[CF-metadata] New Standard Names for Satellite Data

From: Jonathan Gregory <j.m.gregory>
Date: Mon, 6 May 2013 21:51:26 +0100

Dear Aleksandar

> > Why is senor_zenith_angle distinct from the existing platform_zenith_angle? Can
> > you see the sensor separately from the platform (satellite)?
>
> The two names allow better description of the reference used to
> calculate those zenith angles. For some viewing geometries the sensor
> and the platform cannot be assumed close enough to neglect that
> difference.

OK.

> > When I look up "look angle" in Google, its main meaning appears to be the
> > elevation angle for seeing a satellite, rather than the angle of observation
> > from a satellite. I wonder if there is alternative phrase that could be used
> > for this.
>
> It is different in remote sensing, I guess it depends on the vantage
> point. :-) "view" can be an alternative to "look" but don't think it
> really changes much:
>
> platform_view_angle
> sensor_view_angle
>
> My initial proposal used "scan" but a member of the cf-satellite
> community requested a change since some sensors do not scan.

I think that "view" would be better than "look". In fact "view" was the word
which occurred to me as well as a possible alternative.

But it's not obvious that "view angle" (or "look angle") of zero means straight
down. Indeed I suppose someone might want to use these standard names for a
sensor of a different kind that was mounted on the ground. Since you say

> I deliberately did not want to use the term nadir to avoid the debate
> over whether it is geocentric or geodetic nadir. The definition is
> purposefully vague in that regard.

can you think of any other word(s) you could insert to qualify the convention
for this angle i.e. that it's wrt the downward view.

> > What is the use case for the relative angles? It might be helpful to be more specific if possible.
>
> To describe the difference in viewing geometries from two different
> sensors over the same target. I think the definitions are specific
> enough.

The definitions are specific, yes. I was asking the question because I wondered
if these were concepts that needed to be compared between datasets, which is
the main reason for standard names. The definitions could be more specific
by indicating the sign convention, or stating the angle is not signed.
Standard names for quantities where the sign could be open to doubt always
specific it as part of the name.

> > However, I do have a comment on toa_brightness_temperature_bias_at_standard_scene_wrt_intercalibration. I wonder if this could be simpler e.g.
> > bias_in_toa_brightness_temperature_of_standard_scene. That also has the advantage of being the same as the previous one with a bit added.
>
> The reason for that bias is intercalibration so it should probably be
> in the name. Did you mean:
>
> bias_in_toa_brightness_temperature_of_standard_scene_wrt_intercalibration ?

Yes, I meant with bias_in at the start, rather than being in the middle of
toa_brightness_temperature_of_standard_scene. But I also wondered whether
wrt_intercalibration could be omitted. If it's a bias, mustn't it have been
determined by calibration, so that could be taken as read?

Cheers

Jonathan
Received on Mon May 06 2013 - 14:51:26 BST