Many of our TCCON stations are located in time zones where a local day crosses over two UTC dates. We use one a priori profile per local day, so the date in the files represents the local day to allow you to choose the correct a priori profile. The fractional hour lets you know what time it is in UTC.
So, if the UT hour is -1, then the UTC date is the previous day, 11 pm.
This also allows a full day of data to be plotted versus UT hour without discontinuities.
Yes, please see the Data Description page for the details.
Please see the Auxiliary Data page for details.
In the HITRAN database, all line intensities are multiplied by the standard isotopic abundances (from De Bievre et al.[1984]). See the HITRAN Isotopic Abundances (now http://hitran.org/docs/iso-meta/) page for details - HDO is H_{2}O isotopologue 162. This means that retrievals always return the total VMR of the gas, irrespective of the isotopic identity of the spectral lines used. To find the true VMR of HDO, multiply the retrieved value by the standard isotopic abundance, which for HDO is defined by Vienna Standard Mean Ocean Water (SMOW = 3.10693x10^{-4}).
To compute δD, all you need is: 1000*((X_{HDO}/X_{H2O})-1)
The averaging kernel matrix represents the change in the retrieved vmr profile at level i due to a perturbation to the true vmr profile at level j. A perfect measurement would have an averaging kernel equal to the unit identity matrix. But for real measurements with limited vertical resolution, a pertubation at level j will change the retrieval not just at level j, but also at adjacent levels.
where is the column averaging kernel produced by GFIT, is the full averaging kernel (not produced by GFIT), and is the pressure at level i. Because there are 71 pressure levels (i.e. ), the column averging kernel also contains 71 elements.
What does mean, exactly?
It's the effective thickness (in pressure) of the j'th level.