Trying to understand LED output

I am trying to engineer a relatively simple experiment, but I could use some help on understanding lumens. In the EU, the regulations for workplace lighting in medical laboratories calls for 500-1000 lumens so that is the illumination level that I want to duplicate. Typical laboratory lighting is provided by white fluorescent tubes and, if passed through a prism, contains many different color.

What I need to do is to calculate the output required for a given color to match that which is provided at 1000 lumens by typical white light. Obviously I don't need an LED array that puts out 1000 lumens of, say red, since only a portion of the lamp output is in that color. However, I do need to know the illumination level by color provided by typical commercial lighting. Can somebody please point me to a source that gives these details? I doubt I am the first person to have asked this question!

You measuring lux value not lumens ( light source is in lumens - what you get is light reflection from some (what colour ? ) surface

http://www.instrumentsystems.com/fileadmin/editors/downloads/Products/LED_Handbook_e.pdf and

http://www.dfisica.ubi.pt/~hgil/fotometria/HandBook/luz.html

But which wavelengths you will use?
I will tell you one stupid non scientific way.
If you use RGB colors I mean Red Green Blue LED's only you can find this easy using Osram Color Calculator just check the Blue Green and Red checkboxes and put values there, for 1000lm the values can be 950mW blue, 750lm green and 250lm red.
The output is 1036lm, 6633CCT(color temperature) and 44CRI.

But the spectrum of the luminescent light is not from just 3 colors. It is more complex, so to mimic it you have to use more colors.

Thanks for the helpful comments, despite my unartfully worded question.

What I need to know is the illuminance from a particular range of wavelengths. The goal is to provide a quantity of light roughly equivalent to bright early afternoon sunlight on a cloudless day, which is about 100,000 lux, but sunlight is a full spectrum light source with slightly more power on the blue end of the spectrum than on the red.

If I want to know the illuminance from a given source in a particular part of the spectrum, say that coming from 560-580 nm range, what is the best method to go about that? I assume you could eyeball the spectral power distribution for the light source of interest and multiply the total illuminance by the fraction under the curve that represents light from the target range (i.e. 560-580 nm for this example). While I think that will work as a matter of mathematics, since light manufacturers publish the spectral power distribution curves, but not the underlying data used to generate the graph, that seems like a highly imprecise method. Is there a source for the underlying data that is easily accessed? I have looked and not found anything more precise that a line or bar chart. Is there any software that can process a JPG and determine the area under a curve?

The alternative is to determine the values explicitly with a spectrophotometer. That is fairly easy to do with direct sunlight, although there is a seasonal and latitude specific component, but I want to have similar data on various artificial sources and there are too many manufacturers and model of lighting product to handle easily. Any suggestions would be helpful.

In answer to Nichirenshu, I plan to use slices of the spectrum from UV to IR in roughly 20 nm increments, and the LED has to put out a narrow spectrum around a central frequency as I am testing the effects of particular spectrum slices on cell suspensions. The light does not have to be coherent source, as with a laser, but is does have to be close to monochromatic (defined as a central wavelength +/- about 10 nm). I assume I will have to use multiple LEDs for some wavelength slices to match bright sunlight intensities and cannot expect all the LEDs to have the exact same wavelength due to variability in manufacturing.