Flooding and food: using GIS to assess the impacts of the 2014 floods

As the controversy surrounding the floods in southern England has escalated, the debate has often centred on trade-offs between protecting housing, farmland and – to a lesser extent – natural habitats. For some quarters of the agri-food industry the images of flooded farmland is further proof that conservation management has gone ‘too far’ and is putting national food security at risk.

While the flooding is terrible for the individual farmers affected and efforts to support them are much welcomed, the conclusion that UK conservation policy should be reversed needs to be carefully considered. Are such statements just hyperbole? Regardless of who or what was to blame for the flooding, it is useful to put these extreme events in context. Here we quantify the potential impacts on food supply from one high profile area: The Somerset Levels.

In the UK we are lucky to have access to an amazing amount of data on land use. This enabled us to undertake a rapid, high-level spatial assessment of the 2014 Somerset Levels flood with the help of one of 3Keel’s partners, remote sensing specialists geoger ltd. For our analysis, geoger’s Dr Alastair Graham used freely available satellite imagery and other data to assess the areas of land affected by the flooding. For more detail on the techniques used by Alastair see “The Science Bit” at the end of this blog.

The following maps show the extent of the flooding, the quality of land affected and the amount of that land that has a wildlife designation (click images to view high resolution versions).



So what can we say about the affected land from our analysis? Here are some of the key findings:

  • Approximately 12,000 hectares of land was inundated. This is equivalent to 0.07% of the UK’s Utilised Agricultural Area (or 0.12% of the UK’s permanent grassland);
  • Almost no Grade 1 land was affected – approximately equal amounts of Grade 2, 3 & 4 land were flooded
  • Little cropped land has been submerged by the flooding. Grass and uncropped land dominates the whole of the Somerset Levels (about 86 per cent of the total area)
  • This land is worth £170m and supports approximately 21,000 cattle and sheep (0.05% of the UK’s ruminant livestock population)
  • 73% of the inundated land is designated for wildlife (Sites of Special Scientific Interest account for most of this);

The question of balancing food security and non-agricultural land use is the subject of on-going debate and research. For example, it was included in a 2013 review on the restoration of fen peatland in East Anglia for the Committee on Climate Change. This study by Cranfield University estimated the impact on food security of withdrawing 20,500 peat-rich hectares of cropland from agriculture (twice the area we calculate was submerged by flooding in the Somerset Levels). They concluded that this withdrawal would “probably not have a major impact on UK national food supply and food security” and that “it is likely that the production of high value cropping would be for the most part made good by substitution of cropping elsewhere”. Compared to the Fens, the flooded area of the Somerset Levels is less productive (only 2% of Levels land is Grade 1 agricultural land, compared to 49% in The Fens).

Research for Natural England last year also identified the potential benefits to agriculture of environmental stewardship schemes (e.g. pollination, pest regulation, genetic resources, water regulation, etc.). Interestingly, the report highlighted a gap in knowledge on the impact of environmental stewardship on overall agricultural output. This is an important priority for research, given recent events and the potential conservation backlash.

So where does that leave us? From our analysis it appears that the flooded areas in Somerset, although on a scale not seen for many years in the UK, contribute modestly to national food supply. They do however host nationally and internationally important habitats and provide important ecosystem services (such as significant levels of carbon storage in peat-rich soils).

These findings suggest that, rather than abandoning conservation efforts in flood plain habitats, there are probably more important things that we can do to improve UK food security. We suspect that agricultural output is more significantly hit by the myriad of diffuse, less dramatic water-logging and poor weather events that occur far from conservation areas across the whole country (Defra reported a 14% drop in farm income in 2012, linked to poor weather). A better solution would therefore be to focus our efforts on helping all farmers become more climate resilient and economically sustainable.

While it is likely that extreme weather will have increasingly serious implications for global food security we need to be careful not to throw the bittern out with the floodwater. By better understanding landscapes through emerging techniques such as remote sensing, we should be able to better balance the competing demands on land in a way that puts affordable food on the table and protects our unique natural heritage.

The Science Bit

Remote sensing and geographic information systems (GIS) are increasingly being used in our supply chain consulting work, where the location of activities has such a marked effect on an issue’s importance. Investigating spatial patterns and interactions can enhance our understanding of a given issue.

The rapid collection and dissemination of satellite data creates the ability to readily assess the scale of environmental issues over wide areas. The visual nature of the image data helps place context around the issue being studied. On top of this, image analysts have at their disposal an increasing array of processing techniques to maximise the amount of information that can be extracted from the data, such as water availability, land cover type, water quality, etc.

And, for those techies out there – here is a quick overview of how we established the flood extent.

Three Landsat 8 images were obtained through the USGS EarthExplorer portal. These images were for path 203, row 24, and were collected on the 7 January, 23 January and 8 February 2014. The images were clipped to the Somerset Levels area, and then assessed using open source software, with the objective of mapping the estimated total extent of flooding throughout the months of January and February 2014. Extensive amounts of cloud covered parts of the imagery (unsurprising given the weather conditions leading to the flooding). This did make the analysis of the imagery slightly more time-consuming than usual as the cloud had to be removed as best as possible.

The Environment Agency published a map of estimated flood extent for the area using TerraSAR-X (ScanSAR mode) data, collected at 8m resolution for the 8 February. This satellite imagery was made available to official disaster response agencies through the International Charter initiative – a global system that provides geospatial data to help in disaster response. It is rarely required in the UK, but has been activated twice by the Environment Agency in recent months – which says something about the scale of the flooding we have experienced. The SAR data has the ability to image the Earth’s surface through cloud, hence its utility for such applications.

The estimated flood extent published by the Environment Agency using this specialist flood-mapping dataset with a higher resolution was 12,200 ha. Using the Landsat data, we have been able to estimate the flood extent at 11,760 ha, which given the confounding nature of the cloud, the lower spatial resolution and the rapidity of the analysis is acceptably close to the official estimate, and suitable for the purposes of the analyses reported in this post.

The Agricultural Land Classification and Designations data are available through the Natural England data portal and licensed under the Open Government License.

The interpretation of the results was undertaken by 3Keel director Richard Sheane, who has worked with farmers and land managers on conservation management in the Thames catchment.

This developing field of science should be of interest to the business world on two levels: firstly supply chains ultimately depend on the continued functioning of ecological systems for the provision of renewable raw materials. Secondly – I believe that the identification of key principles and characteristics of resilient ecosystems should provide insights to the managers of man-made supply chain ‘ecosystems’.