[Talk-GB] Footpaths - search for the missing ones

Fri May 11 15:40:21 UTC 2018

Quick impressions:

   - There's a fair amount of noise in text, but most are "F.P."
   - Lat/lon could be reduced from 15 decimal places, would make file size
   far smaller. OSM use 7, but I suspect 5 (~ 1 m accuracy) would be fine.
   - Filtering by a buffer round OSM roads does not reduce count enough to
   be useful. 21k points in East Mids goes to 14k with 20 m buffer, 10 with 50
   m buffer.
   - Instead created 1000 m buffer around points and looked for distance
   from OSM highways in that buffer. This allows to focus on points which are
   distant from existing highways.
   - In the main dots which are a long way from highways are clustered in
   areas we already know lack footpaths. Map shows points over 400 m from an
   OSM highway, underlain by a heatmap of total length of missing prows. It is
   apparent that these are coincident (W of Derby, around Buxton, SE
   Derbyshire, Trent Valley in N Notts, much of Lincolnshire). Other areas may
   be simply a result of rather different comparison periods for the data
   (distance from road is 3 years old OSM data).
   https://www.dropbox.com/s/nz0893l9io61vtk/gb1900_fps1.jpg?dl=0
   - Paths which were formerly isolated may now be close to new roads and
   therefore get discarded with use of buffers or short distances.
   - Not clear that searching in urban areas is worthwhile. Using something
   like the OS Urban Area shape files may reduce volume.
   - Even with these filters the total points more than 500 m from a (2015)
   OSM road is nearly a 1000 for the East Midlands
   - The NLS 6 inch maps are needed for good comparison, although I suspect
   many paths will be on 1:25k
   - There are interesting paths which seem to have disappeared entirely
   from the PRoW network, but noting them does require local knowledge rather
   than a bulk comparison. Here are a couple I noted, which also appear on
   1:25k and therefore look like prima facie cases for lost paths:
   - https://openstreetmap.lu/os-ooc-nls.html#16/52.9181/-1.2688/nlsos1
      path N-S from New Farm
      - https://openstreetmap.lu/os-ooc-nls.html#16/52.9503/-1.2603/nlsos1
      path from Noggins Nook to Swanacar Farm

So broadly in conclusion: it doesn't seem to give more than comparison
against rowmaps for identifying missing paths for OSM, but it does have
potential for finding lost paths. For the latter case rather more
annotation of information would be needed.

Jerry

On 10 May 2018 at 22:50, Rob Nickerson <rob.j.nickerson at gmail.com> wrote:

> Many thanks. Now shared with Richard, Nick and Jerry. Wont share publicly
> yet as I wouldn't want to disrupt the project comms plan.
>
> @Dave: Oh yes this is definitely not for OSM import. It's node data for
> linear features for a start!! No, instead this can be used to identify
> possible missing paths which should then be investigated using ground
> survey, aerial imagery and GPS (or Strava) data. See it as a helping hand
> to direct you where to look.
>
> Best,
> Rob
>
>
>
>
>
> On Thu, 10 May 2018, 13:54 SK53, <sk53.osm at gmail.com> wrote:
>
>> Quick correction, as I uploaded heat map to wrong Flickr account. This is
>> the proper link: https://flic.kr/p/JSXgyh.
>>
>> J
>>
>
> On 10 May 2018 1:54 p.m., "SK53" <sk53.osm at gmail.com> wrote:
>
> Quick correction, as I uploaded heat map to wrong Flickr account. This is
> the proper link: https://flic.kr/p/JSXgyh.
>
> J
>
> On 10 May 2018 at 13:07, SK53 <sk53.osm at gmail.com> wrote:
>
>> I just checked on the Vision of Britain site: the core data is currently
>> released under CC-BY-NC. I presume OSM-UK have a waiver from these terms.
>>
>> Undoubtedly there will be rights of way which have effectively fallen in
>> to abeyance. I noted one the other day which was on NPE maps, but no longer
>> visible on the ground nor on modern OS maps, nor in the data available from
>> rowmaps. Broadly speaking such paths fall outside the ambit of OSM, but
>> finding such things is very valuable.
>>
>> Note that we have other sources as well. As a quick experiment I spent 15
>> minutes quickly tracing paths marked on NPE maps for SE Notts and managed
>> just over 200 using JOSM. This is of course what I should have done many
>> years ago rather than adding them to OSM (hindsight is a wonderful thing).
>> The geometry wont be very good, but can be refined using the 1:25 OOC maps.
>> Such data can be more useful than the raw names from GB1900, but could be
>> used in conjunction. Furthermore with suitable tagging this can be added to
>> OHM (I would suggest start_date=1900-01-01 with end_date=1950-12-31 unless
>> one knows path is still in use) which makes it a tad easier for sharing
>> (although OHM overpass instance is not working atm).
>>
>> A couple of other things to note regarding the GB1900 data:
>>
>>
>>    - Many current footpaths will be marked as Bridle Roads (B.R.). It
>>    would be useful to add these names to the available data.
>>    - footpaths and bridle roads often fall well short of their current
>>    entry points because the current right of way will have followed farm
>>    tracks and service roads, which in many cases have disappeared.
>>
>> Returning to use of rowmaps I have a recent geojson file of missing paths
>> in the North Midlands (Staffs, Derbys, Notts, Leics & Rutland) up on
>> github: https://github.com/SK53/osm-prow-stats. I intend to add other
>> areas as time permits. Unfortunately I've never got my comparison process
>> to work on PostGIS so I still use QGIS which is a little unwieldy for
>> automation. I process rowmaps data into a fairly standard form in PostGIS
>> before making the comparisons. This <https://flic.kr/p/25DgebX>is a heat
>> map of missing footpaths in the East Midlands area as of Autumn 2017, I
>> compare length of missing paths with total length in a tetrad (2km grid
>> square). It readily shows hotspots of missing paths. This was done to
>> identify suitable places for our 2018 New Year footpath mapping. The
>> National Forest area in SE Derbyshire still has a lot of outstanding
>> mapping to do: it's not too bad as walking country either.
>>
>> Jerry
>>
>>
>>
>>
>> On 10 May 2018 at 11:34, Nick Whitelegg <nick.whitelegg at solent.ac.uk>
>> wrote:
>>
>>>
>>> I might be potentially interested in developing something with this
>>> data, partly because I already run a site (freemap) which shows OSM maps
>>> for walkers and stores them in a PostGIS database - so it should be an easy
>>> process to filter out the data to find those points which are not close to
>>> an OSM highway. It would also be easy for me to adapt my existing code to
>>> visualise these "FP" points. Presumably they are just points with no
>>> indication of direction of the path? An "FP" label presumably has
>>> orientation so something could possibly be deduced about its course at that
>>> point if orientation was available too.
>>>
>>>
>>> I also already visualise the data so visualising the missing ROWs would
>>> be easily done too.
>>>
>>>
>>> It would be nice to develop features to find nearby locations where
>>> there are lots of these missing paths, e.g. if I am in Southampton, find
>>> the nearest village with 10, 20 (or whatever) missing paths within a 5-mile
>>> radius.
>>>
>>>
>>> Would be nice to have an app too so you can find these footpaths while
>>> you're actually out.
>>>
>>>
>>>
>>> So potentially interested in this, yes. I don't want to commit 100% but
>>> would be nice to have the data.
>>>
>>>
>>> Nick
>>>
>>>
>>>
>>> ------------------------------
>>> *From:* Richard Fairhurst <richard at systemed.net>
>>> *Sent:* 10 May 2018 09:07:49
>>> *To:* talk-gb at openstreetmap.org
>>> *Subject:* Re: [Talk-GB] Footpaths - search for the missing ones
>>>
>>> Rob Nickerson wrote:
>>> > Basically we have point data of historic footpaths (some 300k points)
>>> and
>>> > I think it would be amazing to compare this to OSM to see if we can
>>> find
>>> > more footpaths to map.
>>>
>>> Very cool. Could you post the data somewhere?
>>>
>>> Richard
>>>
>>>
>>>
>>> --
>>> Sent from: http://gis.19327.n8.nabble.com/Great-Britain-f5372682.html
>>>
>>> _______________________________________________
>>> Talk-GB mailing list
>>> Talk-GB at openstreetmap.org
>>> https://lists.openstreetmap.org/listinfo/talk-gb
>>>
>>> _______________________________________________
>>> Talk-GB mailing list
>>> Talk-GB at openstreetmap.org
>>> https://lists.openstreetmap.org/listinfo/talk-gb
>>>
>>>
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-gb/attachments/20180511/772d02a0/attachment.html>