You're conflicting PAr w/ PUR (photosythectically useable radiation)
PUR would need to be calculated based on the actual useage of ind. organisms.. A daunting task..
PAR "as defined" and measured is basically this (the red square):
Li-COr (best of class) quantum PAR meter response:
some of this is just terminology..
As to converting LUX sensors to more accurate "PAR" sensors one just needs to remove the green filter.. and understand the basic response of the underlying sensor.
As to making spectrophotometers.. well that is in another class by itself, though the underlying hardware is "relatively" simple..
directed light path.. diffraction grating or prism, and a linear CCD or CMOS
http://bwtek.com/spectrometer-part-3a-the-detector/
Actually is easier than it looks.. but getting accuracy and precision is the tricky part..
and the software.. always the software..
Main problem is most "sensors" are not very linear in an entire band..
thus the use of filters and "magic" to smooth the graph out to the flat topped response in the 400-700nm range (for visual light) and to "cut" the tails after 400 and 700nm..
http://www.rags-int-inc.com/phototechstuff/pipeline/