Would it be possible to roughly estimate this by using green and yellow filters? If you were to get some photographic filters that block the green and yellow light that aren't all that useful, you might get a better idea of the PUR values of these LEDs.
That's an interesting idea. I honestly don't know if that would work or not, but it sounds reasonable. My thought is: if it was really that simple, wouldnt the manufacturers already have done this to their sensors?
Does anyone have a PAR meter and a few photographic filters? Maybe put a green filter over the PAR meter probe and see what they get with various light sources?