A model of perfection

Angus Middleton assesses the use of tools for natural flood management.

There are many tools out there that help us understand how nature can reduce flooding, but how useful are they? I would suggest not very – not because of problems with the tools, but how we are using them. We are getting lost in the joys of modelling and losing sight of the goal.

The tools can be put into two camps: map analysis (geographic information systems), and mathematical flow calculations (hydrology). The former entails creating maps that show certain characteristics, such as habitats and soil type, to identify locations that possess more or fewer target characteristics. In this way, a location with a steep slope, poor vegetation and clay soil would be identified as unable to store water during heavy rain. This would be given a high opportunity rating for natural flood management (NFM): improving its water-holding abilities, perhaps by planting trees, should lessen local flooding.

But imagine that a large marshy area is at the bottom of the slope. When it rains heavily, the water will indeed pour down the hill almost instantly and deluge the area below. But this area is marshland, so will absorb the water and stop it flooding into the river (for a while). Planting trees on the slope will therefore slow the water entering the marsh, but will have almost no effect on how much enters the stream to cause local flooding. We would have spent considerable time and money on creating a lovely hillside forest, but gain no reduction in flooding.

This may sound simplistic, but it could explain why there are such mixed results from NFM projects, which commonly use GIS alone to identify where to make NFM interventions. Even when hydrology is used to understand the degree of flood reductions that these interventions will make, it is generally GIS that identifies where to place the interventions in the first place.

Hydraulic modelling mathematically calculates how water flows down hills and into rivers, then down those rivers to flood settlements (for instance). The models generally try to say how much water will flow past a given point during a given storm event. This is then combined with other information to determine flood levels in and around the river. It does not show where interventions should be placed, or what they should be. This usually has to be identified using other techniques, then these changes put into the hydraulic model to see what happens.

The equations are complex and require detailed input data: any errors can significantly skew results. There is also a problem with assumptions. Many assessments look at a one-in-100-year storm event, during which it is assumed that rain will fall in ‘these’ locations and ‘these’ volumes over ‘this’ period. It is far from certain how well these relate to real-life weather events.

Hydraulic modelling is sophisticated and accurate, but how meaningful is it given a capricious natural system? Does it make sense to model using ‘these’ assumptions and ‘this’ specific storm? NFM is all about using nature, so the solutions themselves will change over the decades. Surely it is wiser to use models that reflect this and work with the data uncertainties of landscapes, averaging results over the lifetime of the natural solutions rather than suggesting an impossible degree of certainty.

GIS and hydraulic modelling are both excellent tools, but for NFM they should be used in the right ways. GIS works best for broad-brush concepts at large scales – such as where NFM is likely to be valuable – rather than informing local actions. Hydraulic modelling is perfect for scenario analysis – the ‘what ifs’ of specific situations – but is not so good for showing what changes to make to deliver most benefit under real conditions.

The answer is to combine hydraulic and GIS modelling into a single system to prioritise NFM actions across catchments. We should then quantify the expected reduction in flooding as an average over the lifetime of the habitats, since this is where NFM functions most efficiently. Generating more exact results for specific storms will give a false sense of accuracy at unnecessary cost.

This pragmatic approach will make NFM prioritisation modelling sufficiently cheap and accessible for every project to use, and help people understand the likely benefits in meaningful terms.

Angus Middleton is director at Viridian Logic