[MetaCRS] Question about CS-Map coordinate transformation accuracy

[Martin Desruisseaux]

Hello Stefani

Accuracy depends if the coordinate operation involves datum shift or
not. The ISO 19111 international standard (/Spatial Referencing by
Coordinates/) defines two terms - /transformation/ or /conversion/ -
depending on the above.

If there is a datum shift (/coordinate _transformation_/), the main
cause of inaccuracy does not depend on the software (assuming there is
no major bug). The accuracy rather depends on the way the datum shift
has been defined in the real world. This accuracy is recorded in the
EPSG database on a case-by-case basis. The accuracy reported by EPSG is
often 1 meter, sometime less, sometime more. If an application needs
this information, then the application may need a library connected to
the EPSG database or something equivalent. Alternatively, if the
application only need a limited set of coordinate operations, you may
get this information manually from http://epsg-registry.org/

If there is no datum shift (/coordinate //_conversion_/), then the
accuracy is software-dependent. It usually depends on rounding errors
and approximations used in the projection algorithms. To my knowledge,
most major libraries have a good precision (1 centimeter or better) if
the coordinates are inside the projection domain of validity. Those
domains of validity can also be found in the EPSG database.

To test the accuracy of a software, the EPSG authors created a tests
suite: the /Geospatial Integrity of Geoscience Software/ (GIGS) tests
(http://info.ogp.org.uk/geomatics/gigs.html). Those tests are defined as
Microsoft Excel files, but some of them have been implemented in the
Java language in the GeoAPI conformance module
(http://www.geoapi.org/geoapi-conformance/index.html). It does not means
that the library must be written in Java - we can write wrappers using
JNI. We have done exactly that for the Proj.4 library, which allowed us
to run some GIGS tests on Proj.4. If they were some volunteers for
writing JNI/GeoAPI wrappers around CS-Map, it would be possible to run
some GIGS tests on CS-Map.

    Martin




Le 14/08/14 10:43, Stefani Paolo a écrit :
> Hello,
>
> I am in charge of the design of a software tool that will collect survey data and integrate it in a database and am thinking of using CS-Map to handle the transformation from local to global coordinate systems and I have a question that I have not been able to answer looking at the CS-Map documentation that I have found on the internet.
>
> The question is the following: I would like to have some information regarding the numerical accuracy of the CS-Map transformation engine when used to convert between different coordinate systems. I do understand that CS-Map is an open source and commonly used transformation engine but, due to software safety requirements, I do need to have evidence about the accuracy figures that CS-Map provides, in my case when transforming from a local to a global coordinate system.
>
> Thanks in advance and best regards,
> Paolo Stefani

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/metacrs/attachments/20140814/267eeccc/attachment.html>