On 7/26/2011 3:27 PM, Jim Biard wrote:
John,
I think part of what people are saying is that we don't always have a
situation where wavenumber and/or wavelength is unique for every
band. This is what has been leading us to the parametric driver
coordinate. A further example of this is polarization, which doesn't
fit in a nice numeric construct like I was describing in my little
example. (I think polarization would likely work well using flags,
but that's yet another twist that we don't need to tackle just now.)
When Christopher Lynnes is referring to different band numbers, he
means index numbers, not wavelengths or wavenumbers. It seems awkward
to separate out one band into a different variable from the others
because it is not differentiated by wavelength (for example). The
problem is, there is a tremendous variety of differentiators for the
bands. Some of them clearly should cause data to be separated into
different variables (different resolutions, for example), but others
"feel" like they shouldn't provoke variable splitting. The feeling
could be wrong, (I feel a song coming on here... "if loving you is
wrong, then I don't wanna be right!") and we may just need to learn
how to think about this differently.
I think what this boils down to is that the "identification space" for
the bands can often be multi-dimensional, but the filling of that
space is sparse. We are looking for a solution that is somewhat akin
to the CF compression by gathering convention.
Jim
hi jim:
Im thinking the way forward for now might be to follow Chris' idea of
specific examples. Then we can see what the real issues are.
So Id invite anyone to submit examples, and we can work out the CDL.
john
On 7/26/2011 4:49 PM, John Caron wrote:
On 7/26/2011 2:34 PM, Lynnes, Christopher S. (GSFC-6102) wrote:
On Jul 26, 2011, at 3:05 PM, John Caron wrote:
On 7/26/2011 12:26 PM, Tom Rink wrote:
Upendra,
On 7/26/11 1:07 PM, Upendra Dadi wrote:
Jim,
Could you please clarify how to represent data which contain
bands
with multiple spatial resolutions using you scheme? I am thinking of
MODIS data:
http://en.wikipedia.org/wiki/Moderate-Resolution_Imaging_Spectroradiometer#MODIS_Bands
As you can see, not all the bands have same spatial resolution (or
spatial dimensions), even though all of them have same units. Could
we even store all the bands in the same variable?
I think multiple resolutions in same variable would be difficult and
impracticable,
the CF conventions for defining Projections, analytic or lat/lon,
don't work like
this. You'd probably have to define another dimension to index the
different
resolutions for the data, lon and lat variables. I would think this
would be
pretty messy.
Tom
Correct, you need seperate variables for different resolutions.
However,
theres no problem with having multiple coordinate systems in the same
file. So if you chose, you would create multiple groups of variables,
each group with their own coordinates.
The HDF files for MODIS currently do separate the resolutions into
different variables. However, MODIS presents some other issues,
such as having both low-gain and high-gain versions of the same band
in the same 3D variable. Currently, they handle this inside the
file by calling one band "13" and another "13.5", but this is a bit
opaque. I can never remember which is low and which is high.
Im not sure what the difference between the high and low gain, but it
might be a good candidate to separate into 2 variables.
There are also two bands (21 and 22) with different band "numbers"
but the same bandwidth (different Spectral Reflectance).
if you accept that a data variable must have either a wavelength or
wavenumber coordinate, then we need just those coordinate values (or
bounds) to be unique, so they can be named uniquely by coordinate
value. Other auxilary coordinates like bandwidth dont need to be unique.
I have been mentally struggling to understand how these would fit in
the various "band" dimension/coordinate proposals.
Would it be useful to begin a set of examples for actual instruments
so that the less CF-savvy amongst us could see the implications?
yes, very helpful.
_______________________________________________
cf-satellite mailing list
cf-satellite@xxxxxxxxxxxxxxxx
For list information or to unsubscribe, visit:
http://www.unidata.ucar.edu/mailing_lists/
