[CF-metadata] More canonical units or more standard names?

From: Aleksandar Jelenak - NOAA Affiliate <aleksandar.jelenak>
Date: Wed, 17 Apr 2013 10:46:03 -0400

Dear Ted,

On Tue, Apr 16, 2013 at 7:11 PM, Ted Kennelly <ekennell at aer.com> wrote:
> it seems that some good progress has been made to
> tackle the thorny issue of how to represent what is the fundamental product
> of remote sensing observations, i.e, the
> measurement of top of atmosphere radiance incident at the sensor.

To be precise, what is reported as measured is spectral radiance at
the central wavelength, wavenumber, or frequency of the sensor's band.
Radiance and spectral radiance are two different physical quantities
of which only the latter is a function of either wavelength,
frequency, or wavenumber.

> We have struggled with the same issue concerning
> radiance - that is, does the difference between radiance measured as a
> function of wavelength or frequency represent
> physically distinct quantities representing unique standard_names or are
> these fundamentally equivalent quantities
> represented by a simple units conversion and therefore can be represented by
> a common standard_name.
> The remote sensing community is used to dealing with the conversion between these
> quantities as different programs have chosen
> to adopt different units. For my part I am used to radiances as function of
> wavelength for the solar reflectance bands and
> as a function of wavenumber for the emissive bands. For those reading this
> who may no be aware, the conversion between
> these quantities is a function of the wavelength-squared (or
> wavenumber-squared). What is new over the
> last few weeks is the general acceptance that wavelength, or wavenumber (or
> some other physical quantity) can be adopted
> as a coordinate variable. Therefore the radiance conversion in question is a
> function of the coordinate variable itself. I think
> was is needed to close this issue is to determine if this type of
> functionality constitutes a unique quantity and therefore a
> unique standard name. I think we are pretty close to achieving this.

The units of spectral radiance for the three domains -- wavelength,
frequency, or wavenumber -- are not physically equivalent. You cannot
write for example:

X mW m-2 sr-1 um-1 = Y mW m-2 sr-1 (cm-1)-1

but within one domain, wavelength in the following example, you can write:

X mW m-2 sr-1 um-1 = Y uW cm-2 sr-1 nm-1

The conversion of spectral radiance between these three domains is not
a simple units conversion but, as you pointed out, involves the
wavelength/wavenumber/frequency at which the spectral radiance is
reported. The convention allows the use of units that are only
physically equivalent and hence that is why I have introduced more
than one standard name for spectral radiance.

> In terms of the standard_names options you have proposed, I personally
> prefer:
>
> toa_outgoing_spectral_radiance_per_unit_wavelength
> units: mW m-2 sr-1 um-1
>
> toa_outgoing_spectral_radiance_per_unit_wavenumber
> units: mW m-2 sr-1 (cm-1)-1
>
> toa_outgoing_spectral_radiance_per_unit_frequency
> units: W m?2 sr?1 Hz?1

I made a mistake in the above names: spectral radiance is radiance per
unit of either wavelength, wavenumber, or frequency. The correct names
are:

toa_outgoing_radiance_per_unit_wavelength

toa_outgoing_radiance_per_unit_wavenumber

toa_outgoing_radiance_per_unit_frequency

> I hope we can reach consensus on the debate of 1 versus 3 standard names for
> the different flavors of toa_spectral_radiance.

I think the consensus have been reached that there will have to be
more than one standard name.

       -Aleksandar
Received on Wed Apr 17 2013 - 08:46:03 BST