Fwd: [Qgis-user] CRS Transfer MGA94 <-> AMG94

[Ramon Andinach]

(Sorry, meant to send this back to the list)

Begin forwarded message:

> From: Ramon Andinach <custard at westnet.com.au>
> Date: 23 December 2010 22:09:38 
> To: Micha Silver <micha at arava.co.il>
> Subject: Re: [Qgis-user] CRS Transfer MGA94 <-> AMG94
> 
> 
> On 21/12/2010, at 17:50 , Micha Silver wrote:
> 
>> Ramon Andinach wrote:
>> 
>>> Hello,
>>> 
>>> I am wanting to take a set of points stored in csv format that are in the projected CRS GDA94/MGA94 zone 51 (MGAz51) and end up with a csv file with the same points in CRS AGD84/AMG zone 51 (AMGz51). Being able to do this is useful for reconstructing some historical datasets that I deal with.
>>> 
>> 
>> If all you need is the new AMG z51 coordinates, you might want to try using cs2cs to do the conversion. You could script this to create a new CSV file from the original with two new columns for the new coord pair.
>> Here's a one liner to get started (tested on linux):
>> while read line; do echo $line | gawk -F , '{print $13"\t"$12}' | cs2cs +init=epsg:28351 +to +init=epsg:20251;  done < MGAOnlyAC.csv
>> 
>> Columns 12 and 13 are the N and E coords, but in "reverse" order. The gawk output gets piped straigt into the cs2cs utility, which then dumps (to the screen, at this point) the coordinates, transformed from EPSG 28351 (MGA z51 ) to EPSG 20251 (AMG94 z51)
> 
> A script! I've clearly been hanging around it windows too long.
> 
> There doesn't seem to be enough of a shell in windows, even though qgis seems to have brought gawk with it, so I've been playing with it on ubuntu.
> 
> The script runs. Thanks, but it brings up another problem.
> I don't trust the values it's returning. They seem way to close to me.
> 
> So to test, I've collected a point on my GPS (GPS76, 7 sats, 3D, accuracy 4.3m).Then I've compared it to expected results from the GDA94 technical manual and the results from cs2cs.
> 
> Short version cs2cs does not hadle AGD84 and AGD66 (and their projections) appropriately. Badly enough that I'd call it buggy.
> 
> -ramon.
> 
> 
> 
> This is the long version.
> _GPS Data_
> 
> The GPS says that this point is:
> WGS84z51s,3542946,6598129
> WGS84,121d27'49.86"E,30d44'26.46S
> MGA94z51,352949,6598127
> GDA94,121d27'49.98E,30d44'26.46S
> AMG84z51,352810,6597969
> AGD84,121d27'44.50E,30d44'32.14S
> AMG66z51,352811,6597969
> AGD66,121d44'31.14E,30d44'31.14S
> 
> 
> _Expectations from GDA94 Techical Manual_
> 
> I'm not going to claim expertise in the next bit. Refering to the "Geocentric Datum of Australia Technical Manual" (www.icsm.gov.au/gda/gdatm/index.html) the document states that;
> 1. for points at an accuracy of 1m or greater, WGS94 and GDA94 are functionally the same, and
> 2. "GDA94 coordiates of a point appear to be about 200m northeast ofthe AGD coordinates of the same point."
> 
> One of the supporting documents (adg-gda.xls) has approximate east and north differences for GDA94, AGD84 and AGD66 and related UTM projections for each of the Australian 1:250,000 mapsheets. For the relevant mapsheet for this point (SH51-09) the approximates are as below.
> 
> Datums,Long("),Lat("),East(m),North(m)
> AGD66-GDA94,5.23,4.68,137,159
> AGD84-GDA94,5.23,4.65,136,159
> AGD66-AGD84,0.038,0.032,-1.40,097
> 
> These seem to fit the GPS data fairly well (within a metre or two, ie. under half the stated error of the GPS reading).
> 
> _cs2cs Conversion_
> 
> I've taken Micha's suggestion, and built it into a little script around it so that I can apply it to different conversions as needed. I've been given to understand that GPS typically think in WGS84 and so I've run cs2cs against the WGS84 lat/long value.
> 
> Starting point: 121.00d27'49.86" -30d44'26.46"
> 
> Results:
> CRS,East/Long,North/Lat
> WGS84z51s,352946.00,6598128.49
> GDA94,121d27'49.86"E,30d44'26.46"S
> MGA94z51,352946.00,6598128.49
> AGD84,121d27'49.86"E,30d44'26.46"S
> AMG84z51,352945.46,6598116.70
> AGD66,121d27'49.86"E,30d44'26.46"S
> AMG66z51,352945.46,6598116.70
> 
> I can understand AGD66 and AGD84 results being fairly close.
> I can understand GDA94 and WGS84 results being fairly close.
> But the GDA94 results and the AGD* results are way, way to close.
> 
> 
> Summary,
> My GPS's conversions are in line (more or less) with what I would expect from reading the GDA Technical report.
> The conversions using cs2cs and Micha's script are not accurate for AGD84 and AGD66 which are too close to GDA94.
> Can I respectfully suggest that there is a bug in the way the AGD84 and AGD66 datums are handled?
>