Across the UK there exist neighbourhoods where a lack of supermarkets and food stores limit access to food supplies, including fresh fruit and vegetables. This recent study looked to identify where these ‘food deserts’ are by using area units. We wondered what this might look like using an alternative spatial gap analysis and some open data on supermarkets, together with small-area census data on residents, which would in some way overcome the use of aggregating data into defined areas.
By using the population-weighted centroid of a Census 2011 Output Area (around 130 households in England and Wales, 50 in Scotland and 150 in Northern Ireland), falling within the radius of a supermarket we were able to identify the geographic gaps. We used 1080m and 12.5km for 15 minute average walk and drive times respectively.
There are wide regional variations. As we might expect, London is served very well by supermarkets, with almost no ‘food deserts’. Outside London, the worst English region is the South West. Outside England, Wales and Northern Ireland are also particularly under-served.
The results also give us a more detailed picture at the local scale. Neighbourhoods within otherwise well-served cities are also evident as ‘food deserts’ in the context of supermarket access, particularly in the context of walking access. The example below in Newcastle-Upon-Tyne reveals the neighbourhoods of Hedworth, Whiteleas and Percy Main among others. These important local variations can often be lost in headline statistics.
Using open data often has its limitations, and all analysis should identify where the drawbacks lie. Whilst this brief model identifies supermarkets, buried within the larger dataset are other, smaller food stores. Whilst it would be too difficult to identify these for the whole country, a local analysis may make use of this data for a more comprehensive picture of food shops.
Building on these results, analysis could also examine in greater detail which of these neighbourhoods are also affected by poverty and public transport access, potentially exacerbating the food desert problem. To improve the spatial accuracy, a network analysis using multiple transport modes and sensitivity testing on the distances used could also be used. This would be an interesting analysis!
You can examine this data further on our web map.