[HOT] Nepal major road validation

kusala nine kusala9 at googlemail.com
Tue Apr 28 12:22:42 UTC 2015


Hi - just a bit more info on the validation on roads. I concentrated on
Task 6 and got the following results from OSRM vs the UN document. OSRM are
in (brackets). All distances in km.

Kathmandu - Naubise 25 (30)
Naubise - Palung 37 (43)
Palung - Simbhanjyang 15 (18)
Simbhanjyang - Bhainse 41 (41) note: Bhainse = Bhaise Dovan on geonames...
Bhainse - Hetauda 11 (12)
Hetauda - Bharatpur 77 (76)
Palung - Kulekhani 21 (35) I think Kulekhani not yet fully captured and
actual location looks vague to me.

I'm using this to focus on digitising some of the Task 6 squares to get the
main road distances working. The area around Naubise may be dubious as two
of the distances are greater in OSRM. Please note that, as I said before
Kathmandu-Hetauda = 129 on the network map and 221 on the triangle map
(unless I'm suffering from lack of sleep??). Anyway, I'm moving on to task
3 (most of task 5 is outside the area of the network map)... Hope this
helps. I will also keep digitising roads in task 6 according to these
results.

Jon.


On Tue, Apr 28, 2015 at 11:11 AM, kusala nine <kusala9 at googlemail.com>
wrote:

> I agree -  a good methodology is to use the smaller tree diagram to
> validate in dividual distances and look at where they differ significantly.
> The UN doc itself is inconsistent. If you look at Kathmandu to Hetauda
> the tree diagram distances don't add up to the distance in the triangular
> table (89km vs 211km!). The tree distances are much closer to the OSRM
> distances and you can work on smaller ones. I managed to validate the
> distances within task 6 last night fairly quickly. I'll write the results
> up and post them on later today.
>
> Jon.
>
> On Tue, Apr 28, 2015 at 11:02 AM, Pierre Béland <pierzenh at yahoo.fr> wrote:
>
>> Great
>>
>> From your table, I see many significative differences. Taking shorter
>> segments would help to validate further I think.
>>
>> cheers
>>
>>
>> Pierre
>>
>>   ------------------------------
>>  *De :* kusala nine <kusala9 at googlemail.com>
>> *À :* hot <hot at openstreetmap.org>
>> *Envoyé le :* Mardi 28 avril 2015 1h47
>> *Objet :* Re: [HOT] Nepal major road validation
>>
>> I did quick comparison just using names from the distance calculator from
>> kathmandu (results below) but now moved onto checking using the smaller
>> network diagram. this lets you concentrate on areas which are in the
>> current task manager tasks. Just finishd Task 6 last night and they look
>> pretty good covering Kathmandu, Hetauda, Bharatpur, Palung and Bhampali.
>> I've since digitised roads in a couple of key tiles based on km distance
>> discrepancies and when I've finished those I'll move on to task 3. The
>> triangular distance table is good but points to a lot of discrepancies in
>> the longer distances which are out of the target zones.  I'm happy to
>> digitise the triangular table though if anyone wants to try the URL method
>> to identify areas of difference. I'm using NGAs geonames service to
>> identify differences in place naming - what should we do when there's a
>> difference? e.g "Bhainsie = Bhaise Dovan". Can I attribute a synonym?
>>
>>
>> *place**OSM**UN**difference**Comments…*baitadi8421168-326biratnagar396549
>> -153chandragadhi473624-151terhathum505654-149ilam538685-147dharan403545
>> -142birganj135276-141hetauda82221-139rajbiraj317456-139gaighat314452-138
>> dhankuta467595-128janakpur295378-83tulsipur401441-40narayanghat133144-11
>> bhairahawa274279-5dhunche115117-2dhulikhel3132-1lumbini300301-1trishuli
>> bazar6970-1salyan5045031kodari1161133pokhara2021984chautara87825jiri188
>> 17612butwal27225715tansen31129615Birendranagar59557520dipayal83681620
>> mahendranagar70668422dhangadhi67665026nepalgunj52549926taplejung88783552
>> gorkha19714057dailekh66460559krishnagar831334497baglungno route271Baglung
>> not connected - closest major place is pokharaBarhabiseno place111no
>> such place. Closest OSM place is Mangalsen by wikipedia's coordinates.
>> kakarbhittano place618not inOSM. Closest is charali = 1036km
>> (charali-kakarbhitta=11km)sindhullmadhino place387still looking….
>>
>>
>>
>> On Tue, Apr 28, 2015 at 12:53 AM, Andy Anderson <aanderson at amherst.edu>
>> wrote:
>>
>> Hi, Andrew,
>>
>> The documentation says “The results are network distance, i.e.
>> travel-time, in 10th of seconds.” It’s hard to translate that into
>> distances without the speeds assigned to road segments. If one knows the
>> type of road and what the “common” speed is for that type of road one can
>> make an estimate, but it could still be quite far off. Do you have any
>> guidance on how to get road segment information? If one needs to download
>> the road data itself, then the length of the segments should be there, too.
>>
>> — Andy
>>
>> On Apr 27, 2015, at 6:44 PM, Andrew Buck <andrew.r.buck at gmail.com> wrote:
>>
>> > -----BEGIN PGP SIGNED MESSAGE-----
>> > Hash: SHA1
>> >
>> > The OSRM distance table calculator is documented here...
>> >
>> >
>> https://github.com/Project-OSRM/osrm-backend/wiki/Server-api#distance-ta
>> > bles
>> >
>> > Basically you feed it a url with a bunch of lat/lon pairs are
>> > parameters (say for the 10 largest cities in the area) and then it
>> > creates a 10x10 table of the distances from every city to every other
>> > city.  So we can duplicate the table in the top right corner of the
>> > PDF, we just need someone to get the lat/lon of each of those cities
>> > and feed them into the api call documented above.  If our distances
>> > are signifigantly greater than the PDF distances listed for any
>> > entries then that probably indicates a problem in our data on the road
>> > between those cities.
>> >
>> > - -AndrewBuck
>> >
>> >
>> > On 04/27/2015 07:57 AM, Pierre Béland wrote:
>> >> http://un.org.np/node/10028, shows road distance of major cities
>> >> from kathmandu. Look at the pdf files that shows distance of
>> >> various road segments. Any volunteer to use this data and compare
>> >> with OSRM's distance matrix tool? If distances are significantly
>> >> different, this would indicate missing connecting roads or wrong
>> >> tag like path.
>> >>
>> >> Pierre
>> >>
>> >>
>> >>
>> >>
>> >>
>> >> _______________________________________________ HOT mailing list
>> >> HOT at openstreetmap.org https://lists.openstreetmap.org/listinfo/hot
>> >>
>> >
>> > -----BEGIN PGP SIGNATURE-----
>> > Version: GnuPG v2.0.22 (GNU/Linux)
>> >
>> > iQIcBAEBAgAGBQJVPrvkAAoJEK7RwIfxHSXbpKQP/0aQYXGoDWoHJqiDPx0n92sF
>> > wNIQ5nVdS2bOr+5BG7HBr1IIyjnmMWaqwfQa7b1PqI1qAJBI9gY2yyK2Hg0uGrj8
>> > AQvAnitwveBwJYfgL3OBH6KD1j6OQqKgeRDg/j58tPtO+n3egWQhLlxs/XO9cGmX
>> > aL4TLvrkxIlBnbpPsq7/bw1J7c/XidaxNgb4MkRmNV8hx/ICn+km4UTgu/+jlpOY
>> > BHICBmX4M/9t01r9WR91yLG4fYPenb2vytl077+0NItqKSwEh0FlWnfUm1e/W/06
>> > gXKx3hvbYzXEM0juN1QVYJOBi1O+KORLcMdvAlfFhSyZ1Az9pKj3s/bp9j4EwSZW
>> > 6b+lrLUCKLm+JAboJ1gy/Q1/O/0Oo6iI+e8iWLPiCHGl+gO4DFJTZHU25Na3E+TQ
>> > ExN25NGa94947sh10gNWmWkMB3G+7N5xX6qo/5j8Es9G2VfLnHwAIUdzIqI4n8YB
>> > DJ82wDFH3JMZJW2qAwS5FQyzyz6kSsGOqJMVGz96CTLf+A74U597RqCHCjEr3rb1
>> > x/GhUawgPZDeEuPX+TFdAa2ma1kG1DOiIxejCSkgNCOrWvqXL5EK3t1viiy4yC7j
>> > cEIKCGIg6Z4ZD8//SqYjxKUevHV+1jDcRgzD0AnA90vsZGzooF0eCEOy+FsCaDHJ
>> > 3HYcOE6DTu7JhsaPunL/
>> > =ZC61
>> > -----END PGP SIGNATURE-----
>> >
>> > _______________________________________________
>> > HOT mailing list
>> > HOT at openstreetmap.org
>> > https://lists.openstreetmap.org/listinfo/hot
>>
>>
>> _______________________________________________
>> HOT mailing list
>> HOT at openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/hot
>>
>>
>>
>> _______________________________________________
>> HOT mailing list
>> HOT at openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/hot
>>
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/hot/attachments/20150428/0c4247c0/attachment.html>


More information about the HOT mailing list