OR/15/016 Review methodology

McKenzie, A A, Ward, R S. 2015. Estimating numbers of properties susceptible to groundwater flooding in England. British Geological Survey Internal Report, OR/15/016.

name="_TOC_250007"

The Environment Agency have flood risk mapping for three different types of flooding; flooding from rivers, the sea and surface water. These maps are used in conjunction with the National Receptor Database (NRD), a GIS of properties and infrastructure, to calculate the number of properties that lie within areas with a 1 in 100 and 1 in 1000 chance of occurring in any year2. Comprehensive records of households and other properties affected by groundwater flooding are not available, for England and Wales. Some of the reasons for this are discussed below.

name="_TOC_250006"

BGS has prepared GB-wide groundwater flood susceptibility maps3 – on a 50 metre grid, with a nominal resolution of 1:50k. The maps are effectively an amalgamation of datasets on permeability and on groundwater level (either observed or estimated) to identify shallow groundwater.

The dataset recognises two modes of groundwater flooding – ‘Clearwater’ when groundwater levels rise in an unconfined aquifer in response to recharge higher in the catchment (which can be further subdivided into flooding from Chalk and Limestones, and flooding from other aquifers) and ‘Permeable Superficial Deposits’ (PSD) where surface water can move through sands and gravels and cause flooding adjacent to a watercourse.

Four classes of groundwater flood susceptibility are defined:

C: Potential for groundwater flooding to occur at surface, emergent

B: Potential for groundwater flooding of properties situated below ground level, affecting infrastructure

A: Limited potential for groundwater flooding to occur D: Not considered to be prone to groundwater flooding

Assumptions:

Susceptibility doesn’t mean that flooding will occur, and ignores anthropogenic factors that might reduce risk – e.g. drains, cut-off walls or elevation of infrastructure above ground level.

name="_TOC_250005"

Data on actual occurrences of groundwater flooding comes from 3 sources.

· Recorded information. This includes incidents logged by the Environment Agency at household level. Some data may also be collected by Department for Transport (for roads), water companies (for sewers), the fire service, local authorities and insurance companies. At local level this may be complemented by reports from flood wardens and other local groups. With the exception of the Environment Agency data, these reports are not centrally collated.

· Systematic survey. This may include mapping during flood events, either on the ground or using remote sensing/aerial photography.

· Reports. Published and unpublished reports and literature on groundwater flooding and local hydrogeological conditions.

2 Flooding in England: A National Assessment of Flood Risk, Environment Agency, 2009.

3 http://www.bgs.ac.uk/research/groundwater/datainfo/GFSD.html

Assumptions:

Logged incidents may fail to distinguish between flooding from rivers, surface water and groundwater incidents, and incidents may have multiple causes.

Groundwater flooding incidents are relatively rare, and were not systematically recorded before 2000/01. Each episode of flooding is driven by unique spatial distributions of rainfall, and it is likely that there are areas where flooding may occur, but where it hasn’t occurred since detailed records began.

Logs of incidents represent reports, rather than occurrences. Households may not report for a variety of reasons, including familiarity, where incidents are common, assumption that authorities are already aware or perceptions that reporting won’t make any difference.

Aerial survey is likely to be limited to groundwater flooding at the surface, and is unlikely to detect shallow groundwater, sewer surcharge and flooded basements.

name="_TOC_250004"

To estimate the number of properties that may be affected by groundwater flooding groundwater susceptibility mapping has been used, supplemented by data from the Environment Agency on reported incidents during the 2013/14 flooding event.

• ) BGS groundwater flood susceptibility has been used to highlight the areas where groundwater flooding might occur, distinguishing the source of flooding, the geological environment and whether water is likely to reach the surface or affect buried infrastructure. The classes used were:
• ) Clearwater flood susceptibility on Chalk and limestone aquifers, emergent or affecting infrastructure
• ) Clearwater flooding on other bedrock aquifers, emergent only iii) Permeable superficial flooding, emergent only

• ) Properties within the different areas have been summed. This was carried out using counts of residential properties from the NRD, coupled with an allowance for non- residential properties of a further 33% (NRD estimates of properties are more accurate than postcode based estimates, which may be 40% smaller).

• ) Results were compared to the area that is considered to have at least a 1:1000 chance of flooding in any year from river or sea.

• ) The number of properties in a susceptible community was compared to reported incidents where data were available to calculate the percentage of properties vulnerable to flooding.

Assumptions:

This approach works well on outcrop Chalk where the flooding process is straightforward, and good observations are available, on other geologies there will be greater uncertainty.

Estimating the total number of properties affected from reported incidents requires expert judgement.

No account is taken of differences in occurrence and reporting between urban and rural areas. This may lead to an over estimate of affected properties in urban areas where more developed drainage and sewer systems are expected to reduce receptor vulnerability.

It is assumed that groundwater flooding processes are similar across a geological unit, so estimates of susceptibility and vulnerability based on recent events are applicable in areas outside the footprint of the 2013/14 event.

It needs to be emphasised that this approach is designed to estimate the number of properties that might be affected in a statistical manner, not to identify individual properties at risk.