When running a radio-spectrometer side by side with a relatively high end PAR meter, we have found that PAR meters are consistently reading about 30% lower than the actual value in the 445nm range (that's royal blue, or 20,000-24,000K actinic blue folks!), the most important range for coral since that's where the primary alpha (α) photosynthetic peak lays. It also gets saturated in the 460nm (regular cool blue), 430nm (violet) and 520nm (green) wavelengths as well, but to a lesser degree of about 12% reduction. We found negligible saturation (<5%) in the amber (600-700nm) and red (640-680nm) ranges too. These tests were performed on the Philips Luxeon Rebel and Philips Luxeon ES series LEDs. We can expect other high-end LEDs to perform similarly, whereas the no-name chinese LEDs (Epistar, Bridgelux, etc) perform at about 50%, or half as well, or about half as efficient dependent on the individual LED.
