I was very intrigued to read in earlier posts about the possible "shielding" benefit of CYA. The suggestion with regard to this benefit has been that its significance increases with pool depth. My thought is actually the opposite, since I assume the affect of UV rays on HOCl would be less at greater depths anyway, so the shielding benefit would be more significant for HOCl that is closer to the surface. (Is this assumption incorrect? It is certainly not based on physics beyond the fact that refraction is obviously involved.)
Speaking of refraction, I'm conducting an experiment today and tomorrow measuring FC loss with returns directed upward to create maximum disturbance in the water surface (maximum refraction) compared to FC loss with the returns turned downward to minimize surface disturbance. My hypothesis is obviously that the disturbance in the water surface could inhibit UV penetration due to increased (or perhaps more "disordered) refraction.
I've been running the pump basically all day (with returns directed upward) based on this yet to be validated hypothesis, but reading this thread has motivated me to experiment. Fortunately, today is mostly clear skies and tomorrow is supposed to be the same. I'm measuring FC loss from 10:30 am to 4:30 pm both days. In retrospect, since clouds are now in the sky (1:00 pm), I wish I had made today the "low disturbance" day. Tomorrow is scheduled to be clear all day and if I lost more FC on the low disturbance day even with a few clouds in the sky, that would offer more conclusive evidence in support of my hypothesis. I might just measure now since the vast majority of time between 10:30 and now has been clear.
Based on my retrospective wish in the previous paragraph, I decided alter the experiment. I just took a measurement and pointed the returns down (minimal surface disturbance). The rest of the day looks like it is going to be partly cloudy (lots of fluffy cumulus), so I'll compare FC loss this afternoon (calm pool surface) to tomorrow afternoon (disturbed pool surface) which is scheduled to be clear. I'll also reverse the 10:30-1:15 return orientation (pointed down tomorrow since they were pointed up today), which will give me two sets of comparative data. It is obviously difficult to minimize lurking variables in an experiment like this!