Need some advice

I have implimented Vincenty's Direct and Inverse algorithms with both VB6
and Delphi7. I have been distributing the code as open source.

A user has pointed out a problem in the Inverse function which I had not
noticed. The problem exists in both the VB6 and Delphi versions. I have
been trying for several days to fix it. The essence of the problem is:

given that point 1 is at a fixed latitude/longitude (I have been using
89N/0E) and compute a the length of a geodesic at say 20N/0E the length is
1,572,913.307 meters and when doing the same for 20S/0E (trans-equatorial)
the length is the same. Obviously this is not correct. I believe the length
of 89N/0 to 20N/0 is correct. I am using WGS84.

My algorithm is taken directly from Vincenty's paper as published in "Survey
Review" XXII, 176, April 1975 entitled "Direct and Inverse Solutions of
Geodesics on the Ellipsoid with Application of Nested Equations"

I would like to correspond with anyone that might shed some light on the
problem. I am confident that the equations faithfully follow those in the
paper, because they yield good results for known test cases of data that do
not violate the conditions set forth above.

Any enlightenment will be appreciated.

TIA

Vic
--
__________________________________________________ ______

Victor Fraenckel - The Windman vfraenc1@nycap.rr.com
KC2GUI www.windsway.com

Home of the WindReader Electronic Theodolite
Read the WIND

"Victory at all costs, victory in spite of all terror, victory however long
and hard the road may be; for without victory there is no survival."
- Winston [Leonard Spencer] Churchill (1874 - 1965)

Dost thou not know, my son, with how little wisdom the world is governed?
-Count Oxenstierna (ca 1620) to the young King Gustavus Adolphus

I didn't quite follow what the problem was. Does my calculator suffer
the same problem? It is also based on Vicenty.

http://williams.best.vwh.net/gccalc.htm


"Vic Fraenckel" <vfraenc1@nycap.rr.com> wrote in message news:<Wjj1b.1771$lk1.1472@twister.nyroc.rr.com>...
> I have implimented Vincenty's Direct and Inverse algorithms with both VB6
> and Delphi7. I have been distributing the code as open source.
>
> A user has pointed out a problem in the Inverse function which I had not
> noticed. The problem exists in both the VB6 and Delphi versions. I have
> been trying for several days to fix it. The essence of the problem is:
>
> given that point 1 is at a fixed latitude/longitude (I have been using
> 89N/0E) and compute a the length of a geodesic at say 20N/0E the length is
> 1,572,913.307 meters and when doing the same for 20S/0E (trans-equatorial)
> the length is the same. Obviously this is not correct. I believe the length
> of 89N/0 to 20N/0 is correct. I am using WGS84.
>
> My algorithm is taken directly from Vincenty's paper as published in "Survey
> Review" XXII, 176, April 1975 entitled "Direct and Inverse Solutions of
> Geodesics on the Ellipsoid with Application of Nested Equations"
>
> I would like to correspond with anyone that might shed some light on the
> problem. I am confident that the equations faithfully follow those in the
> paper, because they yield good results for known test cases of data that do
> not violate the conditions set forth above.
>
> Any enlightenment will be appreciated.
>
> TIA
>
> Vic
> --
> __________________________________________________ ______
>
> Victor Fraenckel - The Windman vfraenc1@nycap.rr.com
> KC2GUI www.windsway.com
>
> Home of the WindReader Electronic Theodolite
> Read the WIND
>
> "Victory at all costs, victory in spite of all terror, victory however long
> and hard the road may be; for without victory there is no survival."
> - Winston [Leonard Spencer] Churchill (1874 - 1965)
>
> Dost thou not know, my son, with how little wisdom the world is governed?
> -Count Oxenstierna (ca 1620) to the young King Gustavus Adolphus

"Vic Fraenckel" <vfraenc1@nycap.rr.com> wrote in message news:<Wjj1b.1771$lk1.1472@twister.nyroc.rr.com>...
> I have implimented Vincenty's Direct and Inverse algorithms with both VB6
> and Delphi7. I have been distributing the code as open source.
>
> A user has pointed out a problem in the Inverse function which I had not
> noticed. The problem exists in both the VB6 and Delphi versions. I have
> been trying for several days to fix it. The essence of the problem is:
>
> given that point 1 is at a fixed latitude/longitude (I have been using
> 89N/0E) and compute a the length of a geodesic at say 20N/0E the length is
> 1,572,913.307 meters and when doing the same for 20S/0E (trans-equatorial)
> the length is the same. Obviously this is not correct. I believe the length
> of 89N/0 to 20N/0 is correct. I am using WGS84.
>


89N to 20N along a meridian is 69 degrees. 90 degrees along a
meridian is approximately 10,000km (the original definition of a
meter). So 69 degrees should be about (69/90)*10000=7667km. On the
WGS84 spheroid it will be slightly different - I get 7677.9056km using
Vincenty.
So where does 1572.913307km come from - or are we doing a different
problem?

Ed

Ed Williams wrote:

> 89N to 20N along a meridian is 69 degrees. 90 degrees along a
> meridian is approximately 10,000km (the original definition of a
> meter). So 69 degrees should be about (69/90)*10000=7667km. On the
> WGS84 spheroid it will be slightly different - I get 7677.9056km using
> Vincenty.

I got 89 0 Sodano 20 0 = 7677905.396. Close but ... In general is
there an advantage to Vincenty over Sodano?

Ted

BTW, if you are not familiar with the antipodal problem - try the
following (WGS84)

Start at N0, E0 and proceed 19987.13950km with an initial course of
45 degrees (direct problem). You should end up at N0, E179 degrees
34.406 min.
Now try the inverse problem between these points. Vicenty won't
converge. Sodano, I believe, just gets the wrong answer. Bowring gets
it right.

Ed