wrapping around the dateline using geographic coordinates

[percy]

Brent, thanks for taking the time to tell me about your tactics!
This gives me some ideas to chew on... It will take a while to work it 
all out :-)

Brent Wood wrote:
> --- percy <percyd at PDX.EDU> wrote:
> 
>> Thanks Frank!
>>
>> I'd rather stick with something more standard so that we can integrate 
>> with other services.
> 
> Hi Percy.
> 
> We have done this two ways (New Zealand has the same problem). 
> 
> By a custom mercator with a central meridian at 100 & lat of true scale at -41.
> Geographic coordinates are noticeably lacking however :-(
> 
> 
> I have used for some years a 0-360 longitude EPSG:4326 coordinate system.
> 
> Note that EPSG:4326 does not explictly define the longitude range, but most
> implementations assume +-180, although many, including mapserver, are quite
> happy with 0-360 data.
> 
> I have a request in to EPSG to consider tidying this up & providing a 0-360
> WGS84 "projection" with a formally specified EPSG code. I'm not sure where this
> is at, hoping to hear soon.
> 
> Proj.4 now allows users to specify the central meridian of a lat/long
> coordinate system, so you can specify 0 (default) or 180 (ie, 0-360).  
> 
> GMT supports (effectively) a mod360 longitudinal model, which is great, but not
> much use with mapserver.
> 
> 
> With raster tiles, I have used +-180 based tiles, then symbolic linked the
> -180-0 tiles & generated world files which are 180-360 for the same tiles. This
> avoided having to replicate tiles & gives a -180 to 360 cover.
> 
> Using PostGIS for my vector data, something along the lines of
> 
> create table temp as
> select id + (select max(id),
>        ...,
>         setsrid(translate(the_geom,360,0,0),4327);
> 
> insert into table
> select * from temp;
> 
> replicates all the -180 to 0 features with 180-360 longitudes, (where my 4327
> is a custom projection, 4326 with the proj central meridian at 180).
> 
> I'd guess a view on the 4326 tables unioning translated vectors would also
> work.
>         
> 
> Lines & points work fine, polygons require tidying up by selecting
> intersections with a polygon of each longitudinal hemisphere to close them at
> the new limits, but it is quite workable. Hope this makes sense...
> 
> If you have to work with other data which is stuck at +-180, I think stuck
> might describe the situation well.
> 
> If EPSG come to the party it may provide a solution longer term.
> 
> 
> Cheers,
> 
>   Brent Wood
> 

-- 
David Percy
Geospatial Data Manager
Geology Department
Portland State University
http://gisgeek.pdx.edu
503-725-3373