Re: SRID for analyzing a USA national data set in Meters

Right, I should be more specific from the outset.  I did do some searching thru the PostGIS archives, and didn't find the answer I was looking for; is there a PostGIS FAQ somewhere?

As for my problem, my inputs are two sets of geocoded hospitals, and I just want to be able to identify for each hospital in the first set, the hospitals in the second set within approximately 25 miles.  I will the map these sets, with a 25 mile buffer around the first set, using Geoserver.  So, distance and area are both somewhat important, heading not at all.  distance_sphere(oid), sounds good for the calculation, but won't help with the buffering because it doesn't tell me the 'distance' in lat/lng space that would equate to 25 miles (because of course this varies over the globe).  To achieve this I would need to reproject into something that is in meters, and buffer around that.

How egregious would you expect the errors to be if I simply use the projection for the UTM zone that represents, say, Central time?

thanks,
mike

woodbri@... wrote:

Michael,

The Simple answer is that there is no simple answer. You have not given us enough information about what kind of calculations you need to do and what are acceptable errors. Are you worried about distortions in distance, area, or direction/heading for example.

Why don't you do all you analysis in lat/lon and convert to meters based on the fact that 1 deg is roughly 69..... nautical miles and you can convert miles to meters? You this would be problematic over large changes in latitude. Or use distance_sphere() which uses another approximation based on a spherical earth.

Any suggestion we can offer will have problems. You need to look at the problems in light of you analysis constraints and pick a compomise that you can live with. If you give us additional information me might be able to make additional suggestions.

Also variants on this request can be found in the archives as this is a fairly common question.

Hope this helps,
   -Steve W

Michael Frumin wrote:
  

pedro,
    

This appears to suggest that i should pick an SRID based on one point in > my data set.  However, since the data set is national, it includes > points from all around the USA (i.e. at least 4 UTM zones).  So, if I > pick the SRID for, say, Mountain Time, will that horribly distort points > on the east and/or west coast?  is there no SRID that is perhaps less > accurate in any one time zone, but more accurate across all time zones > in the [continental] US?
thanks,
      

mike
    

pdoria@... wrote:
Hello Michael,

You determine the correct SRID with this bit of code (PHP):
(the $loc was obtained from a SELECT with the astext(geometry) function)

The idea behind this is summarized in the UtmZoneFromLong() function.
Pick a Long value and calculate the utm zone for it.
Then use the SridFromUtmZone() function to get the srid from the given >> mask
(see below) --> this queries spatial_ref_sys with the calculated $utm >> param.
SELECT srid FROM spatial_ref_sys WHERE srtext LIKE 'PROJCS[\"WGS 84 / UTM
zone $utm%' LIMIT 1

// extract the POINT geometry
$geometry = str_replace("POINT(","", $loc);
$geometry = str_replace(")","",$geometry);
$coordinate = explode(" ",$geometry);
$lon=$coordinate[0];
$lat=$coordinate[1];

// get utm zone from coordinates
$utm=UtmZoneFromLong($lon);
// get srid from the spatial_ref_sys table for the given utm zone
// add N or S
if($lat>=0) $utm.="N";
else $utm.="S";
$srid=SridFromUtmZone($utm, $connection);


// determine UTM zone from LonLat coordinates
// params $long: longitude
// returns (int) utm
function UtmZoneFromLong($lon) {
    $utm = floor(($lon + 180) / 6) + 1;
    return $utm;
}

// query the spatial_ref_sys table for the srid of the given utm zone
// params $utm: utm zone string (MUST BE TRAILED WITH 'N' OR 'S'!), >> $conn:
connection to the db
// returns (int) srid
function SridFromUtmZone($utm, $conn){
$myresult = pg_exec($conn, "SELECT srid FROM spatial_ref_sys WHERE srtext
LIKE 'PROJCS[\"WGS 84 / UTM zone $utm%' LIMIT 1");
$srid=pg_result($myresult, 0, 0);
return $srid;
}

HTH,
Pedro Doria Meunier

-----Original Message-----
From: postgis-users-bounces@...
[postgis-users-bounces@...] On Behalf Of >> mfrumin
Sent: segunda-feira, 2 de Abril de 2007 14:12
To: postgis-users@...
Subject: [postgis-users] SRID for analyzing a USA national data set in
Meters


I have a data set that includes points all over the United States in >> lat/lng
(say, SRID = 4326) but I need to do some analysis on these data in real
units, say meters.  So, my task is to select an appropriate SRID that >> will
work reasonably well for the whole USA to reproject lat/lng into >> meters.  There are a bazillion SRID's available. Any suggestions?

This analysis doesn't have to be super-precise, so I'm not so >> concerned with
the limitations of a national projection (as opposed to state/region).

thanks,
Michael

-- View this message in context:
http://www.nabble.com/SRID-for-analyzing-a-USA-national-data-set-in-Meters-t >>
f3505664.html#a9790471
Sent from the PostGIS - User mailing list archive at Nabble.com.

_______________________________________________
postgis-users mailing list
postgis-users@...
http://postgis.refractions.net/mailman/listinfo/postgis-users


_______________________________________________
postgis-users mailing list
postgis-users@...
http://postgis.refractions.net/mailman/listinfo/postgis-users
  > _______________________________________________
      

postgis-users mailing list
postgis-users@...
http://postgis.refractions.net/mailman/listinfo/postgis-users
    

_______________________________________________
postgis-users mailing list
postgis-users@...
http://postgis.refractions.net/mailman/listinfo/postgis-users
  

_______________________________________________
postgis-users mailing list
postgis-users@...
http://postgis.refractions.net/mailman/listinfo/postgis-users