[PROJ] Question about transformation using cs2cs

[Even Rouault]

Le 29/09/2021 à 17:03, José Luis García Pallero a écrit :
> El mié, 29 sept 2021 a las 16:42, Even Rouault
> (<even.rouault at spatialys.com>) escribió:
>> Hi,
>>
>> you're actually hitting this very old change in behavior in PROJ that
>> still applies with recent versions:
>> https://proj.org/usage/differences.html#version-4-6-0
>>
>> Just specifying a +ellps= doesn't define a datum. There's no way we can
>> relate the 2 datums that are partially defined by just their ellipsoids.
>> So when considering the geographic coordinates, they are just kept
>> unchanged; Which explains the behaviour of the second cs2cs invokation.
>>
>> Why do you get different results in the UTM projected coordinates ?
>> Because projection computation takes into account the ellipsoid shape.
>> But you won't get a datum shift by itself when doing so.
> Thank you for yout answer
>
> But I understand that in this case the operations are:
>
> 1. Inverse projection step from UTM ellps=intl to geographic
> coordinates in ellps=intl
> 2. Geographic coordinates in ellps=intl to cartesian geocentric
> coordinates referred to ellps=intl
> 3. Helmert transformation from cartesian geocentric referred to
> ellps=intl to cartesian geocentric referred to ellps=WGS84 using the
> parameters defined in PROJ database
> 4. Cartesian geocentric referred to ellps=WGS84 to geographic
> coordinated in ellps=WGS84
> 5. Projection from geographic coordinated in ellps=WGS84 to UTM ellps=WGS84
>
> Am I right? Then, why do not perform the steps 2, 3, 4 in the case
> just specifying +ellps=?
No, they aren't done. You would need to add +towgs84=0,0,0 to do 2,3,4 
but that's unlikely to be what is appropriate.
>
> About the datum definitions for the +datum option, I can see in
> https://proj.org/operations/conversions/latlon.html that there is the
> option -ld in order to see all defined datums. But in my PROJ 7.2.1
> there is no such option. Is the -ld option only present in PROJ 8?

Yes, it was removed per 
https://github.com/OSGeo/PROJ/commit/a955947c1f3279e51d1272df0cb6aa61ce4b85ee

"Remove -ld option from proj and cs2cs

It promotes use of deprecated paramters +datum and +towgs84 which we
don't want to encourage."

PROJ.4 style strings to express CRSs, and in particular datums, are rather limited and we want people to use codes, WKT or PROJJSON instead.

Even

>
> Thanks
>
>> So you'll probably have to specify a +towgs84 or +nadgrids parameters to
>> your source CRS to define how it relates with the WGS 84 datum, and
>> replace on the target CRS +ellps=WGS84 by +datum=WGS84
>>
>> The PROJ debug traces are quite verbose and hard to understand if you're
>> not a developer, so they won't help much.
>>
>> If you know the EPSG codes of the CRS, you'd better using them directly
>> to get more significant transformations.
>>
>> Even
>>
>> Le 29/09/2021 à 16:18, José Luis García Pallero a écrit :
>>> Hello:
>>>
>>> I'm using PROJ 7.2.1 from Debian repositories and I have some
>>> questions about cs2cs. If I run
>>>
>>> echo 1445748.66 5053500.03 | cs2cs +proj=utm +zone=10 +ellps=intl +to
>>> +proj=utm +zone=10 +ellps=WGS84
>>>
>>> I obtain
>>>
>>> 1445704.69      5053404.02 0.00
>>>
>>> so the transformation of UTM coordinates between a point on the
>>> international ellipsoid to WGS84 was performed.
>>>
>>> But if I try to do the same with geographical coordinates as
>>>
>>> echo 111W 45N | cs2cs +proj=latlong +ellps=intl +to +proj=latlong +ellps=WGS84
>>>
>>> I obtain
>>>
>>> 111dW   45dN 0.000
>>>
>>> so no transformation was performed. Why? Setting PROJ_DEBUG I obtain
>>> for this case
>>>
>>> pj_open_lib(proj.db): call fopen(/usr/share/proj/proj.db) - succeeded
>>> pj_open_lib(proj.ini): call fopen(/usr/share/proj/proj.ini) - succeeded
>>> pj_ellipsoid - final: a=6378137.000 f=1/298.257, errno=0
>>> pj_ellipsoid - final:    ellps=GRS80
>>>
>>> Here I understand that only the second declaration of +ellps was taken
>>> into account (althoug EGS84 and GRS80 ellipsoids are not exactly the
>>> same)
>>>
>>> Thanks
>> --
>> http://www.spatialys.com
>> My software is free, but my time generally not.
>>
>
-- 
http://www.spatialys.com
My software is free, but my time generally not.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/proj/attachments/20210929/9a53cde5/attachment.html>