Something really unusual in GSLV...at least, I have never seen !!!
http://www.ndtv.com/video/player/news/india-s-200-crore-space-mission-delayed-after-leak-in-rocket/287475
Look closely at 21 seconds to 24 seconds in the video...You see a liquid gushing out !!!
Not sure if this has anything to do with the leak
I believe you are seeing water sprayed onto the pad from the tower, possibly for acoustic, fire or thermal control to minimise damage.
Look at the top right. The jet's seen at a time of L-1:14:20 (and the clock isn't counting) - so this was
after they stopped the countdown. (At 00:32 in the news report, there's a shot from earlier in the countdown, only showing what I assume is condensation around some cool plumbing in the tower).
<speculation follows. Begs pardons

>
I don't know what the detox process is for hypergols, but given that the jet was aimed at the bottom of the second stage (between the bottom of the engine shroud and the interstage), I'm assuming that the epicentre of the leak was somewhere there, and the water was used to possibly dissolve the toxic UH25 and wash it down - rather than it volatilise, and have vapours hang around the pad.
Also, going through this thread, I found that the second stage propellant loading was completed (latest) by
0919 hrs IST on launch day. (L-07:31:00). From the ISRO Chairman's statement, we know they detected the leak at L-2:00:00. Interestingly, cryogenic propellant loading (LOX first) commenced (no later than)
1352 hours (L-2:58:00).
So that's a gap of ~5.5 hours between completion of 2nd stage prop loading to first leak detection; or, a gap of 58 minutes between cryo loading and the same[1]. (Another 46 min before they stopped countdown) - presumably to rule out instrumentation fault, and to confirm a leak via appreciable change in tank pressures/levels (Media reports of "smoke" were post-hoc).
[1] After reading
this in the brochure:
Modified design of the Fuel Booster Turbo Pump (FBTP), taking care of the expansion and contraction of the bearings and casing at cryogenic temperatures
- I was tempted to blame similar 'unexpected' contraction for the 2nd stage counterparts, for the leak, but the cold stuff is whole stage apart. So, I'm leaning toward some contamination/corrosion/manufacturing defect as a cause.