Computation of Earth scene radiance

[helgaweb]

I was wondering, why the Earth scene radiance (Ne) does not account for linear radiance estimate Nlin as described in the KLM User's Guide. Following 7.1.2.4-6 on page 7-9, Ne is calculated as: Ne = Nlin + Ncor

pygac/pygac/calibration.py

Lines 553 to 558 in 2d7fbaf

	Nlin = (cal.n_s[chan] +
	(((nBB - cal.n_s[chan])
	* (new_space - counts.astype(float)))
	/ (new_space - new_ict)))
	Ncor = cal.b0[chan] + Nlin * (cal.b1[chan] + cal.b2[chan] * Nlin)
	Ne = Ncor

[carloshorn]

Hi @helgaweb, indeed, to me it also looks like Nlin is missing. Currently, I am working on a refactoring of calibration.py, where I want to put more comments on the equations. Hope to push the corresponding commit to #58 today.

[helgaweb]

@carloshorn that will be very helpful, thanks! I think the documentation in the KLM User's guide is quite well done and provides a good basis.
Alright, let's wait for @abhaydd and @mraspaud ideas on this and then I could open a PR (after you finished the refactoring).

[carloshorn]

I checked the code, and it is doing the right thing, but written in a confusing way...
The coefficients are defined here:

pygac/pygac/calibration.py

Lines 304 to 306 in 2d7fbaf

	'b0': np.array([0.0, 4.76, 3.83]),
	'b1': np.array([1 - 0.0, 1 - 0.0932, 1 - 0.0659]),
	'b2': np.array([0.0, 0.0004524, 0.0002811]),

Which shows, that the linear correction 'b1' term is added by one, which therefore accounts for the linear part. However, I agree that there should be a huge alert sign in the code.

[abhaydd]

@helgaweb @carloshorn Yes, the Nlin is of course included properly (but implemented in an unclear way). Otherwise we would have been off by hundreds of degrees. The non-linearity correction is typically only a fraction and translates to less than few Kelvins.

[helgaweb]

@abhaydd Thanks a lot. Sorry I've overseen the change in the b1 coefficient (as clarified by @carloshorn) and was therefore confused about the calculation of Ne.

[carloshorn]

@helgaweb, I pushed the comment lines to #58. I would appreciate if you could have a look at it and leave a feedback telling if the comments are sufficient to understand what the functions do.

[abhaydd]

@helgaweb @carloshorn I can understand why one can get confused. The code clearly needs revamping. And it’s good to have the fresh minds like yours and Carlos’s :)

[helgaweb]

@carloshorn Sure. I hope to give you some feedback by monday!

[mraspaud]

I'm reopening this as I think it's a problem that it isn't documented clearly that the coeffs are taking acount of this. Hopefully #58 closes this.