[Talk-transit] [Talk-gb-westmidlands] NaPTAN bus stop database import - some moreobservations

[Andy Robinson (blackadder-lists)]

As promised my review of a further 50 bus stops in Sutton Coldfield. See the
linked PDF [1] for full analysis.

50 locations evaluated in Sutton Coldfield. All stops except one in the
NaPTan database and the one that wasn't is an unused shelter in a row of
shelters in the centre of town so probably isn't in the NaPTAN dataset
anyway.

Of the remaining 49, 13 (27%) have no physical presence on the ground and in
most case have never been used as a stop as far as I know. 

The average positional error from my survey was 13.7m with a maximum of 84m
and a minimum of 1.6m. One third were within 5m. The standard deviation was
16.84m.

The majority of stops were logically positioned on the correct side of the
road and where it was close I checked the road alignment and adjusted it to
the GPS traces where appropriate. A small number of NaPTAN nodes were not in
their logical positions, eg both stops of a pair on the same side of the
street. The largest displacement values were all longitudinal displacements
along the street.

A question arises what to tag a NaPTAN stop that is not present on the
ground. I've used for now physically_present=no. These stops also don't
currently carry a highway=bus_stop tag.

Cheers

Andy

[1] http://ajr.hopto.org/osm/NaPTAN.pdf


>-----Original Message-----
>From: talk-gb-westmidlands-bounces at openstreetmap.org [mailto:talk-gb-
>westmidlands-bounces at openstreetmap.org] On Behalf Of Brian Prangle
>Sent: 01 April 2009 1:41 PM
>To: Talk-gb-westmidlands at openstreetmap.org; talk-transit at openstreetmap.org
>Subject: [Talk-gb-westmidlands] NaPTAN bus stop database import - some
>moreobservations
>
>Some more views and observations on the NaPTAN data import in Birmingham:
>1. It serves as a great QA on OSM data and shows that in the City Centre
>where we have not been able to get decent GPS traces more accuracy is
>needed so the potential of obtaining aerial photography is great.
>2. Its a great impetus to resurvey streets where earlier work (potentially
>inaccurate from NPE tracing, older GPS devices, less points collected,
>inexperience, not realising that roads have a width greater than the
>rendered line so placing bus stops too close to the road etc.) can be
>improved.
>3. Its also a great impetus to improve practice on surveying bus stops and
>be much more precise and comprehensive also a stimulus to edit all those
>old bus stops where we placed them as a node on the way rather than to the
>side of ways which is our current practice.
>4. QA works both ways and our surveys should help to improve NaPTAN data.
>5. As an exercise(excuse the pun!) I cycled from Acocks Green to Moseley
>and back this morning along the No 1 bus route and surveyed 47 bus stops
>each time standing at the pole(leaning against it you cant get closer than
>that!) or underneath the plate at a shelter: 4 bus stops coincided within
>3-4 m; 17 were good enough coinciding <8-10m. Thats approx 45%. The rest
>were out by anything up to 90m or were just missing.
>6. I think either the pole where there is no shelter or the bus stop plate
>at a shelter should be what we survey thats where the identification of
>what we survey is located
>7. Where a physical stop on one side of the road doubles up for one on the
>other side also I think were OK by surveying and tagging the physical one
>with Andys suggestion of a tag opposite=yes. The NaPTAN untagged node on
>the other side can be left in place to indicate the logical relationship.
>8. I like the idea of tagging where the bus stop is set back in a lay-by
>from the road which might account for some NaPTAN nodes being some distance
>from the road
>9. For our purposes good enough is probably sufficient rather than precise
>positional detail Im of the view that as long the bus stop has more or less
>the right relationships to its surroundings then thats OK.
>10. Surprised that there is no data for either the closed Digbeth Coach
>Station which is being rebuilt or the temporary replacement nearby. I
>thought we were importing off-street bus stops? Perhaps the NaPTAN data
>doesnt exist?
>11. Cant find any nodes for taxi ranks not imported or doesnt exist for
>Birmingham or Im not looking hard enough?
>12. For the few nodes where Ive estimated the fit between OSM and NaPTAN to
>be good enough Ive merged the nodes deleting the unverified tag and editing
>source tag to v=naptan_import;survey
>13. Verifying the data is going to be a long, slow process with a lot of
>resurveying needed.