[mapserver-dev] Call for vote on RFC 78

[Stephen Woodbridge]

Brent,

I think you can create a gdal VRT thingy that allows GDAL to present the 
two files as a single dual band image. This would keep the code simpler 
in mapserver.

Maybe Daniel or Frank can verify this assumption.

Thanks,
   -Steve W

On 11/17/2011 10:21 AM, Brent Fraser wrote:
> Daniel,
>
> Makes sense. As Steve W. points out, it could be done on angles by
> averaging the components so you may want to at least keep it in mind for
> a future enhancement when doing the implementation.
>
> Looking at the FTP site I listed, I see that the U/V components are
> available for wind, but in separate files: nww3.UGRD.grb, nww3.VGRD.grb
> (with 61 bands per file, one band per 3 hr forecast) . Would it be
> possible to adjust the DATA object to allow for such a situation?
> Something like:
>
> DATA nww3.UGRD.grb:1, nww3.VGRD.grb:1
>
> with the ":1" indicating the band to use.
>
> Best Regards,
> Brent Fraser
>
>
> On 11/16/2011 10:43 PM, Daniel Morissette wrote:
>> On 11-11-16 11:58 PM, Brent Fraser wrote:
>>> Alan,
>>>
>>> Any plans to allow for raster files containing direction values
>>> (http://polar.ncep.noaa.gov/waves_old/products.html#grib, esp Mean Wave
>>> Dir) instead of calculating it from components? Note: sample data
>>> available at ftp://polar.ncep.noaa.gov/pub/waves, e.g. nww3.WVDIR.grb
>>> (wave direction)
>>>
>>
>> Brent,
>>
>> The approach of using u and v bands allows for resampling, which is
>> essential to the approach of rendering vector fields at various
>> scales, and I don't think is possible with a raster of direction
>> (angle) values.
>>
>> i.e. as you zoom in and out, with u and v raster bands, you can
>> average the u and v pixel values independently over an area and the
>> average u and average v give you an accurate average vector for the
>> area. However, I could be wrong, but I don't think you can resample
>> and average the direction (angle) values over a block of pixels to get
>> the same result.
>>
>> It's late here, so maybe you can prove me wrong.
>>
>
> _______________________________________________
> mapserver-dev mailing list
> mapserver-dev at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/mapserver-dev