Hi Regina,
Thanks for the follow up.
To be more specific, the application would indeed cover the whole globe
(including bodies of water). It would be based on regions drawn by
users (and/or points selected by users) via google maps. These
coordinates would be stored and then displayed again on google maps.
There's a possibility we may seek to include input from GPS devices
(depending on the added complexity).
The vast majority of use cases will be proximity analyses on points v.
points, regions v. regions, and regions v. points. By proximity, I mean
what locations are either adjacent or enclosed by another. Since these
would be average users, proximity would have to be measured in meters
(or miles) not in degrees of latitude or longitude.
Based on my interpretation of the points you make, it sounds like
perhaps the approach would be to store in both lon/lat and UTMzone: the
former for submitting back to gmaps and the other for doing the above
proximity analysis? Would you agree?
Would it make sense to store coordinates in different regions of the
world using different SRS's? Or, would simply using the Goole
Projection (i.e. http://spatialreference.org/ref/user/6/) for any
point/polygon on the globe be sufficient? Also, how much of these issues
are non-issues given that the display in in gmaps, and the input is
coming from gmaps?
Thanks,
Ed
Paragon Corporation wrote:
Ed,
Depends of your region of interest and which analysis (mostly database or
just user drawing on a map) and how much of it you will be doing. Also
what kind of tool you will be using to render your maps.
If you are covering the whole globe - you might be best to leave your data
in long lat and transform to an appropriate UTM zone or other regional zone
as needed or have your mapping app do the transformation for you to the base
map layer projection. (a lot of mapping software can do this)
For display purposes long lat (4326, 4269) looks squashed so I usually go
with US National Atlas (2163 which provides a nice balance of look and
measurement but really only works for continental US) or Mercator/Google
Mercator (which sucks for measurement but looks decent on a map and covers
the globe) . If you are just dealing with overlaying points on a map (and
not polygons, lines etc then the issue of the skewedness of polygons etc is
a non-issue so you can keep in 4326) but you'll probably want to maintain a
functional index on the based on transform of the UTMzone.
By far a state plane zone or UTM zone is preferable for any kind of intense
measurement analysis and for presentation on a map.
You could transform on the fly but if you are doing enough of that, it will
affect performance a little and is an added hassle since you will also want
to keep a functional index on that transformation.
Anyrate 2 resources I always recommend if you haven't looked at them already
1) http://spatialreference.org - lists both well-known and user contributed
projections and for each provides the PostGIS sql insert state to insert
into the spatial_ref_sys table if the spatial ref doesn't already exist in
PostGIS.
2) http://postgis.refractions.net/support/wiki/index.php?plpgsqlfunctions -
the UTMZone function which given a point geometry will return the SRID of
the UTM zone the point falls in.
Hope that helps,
Regina
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Edward
Q. Bridges
Sent: Friday, June 27, 2008 8:26 PM
To: postgis-users@postgis.refractions.net
Subject: [postgis-users] Selecting an SRID
Hi,
I'm relatively new to PostGIS and to GIS concepts in general, and I'm
working on a project which has a goal of being able to identify points
within regions (basically).
In enabling a geometry column I'm stuck on how to decide on which SRID to
use. The data will come into the system as lon/lat; and it will be used as
both lon/lat (i.e. to show system users points within regions) and in meters
(i.e. to tell users how far away one point is from another). Other sorts of
analyses will be what regions are near a region, what points are near a
point, and what regions and points are enclosed by another region.
Will my choice of SRID affect how efficiently these analyses can be done?
Is it possible to use one SRID for all of these different usage scenarios?
Thanking in advance for any help, or useful ideas.
Best regards,
Ed
_______________________________________________
postgis-users mailing list
postgis-users@postgis.refractions.net
http://postgis.refractions.net/mailman/listinfo/postgis-users
