Dear James,

I work for the Ordnance Survey and forwarded your query to Dr Philip Davies, 
the Ordnance Survey  Geodetic Adviser, below is his reply, I hope it 
provides some clarification for you.

Regards,

Pete, Ordnance Survey, UK.


<<snip>>
I don't know anything about MapInfo, but I can tell you about how the 
Ordnance Survey coordinate system relates to GPS coordinate systems.

OS National Grid coordinates can be represented either as eastings and 
northings or latitude and longitude.  Both forms give exactly the same 
information.  Regardless of which of these forms is used, these coordinates 
are _not_ the same as WGS84 coordinates - the coordinate system is quite 
different.  Simply converting between (e, n) and (lat, long) will not change 
an OS coordinate into a GPS coordinate, or vice versa.  OS coordinates are 
between 88 and 102 metres east and between 50 and 80 metres south of GPS 
coordinates, the exact shift varying depending on location in a complex way.

There are lots of transformations around to convert GPS coordinates to OS 
coordinates, but most of them are very approximate.  The simple 
transformations found in GPS receivers and most software packages will only 
do the conversion with an accuracy of 5-10 metres.

Ordnance Survey has developed the National Grid Transformation OSTN97 which 
is the national standard transformation between GPS and OS coordinates.  It 
uses a 1km resolution grid of shift parameters covering Great Britain which 
gives a transformation accuracy of 20cm (RMS) everywhere in GB - good enough 
to match GPS data to any OS map.

OSTN97 is licenced to software developers in the fields of GPS, GIS and 
navigation for inclusion in their products.  This has only happened very 
recently so at the time of writing there are only a few packages using this 
transformation - these include a simple-to-use windows utility, for example. 
 The OS Geoid Model OSGM91 completes the 3D transformation by converting GPS 
ellipsoid heights to Mean Sea Level (Newlyn) heights.  This is also 
available to software product developers under licence.

A note on GPS coordinates:  at levels of accuracy better than 1m, we have to 
consider that there is more than one version of the WGS84 coordinate system. 
 The European standard version for precise positioning is called ETRS89, and 
this is the version which Ordnance Survey recommends for all GB positioning 
work.

Ordnance Survey will very shortly (by end of March 99) publish a new 45-page 
booklet 'A Guide to Coordinate Systems in Great Britain' which explains all 
the concepts in detail.  This will be downloadable in .pdf format from the 
OS website www.ordsvy.gov.uk by that date.

Specific technical questions concerning the Ordnance Survey coordinate 
system (including height coordinates) are answered by OS Customer 
Information by email - please see our web page.

Phil Davies, OS Geodetic Adviser.

<<snip>>
 ----------
From: [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Cc: [EMAIL PROTECTED]
Subject: MI When is a Lat/Long not ?
Date: 03 March 1999 10:39

Dear all

Can anyone help me with a problem involving Ordnance Survey coordinates,
MapInfo's conversion to Lat/Long and discrepancies with GPS-output ?

Basically the problem is:
   a) We've used MapInfo to derive Lat/Longs from OS-coordinates
   b) These coordinates don't match what the client's GPS says
   c) The GPS matches (nearly) to a transformation done by a 3rd party.

To give an example for OS coordinates 504469,247557 (middle of England):
    MapInfo Lat/Long     = -0.474, 52.116
    GPS             = -0.287, 52.070
    Client's database    = -0.294, 52.133
    Independent 3rd-party     = -0.283, 52.066
I know about the accuracy constraints applied to 'non-military' GPS but as 
the
MapInfo value is so anomalous this is not the issue.

To do the MapInfo 'conversion' I've used MI4.5 and simple done:
    Table->Create Points
    Projection ... British National Grid
    Select CentroidX(obj), CentroidY(obj) ......

Have I used MapInfo correctly ?
Is it's computation of projections correct ?
Could the other sources be incorrect ... and how/why ?

I don't actually know much about projections (it may well be obvious from 
the
above !) and I've only rejoined the list recently so if the answer is 
obvious
(or has
been covered elsewhere) my apologies for wasting time ... but please point 
me
to
such a solution !

Regards
  James

 ----------------------------------------------------------------------
To unsubscribe from this list, send e-mail to [EMAIL PROTECTED] and put
"unsubscribe MAPINFO-L" in the message body, or contact [EMAIL PROTECTED]







----------------------------------------------------------------------
To unsubscribe from this list, send e-mail to [EMAIL PROTECTED] and put
"unsubscribe MAPINFO-L" in the message body, or contact [EMAIL PROTECTED]

Reply via email to