CIE sky for daylighting simulation

Replies to This Discussion

Permalink Reply by Mostapha Sadeghipour Roudsari on February 13, 2016 at 7:35am

Honeybee uses Radiance's gensky to generate CIE skies. You can find more details here; and see the source code: https://github.com/NREL/Radiance/blob/master/src/gen/gensky.c

There are several discussions on Radiance mailing list about the topic like this recent one [reply] and blog posts like this one.

Permalink Reply by Maha Shalaby on February 13, 2016 at 1:23pm

Hi Mostapha,

Thank you for your reply :) it was useful for understanding some things, but I am still confused about the Honeybee component that generates CIE skies,

as in the image, it has six CIE sky types and I understand the difference between them, but for the intermediate one I am lacking the exact type, as far as i understood from the attached paper (table 1) there are several types between the sky with uniform illuminance and the CIE clear sky, do you have information which one is it in the component?

I also have abbreviations from CIE illuminants, like D65, and so, do you know the abbreviations of these skies?

Thanks in advance,

Attachments:

• CIE.jpg, 480 KB
• CIE skies.pdf, 977 KB

Permalink Reply by Mostapha Sadeghipour Roudsari on February 13, 2016 at 5:35pm

Hi Maha,

I don't really know the abbreviations. You should send this question to Radiance mailing list. All honeybee does is calling Radiance gensky or gendaylit to generate the sky.

Here is the exact formula is inside /lib/skybright.cal if this can help you to find the name.

{ RCSid: $Id$ }
{
Sky brightness function for sunny and cloudy skies.

Additional arguments required for calculation of skybright:

A1 - 1 for CIE clear, 2 for CIE overcast,
3 for uniform, 4 for CIE intermediate
A2 - zenith brightness
A3 - ground plane brightness
A4 - normalization factor based on sun direction
A5,A6,A7 - sun direction
}

cosgamma = Dx*A5 + Dy*A6 + Dz*A7;

gamma = Acos(cosgamma); { angle from sun to this point in sky }

zt = Acos(A7); { angle from zenith to sun }

eta = Acos(Dz); { angle from zenith to this point in sky }

wmean(a, x, b, y) : (a*x + b*y) / (a + b);

skybr = wmean((Dz+1.01)^10,
select(A1, sunnysky, cloudysky, unifsky, intersky),
(Dz+1.01)^-10, A3);

sunnysky = A2 * (.91 + 10*exp(-3*gamma) + .45*cosgamma*cosgamma)
* if( Dz - .01, 1.0 - exp(-.32/Dz), 1.0) / A4;

cloudysky = A2 * (1 + 2*Dz)/3;

unifsky = A2;

intersky = A2 * ( (1.35*sin(5.631-3.59*eta)+3.12)*sin(4.396-2.6*zt)
+ 6.37 - eta ) / 2.326 *
exp(gamma*-.563*((2.629-eta)*(1.562-zt)+.812)) / A4;

Permalink Reply by Maha Shalaby on February 14, 2016 at 2:44am

Hi Mostapha,

Thank you for the quick reply I really appreciate it, I will look more into it,

Regards,

Maha

Permalink Reply by Sarith Subramaniam on February 13, 2016 at 7:09pm

To add to Mostapha's replies, here are a few of things to consider:

CIE Illuminants and Radiance/Honeybee

In the context of your question, D65 relates to the color of the sky. More specifically, CIE Illuminants like "A" or "D65" are spectral power distributions that are defined (as per CIE S 014-2/E:2006) for wavelengths ranging from 300nm to 830nm. 

For example, CIE Illuminants A,B and C are defined as :

And D65 is defined as :

For illuminance and luminance calculations, the radiation from such illuminants are converted to Lux or Candela/sq.m by weighing them against the Photopic Luminous Efficiency function (also called as V-lambda):

The equation (1) used for this purpose is

Where y corresponds to the V-lambda function and J corresponds to an illuminant like "D65" or "A".

So, why is all this relevant? Honeybee/Radiance also use a similar method for calculation of luminous flux, illuminance and luminance. However, in the case of Honeybee/Radiance the lighting calculations are limited only 3  (R,G,B) channels (and not the 300nm to 830nm). So the equation (1) from above becomes something like:

F = 47.4*R+120*G+11.6*B

Where (R,G,B) refers to the spectral power of the radiation and the numbers (47.4,120,11.6) relate to the V-lambda function.
So, the bottom line is that an accurate representation of CIE illuminants is not possible inside Radiance/Honeybee as the spectral information is severely restricted. Some studies have proposed using Radiance with more than 3 channels.
For example: http://link.springer.com/article/10.3758%2FBRM.40.1.304 . However, such attempts have been limited.


What is possible with Radiance/Honeybee is to create a fairly accurate representation of brightness of the sky. Although, I can explain that too, I would suggest that you try this link first: http://www.bozzograo.net/radiance/index.php?module=FAQ&func=dis...

By the way, which CIE document are you referring to for CIE sky definitions ?

Permalink Reply by Maha Shalaby on February 14, 2016 at 3:37am

Hi Sarith,

Thank you for your detailed description, I am working on the topic of circadian lighting that is why I was interested in identifying the spectral power distribution curves as you attached in the image because I was doing some manual calculations. I will send you a message with more details about my problem if that is ok with you.

I do not have a good reference for the CIE sky definitions yet, please let me know if you have one.

Permalink Reply by Sarith Subramaniam on February 14, 2016 at 8:58am

Hi Maha,

I just replied to your message. Have you checked out Lark ? http://www.food4rhino.com/project/lark?ufh

CIE 215:2014 is a fairly descriptive document for CIE skies.

Permalink Reply by Maha Shalaby on February 14, 2016 at 2:11pm

Hi Sarith,

This is great, I did not know about Lark, Thanks alot,

Regards