Simple position plotting

Hi, hopefully someone can understand my question, and help.

Is there any reason why I can't do this .......

......take raster maps of small geographic areas (20km x 20km) with
cartesian coordinate systems - i.e., UK Ordnance Survey 1:10000 Landplan
series - and consider that WGS84 latitude/longitude from a GPS receiver can
linearly be converted to pixel positions on that map image, provided the map
image has been accurately geo-referenced using WGS84 lat/long to pixel
referencing?

I ask because this is what I'm trying to do ;-)
But I'm seeing significant inaccuracies which vary in both x and y axis. My
*suspicion* is that the inaccuracies are dominantly due to inaccuracies in
my georeference "calibration" measurements (which I'm taking with a
handheld, retail-quality GPS receiver).

If I use a much higher quality receiver, for example, with WAAS enabled,
will I see my results improve, or am I ignoring some subtely of GIS systems
which I need to take heed of, for example, something to do with coordinate
transformations between different datums, etc.

Any comment will be helpful, I'm sure.

Regards,

GRo

Hello GRo,

I think you need to consider that your 'scale' (pixels per degree) is different
in your x and y direction. Especially if your coordinate system is lat/lon, 1
degree lat is a totally different distance than 1 degree lon. So what you
propose can and will work only if you develop a relation for your x-direction
AND a different relation for your y-direction. (Since your area is small, you
don't have to consider the viariation in your x and y relation within your area)

Willem

GRo wrote:


--
Willem van Deursen, The Netherlands
wvandeursen_nospam@nospam_carthago.nl
replace _nospam@nospam_ for @ to get a valid email address
www.carthago.nl

Oh, added to my earlier answer, you have to make sure that your background map
is well aligned: making sure that your y direction is pointing straight north.
If you scanned you background map, this is generally not the case and you will
have to rotate it to make sure this is actually the case. Otherwise you end up
with relations as
x = f(lat,lon) and
y = g(lat,lon)

Willem

willem van deursen wrote:


--
Willem van Deursen, The Netherlands
wvandeursen_nospam@nospam_carthago.nl
replace _nospam@nospam_ for @ to get a valid email address
www.carthago.nl

since its a small area, and considering the scale, your maximum RMS
acceptable error is: 5

if the error is less then that, and less then 1, dont worry, its a
acceptable map.

of course, if it gets closer to 5, you might have to re cadastrate your
ground control points, or even do another field survey to obtain new
data from gps.



willem van deursen wrote:

On Thu, 06 Jul 2006 21:12:27 +0100, GRo wrote:


The problem is most likely that the standard methods to convert from WGS84
to OSGB36 introduce significant errors, in the order of metres and
sometimes 10's of metres. The OS do some "formulae" that solve this and
have accuracies in the order of 25cm or better. However they are not
freely available, and are piecemeal (that is depending where you are a
different formula is used).


JAB.

--
Jonathan A. Buzzard Email: jonathan (at) buzzard.me.uk
Northumberland, United Kingdom. Tel: +44 1661-832195