The Plot is tropical, so it's ~equitorial. The dips are due to absorption.
Obviously, the dominant dip at 666 wavenumber is the CO2 15mm line.
The spectrum given in your post is unlabeled. Where did it come from?
The grapic came from a NASA website article:
http://earthobservatory.nasa.gov/Study/Iris/
There are plenty of others all around the internet displaying essentially the same information, this one just overlays CO2 & H20 responses in a way one can see the overlap and saturation of the spectal lines clearly in a single graphic.
"If it weren't saturated it would be a vary narrow dip in the graphic not showing skirt response."
The p, r and q branches are obscured.
Yep, mainly by overlapped water vapor response across the band. The branches would be part of the roughness along the left & right slopes on the primary 15mm line response. The fact the branches are not saturated is what allows the increase in absorption to even be measurable as concentration increases and part of the logaritmic response characteristic rather than the more limiting form such as an exponential like a*(1-e-kC).
It looks like the 5.xxln(C/C,sub>o) is a fitted eq. above a baseline of 278ppmv in 1750. I'll have to look into that.
The analysis comes from Myhre et al. 1998, Geophys.Res.Lett., 25:2715-2718.
From my own integrations of the same spectral data for CO2 the fit is quite good from 100ppmv out beyond 2000ppmv. The numbers I am getting indicate a solid fit with an standard deviation of 0.01 w/m2 throughout that range using a 100meter cell length.
I'm currently building a full integration for an 11km cell length version at mean atmospheric temperature and pressure and CO2 distribution, to see how well the relationship holds for the longer cell length representing a column of the atmosphere and plan on doing a similar analysis for water vapor as well. Figure I just as well take advantage of my access to the software & HITRAN database while I have it.
At present I am working only on the wavenumber region from 555->833 cm-1 (12-18mm wavelength). Given time I'll be expanding the integrations out to cover the wavenumber range from 1 -> 2900 cm-1 corresponding to (3.5-10,000mm).