[Proj] Polish Uklad 1965 proper definition

[Maciek Sieczka]

I compared all the benchmarks done by me and Oscar van Vlijmen for 8 
input points mentioned on 
http://bugzilla.remotesensing.org/show_bug.cgi?id=818 and results are 
following.

Compared to Transpol, official GUGiK (Polish geodesy and cartography 
service) software:

[1]
PROJ.4 stere:
mean northing difference: 4,86306 m
mean easting difference:  2,57497 m

[2]
PROJ.4 sterea:
mean northing difference: 0,00065 m
mean easting difference:  0,00137 m

[3]
Oscar's implementation of 
http://www.atomnet.pl/~geodeta/2000/59text1.htm with his corrections to 
the algorithm:
mean northing difference: 0,00001 m
mean easting difference:  0,00000 m

[4]
Oscar's implementation of rouss:
mean northing difference: 0,00121 m
mean easting difference:  0,00189 m


Summarizing:
1. By now, best would be to use Oscar's [3].
2. Or at least replace stere with stere, as proven by [1] and [2].

What are the chances either solution would be applied to PROJ.4 and libproj?

The other thing is the empirical correction on top of the proper Uklad 
1965 mathematical formulae.

Gerald Evendeen wrote:
 > Is the "correction" polynomial (PASCAL code) a datum shift or a
 > correction to a projection???

I *guess* it is the correction to a projection. 
http://www.geonet.net.pl/gfx/pliki/kod_korekta65.txt says, roughly:
" 1. Tranformation between a mathematical grid 1965 and and it's real 
equivalent (empirical, catalogue) xyz65(teor) <=> xyz65(empir) are 
accomplished by algebraic polynomials of 12th degree in the zone 5 and 
of 16th degree in zones 1,2,3,4. Parameters of the polynomials were 
estimated based sets of several dozens of thousands of control points - 
groundwork of I + II class.."

(I hope my translation explains something)

Maciek

--------------------
W polskim Internecie s? setki milionów stron. My przekazujemy Tobie tylko najlepsze z nich!
http://katalog.epf.pl/