Kent downstream of Stramongate Bridge,
• flooding at factories on the left bank of the Mint,
• surface runoff from fields east of Kendal affecting Ullswater Road, Eastgate, Jenkin Rise, and Sandylands Road,
• overflow of the Stock Beck culvert, and
• rainfall-generated flooding around Wildman Street, Ann Street, and Castle Street before river flooding arrived.

These examples show that the CityCAT set-up was not limited to a single river boundary problem; it was intended to represent the interaction of channel overtopping, tributary effects, and urban surface runoff.

4.3 Roughness

Surface roughness in a 2D hydrodynamic model controls how easily water moves across the terrain. The dissertation used Manning's n values of 0.035 for permeable surfaces and 0.02 for impermeable surfaces, and urban flow routing depended on the land surface representation. In practice, roughness would influence:

• flow speed across roads, open ground, and developed areas,
• the timing of inundation, and
• local ponding depths.

Within the model logic described in Using the CityCAT Hydrodynamic Model to Assess Fluvial and Pluvial Flood Risk in Kendal (MSc Dissertation, Newcastle University, 2019), roughness works together with topography and inflows to control simulated inundation patterns.

5. Validation of the Kendal model

5.1 Comparing simulated and observed flood extent

The dissertation validates the simulation against observed evidence from the Environment Agency and reported flood mechanisms across Kendal. A table of key flooding mechanisms shows whether the observed process occurred in reality and whether it was reproduced in the CityCAT model. This is useful because it tests not just the final extent, but the way flooding developed.

The model reproduced many important mechanisms, including overtopping, pluvial flow routes, and the Stock Beck culvert overflow. The dissertation specifically states that the CityCAT simulation “incidentally recreated the real-life occurrence with high accuracy” for the Stock Beck culvert overflow when compared with the observed Environment Agency extent.

5.2 Critical Success Index (CSI)

The dissertation uses the Critical Success Index to assess agreement between simulated and observed flood extents. The CSI is a spatial fit measure based on the overlap between:

• areas correctly predicted as flooded,
• areas predicted as flooded but not observed, and
• areas observed as flooded but missed by the model.

In interpretation terms:

• a higher CSI means better agreement between simulated and observed flood extent,
• the metric penalises both overprediction and underprediction, and
• it is useful for judging how well the mapped inundation pattern matches reality.

In the dissertation, CityCAT achieved a CSI of 0.63, compared with 0.55 for the Environment Agency flood map, and a weighted CSI of 0.71.

The dissertation also explains that validation is sensitive to how the observed and simulated extents are rasterised and aligned. Because buildings were represented in the 2 m DEM, comparison with mapped observed extent was not straightforward. Resampling could still leave under-representation at buildings and over-representation at extent boundaries. Therefore, CSI results must be interpreted alongside the known mapping and rasterisation issues, not treated as a perfect measure of model truth.

5.3 Depth comparisons

Validation was not only about whether a place was wet or dry. The dissertation also considers flood depths. Depth comparison is important because two models can produce similar extents but very different hazard levels.

Interpreting depth validation involves asking:

• did the model produce plausible depths in the locations known to have flooded?
• were the deepest areas located in the right places?
• did the model capture the contrast between river overtopping zones and more local pluvial ponding?

From the evidence discussed in Using the CityCAT Hydrodynamic Model to Assess Fluvial and Pluvial Flood Risk in Kendal (MSc Dissertation, Newcastle University, 2019), the model’s strength is that it reproduced many of the main observed pathways and mechanisms. However, at the four comparison sites (Carus Green, Mintsfeet, Sandylands and Wildman Street) simulated depths exceeded reported depths by about 32% on average, and flood mechanisms tended to occur earlier in the model than reported, so depth comparison remains limited by uncertainties in observations and by simplifications in the urban drainage representation.

6. What the validation tells us

Overall, the validation suggests that the CityCAT model was able to capture the main flood behaviour of Storm Desmond in Kendal reasonably well. In particular, the source supports three broad conclusions:

• Major pathways were reproduced: the model captured important overtopping and runoff routes across the town.
• Both fluvial and pluvial influences mattered: flooding in Kendal was not solely a river problem, and the model reflected that interaction.
• Spatial fit metrics need careful interpretation: CSI is useful, but results are affected by DEM resolution, building representation, and how observed extents are digitised and rasterised.

7. Limitations of the model

Using the CityCAT Hydrodynamic Model to Assess Fluvial and Pluvial Flood Risk in Kendal (MSc Dissertation, Newcastle University, 2019) makes clear that the model has important limitations.

7.1 No explicit real storm-drain network

One of the biggest limitations is the simplified treatment of urban drainage. The dissertation notes that storm drains function as part of the urban flood-defence system. Because they were not explicitly represented, the simulation can be interpreted as something like a scenario in which urban flood-defence function has failed during an event on the scale of Storm Desmond. This may overestimate some surface flooding in areas where drainage would normally remove water.

7.2 Building-related validation issues

The representation of buildings improved hydraulic realism, but it also made direct extent comparison harder. Gaps in the simulated flood raster at buildings can reduce apparent overlap with observed flood polygons even where the model behaviour is physically reasonable.

7.3 Resolution and rasterisation effects

The model used a 2 m grid, while some observed or digitised data had different effective scales. Resampling between resolutions can create artificial disagreement, especially along flood edges and around small urban features.

7.4 Uncertainty in observed flood maps

Observed flood extents are not error-free. Digitised outlines and reported flood locations may contain uncertainty in timing, local depth, and exact spatial limits. This means validation results reflect uncertainty in both the model and the observations.

7.5 Simplified representation of extreme-event infrastructure behaviour

The dissertation notes that floodplain mapping often excludes defences on precautionary grounds, since defences can fail in extreme events. In Kendal, this raises a modelling question: the simulation may represent a very conservative scenario for drainage and defence performance, which is useful for hazard understanding but may differ from actual infrastructure performance in every location.

8. Key takeaways

• CityCAT was used as a 2D hydrodynamic grid-based model to simulate shallow urban flooding in Kendal during Storm Desmond.
• Its physical basis is the conservation of mass and momentum for shallow overland flow, solved numerically across a raster terrain.
• The Kendal model relied on a 2 m DEM, river and runoff inflows linked to the Kent, Mint, and Stock Beck system, and surface roughness to route water.
• Validation used flood mechanism checks, spatial flood extent comparison, and interpretation through the Critical Success Index, as well as depth plausibility.
• The model reproduced many important observed mechanisms, including the Stock Beck culvert overflow.
• Main limitations include the lack of an explicit real drainage network, building-related comparison problems, and uncertainty in observed extents.

9. Summary

According to Using the CityCAT Hydrodynamic Model to Assess Fluvial and Pluvial Flood Risk in Kendal (MSc Dissertation, Newcastle University, 2019), CityCAT provided a useful way to simulate the complex combined flooding that occurred in Kendal during Storm Desmond. Its 2D hydrodynamic framework allowed the model to represent overtopping, tributary interactions, and urban runoff pathways across a detailed DEM. Validation showed that many observed mechanisms were reproduced, and metrics such as the Critical Success Index helped quantify spatial agreement, although such metrics were affected by data-processing and representation issues. The model was therefore informative and credible, but its results must be interpreted alongside the important limitations of drainage simplification, raster resolution, and validation uncertainty.