OpenStreetMap logo OpenStreetMap

Mapping in Thailand

Posted by AlaskaDave on 11 October 2013 in English.

This is my first contribution to the OSM diary section although I’ve been a very active mapper in Thailand for the past year or so. I write a travel blog and this post contains links to my blog posts that are about OSM mapping. The blog is titled I’m Outta Here because I left my home in Alaska to travel all winter back in 2009.

Mapping in Thailand Oct 2013 - Discusses language issues when mapping in Thailand.

A mapping run to Lamphun Feb, 2013 - Describes a typical mapping run I do with motorcycle and GPS in Thailand.

Mapmaking 101 Dec 2012 - My first big mapping effort using Potlatch2

One item that needs explanation in the most recent post is that I used 4,000 nodes to trace the Kiew Ko Ma Reservoir as discussed therein. I don’t know why I did that in retrospect. I wanted it to be accurate and went a bit overboard. The OSM is a street map after all. Then, when I tried to upload it from inside JOSM, the program informed me that 2,000 nodes was the most it would accept for a single relation. Not wanting to throw away any of my work, I split the relation in two and uploaded it that way. The OSM slippy map renders it okay but I’m concerned that this is not the proper way to handle the situation. There is a link to the area on OSM in the October blog post.

There are several islands in the reservoir that need to be added but I’m waiting to see how it all turns out before adding them. The reservoir itself is an inner member of a larger wood polygon already and islands are inner members of the reservoir. It is complicated. I do not understand relations well enough yet to proceed.

Anyway, that’s my story for now.

Location: Chang Phuak, Chiang Mai, Pa Daet, Saraphi District, Chiang Mai Province, 50030, Thailand

Discussion

Comment from dcp on 11 October 2013 at 06:40

About that reservoir: The outer perimeter should be split into sections of less than 1900 nodes and be part of a relation and defined as “outer”. The islands should be mapped as islets and include other attributes such as “scrub” and be defined as “inner” within that same relation-number! Quite easy really.

Next comment from me is that OSM is much more than a road map. Obviously in many parts of the world the priority must be on motorways, highways, roads, tracks and paths in that order. For example, it is also about POIs and their attributes with make daily life easier.

But, just for fun, take a look at:

http://map.f4-group.com/#lon=11.4188180&lat=53.6245511&zoom=19&camera.theta=59.946&camera.phi=0.938

for what is possible!

Comment from dcp on 11 October 2013 at 11:13

I forgot to mention that it is advisable to keep the nodes per sector lower than the 2000 allowed. Reasoning: Even if your work is perfect, some one will come along and add/make changes to it. Keeping the number of nodes well below 2000 will reduce the chance that a sector has to be divided!

Comment from AlaskaDave on 11 October 2013 at 11:42

dcp

Thanks for the info. I’ve already seen the f4 maps and they are indeed impressive

Please clarify how the reservoir can be defined with role outer when it is already enclosed within a larger wood multi-polygon, a polygon whose role is also “outer”

Someone else mentioned the idea of splitting the outline into two smaller pieces, and then making them part of a relation I have not done that before and would be afraid to try it. I can also assure you that I will not attempt to make such a node-heavy polygon again.

Comment from dcp on 13 October 2013 at 10:06

Multipolygons don’t seem to be understood by many contributors: For example: Take a look at:

http://tools.geofabrik.de/osmi/?view=multipolygon&lon=-117.23495&lat=36.70966&zoom=6&opacity=0.87&overlays=invalid_geometry_hull,duplicate_ways,intersections,intersection_lines,ring_not_closed_hull,ring_not_closed,unconnected_end_nodes,touching_inner_rings_hull,touching_inner_rings,role_mismatch_hull,role_mismatch,duplicate_tags_hull,duplicate_tags,multipolygons_type_is_boundary,type_is_boundary,ways,role_markers,way_end_nodes,way_nodes

and this is CA, where many intelligent people live and it is not confined to USA. It is endemic to the whole OSM-World.

You can have multipolygons within multipolygons but you should try to avoid them if possible (newbies won’t be able to understand them), and it usually possible do so.

I, too, once mapped a reservoir: Link to Blog:

http://www.openstreetmap.org/user/dcp/diary/13817

It , too, was done almost perfectly but over time the newbies moved in and ruined it (my opinion), but I a thankful that it still is visible in OSM although not correct to OSM standards.

I assume that you have read the OSM documentation on multipolygons so you should be aware of what you are doing. Should you have need for advice/assistance then do ask for it: The OSM community is very responsive.

Comment from AlaskaDave on 13 October 2013 at 12:02

Wow, your reservoir makes mine look tiny! Good job.

With OSM Inspector I looked at the geometry of the Kiew Ko Ma Reservoir and it all checked out with the exception of the two islands which had the error message “inner_in_inner”. I changed their roles to outer and will wait to see how that checks out. At some point I will work up enough courage to try adding the two halves of the reservoir multipolygon together but using two distinct way segments for its outline in order to get around the 2,000 node limit. Thanks for bringing OSM Inspector to my attention. I have used it before but not enough to remember to employ it in diagnosing this problem.

Yes, I have read the documentation on multipolygons but it is not terribly helpful. The relation editor is another very opaque tool, difficult to use and not very user friendly. While there is much to like about JOSM, IMO it’s help facility is sadly lacking. The Search tool is another I would take issue with but then, it is still vastly better than nothing at all.

You say “You can have multipolygons within multipolygons but you should try to avoid them if possible”. Well, if that’s the case, then how would you map a reservoir inside a pre-existing wood polygon, and then add islands inside the reservoir?

The other advice you gave from your first post:

“The outer perimeter should be split into sections of less than 1900 nodes and be part of a relation and defined as “outer”. The islands should be mapped as islets and include other attributes such as “scrub” and be defined as “inner” within that same relation-number! Quite easy really.”

is fine but I think you’re forgetting that there was that pre-existing wood multipolygon into which the reservoir had to be inserted. How should I have handled that situation?

Thanks the the feedback and the helpful suggestions.

Dave

Log in to leave a comment