Tess Perrin:

Welcome, everyone. This is Tess Perrin with the Pacific Northwest National Laboratory. Today's webinar, The Impact of LED Street Lighting on Sky Glow, is brought to you by the U.S. Department of Energy's Solid-State Lighting Program. This webinar is the first event of a two-part series to present and discuss the results of a recently released investigation of LED street lighting's impact on sky glow. A PDF file of today's presentation and the video recording will be available in approximately two weeks for download on the Solid-State Lighting website.

Lastly, we would like to review the responses to the questions we asked upon your registration. So the first question, have these concerns impacted your local system design? We received responses from about half of the registrants. Of these responses, more than half, 55%, acknowledged that concerns had, indeed, influenced their designs. The next question revealed insight that the majority, as you'll see here in the green section, 68% subsequently changed their CCT, 13% noted delays in installation, and 19% had other unspecified effects, which we'd like to investigate further. These impacts probably have more to do with the health concerns brought up by the AMA than sky glow. But these two sets of concerns have much overlap.

The second set of questions address controls as we see them as a major strategy for helping address the kinds of issues we're discussing today. So the first question, are there any wireless controls installed on your local system? 38% responded yes, of the 63% response rate. As to whether or not wireless controls are being considered, as you can see from this next chart, the responses were pretty evenly split. Now, on to the presentation.

Our presenter today is Bruce Kinzey. Bruce has worked in energy efficiency and renewable energy for more than 30 years as a research engineer at PNNL. He joined DOEs Solid-State Lighting team in 2006, originally managing the Gateway Technology Demonstration Program, subsequently developing a focus on street and outdoor lighting applications and associated issues like blue light content in sky glow. He is currently chair of the Illuminating Engineering Society sky glow Calculations Committee.

Bruce Kinzey:

OK. Thanks, Tess. Again, as she mentioned, this is the first of two webinars we're going to be presenting our results from our recent investigation into street lighting’s effect on sky glow. Today's presentation is going to provide sort of a high-level summary of the results. And I'll try to put these into some general context. And then next week's presentation is going to provide more of a deep dive into the modeling that we did in order to produce these results.

OK. So street lighting, blue light, correlated color temperature, or CCT, continue to be the focus of a lot of talk these days. And actually among their sort of — the concerns that are being raised have actually been the focus of discussions among their respective communities: the health community and especially the astronomical community regarding sky glow. They've been talking about this for literally decades. But I think these issues really just entered the public imagination, I think, at large, just over a year ago when the American Medical Association issued this guidance to reduce harm from high-intensity streetlights.

Now, when you have an authoritative body like the AMA issue this rather, you know, stern type of document of course, as you would expect, it was picked up by several media outlets. And we saw lots of discussion about this on the Internet and blogs and so on. There were lots of public comments coming into cities that were announcing that they were planning to convert lights, and so on. Essentially this went viral, and just like in any other situation where that occurs, you see a lot of people running with information that maybe they don't fully understand. There's lots of misperceptions or mischaracterizations of what was actually established and what hasn't been and so on.

Sometimes you just see information that is just plain wrong. The Department of Energy Solid-State Lighting program's position or role ever since I've been involved with it, which is more than 10 years now. One of the roles anyway has been to act as an independent third-party provider of objective information. There's always been lots of hype, lots of claims being made that are maybe unsubstantiated or you don't have any way to easily compare them and so on.

Our position has always been to try to provide the best information that's available today to enable decision makers, people that need to make a decision today, give them the best information that's available for them to move forward. And these current issues that we're talking today, I think, fall right in line with that sort of existing and ongoing role. On first glance, sky glow and health issues might seem to be a little bit separate. But, of course, there's actually a lot of overlap between them just from the get-go. They're both related to light at night. There's a lot of the same kind of issues, discussion about blue light content and so on that's related to both of these.

And also some of the deficiencies that I'm going to talk about occur in sort of both spheres of discussion. So I'm actually going to be presenting a few of the same slides I've presented before, for people that have seen me present before. Whereas, before I was referring more to the — responding to the health issues that have been raised, today I'll be tying these into more of a sky glow context. So to begin, what is — let's just come up with a working definition of what is anthropogenic sky glow. And I put that word "anthropogenic" in there to distinguish it from natural sources of sky glow. And you can see a little bit of both in this photograph.

I think it's pretty well understood we're talking about this sort of increase in night brightness that creates sort of a light dome over a populated area, like you can see here in the lower right. This is an amalgamation of virtually all light at night that's being exposed to the night environment that winds up in the night sky. It all gets mixed up in there together, and it ends up creating this kind of a light dome that serves to obscure the view of the night sky, of stars and other celestial objects, and so on.

You can also see some natural sources here. There's a little bit of the Milky Way visible there in the middle of the photograph. And there's other natural sources like reflection off space dust and the moon and other kinds of things like that. But what we're talking about here is this kind of light dome that I think we're pretty familiar with. This is a time-lapsed photograph, so of course it doesn't quite look quite as dramatic. But I think we're all familiar with this concept. This is what we're talking about today.

And not this issue, which is glare, which causes some of the similar problems or issues. This is not what we modeled and not the focus of today's discussion. All right, so just kind of moving forward with a general working definition. This isn't from any official source, this is just something I put together. It's an increase in human-induced night brightness resulting from use of supplemental illumination for any purpose. Now that's probably not perfect, but I like it because it's all-inclusive.

Just like the photograph that we see up there, that light dome. Like I said that's all — that's kind of the sum of all the light sources contributing. This definition is also similarly wide and accommodates things like vehicular lighting or building lighting escaping through windows or someone walking around with a flashlight or whatever, what have you. All of that light went into the night sky and contributes to this kind of a light dome that you see in the photo.

All right. So now just for purposes of background, I want to step to some of the deficiencies in the ongoing discussion to kind of bring you up to speed as to why we got involved with this in the first place. So there seems to be this very singular focus on correlated color temperature, or CCT. Everybody has probably heard this stuff, don't install anything over 3,000 K, for example. Color temperature really only describes the appearance of a light source. Like you see in the photographs there, that's the same classroom lighted with a 3000 K color temperature on the left and a 5,000 K color temperature on the right.

Often you don't even get that much detail when you’re looking at products by manufacturers. They're describing them, this is a soft white product, or this is a day light color product or so on. But even at the actual value there, like the 3,000 K or 5,000 K, these are essentially results from a weighted average of the actual spectral content emitted by a source, multiplied by the relative powers in each wavelength. And you add all those up and you end up with this sort of final score if you will, this nominal color temperature like you see here.

The issue here is that, like any weighted average, it actually doesn't tell you very much detail about what the actual combination of wavelengths are. There's virtually an infinite number of wavelengths and corresponding power levels that could — you could put together to produce this same final result. And I'll have some examples of this in a little bit. And here's my example, here's three products that are all listed at a nominal 4,000 K color temperature. There's a metal halide on the top, that's a fluorescent underneath, and a 4,000 K LED there on the right.

If we were looking at corresponding photos of this classroom, for example, each illuminated by these sources, you probably would not be able to distinguish among them. You wouldn't be able to tell the difference. And, yet, if you look at the actual spectral content, which these graphics are depicting here, you can see some pretty significant differences between them. The whole crux of this issue is that the concerns that are being raised are not caused by colors, they are caused by spectrum or spectra, or individual bands of spectra. And, again, you just don't have enough detail about what the actual content is of these sources in order to really sufficiently describe either the impact or to talk about means of mitigating them.

Similarly, this exclusive focus on blue is also fairly misdirected. And I'm sure everyone's familiar with this. We hear about these blue rich LEDs, and this harsh blue light, glaring light, blue this or that. Both of these terms, CCT and blue, are really just shorthand forms that are being used to describe often a much broader range of wavelengths.

I understand that the purpose in that it's easier to talk in those terms than to actually talk about SPD. But the problem is that they're just not very precise. Here we have a depiction of, I think, a pretty well-known scotopic luminosity function. This is reflecting the fact that our eye does not see all wavelengths equally across the visible spectrum, in low lighting conditions especially, what we call scotopic lighting conditions. And you can actually see here what the relative sensitivity of the human eye is to these different wavelengths.

I put a color bar across the top there just to help sort of illustrate, kind of put us on the map here of where we are in terms of which wavelengths we're talking about. You don't actually see much color down at these levels because you're mostly seeing with the rods in your eyes. But, for example, if you were in that light zone, that we saw a photo of earlier, that's a mixture of all kinds of different spectrum light and so on. The light at — in this case the peak is at about 507 nanometers.

So light at that wavelength that was in that light zone would have, for example, probably the greatest ability to obscure the view of the night sky to a naked human eye, to someone standing out in their backyard in that populated area right at 507 would probably have the most obscuring ability of any of the wavelengths out there. Now, the flip side of that is that's because you see best at that wavelength. Your eye is most receptive to that.

So at the same time if you're a driver on the street, right at 507 nanometers basically is what provides you the best vision at those low lighting levels. So seeing a rock in the road or a pothole or something like that. So there's always two sides to every story. But today we're just going to focus on basically the ability of these different wavelengths to obscure the view of the night sky. OK. Now, what you may have noticed about this function is that it extends across a much broader range of wavelengths than just blue.

And even a generous definition, and this was another issue there isn't an actual official definition of blue. Often it's just defined for the context in which it's being used. But even a generous definition of blue covers less than half of that sensitivity function. And it doesn't even cover the peak wavelengths, as you can see there. Furthermore, you're going to see on the next slide we have a cable that has a column that's labeled % blue.

We've calculated those values using an astronomical source that was really based on the ability of those different wavelengths to affect sky glow. See, they're labeling — that column is the labeled % blue, but actually this extends way beyond any standard definition, and that's part of the issue. So the bottom line in all of this is to really accurately describe these problems or to talk about mitigating them we really need to talk in terms of the actual spectral content of the different products. And you're going to see several examples of this as we go forward.

So here's the table I just mentioned. This was issued in its original form in direct response to the AMA release that came out just over a year ago. That release was really focused on street lighting and also on LEDs in particular. And so we issued this table as sort of like, hey, wait a minute. You know, these issues that are being raised are shared in all broad spectrum sources. So here on the table we left several sources for different lighting applications, this includes both exterior as well as the interior applications.

We have broad spectrum sources, or white light sources, as well as some narrow band sources. These are all normalized for lumen output. That means these are all apples-to-apples comparisons in this table. Then you see the next column lists a variety of correlated color temperatures. And that's particularly relevant for the eight different LEDs listed at different CCTs there. Then you see that column I just mentioned, that % blue.

Now, in its original form, I think there were — the total number of LED products that underlied the values calculated that are reported here on the table. I think there was a total of 76 products, 76 LED products underlied those eight different correlated color temperatures. Just recently we have undated this table to where now there are more than 450 real products, spectral power distributions underlying these values. So I think that we're actually — now we're pretty representative of the type of products that you would find out on the market today.

I don't know that all of these SPDs would still be available, but they were actually listed as real SPDs at the time that data was taken. Here's how it actually breaks down per color temperature. So that count, that just appeared there is the actual number of products underlying each of the ranges that you see in the table. The reason there's a range of those last three columns is this is reflecting the fact that they're far from being a monolithic product or monolithic technology. LEDs come with any number of variations in the actual spectral content.

And because you have different spectral content that are still adding up to that, like I was talking about earlier, that weighted average is still being rated at the nominal color temperature. They actually have really some pretty high variation in that spectral content. And that's what you're seeing the results of here. OK. So this is a relative table where we're speaking in the context of street lighting. The 800-pound gorilla there, even today continues to be high-pressure sodium. That's by far the greatest incumbent technology.

If you look back on or think back on scotopic function that covered a very wide range of wavelengths, including several that are produced by high-pressure sodium. So it certainly has some ability to obscure the view of the night sky. So we're — in this table we've normalized the high-pressure sodium. So we're saying that value is one. And then all of the other products in this table are being recorded relative to high-pressure sodium. OK?

So let me give you a couple of examples and I think that will clarify. So if we look at the LED products at nominal 4,000 K in this table, if we use that for an indicator I spoke of before, they have a relative ability to obscure view of the night sky of about 2.1 to about 2.8 times that of high-pressure sodium. OK. I hope that's clear. Let's go to a second example, at 2700 K the products in the table show a relative ability of about 1.7 to about 2.3 times that of high-pressure sodium. If you just look at those two ranges, you see there's a significant amount of overlap between 2.1 to 2.3. What that means is that there are some LED products at 4000 K that have less scotopic or lower scotopic potential than some products at 2700 K.

If you go back over to the left there and look at that count again, between those two CCTs we have about 110 products total. So we're not talking about one or two oddball products in there. I don't know what the actual number is, but there's probably at least a dozen that fall into that overlap zone. So if you're trying to follow some simplified advice like, well, you can solve this problem by not installing anything over 3000 K, that is not guaranteeing that you're actually going to achieve that goal.

You really need to understand what the specific spectral content is of the products you're considering if that's your — and I'm ignoring the fact that there also are upsides, like we talked about earlier, to having scotopic content in there. We'll ignore that, and we're just focusing on sky glow potential here. Right, and I also using this table, I always like to pull out everyone's favorite old-timey technology, incandescent, listed at about 2,800 nominal CCT. It's corresponding scotopic potential is about 2.2 times that of high-pressure sodium, which puts it firmly in the middle of that overlap zone of between 2,700 and 4,000 K LED products.

So you might wonder, that might come as a surprise, this is not something you normally hear. You could actually have more of an issue using incandescent than using some 4000 K LED products, right? You might wonder how that's possible. I hope to clarify that with this somewhat less than clear graphic. This is actually showing the spectral power distribution for three different products. These are all nominal 2,800 K. There are two LEDs there, as shown in the blue, and red is an incandescent. These are all normalized for lumen outputs so again, apples-to-apples comparisons here. If you recall the scotopic function, and we superimposed that curve over the top here.

Basically the scotopic potential here is the weighted sum of the area under each of those curves. And it's weighted by where it falls on that scotopic efficiency function depicted there. So if you just look at the area of maximum sensitivity of the human eye and you follow that down, you see the LEDs are actually sort of recovering from a relative toss there, if you will. Whereas the incandescent being almost linear has a much higher value through most of those wavelengths of maximum sensitivity.

So I think just mathematically looking at this I think you can understand how an incandescent would have potentially more scotopic content than LED products at similar color temperatures, and even some higher color temperatures like we just saw in the table. OK. So up to now I've been talking about some of the deficiencies that we saw with the way color is being treated in the public discourse on these issues. One of the other issues that we saw was that there was a complete lack of or general lack of discussion of other factors that we know are critical in terms of contributing to what occurs at an actual location during an actual conversion.

We saw this missing from the discussion, and this is really what kind of got us into it in the first place. We decided, you know, DOE should investigate, we should run our own investigation of this to try to get some insight on what's actually happening out there during a typical conversion in the United States. So I want to cover a couple of topics now. The first of these is the greatly improved control over the distribution of light that's produced by LED products relative to the previous incumbents.

I have a couple of, you know, very well-worn photos I'm going to show here. I've seen these lots out in the media, I've used them before myself. This is Hoover Street in Los Angeles near Hollywood. I think everybody knows Los Angeles has largely converted its streetlights over to LED by this point. This is obviously the pre-installation photo. High-pressure sodium, like just about any, I guess, glass lamp based technology is what we call an omnidirectional emitter.

So light is coming out of that glass lamp in virtually every direction. You place that lamp inside a fixture, and this case these are cobraheads. You can't see them, but these are cobrahead fixtures whose partial duty anyway is to reshape that omnidirectional emission, is to fold that light back around and send it back down towards a rectangular target that runs underneath the light poles, right? That road surface, this rectangular target they're trying to take that light that's initially going out, and this is more than half of the light if you think about it.

More than half the light starts out going in the wrong direction and they're trying to reshape that and send it back to that rectangular target. Despite having great success or great scale in this regard, there's only so far they can take this, especially if you're talking about trying to remain cost-effective here. And you can see the results for the imperfections in these areas that I've highlighted here. You can see these, what we call hotpots, these areas of very high excess illumination hitting areas of the road.

You see other un-uniformity between the distribution across that road surface and all. This is light that you're paying for, you're using energy to produce, you're also paying money obviously to produce it. But it's not doing any good, and in fact it's really only causing problems. This light is reflecting off the pavement, going up into the atmosphere. Furthermore, we're above these lights and we're looking down on them. So everything we see here, the fact that you can see those lights this is — everything you can see is uncontrolled uplight or above the horizontal.

And it turns out this is about the most offensive angle for light in terms of resulting sky glow, because this light is traveling the farthest distance through the atmosphere. and, again, scattering off, you know, air particles or dust or what have you, causing sky glow issues before, you know, it ultimately dissipates. Now, if we move to the post conversion you see a pretty dramatic difference here between these two photographs. So not only have a lot of those hotspots, you know, representing inefficiency. You’ve seen those disappear. Also these are, of course, zero percent uplight luminaires.

So you barely can pick them out of this photo. If they weren't actually — if there wasn't some reflection off the supporting poles there you probably would have trouble even finding these luminaires in the photograph. Collectively, once you get rid of all this other that I just pointed out here on the left, it basically represents waste. It's light, again, that you're producing and it's not doing any work for you. If you can get rid of that, like you see on the right-hand side here. now this isn't the result of any official survey, this is more anecdotal. But I've worked with a lot of cities that are converting their lights. I've seen a lot of results.

It seems to be a very typical result during a conversion is that the replacement LEDs are putting out something between 40% to 60% less light, less lumens, fewer lumens, than the high-pressure sodium systems that they are replacing, 40% to 60%. And that was certainly the case here. We're able to meet, LA is able to meet the minimum illumination requirement with less than 50% of the original lumens. So the photo on the right, has fixtures there, they're putting out less than 50% less than half of the light that you see in the left-hand photo. I mean, of course, this has significant ramifications for resulting sky glow, which we'll talk about a little bit more. Right now I just want you to keep that 50% reduction figure in mind. This is how getting rid of that uplight looks from the air.

Now this isn't a photo of Los Angeles; this is Portland, Oregon, where I live. You might know that Portland converted its street lighting to 4,000 K LED a couple of years ago. This is a photo taken by my colleague Tess, as a matter of fact, on an early morning flight out of Portland. Looking down on some residential areas here that have been converted to LED street lighting. And you can just pick out the little points of light there. I suspect these are 0% uplight luminaires. So what we're seeing here, those little points of light, are actually the reflection off the supporting poles, just like we saw in that photograph of Los Angeles.

These are not — this is not direct uplight. In fact, you know, if you look here carefully, you might also pick out some collector roads here where you also cannot directly see the luminaires, but instead you can just faintly see some reflected sort of splotches of light under each pole that's being reflected off the road surface. So there is a little bit of this light winding up in the night sky. We are looking at it from the night sky.

So there's a little bit of light reaching us up here from this aircraft, but compare that with the incumbents that it's surrounded by. You know, there's some big commercial area along the left-hand side. You can count every single light there. If you go over maybe say 3 o'clock over there you see there are a couple of parking lots there, what look like probably metal halide parking lots. Above that is a building that's surrounded by wall packs. And if you look up even above that up in the upper right-hand corner, you see there's probably some kind of shopping center or something that looks like those are probably post top lights. Given their close spacing those are some incumbent HPS lights.

You know, what's clear here is that the contribution to the sky glow coming from this area, these streetlights, having been converted, are actually a minor contributor at this point. So I like to ask the question, I mean, how much more emphasis should Portland have placed on these? As I said they installed 4,000 K. Should they have gone to 3,000 K? From the streetlighting — or from a sky glow perspective, you know, I'd say these are down in the noise. But I'll just leave that question hanging out there for you to decide.

OK. So finally now, I want to start talking about our modeling effort. I'm not going to spend any time talking about the model. Again, this is the focus of Tess' presentation next week. But ultimately we arrived at the use of this sky glow simulator, that was developed by Miroslav Kocifaj. One of the features of this model that was particularly attractive to us was the ability to tweak the different characteristics of the lighting fixtures, like I've just been talking about. The output for fixtures, the percent uplight associated with those fixtures, as well as the spectral power distribution.

We came up with a number of scenarios. There's a lot of variables here that we were testing. And I'll be describing some of these in more detail in just a minute. But one thing I want to point out, we tested 11 spectral power distributions. Only seven of those actually corresponded to LED products. We also had some incumbents in there. We also ran a equal energy spectrum across the entire visual spectrum in five nanometer increments so that we could test the sensitivity of sky glow to spectrum at that granular level.

So one scenario is basically one setting on all of these variables. We run a model, then we intermit one value, one of these variables, one value, that's scenarios, run the model again, etcetera. We stepped through all of these variables. So in the end, we have a scenario or a model run for every single combination of every one of these variables. it may not look like that much, but as you see at the bottom, you add all these up, it's kind of a factorial thing going. You end up with more than 200,000 runs of this model. And that's what we're reporting the results from here.

One other thing I wanted to mention was that during the course of this, now we pretend that the typical incumbent product that's still installed out there, like we've seen here in a couple of the photographs, is a high-pressure sodium, dropped lens fixture. So it has glass that appears below the horizontal, right below the fixture, you can see the glass on the fixture. And that has some off-white characteristics. A lot of those incumbent products out there are probably 25 or 30 or 50 years old by now. But we went out and did just kind of a cursory survey of 10 products currently available, modern products available from major manufacturers.

We found a fair amount of variability with respect to their actual uplight percentages, but we found an average of about 2.4%. So we have modeled 2.2%, as shows up in the list there. And this is what we're calling as a typical incumbent product out there from a typical conversion. OK. I'll talk a little bit more about in a moment. But first, I'm compelled to say a few words, add a few caveats and qualifiers here about modeling.

I think this one is very well-known to everybody, basically garbage in, garbage out. I don't think I need to describe that any further. All models, but especially atmospheric models, incorporate many, many assumptions and simplifications that we all experience. We've been watching the weather forecast for a week, Saturday it's going to be sunny. It's going to be sunny. We went to the ballpark on Saturday, and it rained. You know, because there's just too many variables in the atmosphere. Even with these fancy weather forecasting models and decades of experience, and they even have real time data coming from other locations. They still can't get it right 100% of the time.

We are leagues away from those kind of abilities with our sky glow modeling. So I hope we can all agree that we're just ballparking the results here, trying to gain some insight on what occurs during a typical conversion of high-pressure sodium to LED products. Similarly, you know, we can't model every city. They haven't even come close to modeling every city in the United States. So if you try to take these results and apply them to your local situation or some other city you're familiar with, you'll say, oh, wait. these don't fit because that assumption doesn't match what the situation was here. You know, your mileage is definitely going to vary. Again, we were just trying to look at a typical conversion in the United States.

Now, if it seems I'm belaboring this point it's because we've received a couple comments to the published report saying that, hey, there are a lot of 0% uplight high-pressure sodium overheads that have already been installed. So you're really overstating the impacts that are occurring during a conversion to LED. It's true, there has been a lot of these products installed. But we will contend that as a portion of the overall population, they remain in the distinct minority. And, in fact, we will contend that the typical installed product is more represented like what you see here in the general Chicago area.

Now, most cities are not going to look like Chicago, they don't have the same kind of lights, the same number of lights that Chicago has. But, again, we will contend that this is more representative of the existing inventory. OK. So now one more photograph. Now, this is a photograph of Los Angeles. I took this one myself just this May as I was flying in. Again, they've converted most of their streetlights to LED. Again, we're looking down in some residential areas here where you can just see little points of light in the foreground.

Again, I think this is — these are 0% uplight pictures, so what we're really seeing is reflection off of these supporting poles. But if you look right in the middle of this photograph, now you see there are some parking lots there where, in fact, there have been some already conversions to 0% uplight, high-pressure sodium. You can no longer see the individual fixtures but what you can see is the little splotches of light reflecting off of the pavement. And, in fact, there are also some — looks like some major roads leading up to those parking lots coming from the right where the same situation has occurred.

So there are — in fact, there are a number of installations out there where 0% uplight has been installed. But if you go further up in this photograph, up and to the right, kind of the right-hand corner, you will see an older apparently residential neighborhood where not 0% uplight fixtures have been installed. They've not been converted, so those remain. And so, again, we're going continue to argue that that represents more of the existing inventory.

But the good news here is — let's focus on the prize. Whether or not we're sort of arguing over where do you get credit to this, but the bottom line is that these lights are being converted and getting rid of that uplight is having this tremendous effect on reducing the previous contributions to sky glow from street lighting systems across the United States. Whether that occurs, you know, from HPS to another HPS or from HPS to LED the good news is, in the end, we have greatly addressed this previously large contributor to sky glow. So I think this is all good news in the end. So that's what I prefer to focus on.

OK. So the results that we're about to see now, again, we've ran 200,000 runs. You see literally the result of thousands of runs are combined into these very few charts. They include impacts of changing the spectral power distribution, the light output, uplight for near and distant observers. And what we're actually measuring, what these are based on, if you think about we make some changes to the streetlighting system, there is some change, an increase or decrease to the sky glow produced. That sky glow, in turn, throws a little bit of illumination on the ground on a horizontal position.

And what the model is measuring is that change in horizontal illuminance at that observer position. Of course, we're talking about tiny, tiny amounts here. So, again, we put these results into relative tables. So we're comparing against what was there under the incumbent system, the incumbent high-pressure sodium system, and then how has that changed. and it's a relative number. So this is, you know, 0.5 is the 50% increase or whatever or 50% decrease or what have you.

OK. There's two observer positions, one is located right on the perimeter of the city, basically inside the city but right on the edge of it. The other observer position is 40 kilometers, or about 25 miles, outside of the city center. We're calling that the distance observer. Finally, we're displaying these both in terms of unweighted results, just kind of the raw results. And then we are scotopically weighting those values, again, to get an indication of their relative ability to obscure views of the night sky to somebody who was just standing there looking up at night sky.

OK. So, finally, after all of that, this represents the change that occurs, or the impact of simply substituting the LED SPDs for the high-pressure sodium. I mentioned we did 11, we had a couple of incumbents in there. We're not reporting those here. This is just the LED products that we're looking at results from here. The baseline in this case is depicted, in all of these charts, is depicted as this red line going across. At one point though, you'll also see there, a little box that I highlighted, what currently comprises the base case here.

So reading from the bottom we're looking at LEDs with 0% uplight and 100% output versus high-pressure sodium with 0% uplight and 100% output. So this is substituting the spectral power distribution holding all else constant. And this happens to be at the near observer, so this position for this is right on the edge of the city. So it turns out that the increased short wavelengths content of LED products does contribute to an increase in sky glow. And it's across the board. They're both in unweighted, as shown on the left, and scotopically weighted shown on the right.

However, you can see the initial point that I was making that the correlated color temperature is not a very necessarily reliable predictor of that impact. These are listed in order of increasing CCT. And as you look at the unweighted results you actually seeing the first three there are decreasing. And then you see some overlap between the next ones. It's just fairly — it's not a very reliable indicator of what the — or predictor of what the impact is going to be. Now, if you look at the scotopically weighted we're looking at say the ability to obscure the night sky. Now you have a much stronger correlation between CCT and that effect.

This is for the near observer, again, so this is for the person standing in the city. Now, if we move out to that distant observer position, about 25 miles out of the city the biggest change we see to these results is an increase in the variability of the range that's impacted here. And that's because the light is traveling a much greater distance through the atmosphere so you're getting other things in the atmosphere, aerosols, like dust particles and water droplets and smog and other sorts of things like that.

You're giving it more opportunity to influence the results. So we just see an increase in the variability of the range of results there. That's the main change that we see here. But now these two charts this reflects most of the early discussion, when this first came out right after AMA, we saw a lot of this discussion out there. They were really looking at — most of the discussion was just revolving around looking at the change in spectral power distribution, a substitution of LED spectral power distribution, or color temperature is how it was usually described, for high-pressure sodium color temperature. It's not that these results are incorrect. You can see the results right here, it's that they are incomplete.

It doesn't reflect the full complement of actions or results that occur as a result of a typical conversion of streetlighting in the United States, as you see me talking about. So now we're going to go and look at the impacts of those other factors. So now I've added in this 50% reduction in output for the LEDs, which you've seen as being typical, we're calling typical. This has a scalar impact in terms of reducing the sky glow. If you reduce the light output by 50% you get a 50% reduction in that contribution to sky glow. I think that makes intrinsic sense. So now you see in terms of the unweighted result.

All of the LED products across the board are now producing less sky glow than the high-pressure sodium system baseline system that they replaced. However, if you scotopically weight these, again, some of these LED products, just barely, but they do reduce sky glow relative to the baseline. Others corresponding to higher CCT still do have a higher potential impact for obscuring the night sky, although increment has greatly reduced. Again, this has been cut in half now. And, again, this is the observer in the near position.

If we go out now to the distant observer with the same conditions, 50% reduction in output, again, the major impact we see here is just a greater increase in the range of variability among those results. OK. But wait there's more. Now, if we look at — if you recall this 2% uplight characteristic, which we have now added to the baseline. So we have changed the baseline now in these last two charts. Everything with the LED is still the same: 0% uplight, 50% output. But now the high-pressure sodium has a 2% uplight characteristic and it's still at 100% output.

So you see, again, unweighted results across the board for all the LED products are pretty significant reductions in the sky glow of that result. However, if you scotopically weight, again, you see that some of them result in higher sky glow while others reduce the sky glow. And I think this probably fits with the photos that we were looking at. As you recall, there was a little bit of light still winding up in the night sky from the reflection of those, off the road surface and off the poles and such. You still have a little bit of that light going into the local sky, and that's where this person is located right near the light. So they are still seeing that little bit of interference from that light.

But now the real punchline comes as soon as you get outside of the city, now we're 25 miles out of the city. That little bit of uplight being reflected from the street just doesn't travel very far. And now we see deductions of 95% or more across the board, all of the LED products. So kind of the take away from this, I want to point out for people in the city the biggest hit comes, according to our results here, the biggest hit comes from reducing the output from it on a per-fixture basis.

So you get this big change typically in a typical conversion, but also this really supports the installation of a dimming control system. Because if you're a city like Cambridge, Massachusetts, for example, that dims their lights by 50% after midnight. They're getting a direct additional 50% reduction on the contribution to sky glow coming from their street lighting system. OK. But to make the maximum impact for a distant observer, the big hit comes from getting rid of that uplight, as you can see here. Now does this mean that somebody in, for example, Chicago who is getting ready to undergo a major conversion of about 300,000 of their lights, and that's supposed to be taking place over the next four years. When they've completed that conversion does that mean that something looking back towards the city, like that photograph we saw at the very beginning, does that mean there's not going to be like a light dome over Chicago?

Well, of course not, because streetlighting is only one component of sky glow. I threw up a couple of pie charts out of the paper by a noted astronomical researcher, Christian Luginbuhl, back in 2009. Looking at Flagstaff, Flagstaff is not typical, by no means a typical U.S. city. They've installed a lot of low-pressure sodium there. They've also had this concerted effort to preserve their night sky for years now, for decades literally. But what I really like about this is that, as far as I know, it's the only set of actual estimates of uplight coming from these different applications based on an actual inventory in the city.

So this is the best data that I know of out there. Even though this isn't representative, I think it still supports the main point. You can see here that roadway lighting or streetlighting goes somewhere between 8% and 12%, depending on apparently there's some big sports field there, or there was in 2009, that comprised fully a third of the uplight being emitted in that city by itself when that lighting was on. When that lighting is turned off then the dominant source of uplight in Flagstaff in 2009 was commercial lighting, which is producing more than half. And so if you recall from that picture of Portland, we saw that big commercial area.

This probably includes parking lot lighting and signage lights going out from the front of the windows and so on. But what all this means is that if we decide that these issues like sky glow is something that we really as a society want to address, we are going to have to invoke some sort of collective action across really our entire economy. All these sectors are contributing. Again, that light dome occurs as an amalgamation of all the different light sources.

We're really going to have to take some actions across the board if we really want to achieve any significant reductions. And that goes for the health impacts as well that have been raised. So I'll just leave you with this photograph of Chicago. Again, you know, they're going to be replacing a large portion of their streetlighting. I don't know exactly how far out towards the horizon their purview extends. I don't know how many of those lights are going to be changed, but we can expect to see a pretty big change in this view of Chicago, say four years from now. But in the meantime, in the foreground, you wouldn't really expect any change without some other type of concerted effort. So we have a lot of work ahead of us.

OK. So here I've listed the URL that you can find the posted sky glow report if you want to read the whole thing. We also have several other materials posted there. These were mostly produced in response to the AMA issues. But, again, there's a lot of overlap, So you find probably if you poke around in there you'd find some other information of interest as well. I also want to put out a reminder for next week's presentation webinar by Tess that will be doing a deeper dive into the actual modeling effort. You can register for that on the same website that you registered for this webinar. OK. With that let me turn it over to Tess. Hopefully we've gotten some good questions. Tess.

Tess Perrin:

Thank-you, Bruce. We have received several questions concerning our presentation. And we're planning to leave the lines open here for the next half hour or so to get through as many of them as we can. So perhaps we'll start with this first question. In this study and issue, a single reduced level of luminance was used for all the various LEDs regardless of spectrum. If the purpose of the setting was to determine the optimum approach to street lighting, it would seem necessary to address each SPD luminance to account for visual efficiency. So why was only one reduced luminance level used? So, Bruce, I'll take the first one. So in our study, while nothing below a 50% reduction in light output was modeled, as Bruce noted, sky glow directly scales up or down with light output.

So for example, during certain hours of the evening when lower light levels could be acceptable depending on safety or other criteria, then the lighting system could be dimmed down to any level. It could even be turned off, which would then further decrease any sky glow contributions. But, the data generated by this report could indeed be further adjusted to account for visual efficiency given the linear relationship with light output.

You could essentially just take the relative impact value and then scale it for whatever reduction or increase you'd like to add to that SPD for visual efficiency. Bruce, if you're ready for a question, in the conclusion of the paper regarding improving sky glow conditions, can you provide us with documentation of new or retrofit installations that recommend a 50% reduction in lumen output?

Bruce Kinzey:

OK. So there are lots of case studies and so on available out there on the web. It's just an easy web search away. You can find lots of examples and you can look at the before and after information. We never recommend a 50% reduction. Basically this is driven on an individual basis. So most sites I'd say are using a rendering software. There's one, a big — I don't do a product endorsement here, but there's one big one that everybody uses. you know, you're basically drawing up a lighting design.

What's driving that is meeting the minimum required illumination everywhere in the target space. And I see there's another question later that kind of referred to this as well. And that's what's really driving this. It doesn't really — we're not focused on average lighting levels. This gets more complicating than I can really explain during this session. But when you have hot spots like you saw in that one photograph, that is driving up the overall average in the area but that is not doing any good in terms of lighting.

If you really need to just meet average or minimum lighting level plus a uniformity type criteria. You can meet those much more easily with LED products that are putting out — and, again, this isn't — we didn't go out and do an official survey across different sites, it's more anecdotal. It's just I work in numerous cities, numerous cities that have been undergoing conversions. And this seems to be not always the case, it's not going to be the case in every single situation, but it's frequently the case that they are putting in products that produce 50% on a per-lumen basis, per fixture basis 50% or less lumens. And actually typically it runs between like 40-60% is what I said.

Tess Perrin:

Thanks, Bruce. Thank-you so much, Tess. Yeah. With the Portland, Oregon 0% uplight slide, how much of the different things that we see is due to differences in average illumination that make some areas appear brighter versus the impact of using luminaires with no uplighting?

Bruce Kinzey:

You know, it's tough to answer that because we're just looking at a photograph taken as someone was flying over. You actually were — I should throw that question beck to you, Tess, since you took that photograph. You know, it's pretty clear when you look at the photographs. It's pretty clear when you've got a site that has 0% uplight versus ones that are emitting directly. I mean, there's really no question between them. The luminaires are very visible in those situations. I can't really comment about average illumination levels or anything. I can't tell you exactly where that photograph was taken, obviously. I just know that happened to be right after Tess had taken off, just right here in the Portland area somewhere.

But, you know, I think the results there — I guess the results basically speak for themselves. But I can't specifically get into further detail because I don't know what that is.

Tess Perrin:

OK. How do we explain the dominant role of uplight relative to reflected light when the issue is sky glow at a distance? So, Bruce, I can start to answer this one.

Bruce Kinzey:

OK.

Tess Perrin:

So as modeled in our study, and as Bruce clearly explained, reducing uplight really has the greatest potential to decease sky glow for just an observer. Actually, reducing uplight reduces sky glow under almost all conditions but only one. As Bruce showed earlier, that it doesn't is under the cloudy conditions when the observer is in the city. So I guess getting into the nuances of this question regarding reflective light, reflective light is — and we'll get into this next week, but the reflected light is emitted based on the cosine function. So if there's no uplight then the reflective light is the only contribution to sky glow. But light that reaches different observers have to fall within that solid angle of observation.

So it really is essentially emitted at low elevation angles, otherwise it won't reach an observer. So there's a greater prominence of these angles when there is uplight compared to reflective light. So that's why you'll see in our conclusion there's that dominant impact of reducing uplight when there's an observer at the distance. I hope that makes sense, but we'll get into this more next week.

Bruce Kinzey:

But, yeah, and I pointed out in that picture of Los Angeles where we were just above the horizontal looking at those lights that is really the most offensive contributor to this sky glow light that's being emitted just above the horizontal there. And when you have direct we're talking about the intensity of that light is much, much higher than what's being reflected. We modeled an average 15%, again, not to steal Tess' thunder for next week, we modeled a 15% albedo reflectance from the average ground area.

And so you're already cutting the — if you just take those two and look at direct light versus reflected light you're only looking at 15% of the intensity if you're talking about reflected light. Then when you spread that light out over this cosine function, like Tess is saying, you wind up with a lot less light going in that most offensive direction compared to the direct light that is emitted in that direction. So I think hopefully that helps explain why that is so much more critical in terms of getting rid of that corresponding uplight is so much more effective in reducing sky glow at distance.

So sky glow —

You want to give me another one, Tess?

Tess Perrin:

Staying with the topic of uplight, one of the questions is have you run the model looking at different levels of uplight in LED luminaires?

Bruce Kinzey:

No. Right? This is really a question for you, but they're virtually all — unless we're talking about a post top fixture like some kind of one of these sort of corncob type things that's replacing an existing lamp in a globe or a decorative fixture. Virtually all of the products now, all of the cobrahead replacement products, I can't even think of a single one that does not have a 0% uplight characteristic. So unless somebody is tilting them, which we have seen actually. Sometimes they're trying to throw the light out farther across the roadway or something.

You see this in very few instances, but occasionally. Unless you're titling it or something like that they're always 0% uplight. And in fact, I think a lot of this earlier comment that I talked about where a lot of communities started replacing even their high-pressure sodium. A lot of communities have adopted that some years ago that any new installations, even the high-pressure sodium, will be 0% uplight, for a variety of reasons. It also helps, I think, reduce the glare probably, but also helps preserve night sky, and for other reasons. That seems to be more of a modern standard now, and virtually all the LED products now are 0% uplight.

Tess Perrin:

OK. Here's another one. In Rio de Janeiro, where one of our webinar participants lives, it's impossible to use dimming after midnight due to violence issues. So what's the alternative? Can we consider correlated color temperature more than other developed countries?

Bruce Kinzey:

Well, yes. I mean, ultimately, I don't want any — I don't want to imply that we're saying do this and don't do that. What we want, our whole role in this is to try to put all the information on the table so that a user, a decision maker, can move forward with what's going to be best for them. We don't want to tell them what's best for them. We just want them to have a full set of information at their fingertips for making their decision.

They may go ahead — you know, we don't have any problem if sites say, we're going to install 2,700 across the city anyway. We've looked at all the evidence, and we decided that's the best approach for us. That's great, but what we want to prevent is people saying, well, we want to try to minimize, you know, this issue that's been brought up. So we think that putting up 2,700 K is what's going to solve the issue. That's not necessarily going to achieve that result. And so, ultimately, like I said, if they decide to go that way anyway we're not going to try to argue with them or anything. We just want to have all the information on the table so that they make the best educated decision that they're able to based on what we know today.

So I guess following up on that specific question, if they want to say keep illuminance levels a little bit higher, but you want to go with a warmer color temperature because that's what fits best in Rio. And I'd say, you know, go for it. We're ecstatic that you're actually replacing those. You're still getting a lot of the other benefits that come along with the conversion to LED in the first place. So good. Great. Thanks.

Tess Perrin:

So I guess following on sticking with CCT, one of the questions is we know that CCT is a parametric for evaluating short wavelengths emissions. So is there a new metric being developed by an organization such as the IES, CIE, the DOE or NIST that is addressing the shortcoming?

Bruce Kinzey:

That's a very good question, and there is not any, to my knowledge, but that doesn't necessarily mean that people aren't working on something, I just may not being aware of it. I'm not aware of any activity right now. The problem is that you take a pretty big step, well, a bit of a step in complexity. And I think really all that's needed is that the manufacturers already have the SPD data for their specific files. When they send these off to testing laboratories, they would typically get the SPD of the product returned to them as part of that data file. They're not necessarily accustomed to providing this to the public.

But there's no reason that they couldn't. And in particular, if the public started demanding this, if they said, we really want to get the SPD file for the products because we're doing an evaluation. We're looking at various products, various products options, color temperatures, as well as products available from different manufacturers. As I pointed out in that table, you can have two products at the same color temperature that had very different characteristics. So we're looking at we want to do an accurate comparison among all these different products, so we really need the SPD file from you all for this particular model that we're considering.

The manufacturer should readily provide that. It's not a big deal for them to have that. They should readily provide it, and I'd say if this is important to you and they refuse to provide it then you just move on, and eventually they'll be providing it just like everybody else. So it's a little bit more complex. It's not as easy as just looking at a single value, oh, well, 3,000 K that's an easy kind of thing to work with. But unfortunately, it's just not very precise. Thanks.

Tess Perrin:

So I guess just immediately following up from that, there's been a couple questions around whether or not it's a recommendation from our study that manufacturers publish the scotopic potential for their products. So I guess just echoing what you just said. The real push would be to provide the SPDs, and then from there you could calculate essentially any other weighting factor that you wish to, including scotopic weighting.

Bruce Kinzey:

Yeah, exactly. At that point it's just a pretty simple mathematical calculation. I mean, you'd want to do it in a spreadsheet or something, you don't want to go through manually adding up all the different. But it's really the basis of that whole thing is the actual SPD file. So if we get manufacturers accustomed to providing those as part of their spreadsheets or at least down-loadable from their website, then we will address a lot of these issues.

Tess Perrin:

Thanks. So then sticking with discussions around the scotopic function. There are a couple questions around whether or not mesopic vision was considered. And some of the statements were, it seems to me, that scotopic vision within the urban environment does not really exist and perhaps not even in the distant viewing. So how would scotopic vision [INAUDIBLE]?

Bruce Kinzey:

Tess, you want to take that one?

Tess Perrin:

Sure. So essentially we had some pretty extensive conversations with the astronomical community about this very topic. And we essentially decided upon scotopic weighting because we wanted to show the two extremes eventually between unweighted results, which would then allow you to weight them by any function you're interested in, and then scotopic weighting, which would show essentially the greatest weighting on the shorter wavelengths.

So then essentially you could look at impacts on human vision, adaptive vision, instruments, or other biological impacts. So while we could have looked at the mesopic function, given we — and we'll talk about this more next week — but given that we ran an equal energy spectrum in the model, we were able to come up with this sky glow SPD for all of those results, which you could then weight by either the photopic function, mesopic function, or the scotopic function to then see what the impact was. So I hope that answers the question.

Bruce Kinzey:

Yeah. Thanks for priming the pump there. So just to follow up on that, if you're standing in a city — if you're actually in the mesopic range, that's probably not typically what the typical astronomer would recommend if you really want to see the night sky. So I guess that's where we're really talking about the dark adapted eye.

If you're going to be outside, you want to go outside the city, you're out in the mountains, you really want to do some stargazing or something, most people, I think, would acknowledge that you want to go outside the city where you're going to be in the region generally of scotopic lighting. We're talking about very faint differences between some of these stars that you actually need to be out there for half an hour or something to actually see the full complement of them.

And so I think — I guess we feel that scotopic vision is probably more relevant on that particular — you know, as a metric for that particular, say, value.

Tess Perrin:

Thanks. So did you use any pseudo color or false color images to visual luminance distribution? I can start this one. So we'll actually talk about this next week as well, but one of the outsets of the model is a polar plot which shows essentially the radiance or luminance values, depending on whether you scotopically weighted — so scotopic luminance values.

And it shows them for every zenith and then as a musal angle, based on the observer. So you can then see as you're doing all of the different runs what the impact has been on the luminance distribution. So we actually did not — we didn't use the polar plots in our study, given we were really just interested in looking at that average horizontal irradiance or illuminance value to look at the relative impact, but you could indeed use the model to visualize changes in distribution. Bruce, do you want to add anything?

Bruce Kinzey:

No. I think that was good. Thanks.

Tess Perrin:

In calculating percent blue, you chose a band of the spectrum that leaks into the green, so what was the basis for your calculation of percent blue?

Bruce Kinzey:

So yeah, and that was actually taken from an astronomical source where they calculated — they used this for modeling. And I guess we took it because — I mean, I guess we could have chosen some other measure as well. But I think it well serves the point that it seems that terms like blue, percent blue, are being defined by the context in which they are being used rather than by some official definition of blue.

Everybody kind of turns to that as the shorthand term. As I was saying, you know, a blue light, for example, there is no official definition of blue, so that's another inaccuracy in the public discourse. You don't actually know when someone says blue light, what band of wavelengths are you actually talking about.

It could be a very narrow band or it could be some large band like we found in this astronomical site, which is really — was calling it, labeling that percent blue, but really they're looking at the potential ability to affect sky glow. That's how they defined that band. And yes, it's much beyond any traditional definition of blue.

And this is all part of the issue. We don't like using — talking about just CCT or talking about blue light. We are strong advocates of talking in terms of the actual spectral content, because that's what you need to talk in if you want to talk about the effects of these — just talk about these different effects that are of concern or talk about mitigating them. You really need to be talking in terms of the actual spectral content.

Tess Perrin:

I would just add that that percent blue calculation was actually developed by Miroslav Kocifaj, who developed the model that we ran, and then also Martin Aube. And so it's actually based at the university that Martin is at. And so it was developed by astronomers. And they're mainly looking at [INAUDIBLE] because those were the radiant bands that they were — that they calibrated impact in for their purposes. Let's see.

Bruce Kinzey:

We're running out of questions, sounds like.

Tess Perrin:

I'm just trying to sort them. So the output from your report for the near observers show variation in the contribution to sky glow for each of the different SPDs used. So there's not difference in being shown the SPDs associated with higher consistency have a higher contribution to sky glow, even at the 50% output reduction.

Bruce Kinzey:

So the results — maybe you can weigh in here, too, Tess, of course. So the results — there is a stronger correlation between higher CCT and higher short wavelength content in general. So there's a general trend. There's a general association. The higher the CCT, the more short wavelength content they tend to contain.

However, it's not a precise measure. So it's not a very good predictor. There is a general trend, however, especially if you scotopically weight the results, if you're looking specifically at the ability of those wavelengths to obscure the view of the night sky to somebody exposed to them, then there is a general trend.

Higher CCT will generally lead to that. You know, I feel compelled however to also point out that, again, there are visual acuity benefits that also accompany that higher short wavelength content. So it's also enabling you, as I've pointed out in the presentation, to see objects in the road or other things at those low lighting levels that are typically associated with streetlighting.

Tess Perrin:

I would just add that even for the highest CCT that we considered, if you're looking at those kind of combined results charts for the near observer, when you're not taking uplight into account, you still see that highest CCT only has not even twice the impact compared to high-pressure sodium. And then when you're taking a reduction in uplight into account, you have an even greater range in how CCT can perform.

Bruce Kinzey:

What's next?

Tess Perrin:

Is the 50% lumen reduction of LED from each [INAUDIBLE] based on lamp lumens or luminaire lumens?

Bruce Kinzey:

I'm going to say luminaire lumens, but you caught got me here. I'd have to — I would say luminaire lumens, but I'll maybe have to issue a retraction if — I'll check that afterwards, after we get off here, and we'll put a note up there if I'm incorrect. I believe that's luminaire lumens. It's a per fixture basis.

Tess Perrin:

I believe we were looking at the lumens exiting the fixture.

Bruce Kinzey:

So it was luminaire lumens, according to my memory, yeah. And then we're running out of questions. So one other question getting back to the 50% reduction in lumen output. The question is around whether or not the statement around that there's 40% to 50% less lumens coming out of LED luminaires, if that's correct or that always takes into account that you're lowering light levels.

Tess Perrin:

And this person goes on to say that the LED would need fewer lumens to produce the same footcandles due to less depreciation and better CUs. And if, in case you're not familiar with CU's, they are coefficients of utilization, which essentially in general lighting calculations, it's the fraction of initial lamp lumens that reach the test area. And there's often a function of reflectance, shape, efficiency and so forth.

But essentially the question is asking, is the actual difference between lumens coming out of the luminaire 40% to 50%?

Bruce Kinzey:

I think that's what we just answered after all that. Yes. Typically — and this was an issue that 10 years ago when I was working with the very early stage products coming out, this was something that the engineers in different cities and so on were saying, well, wait — we're not going to consider this product, this LED product, because it's only putting out half the lumens or less.

You're not going to be able to meet the lighting requirement with that. They were only looking at output and they were finding it hard to believe that they would actually be able to meet their lighting specification with those products. But since then actually I think people have gotten familiar with the fact that the distribution is much better. You're able to meet the lighting specification.

You are actually reducing the average, for example, and in some cases reducing it fairly significantly because you've eliminated those hot spots, which are serving to raise the average. It's like if you think about it like a laser beam focus on the plot. There's this huge dump of lumens in this one area.

It's raising the average in that area, but it's not doing any useful work. And so when you get a much more uniform distribution, you're able to tailor that much more closely, including putting in additional — whether you want to oversize a luminaire at the beginning to compensate for future lumen depreciation, or if you want to do — like I mentioned Cambridge, Massachusetts.

They put in lights that have additional capacity, but they've also got a dimming system. So they're actually running their rights, when they first turn on at dusk, at 70% of the maximum output. So they're actually delivering exactly the amount of light that's required to meet their lighting spec and no more.

So you could look at that and say, well, the average lighting level is much lower than if we went back with a traditional high-pressure sodium system. That's true. But the fact of the matter is that they are meeting their lighting specification with many fewer lumens, much less light than the previous incumbent system.

They're still meeting the specification, still meeting whatever their regulation, what they've said they were required to provide or what have you. So this stuff gets into some nuanced issues that it's not necessarily easy — I hope I articulated that so it's clear enough. Thanks.

Tess Perrin:

Thanks, Bruce. And then I guess just one final question that follows on from that one. In the study that we did was the sky glow evaluated on their lighting conditions for streets in, for example, the EU. So I guess I would just follow on from what you just shared, Bruce, and say that if you had different lighting conditions and different requirements for light output levels, then you can just take the results, since they're all relative, and then just adjust them according to any changes in light output.

Bruce Kinzey:

We didn't expressly look at — again, we we're focused on typical conversions in the United States. And so we were really just looking at what's occurring here. I expressly stayed away from — I know different countries have different lighting levels and also employ different color temperatures and other kinds of things like that.

We didn't want to try to cover the whole waterfront here. We really just wanted to look at what was happening here in our own region. Was that the last one, Tess?

Tess Perrin:

That's it. Thank-you, Bruce.

Bruce Kinzey:

Thank-you.

Tess Perrin:

So thank-you for participating in today's webinar brought to you by the U.S. Department of Energy Solid-State Lighting Program. We are looking forward to your participation in the second event of this two part series — a technical discussion of DOE's sky glow study, modeling methods, and key variables scheduled at the same time next Thursday, August 3. Thank-you.