I've done a bit of initial feasability testing for this project. Last week I used the four 50mcd UV LEDs and two 630nm detection LEDs. The signal generated from the op-amp was around 0.15V. Background noise was measured at about 0.05V.
I've since received the 2000mcd UV LEDs, and wow, are they bright! They also don't blind the detector LEDs during direct exposure. The less powerful 50mcd LEDs would.
A picture of one of the longer distance test setups: