[SoC] Netcdf for uDig - status report #7

[Jody Garnett]

On 12/07/2009, at 5:56 PM, andrea antonello wrote:

> Regarding the CRS format, I think we will not get out of this
> successfull. The geotools and netcdf libraries really deal in a
> different way with the formats and not all the same crs are supported.
> So I really think that the lat/long choice is the best. I also is done
> like that in the ncWMS project, which uses netcdf for reading and
> geotools for rendering  the WMS map. In that case data are always read
> in whatever projection and first reprojected to 4326. Then things are
> projected back to whatever the user asked for. Not so nice, but it is
> the interface point between the two libs I guess. Will try to ask some
> expert about that.

That would be good; I am very confident in the ability of the GeoTools  
referencing model to represent anything (we have not implemented  
everything but whatever); so for me I really want to know what the  
netCDF library is representing in order to see if we can translate  
between the two.  Note: I suspect that it may have an internal grid  
mapping pixels to lat/lon - but we should be able to supply that grid  
to GeoTools and let it work with the raw data directly and not lose  
anything more than we have to...

Jody