[Qgis-user] On format and coordinate systems transformation

[Alex Mandel]

On 06/27/2012 06:45 AM, Mats Elfström wrote:
> Hi Group!
> 
> I will begin this post with a disclaimer. This is no advertisement, or
> endorsed statement. I intend it as friendly advice to fellow professionals,
> for the good of the industry and our reputation.
> 
> 
> Many posts on this list come from users who struggle to move geodata
> between software systems or projections and coordinate systems. Several
> times a week or even per day, more or less tricky conversion problems are
> aired. This is a good thing. It shows that more people are using more
> geodata, and wants to combine data from various sources. This is one
> important purpose of a GIS.
> 
> 
> However, format conversion and coordinate transformation can be tricky
> business. If performed improperly, or with inadequate tools, mistakes can
> result in loss of information, or distorted or misplaced features. Such
> errors may be apparent, but can also go unnoticed, and later reflect badly
> on the person or institution that supplied the data.
> 
> While I appreciate that all data handling tips on this list are given in
> good faith, I have seen those that are outright wrong, difficult or even
> dangerous, and not in line with what I judge to be Good Practice of our
> trade.
> 
> 
>  There is really only one professional and system independent tool for all
> kinds of spatial data conversion, and that is FME from Safe Software. I
> would recommend no other. Only FME can safely handle the multitude of
> coordinate systems and the ever-evolving types and versions of file
> formats. Add to that the ability to extract, validate and combine data from
> different sources into a single workflow. The Workbench, where you
> graphically build your workflow is one of the best GUI:s I know.  You can
> build very complicated workflows and save them for later use. For recurring
> conversion tasks, this ensures that each data transfer is performed exactly
> like before.  Workbench files can be built by FME and GIS pros, and
> distributed for the use of others who just needs a task done.
> 
> 
> Quality does not come cheap, and I realize that the cost of FME can seem
> prohibitive. However, if you take into account the time and effort that
> people must use to try to tweak this tool here, or find that script there,
> with perhaps unpredictable results; you may find that the money for an FME
> license is well spent. On the other hand, consider the risk that lost or
> distorted information leads to costly damage further on in the chain.
> 
> Whenever I attend a seminar or conference on GIS/GIT/mapping and surveying,
> people mention FME as an everyday tool:  “Oh, and then we of course have
> some FME scripts to process the data.”
> 
> The support from Safe Software is excellent. They depend on that their
> tools can be used, so problems and suggestions for improvement are dealt
> with promptly and kindly. There is always a fresh beta build available to
> try out the latest fixes or additions.
> 
> Visit www.safe.com to find out more, and note that any flavor of FME
> Desktop can be downloaded for a free 14 day trial.
> 
> 
> Best Regards, Mats.E


Thanks for the info I have heard great things about FME, but as a word
of caution I have also heard of errors the occur only in FME with
certain formats being unusable as the output (something related to
postgis if I'm remembering correctly).

So while I do think it's good for people to pay attention to this, do
not trust software just because you bought it (use that help you paid
for) or alternately because it's open source. In regards to
transformations, when it's critical data, find a way to verify your
results, possibly doing the same transformation using 2 different libs.

As I discovered helping someone the other day GDAL based apps have
issues with Oblique mercator projections (outstanding bug) and UDig does
the conversion correctly.

Thanks,
Alex