[postgis-users] ST_area returning wrong results (and stretched polygon in the screen)

[p valdes]

Hi,
I'm trying to calculate the area of a polygon, with strange results...

First a description of the problem, the ugly details at the end of the post

I have measured the perimeter of a pond. An irregular
polygon vaguely 'chop shaped' without holes or spikes
and closed (beginning and ending in the same point).
For simplification purposes let's suppose that I was
expecting something like the left shape...
well, I'm obtaining instead this:

expected    obtained
-------            ---
|      |           |  |
|      |           |  |
|      |__        |  |
|          |       |  |
-----------        |  |
                   |  |
                   |  |
                   |  |_
                   -----

The right shape is "not right". The polygon is clearly
stretched in the vertical North-South axis in the screen. The
satellite google map support that the real shape of
the pond is the first represented and I think the same...

Well I could live with this, but there's two more nasty consequences.

1) If I take a point inside or related to the polygon
in a separated map layer the point is drawn OUTSIDE
and far away below the figure of the pond (note also that the
pond is printed much more big than expect, because the relative distance
between a and b is the same in both, and seems right). Graphically:

expected     obtained

--*a               ---
|  |                |  |
|  |                |  |
|  |                |  |
|  |                |  |
|  |                |  |
|  |                |  |
|  |                |  |
|  |_              |  |_
-----*b           ------






                      *a
                      *b

Maybe a problem of the map viewer software, yes but the second
question is more serious:

ST_distance calculates a distance between the "a" and "b" points of
about 160 m for this rectangle. A max length of 160 m for this pond
seems reasonable to me.

but ST_area calculates an area for this polygon of almost 7 ha!, for a
polygon that could fit in a rectangle of about 160x50 m!.
 I'm obtaining a faked area about seven times BIGGER than real, uh-oh...

Its clear to me that I'm doing a very obvious and stupid mistake. I
greatly appreciated if you could take a look at the problem or suggest
any possible source of the error

Thanks in advance

pvaldes.



.....................................................................
Details
..................

The points were registered with a garmin GPS as coordinates in degrees
(ie: 43.254 N -4.839 W)

I'm using PostgreSQL 9.0.4
          on i486-pc-linux-gnu, compiled by GCC gcc-4.6.real (Debian
4.6.0-6) 4.6.1 20110428 (prerelease), 32-bit
          (obtained as Debian package from an official repository)

postgis-1.5.3
      (downloaded and compiled from the tarfile at postgis.refractions.net)

This is the table

CREATE TABLE mytable(
ref serial PRIMARY KEY,
name varchar(80),
perimeter geography(POLYGON,4326)
);

The polygon with lat/lon coords was loaded with something like:

INSERT INTO mytable (perimeter, name)
VALUES (
ST_GeogFromText(
'POLYGON((
-5.3850 43.230,
-5.3858 43.231,
-5.3859 43.245,
   ...
-5.3850 43.230
))', 4326),'myname');

And this is the very simple query returning strange results

SELECT st_area(mytable.perimeter)::double precision AS square_meters
FROM mytable
WHERE name = 'myname';

I use pgsql2shp for obtaining a shapefile that I'm loading with thuban