Hey All!
Noob alert! (duh). I'm a PhD student, studying biology (with a physics & mathematics background), though I'm not a botanist. The pr0f's approach really appeals to me! I'm going to try and advance things a little bit on the experimental side, once I'm satisfied that I have something to add. For now, a bit of an analytical contribution, and some more points for discussion.
I used bitmap-images of emission spectra of various bulbs (found these by following the pr0f's lead), and imported them into MATLAB (a piece of software used primarily by scientists and engineers for numerical computing). I was able to extract reasonable estimates of all the curves (x & y data). These are plotted here:
View attachment 1687906
Then, I estimated the over-all illumination spectrum for the combination of bulbs that the pr0f indicated he uses for veg & flowering (respectively). This was achieved by simply summing the emission values at each wavelength, and dividing the resulting intensity vs. wavelength curve by its maximum value, yielding a relative intensity vs. wavelength curve:
View attachment 1687903View attachment 1687904
I think that pr0f has done a really excellent job in putting these sets of bulbs together. However, I thought I would validate this opinion with some analysis.
What I've done (above) in combining the spectra from multiple bulbs, suggests a route to searching for an "optimal" combination of bulbs (one that most closely matches the action spectrum of photosynthesis). This is quite a straightforward thing to do, as we have the spectra from all the bulbs, and the desired spectrum. This is a simple linear equation which can be solved using linear algebra.
The only other wrinkle is which action spectrum to use? There were two images of action spectra that pr0f posted early on in this thread, so I decided to use both. Here is a plot of those action spectra, the pr0f's veg/flow spectra again (without components, just for comparison) and the two fits that I came up with (one for each action spectrum):
View attachment 1687905
The bulb combinations that this analysis suggested were: (for action spectrum 1) 1.0373 x ATI Actinic, 1.3633 x KorallenZucht Fiji Purple, 0.2249 x UVL AquaSun, 0.1560 x UVL RedSun & (for action spectrum 2) 1.0047 x ATI Actinic, 0.1991 x ATI Blue Plus, 0.2575 x ATI Purple Plus, 1.2164 x KorallenZucht Fiji Purple, 0.3465 x UVL AquaSun, 0.1698 x UVL RedSun.
Note that I don't consider these reasonable or recommended in any way! In fact, to achieve these sorts of fractional illuminances, one would have to use filters, or a very large number of bulbs with small fractions in each of the categories corresponding to the fractions indicated in the analysis. Rather, I was trying to keep in the academic spirit of this thread and explore what the math had to tell us. What the math DOES show is (again) that the pr0f's setup is just great, note that neither of the "optimal" spectra do much better (if at all) approximating either of the action spectra. Why did I do this then? To satisfy my suspicious mind! And for the fun of it (yep, I'm a huge nerd, and proud of it).
Finally, some thoughts about action spectra. An action spectrum is obtained by keeping a plant in a sealed environment where CO2 use can be monitored. The use of C02 being a good approximation of the rate of photosynthesis. We did a very simple version of this in a physiology lab that I was the TA for several times, using spinach, and comparing the rate of photosynthesis under blue light to the rate under red light (using colored filters). The reason that I mention this is that it is important to understand that without actually performing this sort of empirical test of the effects of various wavelengths of light on a cannabis plant, it is essentially impossible to know the action spectrum. This is one reason why it is very likely to see images of "the action spectrum of photosynthesis" that are very different: they're derived from empirical observations of different plants. I've been scouring various journals of botany to find out if somebody has actually collected data that would reveal the action spectrum of photosynthesis in cannabis, but to no avail. I'm particularly interested in the possibility of non-linear interactions between wavelengths (this has been mentioned above, the idea that green light can stimulate photosynthesis in the presence of high-intensity red light).
I have found some nuggets of information, however, that are worth-while in relating:
(1) "Action spectra for the promotion of flowering by long periods of irradiation in the red and far-red regions of the spectrum have been determined by the use of interference filters. The percentage floral initiation was greatest at 710720 um for both wheat and rape [seed]"
http://onlinelibrary.wiley.com/doi/10.1111/j.1399-3054.1968.tb07348.x/abstract
(2) "The concentration of cannabinoids in Cannabis sativa L. is correlated with high ultraviolet-B (UV-B) radiation environments."
http://www.osti.gov/energycitations/product.biblio.jsp?osti_id=6696642
(3) "The concentration of Δ9-tetrahydrocannabinol (Δ9-THC), but not of other cannabinoids, in both leaf and floral tissues increased with UV-B dose in drug-type plants."
http://onlinelibrary.wiley.com/doi/10.1111/j.1751-1097.1987.tb04757.x/abstract
Thus, finally:
It seems like perhaps it would be worthwhile to get some far-red in the spectrum to promote flowering (although the data in ref 1 are obvs not from cannabis, and I know that Fiji Purple definitely gets some in there). It also seems like including some UV-B (315 nm280 nm) during flowering might increase THC production. Thoughts?