Use of Strava Heatmaps in Search and Rescue
Finding missing people is a hard job and one that is made easier with better data about the area around where they’re missing. Strava Heatmaps are a good source of additional data.
Earlier this week, a link popped up in my Twitter stream, pointing out how Strava Heatmaps can be used to locate military bases. There’s now a lot of news coverage around the negative aspects the data that they’re exposing, and that’s rightly feeding in to the wider debate about privacy and increasing awareness to ensure we’re all able to make informed choices about what data we choose to share online. But, I’d like to explore a different angle here — how that data, and data like it, could save lives.
I volunteer with a Search and Rescue team in Oxfordshire. We’re a Lowland Rescue team and we primarily search for vulnerable missing people — they’re people who our local police force have decided are likely to come to harm if they’re not found quickly. “Time = survivability” is one of the mantras we’re taught in our initial Search Technician course. As a Search Manager, I’m often in the control van planning where to send the search teams, which areas are the highest priority and whether I’ve got the right resources. As part of our planning we use the Six Step Process, developed by the Centre for Search Research, which revolves around creating scenarios of where the missing person may have gone. Those scenarios need to be based on data. The more data, the better the scenarios can be, and the better the scenarios are, the higher the probability of finding the missing person quickly. We have lots of data sources for missing person behaviour, like Koester’s Lost Person Behaviour book, the iFind booklet, or the CSR studies, but maps are key to deciding where to send teams.
I love maps. I know lots of other people have a strong affinity to maps too, but I think maps have a special place in the hearts of those involved in search and rescue. As a Lowland Rescue team, our operating area is the whole county of Oxfordshire — 2600 square km of English countryside. We can’t know it all as intimately as teams with smaller patches can. Accurate, detailed maps are vital to our ability to identify potential search areas and understand the ground around us. We use the excellent Ordnance Survey maps for our planning, and they remain my favourite overall maps. We supplement that base mapping with information from sources like Google Earth to give us a different view on the area. A few years ago we started using OpenStreetMap to give us an additional view on paths that may exist — we’ve found that OSM often includes paths that OS doesn’t. Now, with this data from Strava, we have an extra source we can use.
When we look at a map, what we’re trying to infer from it is not just what the terrain looks like, but how that area might be used. The Strava Heatmaps give us an additional datapoint we could consider when we’re developing scenarios or tasking teams to areas.
If we take a look at the three map segments below, you can hopefully see what I mean:
This is the base map from OS, giving us a great overview of the area. It shows us the woodland, including type of trees, along with location of water courses, and other major features:
Next up, we have the OpenStreetMap option, which shows us less general detail about the area, but does feature some additional paths that aren’t shown on the map above:
Next up, we have the Strava Heatmaps, and this is where you can see this otherwise empty expanse of trees comes alive with networks of routes:
Something to bear in mind is that we don’t have access to the raw data behind the Strava maps, and therefore can’t rule out problems like inaccurate GPS readings or people recording walks they (or their dogs) take when they’re not on paths — although there is technical blog that goes into some detail about some data processing they do. With that in mind, though, there appears to be some frequently travelled routes that don’t show on the other mapping sources we use and those paths may change the type of team I can deploy into an area — for example, depending on the missing person, I might choose to prioritise routes and paths over areas of woodlands, and knowing in advance that there are potentially extra paths makes it easier to plan for that. Alternatively, I might be able to more effectively use a level 2 search dog that specialises in searching routes and paths rather than waiting for a level 3 search dog to arrive and conduct a more time consuming area search.
OxSAR doesn’t have a cycle team, but I know other Lowland Rescue teams do use bike-based searchers regularly, and being able to scope the Strava Heatmap to just just show data from bike rides could help identify which paths might be suitable to deploy a bike team on to.
My day job as architect of software designed to help companies comply with information governance regulations (like GDPR) means that I understand the data privacy concerns. However, I’m also big fan of open data, and once appropriate levels of consent and anonymisation are in place, I believe that opening up data that large organisations have is in the public interest. As this example highlights, organisations often won’t know the practical application of insights that can be gained from their data until they share it.
Please let me know if you’re aware of any other data sources that could be useful for search planning, and also: if you like the idea of being woken up in the middle of the night to look at maps or relish the prospect of actually getting your boots dirty to go walk those potential paths looking for someone who needs help, then get in touch with your local search team.
