Malc, Robin, Tasso,
There seem to be two explanations - or two possible causes - as to why we see an increasingly blue-shifted set of absorption lines in this phase of nova evolution. One (the simpler to understand, at least by me!) is advocated by Bob Williams (see Williams and Mason, 2009, Nova ejecta as colliding shells,
http://arxiv.org/ftp/arxiv/papers/0908/0908.3810.pdf):
"[We see] multiple absorption systems having different radial velocities in the spectra of a large majority of novae and interpreted the observations in terms of discrete clouds of gas that are progressively accelerated outward. The innermost ejecta, having the higher velocities, correspond to the surface layers of the white dwarf (WD) that have undergone nuclear reactions. The outermost ejecta are circumbinary gas having lower velocities, and their presence at maximum light surrounding the primary WD ejecta strongly suggests that the outer circumbinary transient heavy element absorbing (THEA) gas has existed before the TNR." (Williams and Mason, 2009 p.1)
Basically he argues that previous outflow from both the WD and in particular the secondary star creates a circumbinary gas envelope moving outwards from the pair at relatively low velocities. This is successively energised (ionised, heated, and collided with) by the initial nova outburst hence is the first set of absorption (and emission) lines we see. Later, as the fireball become less optically thick, we see faster moving lines from the actual ejecta themselves. The attached diagram is from p.7 of that article.
One of the problems with this theory seems to be that the energy required to push this much material out into the circumbinary envelope is substantial, and we see little evidence of this sort of emission from the system prior to the nova outburst. But it seems true to say that the recent literature seems to be dominated by discussion of the evidence for and against some sort of steady-flow stellar wind (as opposed to a ballistic expansion) as a contributor to novae spectral evolution.
Steve Shore posits another explanation for Discrete (blue-shifted) Absorption Components based on his study of the T Pyx outburst in 2011 (Shore et al, 2011
http://arxiv.org/pdf/1108.3505.pdf). Shore argues that the velocity structure observed in the ejecta is "frozen" in the structure of the initial explosion - that is, it's all there from the start. However, as Robin suggested, different absorption components become visible as specific transitions, particularly in the UV, become less optically thick, while a recombination front moves outward through the ejecta. However, the point is, we're not "seeing further in", as much as, as things thin, more differentiated structure is able to *get out* - gradually picking up structure in the outer/faster envelope.
I won't begin to suggest that I understand all the physics in this latter case. But a few things it helped me to remember when thinking this through. One is that optical thickness varies by wavelength, so the velocity structure we observe down near the UV will not be the same as that up in the IR. The other is that we are seeing the cumulative effect of superimposed shells of absorption. So, let's assume for simplicity that the inner (slower - let's say -400 km/s) ejecta start off absorbing 100% of the photons between 4830 and 4861 - we'd therefore see no faster absorption line sat all, just a band at -400 km/s. As these ejecta thin, more photons will get through to be absorbed by the outer (faster) ejecta, thus we'd begin to gradually see more, thinner, and faster lines as the ejecta thin. And this is exactly what we see.
Finally, the recombination front. This is particularly relevant to hydrogen, but that's what we're looking at. Obviously, ionised hydrogen can't create absorption lines. So as the recombination front (which I'm assuming is an expanding circumference in the ejecta which is the point at which energy levels have dropped enough to allow significant recombination of the hydrogen plamsa to occur) moves outward from the WD, it begins to reach parts of the ejecta that are further out (and thus faster-moving). As the hydrogen in successively further and faster regions recombines, it will create emission lines; but it will also therefore provide a supply of neutral hydrogen to create absorption lines.
I must say I found this notion of the "recombination front" very helpful in understanding why the absorption lines velocities appear "inside out" and get faster with time.h
Cheers
Jonathan