It is decimals of a minute. So e.g. 10:47.7 is 10:47 and 7/10 of a minute, which is 42 seconds.
Some software packages used in astronomy produce this output.
I observe Jupiter a lot and use the WinJupos package by Grischa Hahn, and that gives everything in decimals of a minute. So also every piece of software that feeds into this (e.g. the image capture software) has to use that time format as well. In this case I think the reason is just historical, that visual observations could not be accurate to better than this, and in early days of programming it was easier to do it this way, and it has carried on. I don’t really like it.
Occultations of course can be timed to at least a tenth of a second, but I suppose the prediction does not need to be so accurate, so I suspect this is really due to both trying to save space (in print) and software that is a bit long in the tooth.
But someone who works with occultation predictions might care to comment.