[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [APD] light and the w/gal rule

S. Hieber wrote:
> Well, if you want to go whole hog, use Pearson's R to do a
> partial regression on all the relevent factors, and don't
> forget to take into account the spectral shift that occurs
> with age and variations that occur coming off the
> production line, etc. When you're all done, most folks will
> find that they wpg still works well, is close at hand and
> convenient, and works as anything else. It's not like there
> is some critical value of light that has to be obtained or
> else plants won't grow well. 

There is such a value, but it's probably different for individual plant 
species and aquarium conditions.

One neat thing about regression analysis is that it doesn't matter *why* 
something is correlated, it just is or isn't. I can explain 96% of the 
PAR output using lumen data alone, and I will be within 3.5 PAR of the 
true PAR value 89% of the time. That's pretty good to me. It doesn't 
really motivate me to go find the source of that other 4%.

Your suggestion gave me an idea. Since lumens are based upon almost the 
same spectrum of light that PAR is, but uses a weighted curve of 
perceived brightness as seen by the human eye, I wondered if lumens and 
watts taken together would make a more accurate predictor of PAR than 
lumens alone. Thank you for the idea. I now have a logarithmic equation 
with an r-squared of 0.98!!! Here's the equation:

PAR = 0.13 x sqrt(lumens x watts)

So, for an example let's use an AH Supply 55W bulb. We know it's 55 
watts and has about 4700 lumens. That comes out to a PAR of 66. Who 
knows how true this is?

Jerry Baker
Aquatic-Plants mailing list
Aquatic-Plants at actwin_com