Personal space exploration – pavement widths and walking space for social distancing

As UBDC continues to provide evidence of the effects of the COVID-19 pandemic and support the recovery process with data and analysis, one area of uncertainty concerns mobility and space.

With many city dwellers rediscovering the personal and environmental benefits of walking and cycling during the lockdown period, it’s widely hoped that such trends will continue when we return to more normal working and movement behaviours. Nervousness at returning to crowded public transport may encourage people to walk or cycle rather than hop on a train or bus. With ambitious environmental targets remaining despite the current crisis, local, devolved and national government organisations are keen to dissuade commuters from returning to private cars. Many UK cities and administrations have announced investments in temporary and permanent infrastructure improvements and additions to enable suitable physical distancing.

We have by now become quite accustomed to the two-metre gap that should be maintained between ourselves and others – one of the key social distancing conventions intended to limit the virus’ spread and keep each other safe. But with the volume of people on our pavements and footpaths expected to grow as lockdown measures ease, we know that there will be some areas where social distancing is more challenging.

One important consideration is the width of pavements and walkways that are accessible to pedestrians – where these are narrow, social distancing is less achievable. But how do we measure the width of pavements across large areas reliably and efficiently? Fortunately, data and methods are available that mean we don’t have to rely on just shoe leather and tape measures to do so!

To measure pavement width at scale, we turned our attention to our local city, Glasgow. UBDC is working with Glasgow City Council on issues relating to mobility and neighbourhoods so it seemed an obvious setting to develop an initial proof of concept. Our first step was to identify the paved ways, sidewalks and paths themselves. We used the Ordnance Survey MasterMap Topography Layer, accessed via Edina’s Digimap service as our starting point. We filtered four 10km tiles covering most of Glasgow to produce a dataset comprising man-made roadside features, although different filters could be used to exclude or include other features.

Having isolated these polygons we needed to be able to identify the central line of each polygon, establish a means to calculate the distance between two points and establish a value for pavement width – essentially the sum of the lengths of the shortest lines between the centre line and the left/right boundaries of the parent polygon.

Figure 1 - From polygon to pavement

We use a built-in transformer within FME called CenterLineReplacer to calculate the centre line. FME is a commercial tool that supports manipulation of (especially spatial) data. The software is quite accessible, available for non-commercial work under a free personal license.

Figure 2 - Centre lines and width vectors

The transformer does a pretty good job of identifying approximate centre lines, our basis for calculating width. Subsequently, each centre line is densified to individual points along that line, attaining sub-metre resolution. For each one, we measure the shortest distance between that point and the approximate diametrically opposed points on the polygon – i.e, from the centre to the edges of the pavement. To ensure the width is represented correctly, we enforce a rule that the angle formed between the lines connecting the centre point with each edge can be no less than 140 degrees. Finally, for each join between adjacent centre points we reconstruct a line segment and apply to it a width value – the average distance between the two connected points.

Figure 3 - Calculating average pavement width by segment

The applications for more informed social distancing investments and policy are quite clear - Glasgow City Council has already expressed an interest in using the resultant data to inform its policy and operational decision making. Because of the scalability of the method and the full UK-wide availability of the OS topography data, we can calculate equivalent pavement width datasets for any location, customised to include only those features of interest.

Some very light analysis of the data has revealed that around 31% of Glasgow pavements included in our pilot sample are less than 2 metres wide. About 63% are between 2 and 4 metres wide with just 6% or so in excess of that. We’ll be interested in looking in more detail where the narrowest pavements are – particularly where they are collocated with areas with high footfall or traffic volumes, which make it more difficult to simply step onto the street to maintain social distance.

We’re currently in the process of producing additional extracts and hope to be able to provide access to these data and services in the very near future. If in the meantime you are interested in discussing the methodology or exploring how it might be used in your location, please get in touch with us and we’ll be very happy to chat!

Authors: Nikos Ves, Data Scientist and Dr Andrew McHugh, Senior Data Science Manager.