'Quantifiers' are essentially tied to a particular unit, for expressing the idea that a unit is related to one part of a system. This is not the way these things are 'officially' supposed to be dealt with, but are common enough to be useful. On the other hand, there is a limit to the amount that can be covered by an automated system. In particular, raising a 'qualified' unit to a power is normally taken as applying the power to the entire thing, for example
\documentclass{article}
\usepackage{siunitx}
\DeclareSIQualifier{\cat}{cat}
\begin{document}
\SI{10}{\mmol\per\kg\cat}
is the same as
$10$\,mmol\,kg$_{\textrm{cat}^{-1}}
\end{document}
Dealing with a case where you want to 'qualify' a unit raised to a power is probably beyond the boundaries of what can be squeezed into the standard interface. I would therefore set up a special 'unit' for this
\documentclass{article}
\usepackage{siunitx}
\DeclareSIUnit{\mcubedair}{m^{3}\text{ of air}}
\sisetup{per-mode = symbol}
\begin{document}
\SI{10}{\kg\per\mcubedair}
\end{document}
of air, it seems to me that the correct thing is them of airthat is cubed, hence the10 kg of water/(m of air)^3make sense, which is whatsiunitxis producing. – Peter Grill Jun 11 '12 at 4:45qualifier-mode = spacesetting, or simply use\SI{10}{\kg\water\per\m\cubed} of air. Perhaps someone more knowledgeable can help you further. – Peter Grill Jun 11 '12 at 5:16of airafter the unit. This isn't SI notation, or commonly used notation. I've been trying to find a journal article where "<something> / m3 of <something>" is used, but found only one from 1983. The best thing might be to look at publications from your field and see how they do it. I would probably go with "The water content of the air is 10kg/m3", which is correct SI notation. – Jake Jun 11 '12 at 6:06