sky definition part 2

radiance-general
2

Hi Zack,

It was great to meet you at the workshop, and I really enjoyed your presentation.

I think you're making trouble for yourself by trying to adjust the turbidity factor in gensky. If you have the horizontal and diffuse irradiance, or facsimilies thereof, you should be giving those to gensky directly using the -B or -b and -R or -r options as we have been discussing.

Just to clarify, gensky ALWAYS produces an accurate CIE sky distribution. Adjusting the turbidity or these other factors only affects the absolute levels and the ratio between solar and sky components. It does not change the distribution of skylight.

This is not to say that the CIE sky model is the best. It is just an agreed upon standard, and therefore serves as a reasonable point for comparison. The Perez sky is probably a better approximation to actual skies, particularly for the intermediate case.

As I said in the bit of e-mail you so aptly quoted, the absolute values produced by gensky with the default settings are unreliable because they are not based on weather data for your area. If you have such measurements, by all means use them, but don't adjust the turbidity -- go straight for the horizontal and diffuse specifications to gensky. This overrides the turbidity approximation and makes it irrelevant.

I hope this helps.
-Greg

P.S. I got an e-mail from Martin Moeck on the related topic of solar efficacy, and include it here for discussion. I don't know the right answer, but I'm sure someone out there does...

···

Begin forwarded message:

From: Greg Ward <[email protected]>
Date: Wed Oct 15, 2003 2:33:06 PM US/Pacific

Hi Martin,

Now that you mention it, I'm not sure at all. I got my number from the efficacy of Standard Illuminant B, which is simulated (rather than real) sunlight, and it could be way off. Where do you get your numbers? If you are sure of them, please forward this message with my apologies to radiance-general.

Thanks!
-Greg

From: "Martin Moeck" <[email protected]>
Date: Wed Oct 15, 2003 2:26:55 PM US/Pacific

Greg,

are you sure? The sun's luminous efficacy above 20 deg. altitude is somewhere around 95-100 lm/W, overcast skies are around 120 lmW, clear skies around 150 lm/W. Therefore, your multipliers for the -R/r options should be smaller than the multipliers for the -B/b options?

Martin

3

Gidday Greg, Zack & others...

Just to chip in about measurement and efficacies -

Paul Littlefair wrote a great review paper on sky efficacies, with overall results:
144+-7 lm/W clear skys
115+-8 lm/W overcast skies

Littlefair, PJ Measurement of the Luminous Efficacy, Light. Res & Tech, 1988 4:p177-188

This might also be of interest
Fontoynont, M, Perceived Performance of Daylighting Systems: Lighting Efficacy and Agreeableness, Solar Energy 2002 73(2) p83-94

And it should be remembered that measurements are often not of the actual sky spectral power distribution but are taken with equipment calibrated to Illuminant A with 160 lm/W.

And if calibrated to D65 then efficacy is 203 lm/W.

hopefully all this doesn't make things even more 'hazy' :slight_smile:

cheers
alex

···

*******************************************************
A. J. Summerfield [email protected]
Faculty of Architecture, University of Sydney

Outside of a dog, a book is man's best friend.
Inside a dog, it's too dark to read. Groucho Marx
*******************************************************

On Thursday, October 16, 2003, at 04:02 PM, Greg Ward wrote:

Hi Zack,

It was great to meet you at the workshop, and I really enjoyed your presentation.

I think you're making trouble for yourself by trying to adjust the turbidity factor in gensky. If you have the horizontal and diffuse irradiance, or facsimilies thereof, you should be giving those to gensky directly using the -B or -b and -R or -r options as we have been discussing.

Just to clarify, gensky ALWAYS produces an accurate CIE sky distribution.
  Adjusting the turbidity or these other factors only affects the absolute levels and the ratio between solar and sky components. It does not change the distribution of skylight.

This is not to say that the CIE sky model is the best. It is just an agreed upon standard, and therefore serves as a reasonable point for comparison. The Perez sky is probably a better approximation to actual skies, particularly for the intermediate case.

As I said in the bit of e-mail you so aptly quoted, the absolute values produced by gensky with the default settings are unreliable because they are not based on weather data for your area. If you have such measurements, by all means use them, but don't adjust the turbidity -- go straight for the horizontal and diffuse specifications to gensky. This overrides the turbidity approximation and makes it irrelevant.

I hope this helps.
-Greg

P.S. I got an e-mail from Martin Moeck on the related topic of solar efficacy, and include it here for discussion. I don't know the right answer, but I'm sure someone out there does...

Begin forwarded message:

From: Greg Ward <[email protected]>
Date: Wed Oct 15, 2003 2:33:06 PM US/Pacific

Hi Martin,

Now that you mention it, I'm not sure at all. I got my number from the efficacy of Standard Illuminant B, which is simulated (rather than real)
sunlight, and it could be way off. Where do you get your numbers? If you are sure of them, please forward this message with my apologies to radiance-general.

Thanks!
-Greg

From: "Martin Moeck" <[email protected]>
Date: Wed Oct 15, 2003 2:26:55 PM US/Pacific

Greg,

are you sure? The sun's luminous efficacy above 20 deg. altitude is somewhere around 95-100 lm/W, overcast skies are around 120 lmW, clear skies around 150 lm/W. Therefore, your multipliers for the -R/r options should be smaller than the multipliers for the -B/b options?

Martin

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

4

Gidday Greg, Zack & others...

Just to chip in about measurement and efficacies -

Paul Littlefair wrote a great review paper on sky efficacies, with overall results:
144+-7 lm/W clear sky
115+-8 lm/W overcast sky

Littlefair, PJ Measurement of the Luminous Efficacy, Light. Res & Tech, 1988 4:p177-188

Also:
Fontoynont, M, Perceived Performance of Daylighting Systems: Lighting Efficacy and Agreeableness, Solar Energy 2002 73(2) p83-94

It should be remembered that measurements are often not of the actual sky spectral power distribution but are taken with equipment calibrated to Illuminant A with 160 lm/W.

And if calibrated to D65 then efficacy is 203 lm/W.

hopefully all this doesn't make things even more 'hazy' :slight_smile:

cheers
alex

···

*******************************************************
A. J. Summerfield [email protected]
Faculty of Architecture, University of Sydney

Outside of a dog, a book is man's best friend.
Inside a dog, it's too dark to read. Groucho Marx
*******************************************************

On Thursday, October 16, 2003, at 04:02 PM, Greg Ward wrote:

Hi Zack,

It was great to meet you at the workshop, and I really enjoyed your presentation.

I think you're making trouble for yourself by trying to adjust the turbidity factor in gensky. If you have the horizontal and diffuse irradiance, or facsimilies thereof, you should be giving those to gensky directly using the -B or -b and -R or -r options as we have been discussing.

Just to clarify, gensky ALWAYS produces an accurate CIE sky distribution.
  Adjusting the turbidity or these other factors only affects the absolute levels and the ratio between solar and sky components. It does not change the distribution of skylight.

This is not to say that the CIE sky model is the best. It is just an agreed upon standard, and therefore serves as a reasonable point for comparison. The Perez sky is probably a better approximation to actual skies, particularly for the intermediate case.

As I said in the bit of e-mail you so aptly quoted, the absolute values produced by gensky with the default settings are unreliable because they are not based on weather data for your area. If you have such measurements, by all means use them, but don't adjust the turbidity -- go straight for the horizontal and diffuse specifications to gensky. This overrides the turbidity approximation and makes it irrelevant.

I hope this helps.
-Greg

P.S. I got an e-mail from Martin Moeck on the related topic of solar efficacy, and include it here for discussion. I don't know the right answer, but I'm sure someone out there does...

Begin forwarded message:

From: Greg Ward <[email protected]>
Date: Wed Oct 15, 2003 2:33:06 PM US/Pacific

Hi Martin,

Now that you mention it, I'm not sure at all. I got my number from the efficacy of Standard Illuminant B, which is simulated (rather than real)
sunlight, and it could be way off. Where do you get your numbers? If you are sure of them, please forward this message with my apologies to radiance-general.

Thanks!
-Greg

From: "Martin Moeck" <[email protected]>
Date: Wed Oct 15, 2003 2:26:55 PM US/Pacific

Greg,

are you sure? The sun's luminous efficacy above 20 deg. altitude is somewhere around 95-100 lm/W, overcast skies are around 120 lmW, clear skies around 150 lm/W. Therefore, your multipliers for the -R/r options should be smaller than the multipliers for the -B/b options?

Martin

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

5

Hi Greg,

Thanks for the quick response.

I think you're making trouble for yourself by trying to adjust the turbidity factor in gensky. If you have the horizontal and diffuse irradiance, or facsimilies thereof, you should be giving those to gensky directly using the -B or -b and -R or -r options as we have been discussing.

Yes, the turbidity does seem to be trouble since it doesn't affect the sun component. I don't have measurements I am trying to match, I am just trying to make sure I have the correct IESNA recommended sky (CIE sky) and sun components throughout the year.

Just to clarify, gensky ALWAYS produces an accurate CIE sky distribution. Adjusting the turbidity or these other factors only affects the absolute levels and the ratio between solar and sky components. It does not change the distribution of skylight.

I don't doubt the relative luminance distribution of the sky is accurate, I'm just questioning how Radiance determines the correct overall level.

I should clarify what I mean when I say the CIE sky function tells me....I am refering to both the sky and sun calculations laid out in the latest IESNA lighting handbook which uses Kittlers CIE adopted sky function (which I understand is from his 1967 studies) to determine the sky component. So I probably should have said that IESNA sky/sun functions tell me. I am not sure if this is the accepted method used by others. The direct component is calculated as this;

Edn = Ext * e ^ (-cm)

where;
Edn = direct normal solar illuminance
Ext = extraterrestrial solar illuminance = Esc * (1 + 0.034 * cos ( (2pi/365)*(J-2)))
                where;
                Esc = solar illumination constant = 127.5 klx (11,850 fc)
                J = julian date
c = atmospheric extinction coefficient (given as c = 0.21 for clear ; c = 0.8 for partly cloudy)
m = optical air mass = 1 / sin (solar altitude)

and Direct Horizontal (Edh) would then be;
Edh = Edn * sin (solar altitude)

This gives the Direct Horiz to be 8,821 fc on a clear summer solstice at 40 lat. This is what I would like Radiance to give me without having to specify with the -r parameter. Currently it is giving me 6,836fc.

The IESNA recommends calculating the diffuse component like this;

Ekh = A + B * (sin (solar altitude))^ C
where;
Ekh = horiz. sky illuminance
A = sunrise/set illuminance = given as 0.8 for clear sky
B = solar altitude illuminance coef. = given as 15.5 for clear sky
C = solar altitude illuminance exponent = given as 0.5 for clear sky (all in terms of klux)

This gives a diffuse horiz of 1484.2fc. Radiance gives me this with a turbidity of 3.22. But as I illustrated before it can range dramatically. Ekh is then fed into the CIE sky distribution function like so;

Lz = Ekh * ZL
where;
Lz = zenith luminance
ZL = zenith luminance factor - given as a table of values

Then Lz is part of Kittlers distribution function. I am not going to attempt to type this function but I think we are on the same page here. Although I understand they, CIE, are about to adopt a new sky function with 15 different sky types (Thanks Karen for the info!). I have not attempted to wrap my head around this yet...for now I am happy enough with the old function.

This is not to say that the CIE sky model is the best. It is just an agreed upon standard, and therefore serves as a reasonable point for comparison. The Perez sky is probably a better approximation to actual skies, particularly for the intermediate case.

Agreed, although the calcs I have outlined here have generally matched up well with my measurements with enough accuracy for my uses, but I am usually not in need of an extremely accurate sky just good average representations.

As I said in the bit of e-mail you so aptly quoted, the absolute values produced by gensky with the default settings are unreliable because they are not based on weather data for your area. If you have such measurements, by all means use them, but don't adjust the turbidity -- go straight for the horizontal and diffuse specifications to gensky. This overrides the turbidity approximation and makes it irrelevant.

Again, I don't necessarily have measurements and I don't want to have to give gensky horiz. or diffuse specifications for annual calculations. Although, I guess these could be pulled from annual weather data (TMY2). Is this what others do??? I guess I am just wanting Radiance to give me these IESNA recommended results throughout the year without having to specify the correct turbidity, diffuce irradiance or solar radiance values which are not required inputs to these equations. It would be nice to have an elevation factor as well or a turbidity factor that adjusts both the solar and sky accordingly.

Please give me feedback, let me know if I have misintepretted any of this. Since daylighting design puts food on the table I need to make sure I am confident in my results. I am especially concerned since one of my current projects is getting pretty detailed discussing fc-hours underneath a tree canopy in an atrium in Chicago, given all the other approximated variables that will be going into this model it would be nice to be confident I am starting with good data. In this case though I am just creating DF's with Radiance and feeding that into TMY2 data. And so I guess as long as my sun to sky ratios are correct it should be fine.

Huh, maybe I just got it…Does the 2.75 default turbidity give the correct ratio for the lower sun component Radiance provides? I guess this is what you meant when you said they are not based on measurements for my area. But these IESNA calcs are not for my area either, they are for a generic sea level site. It seems more and more its a discrepancy in where the c, A, B, C values come from (these variables essentially determine the absolute magnitude for the sun and sky components) What other standards are there in determing the magnitude of the sun and sky throughout the year? Is weather data just used? I guess I could just give my skies a mulitplier outside of gensky that adjusts my skies to the levels I want. Did I just answer my own questions? Still it would be nice if Radiance did this on its own. Are people really seeing7970 fc on a clear summer day??? Where are you living? I guess I am just used to our clear Colorado skies where we will see up to 11,000fc regularly in the summer.

Well, sorry for typing your eyes off? Just describing this problem has helped me understand much better what is going on. Hopefully others find this discussion useful. If not, I aplogize.

Thanks!
Zack

···

--
Zack Rogers
Staff Engineer
Architectural Energy Corporation
2540 Frontier Avenue, Suite 201
Boulder, CO 80301 USA

tel (303)444-4149 ext.235
fax (303)444-4304

6

From: Zack Rogers <[email protected]>
Date: Thu Oct 16, 2003 5:43:13 PM US/Pacific

Just to clarify, gensky ALWAYS produces an accurate CIE sky distribution. Adjusting the turbidity or these other factors only affects the absolute levels and the ratio between solar and sky components. It does not change the distribution of skylight.

I don't doubt the relative luminance distribution of the sky is accurate, I'm just questioning how Radiance determines the correct overall level.

In my opinion, and others may differ, there is no "correct overall level" for daylight. You can come up with a ballpark value, but getting within 50% without some more specific weather data is overly ambitious.

I should clarify what I mean when I say the CIE sky function tells me....I am refering to both the sky and sun calculations laid out in the latest IESNA lighting handbook which uses Kittlers CIE adopted sky function (which I understand is from his 1967 studies) to determine the sky component. So I probably should have said that IESNA sky/sun functions tell me. I am not sure if this is the accepted method used by others. The direct component is calculated as this;

Edn = Ext * e ^ (-cm)

where;
Edn = direct normal solar illuminance
Ext = extraterrestrial solar illuminance = Esc * (1 + 0.034 * cos ( (2pi/365)*(J-2)))
               where;
               Esc = solar illumination constant = 127.5 klx (11,850 fc)
               J = julian date
c = atmospheric extinction coefficient (given as c = 0.21 for clear ; c = 0.8 for partly cloudy)
m = optical air mass = 1 / sin (solar altitude)

and Direct Horizontal (Edh) would then be;
Edh = Edn * sin (solar altitude)

This gives the Direct Horiz to be 8,821 fc on a clear summer solstice at 40 lat. This is what I would like Radiance to give me without having to specify with the -r parameter. Currently it is giving me 6,836fc.

The formula above for Ext must be specific to the northern hemisphere. I don't see how the same seasonal correction factor could apply everywhere on the globe. I don't think this is the declination angle, as the solar altitude includes that. I'm not sure what this is about.

The variation of atmospheric extinction is quite large. I wouldn't trust this, either. In other words, this gives you a value, but I'm not sure it's any better than the guess provided by gensky. It's still just a guess.

The IESNA recommends calculating the diffuse component like this;

Ekh = A + B * (sin (solar altitude))^ C
where;
Ekh = horiz. sky illuminance
A = sunrise/set illuminance = given as 0.8 for clear sky
B = solar altitude illuminance coef. = given as 15.5 for clear sky
C = solar altitude illuminance exponent = given as 0.5 for clear sky (all in terms of klux)

This gives a diffuse horiz of 1484.2fc. Radiance gives me this with a turbidity of 3.22. But as I illustrated before it can range dramatically. Ekh is then fed into the CIE sky distribution function like so;

Lz = Ekh * ZL
where;
Lz = zenith luminance
ZL = zenith luminance factor - given as a table of values

Then Lz is part of Kittlers distribution function. I am not going to attempt to type this function but I think we are on the same page here. Although I understand they, CIE, are about to adopt a new sky function with 15 different sky types (Thanks Karen for the info!). I have not attempted to wrap my head around this yet...for now I am happy enough with the old function.

What is the table based on? Is Kittler's function the same as the CIE clear sky? I'm confused. This all seems a bit arbitrary to me, especially in the slection of constants for clear vs. overcast skies.

However, if you like this calculation, you could certainly add it as a wrapper to gensky using a simple shell script that computes the -r and -b options based on these factors and I'd be happy to add it to the distribution. To my sensibilities, it seems like a lot of effort for what must be a pretty crude estimate.

As I said in the bit of e-mail you so aptly quoted, the absolute values produced by gensky with the default settings are unreliable because they are not based on weather data for your area. If you have such measurements, by all means use them, but don't adjust the turbidity -- go straight for the horizontal and diffuse specifications to gensky. This overrides the turbidity approximation and makes it irrelevant.

Again, I don't necessarily have measurements and I don't want to have to give gensky horiz. or diffuse specifications for annual calculations. Although, I guess these could be pulled from annual weather data (TMY2). Is this what others do??? I guess I am just wanting Radiance to give me these IESNA recommended results throughout the year without having to specify the correct turbidity, diffuce irradiance or solar radiance values which are not required inputs to these equations. It would be nice to have an elevation factor as well or a turbidity factor that adjusts both the solar and sky accordingly.

It seems that the correct input to these equations is every bit as difficult as turbidity to obtain, if not moreso, and just using the values in the book is equivalent to using some rule-of-thumb value for turbidity in most respects. It's just as likely to give you a reasonable value, not!

Please give me feedback, let me know if I have misintepretted any of this. Since daylighting design puts food on the table I need to make sure I am confident in my results. I am especially concerned since one of my current projects is getting pretty detailed discussing fc-hours underneath a tree canopy in an atrium in Chicago, given all the other approximated variables that will be going into this model it would be nice to be confident I am starting with good data. In this case though I am just creating DF's with Radiance and feeding that into TMY2 data. And so I guess as long as my sun to sky ratios are correct it should be fine.

I really think TMY data is a better starting point than the IES handbook or gensky's default calculation.

Huh, maybe I just got it…Does the 2.75 default turbidity give the correct ratio for the lower sun component Radiance provides? I guess this is what you meant when you said they are not based on measurements for my area. But these IESNA calcs are not for my area either, they are for a generic sea level site. It seems more and more its a discrepancy in where the c, A, B, C values come from (these variables essentially determine the absolute magnitude for the sun and sky components) What other standards are there in determing the magnitude of the sun and sky throughout the year? Is weather data just used? I guess I could just give my skies a mulitplier outside of gensky that adjusts my skies to the levels I want. Did I just answer my own questions? Still it would be nice if Radiance did this on its own. Are people really seeing7970 fc on a clear summer day??? Where are you living? I guess I am just used to our clear Colorado skies where we will see up to 11,000fc regularly in the summer.

Bottom line is that the calculation of absolute levels is suspect in gensky and also suspect in the equations you've listed. Weather data is more reliable, and that will show you that absolutes vary quite a lot even in a particular locale, especially for overcast skies. If there is a consensus as to what the default calculation should be in gensky, I'd be happy to implement it. However, a change to the output from one version to the next can throw people off just as badly as leaving a slightly wrong calculation in there.

-Greg

···

Begin forwarded message:

7

Hi Greg,

Just to clarify, gensky ALWAYS produces an accurate CIE sky distribution. Adjusting the turbidity or these other factors only affects the absolute levels and the ratio between solar and sky components. It does not change the distribution of skylight.

I don't doubt the relative luminance distribution of the sky is accurate, I'm just questioning how Radiance determines the correct overall level.

In my opinion, and others may differ, there is no "correct overall level" for daylight. You can come up with a ballpark value, but getting within 50% without some more specific weather data is overly ambitious.

Huh, I've always seen a little better correlation between these calculations and both weather data and measurements I've taken in Colorado (when taking into account elevation factors). I know you can never count on it to be exact but from the limited data I've looked I say it was always atleast within 20% both + and -. And I guess all I'm looking for is a good average - so half the sites at that lat. might have higher and half, lower.

I should clarify what I mean when I say the CIE sky function tells me....I am refering to both the sky and sun calculations laid out in the latest IESNA lighting handbook which uses Kittlers CIE adopted sky function (which I understand is from his 1967 studies) to determine the sky component. So I probably should have said that IESNA sky/sun functions tell me. I am not sure if this is the accepted method used by others. The direct component is calculated as this;

Edn = Ext * e ^ (-cm)

where;
Edn = direct normal solar illuminance
Ext = extraterrestrial solar illuminance = Esc * (1 + 0.034 * cos ( (2pi/365)*(J-2)))
               where;
               Esc = solar illumination constant = 127.5 klx (11,850 fc)
               J = julian date
c = atmospheric extinction coefficient (given as c = 0.21 for clear ; c = 0.8 for partly cloudy)
m = optical air mass = 1 / sin (solar altitude)

and Direct Horizontal (Edh) would then be;
Edh = Edn * sin (solar altitude)

This gives the Direct Horiz to be 8,821 fc on a clear summer solstice at 40 lat. This is what I would like Radiance to give me without having to specify with the -r parameter. Currently it is giving me 6,836fc.

The formula above for Ext must be specific to the northern hemisphere. I don't see how the same seasonal correction factor could apply everywhere on the globe. I don't think this is the declination angle, as the solar altitude includes that. I'm not sure what this is about.

I'm confused. What seasonal correction factor, c? The declination angle does go into the solar altitude angle calc, but it is not used elsewhere in this calc.

The variation of atmospheric extinction is quite large. I wouldn't trust this, either. In other words, this gives you a value, but I'm not sure it's any better than the guess provided by gensky. It's still just a guess.

I'm sure this atmospheric extinction factor was derived from weather data. I don't think IES would publish unless there was atleast some verification. I think this variable is trying to take into account the different turbidities for clear and partly cloudy skies.

The IESNA recommends calculating the diffuse component like this;

Ekh = A + B * (sin (solar altitude))^ C
where;
Ekh = horiz. sky illuminance
A = sunrise/set illuminance = given as 0.8 for clear sky
B = solar altitude illuminance coef. = given as 15.5 for clear sky
C = solar altitude illuminance exponent = given as 0.5 for clear sky (all in terms of klux)

This gives a diffuse horiz of 1484.2fc. Radiance gives me this with a turbidity of 3.22. But as I illustrated before it can range dramatically. Ekh is then fed into the CIE sky distribution function like so;

Lz = Ekh * ZL
where;
Lz = zenith luminance
ZL = zenith luminance factor - given as a table of values

Then Lz is part of Kittlers distribution function. I am not going to attempt to type this function but I think we are on the same page here. Although I understand they, CIE, are about to adopt a new sky function with 15 different sky types (Thanks Karen for the info!). I have not attempted to wrap my head around this yet...for now I am happy enough with the old function.

What is the table based on? Is Kittler's function the same as the CIE clear sky? I'm confused. This all seems a bit arbitrary to me, especially in the slection of constants for clear vs. overcast skies.

I'm not sure what the ZL table is based on, again I am guessing it is derived from weather data. Yes, from what I understand the CIE clear sky is Kittler's function from the 1967 studies. Yeah, I'm not sure where the A, B, and C variables are derived from either. I just do what IES tells me to. Actually, I tend to trust these calcs for several reasons; 1) they have never seemed to be off more than 20% from weather data and measurements I've looked at (but I've done no extensive study by any means) and 2) while I'm not one to question master Greg Ward, I am also not one to question David DiLaura or the IES. One of the other sky calculation programs I'm using for these comparisons is one he wrote based on these equations, it was discussed as one of the best methodologies to estimate daylight levels, given that exact predictions are always impossible.

However, if you like this calculation, you could certainly add it as a wrapper to gensky using a simple shell script that computes the -r and -b options based on these factors and I'd be happy to add it to the distribution. To my sensibilities, it seems like a lot of effort for what must be a pretty crude estimate.

I do like this calculation. If it is not too much trouble, I for one, would like to see it added to gensky. Otherwise, I can always apply a multiplier outside of gensky to adjust. BTW - I think what this has shown is that it is dangerous to apply the -r option without applying the -b options or adjusting the turbidity and vice versa. Since that would change the sky to sun ratio and give you a false sky/sun condition? When I was only specifying a -R the change in the sky illuminance was less than a 1 fc while the change in the solar illuminance was roughly 2000fc.

It seems that the correct input to these equations is every bit as difficult as turbidity to obtain, if not moreso, and just using the values in the book is equivalent to using some rule-of-thumb value for turbidity in most respects. It's just as likely to give you a reasonable value, not!

Well, I guess since the difficult input has be done for us with the constants, and if you trust the constants its not as difficult.

I really think TMY data is a better starting point than the IES handbook or gensky's default calculation.

Yeah, I agree. And when it matters (for annual calculations) I use TMY data. Just sometimes in daylighting design you want to show those worst case conditions - a sunny summer solstice and a cloudy winter solstice. These are all theoretical studies where it is okay to use an average sky and not actual data, since TMY weather data might not give you those conditions. Also, with TMY data, I have never found the direct normal (converted to horiz.) and diffuse horiz. data to add up to the global data reported, they are typically close but never exact. And it seems like the global measurement must be the most accurate of the three.

Bottom line is that the calculation of absolute levels is suspect in gensky and also suspect in the equations you've listed. Weather data is more reliable, and that will show you that absolutes vary quite a lot even in a particular locale, especially for overcast skies. If there is a consensus as to what the default calculation should be in gensky, I'd be happy to implement it. However, a change to the output from one version to the next can throw people off just as badly as leaving a slightly wrong calculation in there.

Perhaps if you do the change, it could just be an add on specified by a -ies switch or something.

Thanks so much for the quick responses and helpful information!
Zack

···

--
Zack Rogers
Staff Engineer
Architectural Energy Corporation
2540 Frontier Avenue, Suite 201
Boulder, CO 80301 USA

tel (303)444-4149 ext.235
fax (303)444-4304

8

Hi Zack and all,

Just to add to the discussion (I hope not to the confusion) T. Muneer gives
a very detailed description of efficacy models in his book "Solar Radiation
& Daylight Models" which I think is very recommendable.
Or you can also look at:
Luminous Efficacy: Evaluation of models for the UK, Lighting Research &
Technology, 27(2) 71-77 (1995)
Solar irradiance and illuminance models for Japan II: Luminous efficacies,
Ligh. Res. & Tech., 27(4) 223-230 (1995)
both also by Muneer.

For beam luminous efficacy, he says that taking a single average value is "a
simple yet robust model", but he recommends using a value between 93-115
lm/W.

For global and diffuse efficacies there are several models. I used the Perez
model for my masters thesis, but I found it had to be corrected for Tokyo,
which I did with local weather data. There is a short description of what I
did in last year's Radiance Workshop CD. The Perez model considers many
factors like atmospheric turbidity, precipitable water content, etc, etc
(the list is long) which can be derived from TMY data, and since the model
was originally calculated from US and european data, you can probably use it
for your location without re-calibration.
I think the reference is:
Perez et al., Modeling daylight availability and irradiance components from
direct and global irradiance. Solar Energy, 44, 271 (1990)

Hope this helps. On the other hand, if it`s not a problem I would like to
see your study on different heights. I`m very interested in the subject as
we usually have to interpolate-extrapolate data in Argentina, due to the
lack of measuring stations.

Please keep posting your findings, I for one, get a lot from it. Thanks and
regards!

Santiago.

9

Hi Santiago,

I've been buzzing with thoughts of Image Based Rendering and HDR displays ever since the conference, I've been showing Greg's Siggraph animation and your studies to all my colleagues. I am very interested in getting into both of these in my daylighting design work.

Yeah, the IES recommends 94.2 lm/watt for direct beam -- IES is my master, but google does reign supreme;-) -- but I don't know how they think they can be so accurate when everyone else reports these relatively large ranges.

I plan on looking into the Perez sky model in more detail. So you found it gave better correlation than the CIE model in your studies? Is gendaylit available in any windows form yet? If not I have I do have access to linux machines.

Georg, are there any plans to implement the Perez sky model (gendaylit) in Rayfront?

Thanks for the info!
Zack

PS I will send the elevation study separately as i think it will just turn binary here.
PPS but i need your e-mail address. I can't seem to get it from the site. Mine is [email protected]

···

--
Zack Rogers
Staff Engineer
Architectural Energy Corporation
2540 Frontier Avenue, Suite 201
Boulder, CO 80301 USA

tel (303)444-4149 ext.235
fax (303)444-4304

10

Hi Zack,

From: Zack Rogers <[email protected]>
Date: Thu Oct 16, 2003 10:25:29 PM US/Pacific

I should clarify what I mean when I say the CIE sky function tells me....I am refering to both the sky and sun calculations laid out in the latest IESNA lighting handbook which uses Kittlers CIE adopted sky function (which I understand is from his 1967 studies) to determine the sky component. So I probably should have said that IESNA sky/sun functions tell me. I am not sure if this is the accepted method used by others. The direct component is calculated as this;

Edn = Ext * e ^ (-cm)

where;
Edn = direct normal solar illuminance
Ext = extraterrestrial solar illuminance = Esc * (1 + 0.034 * cos ( (2pi/365)*(J-2)))
               where;
               Esc = solar illumination constant = 127.5 klx (11,850 fc)
               J = julian date
c = atmospheric extinction coefficient (given as c = 0.21 for clear ; c = 0.8 for partly cloudy)
m = optical air mass = 1 / sin (solar altitude)

and Direct Horizontal (Edh) would then be;
Edh = Edn * sin (solar altitude)

This gives the Direct Horiz to be 8,821 fc on a clear summer solstice at 40 lat. This is what I would like Radiance to give me without having to specify with the -r parameter. Currently it is giving me 6,836fc.

The formula above for Ext must be specific to the northern hemisphere. I don't see how the same seasonal correction factor could apply everywhere on the globe. I don't think this is the declination angle, as the solar altitude includes that. I'm not sure what this is about.

I'm confused. What seasonal correction factor, c? The declination angle does go into the solar altitude angle calc, but it is not used elsewhere in this calc.

I'm referring to the expression for Ext = Esc * (1 + 0.034 * cos ( (2pi/365)*(J-2))), which looks like a seasonal correction much like declination angle to me. I really don't know what it's doing in there.

The variation of atmospheric extinction is quite large. I wouldn't trust this, either. In other words, this gives you a value, but I'm not sure it's any better than the guess provided by gensky. It's still just a guess.

I'm sure this atmospheric extinction factor was derived from weather data. I don't think IES would publish unless there was atleast some verification. I think this variable is trying to take into account the different turbidities for clear and partly cloudy skies.

Your trust is greater than mine if you believe you can take weather data and come up with one value for clear and another for cloudy.

I'm not sure what the ZL table is based on, again I am guessing it is derived from weather data. Yes, from what I understand the CIE clear sky is Kittler's function from the 1967 studies. Yeah, I'm not sure where the A, B, and C variables are derived from either. I just do what IES tells me to. Actually, I tend to trust these calcs for several reasons; 1) they have never seemed to be off more than 20% from weather data and measurements I've looked at (but I've done no extensive study by any means) and 2) while I'm not one to question master Greg Ward, I am also not one to question David DiLaura or the IES. One of the other sky calculation programs I'm using for these comparisons is one he wrote based on these equations, it was discussed as one of the best methodologies to estimate daylight levels, given that exact predictions are always impossible.

It's good to get your experience, but I'd like to hear others. You should *always* question these calculations, no matter what the source. Especially if it's from me!

However, if you like this calculation, you could certainly add it as a wrapper to gensky using a simple shell script that computes the -r and -b options based on these factors and I'd be happy to add it to the distribution. To my sensibilities, it seems like a lot of effort for what must be a pretty crude estimate.

I do like this calculation. If it is not too much trouble, I for one, would like to see it added to gensky. Otherwise, I can always apply a multiplier outside of gensky to adjust. BTW - I think what this has shown is that it is dangerous to apply the -r option without applying the -b options or adjusting the turbidity and vice versa. Since that would change the sky to sun ratio and give you a false sky/sun condition? When I was only specifying a -R the change in the sky illuminance was less than a 1 fc while the change in the solar illuminance was roughly 2000fc.

The ratios between skylight and sunlight are no more reliable than the absolutes of the two in gensky. It is indeed a very crude estimate I'm using. If you have a look at the source, you'll see what I mean. I do recommend specifying -r and -b options together, though -- if you're using one, you should use the other. I would do this rather than applying a multiplier outside of gensky.

It seems that the correct input to these equations is every bit as difficult as turbidity to obtain, if not moreso, and just using the values in the book is equivalent to using some rule-of-thumb value for turbidity in most respects. It's just as likely to give you a reasonable value, not!

Well, I guess since the difficult input has be done for us with the constants, and if you trust the constants its not as difficult.

As I indicated, I am agnostic towards these absolute sky level estimates, leaning towards athiest. I am happy to implement the IES calcs in gensky, as I see them as being no worse than what's in there, but I would like to hear a general "huzzah!" from the group before doing so.

-Greg

11

I really don't care too much about changing gensky, since horizontal illuminance values for overcast and clear sky conditions as well as solar illuminance can be looked up quickly, and gensky output can be adjusted accordingly. What really matters for green buildings and daylight availability calculations is not the IES handbook formulae and turbidity factors, but weather tapes with yearly diffuse and beam data, and I'd like to repeat that gendaylit based on the Perez sky model is a good program for that. The CIE clear sky model overestimates light from the horizon too much for my taste.

Martin

12

Hi again,

I've been buzzing with thoughts of Image Based Rendering and HDR

I`m happy to hear you found it interesting. I promise to keep the list
posted about new results.

I plan on looking into the Perez sky model in more detail. So you found
it gave better correlation than the CIE model in your studies? Is

In fact I was referring to the Perez model for luminous efficacy. I should
have been more clear perhaps.
I haven`t tried the sky model (nor gendaylit), but I plan to compare results
between models in the future. If anyone knows of such studies, I would
really appreciate letting me know.
The efficacy model is a bit awkward, but all the input can be derived from
quite simple weather data (when they don`t include illuminances)

Regards,

Santiago.

13

Just a short idea

I'm referring to the expression for Ext = Esc * (1 + 0.034 * cos

(2pi/365)*(J-2))), which looks like a seasonal correction much like
declination angle to me. I really don't know what it's doing in there.

That`s probably to compensate for the earth`s orbit excentricity. Being an
ellipse, the distance to sun is not allways the same. I think it`s closer
during southern summer, and only a little bit, that`s why the compensation
factor is so small (0.034). Extraterrestrial illuminance is the illuminance
measured outside the atmosphere, so it`s the same for all the world at any
given time. The solar illumination constant is probably the average value.

14

Thanks, Santiago.

This sounds like a good explanation for this expression in the formula. That's an amazing coincidence that the eccentricity so closely corresponds to the seasons -- within a day! It would have never occured to me.

-Greg

···

From: "Santiago Torres" <[email protected]>
Date: Sat Oct 18, 2003 3:39:53 AM US/Pacific

Just a short idea

I'm referring to the expression for Ext = Esc * (1 + 0.034 * cos

(2pi/365)*(J-2))), which looks like a seasonal correction much like
declination angle to me. I really don't know what it's doing in there.

That`s probably to compensate for the earth`s orbit excentricity. Being an
ellipse, the distance to sun is not allways the same. I think it`s closer
during southern summer, and only a little bit, that`s why the compensation
factor is so small (0.034). Extraterrestrial illuminance is the illuminance
measured outside the atmosphere, so it`s the same for all the world at any
given time. The solar illumination constant is probably the average value.

15

Hi all!

My apologies if this message is much too long ...

Greg was asking about adding the efficacy correction factors into gensky:
as he already noticed this would cause many troubles with old scripts,
so I think that it is better not to make this change. Moreover, as
already pointed out, the sun/sky luminous efficacies do vary with solar
position and there can't be any agreed average standards ... they also
depend on site location ...

I have a few thoughts about gensky and the CIE standard skies that it models.

First I'll give a description of the gensky behaviour and of the
formulas that it uses (as far as I can understand from the gensky.c code),
and then I'll make a short proposal about some integrations that can be
done. I would like to receive some feedback about my ideas.

Part 1: gensky

gensky at the moment behaves this way (I hope not to make any error ...):

1. it can model uniform skies

2. it can model CIE standard clear skies (as described in CIE 110-1994)

3. it can model CIE standard overcast skies (as described in CIE 110-1994 and
   previous standards)

4. it can model non standard intermediate skies, whose definition is contained
   inside src/gen/gensky.c and /src/gen/skybright.cal
   (I've already asked this question 1 year ago: is it possible to give a
   reference about where the intermediate sky definition was taken from?)

If it is given zenith radiance (-b) and solar radiance (-r), gensky uses
directly the CIE formulas or the intermediate sky formula, since all the
sky distribution parameters are already defined, but it is not very usual to
have measured sequences of such weather data (at least not in Italy where
there isn't any IDMP station and sky scanners have only been watched in
photographs ...).

DIRECT CONTRIBUTION from the sun

If we provide gensky with horizontal direct irradiance (ees),
solar radiance (solarbr) is computed with the following formula:

solarbr = ees/(6e-5*sin(sunaltitude))

If we only specify location and time, gensky uses the following formula:

solarbr = 1.5e9/SUNEFFICACY*sin(sunaltitude)
                         if (1.147 - .147/sin(sunaltitude) > .16), otherwise
          1.5e9/SUNEFFICACY*.16
          (could anybody give any reference about this equation?)

where
SUNEFFICACY = 208 /* illuminant B (solar dir.) */

Additionally, if the sky is intermediate, the previous value is multiplied
by 0.15 (i.e the so called /* fudge factor! */)

DIFFUSE CONTRIBUTION from the sky

If neither zenith radiance (zenithbr) nor horizontal diffuse irradiance (eed)
are given, the Krochmann equation is used for the CIE OVERCAST SKY:

zenithbr = 8.6*sin(sunaltitude) + .123 [kcd/m^2] (Krochmann)

CIE 110-1994 also reports equations by Kittler and Nakamura, Oki et al. for this
purpose.

For the CIE CLEAR SKY the LBL equation (Karayel, Navvab, Ne'eman, Selkowitz) is used
(here the Linke Turbidity appears for the first time):

zenithbr = (1.376*turbidity-1.81)*tan(sunaltitude)+0.38 [kcd/m^2]
(Karayel, Navvab, Ne'eman, Selkowitz)

CIE 110-1994 also reports a lot of equations that should be chosen according to
the climate condition in the location of interest (Kittler, Dogniaux, Krochmann,
Liebelt, Gusev, Nagata, Nakamura Oki et al.), but gensky only uses the LBL one.

For the INTERMEDIATE SKY, gensky computes the average of the overcast and CIE clear
zenith radiances.

At the end, the computed resulting value is divided by SKYEFFICACY

D65EFFICACY = 203 /* standard illuminant D65 */
SKYEFFICACY = D65EFFICACY /* skylight */

If the horizontal diffuse irradiance (eed) is provided, then gensky
computes the zenith radiance by using the following formula:

zenithbr = eed/(normfactor*PI)

normfactor = 7/9 for CIE OVERCAST SKY (as we all know by heart!)

while for CIE CLEAR SKY and the INTERMEDIATE SKY, normfactor
is computed by using two different polynomial approximations
(I'd like to have some references about these approximations, too).

For the CIE CLEAR SKY, CIE 110-1994 suggests two different polynomial approximations
by Kittler (as said, "for practical purposes") and by Gusev (for polluted
atmosphere):
both of them are not used by gensky.

Part 2: proposal(s)

A)
As we've seen, gensky is compliant with the CIE standard for the CIE CLEAR and
CIE OVERCAST skies when the user gives solar and zenithal radiances.

When the user only specifies time and location, gensky uses only some of the
equations suggested by CIE. It would be useful to be able to switch between
those equations for practical reasons.
As I've already said, in Italy there is only a few directly measured illuminance
data:
some illuminance data series have been derived by using the Perez model.
Berin and Vio (University of Venice) have shown that the Doginaux equation
approximates very well the Italian luminous conditions, that's why I'm proposing
this integration inside the gensky code ...

B)
There is no standardised intermediate sky luminance distribution model at the
moment, but
CIE 110-1994 cites the Nakamura, Oki et al. intermediate sky (does anybody know
about
recent works of CIE commissions about this topic?)
Zack (hi Zack!) showed interest in being able to use the IESNA (Kittler) sky model.

It would be very useful to have a sort of "plug in" method to insert new sky models,
for instance simply via .cal files (at the moment sky definition is splitted between
a program and a .cal file).

Does anybody else think that this could be useful and feasible?
Probably it would be better to develop another program to generate skies and
leave gensky
as it is now ... and to add a sky library of .cal files to it ...

Any ideas?

Thank you for having the patience of reading it all!

···

--
Francesco Anselmo
[email protected]
[email protected]

-------------------------------------------------
This mail sent through IMP: http://horde.org/imp/

16

Thank you Francesco for summarizing so nicely how gensky generates skies. I
agree that it would be useful to have a method that allows one to pick sky
models via .cal files

This might be particularly interesting since the CIE recently approved a new
"Standard General Sky" Model that replaces all previous CIE Overcast and
Clear Standard Sky definitions. It is generalized from the old CIE Clear Sky
formula (Kittler) and consists of 15 sky types of various luminance
gradations and scattering characteristics - one of them being the new
Standard Overcast Sky, and two others representing the new Standard Clear
Skies (one for low luminance turbidity, one for polluted atmospheres).

I have listed a few links below, if anyone is interested .....

http://www.cie.co.at/publ/abst/s011.html announcement of new 2003 standard
http://www.cie-usnc.org/images/CIE%20DS011_2.pdf copy of the draft standard
(I could not find the final version online - if anyone does please let me
know)
http://www.esim.ca/2002/documents/Proceedings/other2.pdf presentation by
Kittler with information on General Sky Standard (Esim 2002)

- Karen

···

_________________
Karen Carrier
M.ARCH
UC Berkeley

17

"In fact I was referring to the Perez model for luminous efficacy. I should
  have been more clear perhaps.
  I haven`t tried the sky model (nor gendaylit), but I plan to compare results
  between models in the future. If anyone knows of such studies, I would
  really appreciate letting me know.
  Regards,
  
  Santiago."
  
  Vartianinen, E.: "A comparison of luminous efficacy models with illuminance and irradiance measurements", Renewable Energy 20 (2000), pp. 265-277

  Abstract:

  "Several luminous efficacy models have been tested against simultaneous illuminance and irradiance measurements in Helsinki ... The Perez luminous efficacy model had the lowest relative mean square difference of 6.7%. ... For all models, the error was smaller for high solar altitudes but greater for the low altitudes ..."

  Martin

18

Hi all,

The new CIE Standard General Sky is based on the Standard Sky Luminance Distribution of Kittler, Perez and Darula (1997). I created a cal file and technique to model these skies a while ago, and have been using it wherever possible, as I believe it provides more representative sky models than the old, limited, CIE standards. The cal file is not ideal, as it takes a few calculations and a bit of stuffing around to create the models. the difficulty is in setting the relative intensities of sky and sun. but, it does the trick.

If anyone wants it, i can send it and a document describing its use to you. Otherwise, can i upload it somewhere?
Regards,
Phil.

···

_______________________________
Phillip Greenup
Specialist Engineer
Arup Australasia
Level 10, 201 Kent St
Sydney NSW 2000
Ph (02) 9320 9426
Fax (02) 9320 9321

[email protected] 19/10/2003 3:55:48 am >>>

Thank you Francesco for summarizing so nicely how gensky generates skies. I
agree that it would be useful to have a method that allows one to pick sky
models via .cal files

This might be particularly interesting since the CIE recently approved a new
"Standard General Sky" Model that replaces all previous CIE Overcast and
Clear Standard Sky definitions. It is generalized from the old CIE Clear Sky
formula (Kittler) and consists of 15 sky types of various luminance
gradations and scattering characteristics - one of them being the new
Standard Overcast Sky, and two others representing the new Standard Clear
Skies (one for low luminance turbidity, one for polluted atmospheres).

I have listed a few links below, if anyone is interested .....

http://www.cie.co.at/publ/abst/s011.html announcement of new 2003 standard
http://www.cie-usnc.org/images/CIE%20DS011_2.pdf copy of the draft standard
(I could not find the final version online - if anyone does please let me
know)
http://www.esim.ca/2002/documents/Proceedings/other2.pdf presentation by
Kittler with information on General Sky Standard (Esim 2002)

- Karen

_________________
Karen Carrier
M.ARCH
UC Berkeley

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

19

Slight correction to my last post. I don't think solar efficacy is a function of eccentricity, only the extraterrestrial solar irradiation and illumination constants, since a vacuum does not change the efficacy of radiation. But a change in the distance to the sun does change extraterrestrial irradiance and illuminance. Is this correct?

···

--
Zack Rogers
Staff Engineer
Architectural Energy Corporation
2540 Frontier Avenue, Suite 201
Boulder, CO 80301 USA

tel (303)444-4149 ext.235
fax (303)444-4304

20

Hi Phillip and all!

The cal file is not ideal, as it takes a few
calculations and a bit of stuffing around to create the models. the
difficulty is in setting the relative intensities of sky and sun. but, it
does the trick.

That's the power of Radiance:
if you give me a proper .cal file then I'll solve my problem! :wink:

Thanks for sharing it!

···

--
Francesco

21

The Standard Sky Luminance Distribution algorithm, to model the current CIE standard skies, is now available from the Radiance-online web site. The cal file and a pdf describing its use can be downloaded from www.radiance-online.org/patches/ssld/SSLD algorithm with PDF.zip. Thanks heaps to Peter for your help getting that uploaded.

The SSLD became the new CIE standard in Feb this year - Spatial Distribution of Daylight - CIE Standard General Sky, S011/E:2003. To find out more about the skies, follow the link to the paper by Kittler, Perez and Darula. to see a pile of different sky models, go to Geoff Roy's SDF client web site. both of these sites are listed in the pdf attached to the algorithm.

I hope someone finds it useful!
Phil.

···

_______________________________
Phillip Greenup
Specialist Engineer
Arup Australasia
Level 10, 201 Kent St
Sydney NSW 2000
Ph (02) 9320 9426
Fax (02) 9320 9321

[email protected] 21/10/2003 1:46:58 am >>>

Hi Phillip and all!

The cal file is not ideal, as it takes a few
calculations and a bit of stuffing around to create the models. the
difficulty is in setting the relative intensities of sky and sun. but, it
does the trick.

That's the power of Radiance:
if you give me a proper .cal file then I'll solve my problem! :wink:

Thanks for sharing it!

--
Francesco

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general