Bruce Kinzey:

So let's get started. Again, welcome. This is Bruce Kinzey of the Pacific Northwest National Laboratory. Today's webinar on the Sky Glow Comparison Tool is being brought to you by the U.S. Department of Energy Solid-State Lighting program and is intended to be a general tutorial for getting to know the tool and how to use it.

I also hope to effectively convey the kinds of uses for which it's best suited. First, a little bit of background. I imagine that most of the folks listening in are familiar with the concerns that have been raised recently over the last few years, anyway, about the streetlighting conversions taking place in the United States, as well as internationally, replacing the previous high-pressure sodium technology with LEDs.

The concerns revolve largely around perceived issues from these products, higher content of short wavelengths, usually referred to as blue wavelengths compared to the sodium lights that are being replaced. These blue wavelengths are seen as potentially raising issues like increased sky glow, along with other related issues, such as health and environmental impacts and so on.

However, the concerns raised often refer to projections of impacts that are based on assumptions that don't match what we've seen from actual field experience, at least, in the United States to date. Furthermore, the most commonly suggested remedy limiting the color temperature or CCT of the light source carries some energy penalty and other likely tradeoffs, while at the same time, it's really of little or no actual effectiveness in terms of achieving the goals for which it's intended.

I think this has become fairly well-known by this point. And yet, we still see an overreliance on CCT as a suggested measure out in the wild. I expect you'll get an idea of how ineffective a means that control CCT is for these issues during today's webinar.

For these reasons, the Department of Energy Solid-State Lighting program decided to conduct its own investigation — streetlighting and sky glow — starting in 2016 and spent more than a year studying it. One aspect of the study was that we took a well-published model that came out of the astronomical community and ran it more than 215,000 times. And I'll talk a little bit more about that in a minute.

These runs created a data set that enabled the subsequent development of this simple tool. And I'm emphasizing the word, simple, here. This tool is not a model. It's a simple spreadsheet that enables the user to get some use out of this big data set that was created from a model to investigate the relative impacts to sky glow within this sort of contained universe of results that were generated. The tool itself is not a model. And I'll probably repeat that multiple times today.

So what's the primary purpose of the Sky Glow Comparison Tool? Well, up to this point, the outdoor lighting community has had no simple means of evaluating their impact on sky glow, whether that's positive or negative, no way to test the validity of assertions by others, such as that this 2,700K product will cause less of an issue than that 4,000K product.

And why we've been conveying that message that CCT is an inadequate metric for these kinds of effects and health effects, as well. And, therefore, people have to look at the actual spectral content. All that while, we've known that there really haven't been any good simple tools available for doing that. So this tool's intended to help address these gaps and has been designed specifically to be easy to use by the lighting community.

It enables comparisons of the relative impact of different lighting system characteristics, specifically, percent uplight, relative light output, and a spectral content as defined by the spectral power distribution or SPD under a limited set of input conditions. It's intended to help identify the most effective means for addressing sky glow associated with an individual system or design.

We're hoping it will help provide useful guidance to lighting designers and municipal streetlighting facilities people and the like, regarding the palette at different options that are available to them should they want to include considerations of sky glow in their designs. The outputs are always provided relative to the baseline that the user has selected. And you'll see this shortly.

But the tool isn't a predictor of sky glow. As I said, it's not a model. It's providing insight on the different influences on sky glow from the different system characteristics being compared. Very briefly, these are the variables under which our little contained universe that is used for conducting these comparisons was created.

These are all variables available in the model used in that sky glow simulator that was developed by Miroslav Kocifaj at the Slovak Academy of Sciences starting back in 2007. And he's continued to refine it since then.

Basically, every run of the model involves a single set of values for each of these variables listed. Within increment, one of those variables, one value and run the model again, stepping through each of the variables in turn. So in the end, we have one model run for every combination of every one of the variables you see here, totaling, more than 215,000.

This was the variable that really enabled us to build this tool, an equal energy spectrum from 380 to 770 nanometers in 5 nanometer increments. Now, we have a lot more information available on both the original study and on the modeling effort associated with it. We've done previous webinars on both of those topics, and they're both available on the DOE Solid-State Lighting website.

If you're interested in going into further details, I really recommend those to you. But I'll just warn you that, especially, the modeling one is not for the faint of heart. Using those equal energy spectrum results, we were able to create individual scattering functions from the combination to the other variables that were identified as being the most relevant for purposes of this tool.

Essentially, when the user makes their selection of the input conditions from this table — this is the input table, and you'll see this again — they're identifying which of these scattering functions they're going to use. And there are 48 relevant combinations among the dropdown menus that you can see there. In addition, you have the relative lumen output that you see at the bottom, which has a scalar impact.

So it's just a straight multiplier corresponding to the percentage that the user selects. Probably most useful is the ability to input any SPD that the user has obtained from, say, a manufacturer or from their own photometric testing or from some other database, which is where most of the SPD currently in the tool came from, the database from the CIE.

Each scattering function has been calculated based on its influence to an equal energy spectrum. And then those results are adjusted by this spreadsheet according to the corresponding wavelengths in each SPD that the user has input or selected for comparison. OK, simple. Fortunately, we had a couple of mathematicians involved in the tool’s development.

OK, this is the main user interface. I think everything here will quickly become second nature after a few uses of the tool. At upper left is that box as the input conditions I was just talking about. To the right of that is a graphic depiction of the source SPDs. I think this would be probably very familiar to the folks watching this.

The results box here at the lower left reports a sky glow score, if you will, that indicates the relative contribution to sky glow of each of these products being compared to that of the baseline that the user has selected under these input conditions.

Let me take a step back for a moment, because I'm using the phrase, contribution to sky glow, because whatever system the user is investigating is likely to be only one contributor to the overall sky glow produced by a populated area. There'll also be contributions from buildings and signage and vehicles and so on.

So if the tool, for example, indicates something has a contribution that’s say 20% greater than the baseline, it doesn't mean that the sky glow over the city increases by 20%, because we're not talking about all those other sources. We're only talking about the contribution from whatever system the user is investigating.

Continuing with that example, the baseline system always has a score of 1.0 like you can see here. A comparison product with a score of 1.2 would mean that that product has a 20% higher contribution to sky glow and the baseline. Or conversely, a score of 0.8 would mean it has a 20% less contribution to sky glow than the baseline. And then this histogram to the right — just displays those results graphically to facilitate rapid comparisons between them.

Next, I'm going to step through some of the input procedures. And I'll be referring to the different tabs that you could see down here along the bottom of the screen. And I'll also talk about this bottom sky glow SPD chart that you can only see about half of right now in just a minute.

So we're going to get into a live demonstration in just a bit. But first, I just want to make it very clear what I'm doing. So the SPD Input Tab is where you select the various SPDs that you want to compare in the next run. These need to be in a specific format for the tool to use them correctly. And the tool does this for you through this input procedure on the left.

So, for example, you might have SPDs from somewhere from different sources. One of them's in the 1/2 nanometer or maybe 1 nanometer increments or 5 nanometer increments. You might notice here that the tool uses them in 5 nanometer increments.

So to correctly get these into the tool, you simply copy the range of wavelengths from your source for your SPD over here, along with the corresponding — you paste them using the paste values button. Then you put in the corresponding measured values with that corresponding button and then press the add pasted SPD.

Next, there will be a pop-up box that will come up that will ask you to type in your label, and you can type in anything, though, it only displays three lines like you can see here. But you can put any level of detail in there you want. And then the tool does its thing to put the values into this right format. And it appears your new latest SPD will be appended to the end of this. It will appear over here on the right.

You can have up to 30 SPDs compared in a single run, though, I'll just tell you it starts getting pretty messy after about 8 or 10. So it's probably better to do multiple comparisons if you have that many to look at. The tool also has this useful little feature for calculating the CCT of your SPD.

So, for example, you might have an example where you've typed something in, and you didn't include the CCT. Like if you look over here on this dropdown menu, you just see LED example five. You might have various reasons that you don't want to put the CCT in there. You might have a SPD that your manufacturer — your vendor gave you. That's just in a nominal value like I said.

This is a 3,000K product. And you want to know actually what the specific color temperature is. You want to know what the precise value is. So this turns out to be a very handy little feature, I think. Above that, you'll see that each of these columns has a Clear and a Save button.

There's a saved SPDs tab that basically archives your SPD. So you can immediately access them using this dropdown menu. You don't have to go back through that input procedure every time if you're going to use them again in the future. It basically preserves them in this format, along with your label. And you recall them using that dropdown menu.

The Clear button simply eliminates the corresponding SPD for this product from the next comparison. It does not remove them from the archive, from the saved SPDs tab. Just be sure that if you're going to use it again, that you have saved it to the archive tab before you clear it, because that's permanent. Otherwise, you'll just have to go back in and re-enter it using this procedure.

OK, so here we are back at the main user interface. After selecting the SPDs that they want to compare, the user selects their set of input conditions from the dropdown menus and then presses calculate. And then the tool immediately produces a graph of source SPDs, the relative sky glow scores, and the histogram that displays these.

Below that if you recall, the tool also displays the SPD of the sky glow that has been produced by each of the sources at the location and the input conditions under the input conditions that the user has selected. Right now, I just want to show you another neat little feature that comes embedded in Excel.

So, first of all, of these are scaled in the graphic along the same scale. So these are all apples-to-apples kinds of comparisons. But it's controlled by the biggest peak. Among those, wavelengths that's being represented. So in this case, this is a mercury vapor. It's got that high peak. So that's what everything else has been scaled against.

And you can see everything else, particularly, when you have something like a very high peak like this, everything else kind of gets scrunched down at the bottom. And this especially the case if the more of these that you add, this starts getting pretty messy in here.

But you can change this to only view as little as one of these SPDs or as many as you want up to the total number that you've included in the comparison. And I'll show you this again in just a little bit. But the way that you do it is you will click on the graphic.

There's a little set of buttons that come up. One of them looks like a little funnel — it's a Filter button. You can click on that and then just select the individual ones that you want to display. You hit again or apply, sorry, at the bottom. And that's all there is to it. And this feature really makes it much easier to look at just those SPDs you're interested in.

And I will also note here that, as I mentioned earlier, you'll notice this graphic has been re-scaled now because we removed that earlier peak. And here's the high-pressure sodium, which looks pretty small in that graphic. And this is the same thing. And the whole thing's been re-scaled now to fit that, because now that has the highest peak,

So finally, I know this has been a lot to absorb at a single sitting. But I just want to go through it once, so I could step through each of the individual steps with some careful explanations. Now we can jump through a few live examples, a little bit higher speed, so you can see really how easy the tool is to use once you get a little bit more familiar with it.

So let's go over to the tool. Now the first time you open the tool, first of all, you'll have to enable the macros in it. To run, it'll come up with a little warning. Your Excel package will come up to the warning message, and you have to accept that.

It also will open to this instructions tab. And you can find a lot of what I've said already here in the text. We set up a mailbox where you can send in, ask questions or report any bugs that you found or send in notes of gushing praise. Those are also accepted, by the way, what have you.

And, of course, this instructions tab is always in here for your future reference. Once you've — like any other spreadsheet — once you've saved if you're on another page, the next time you open the tool, of course, it will open to that page wherever you saved it the last time.

So let me go down to the SPD input tab here to show you how to introduce a new SPD to your collection, so just because I don't want to be switching back and forth between other apps here on the webinar. Instead of a new SPD, I'm going to show you the exact same procedure. But in this case, it's also how you can rename one of your existing SPDs, should you want to do that.

And this actually is a particular example that I've used myself. So this procedure for adding a new SPD is exactly the same step you might be copying and pasting from some other source. So in this case, I have this example. That's labeled without the CCT. We saw this on the other slide.

And let's say I want to change that. I want to display the CCT for that. So the first thing — now these are already all formatted, but this would be exactly the same. You copy the wavelengths. And your source, again, these might be in 1 nanometers or half nanometer or 5 nanometer or what have you. It's all the same thing.

You want to go down, copy, copy the wavelengths. Then you go over to the appropriate column here, paste those values in. Then you go to your measured values. You can copy all of those the same way. Paste those in the appropriate column.

And then as I said, when you hit Add Pasted SPD, it will pop up with a request for you to enter a label. And we wanted to put the CCT in here, so we look up here. And that was at 5,196K. So let's call this just LED 5,196K. Now keep your eye on that right-hand part of the screen when I hit OK.

And you see there's where they, how it got appended to the end over there. Now, of course, we have two here. We have duplicates here. This is the same. So I don't need this one. So I'm just going to clear that one. And we're ready to go.

OK, so let's move over to the calculate tab here at the bottom. And let's go ahead and press Calculate to populate the different items here in the tool. Now, in this particular comparison, we have a range of sources. So there's the high-pressure sodium, which I've also selected as the baseline.

Then there's a metal halide here. There's a mercury vapor. Looking at the mercury vapor, I don't have the CCT listed there too, so we can just go quickly back over here to the SPD input. We'll go over and look at the mercury vapor. And we see that that one is about 3,430K.

I'm not going to bother to go through that procedure again. To rename it, we'll just keep that one in mind. That one's 3,430K. Then I have a number of LED products here. There's a couple at 2,700K. This one, number five, is a violet pump that I've included. Then we a couple of 4,000th, and then, of course, 5,200K that we just modified.

Our input conditions are all at the very basic level here. Let's step through these. We're looking right now, we're looking at the near observer position, which is one of two choices. The near was modeled at the perimeter of the city. So this viewer would be underneath the light. He's right on the edge of the city.

So this is underneath the light dome that's being created by the city. The other option is the distant observer position, which was modeled at a point, 40 kilometers or about 25 miles from the city center. We'll just keep that one. That setting next is the atmospheric conditions.

We're currently in the clear sky, low particulate matter — sounds like a place we probably all want to be most of the time. We also have two other options. One is a clear high particulate, so that might be where the atmosphere is dusty or smoky. Or you might have high levels of pollution, et cetera.

And then a third option there is complete cloud cover. We have two waiting option functions. So the first here is unweighted, which is just the raw power in watts emitted across the spectrum. So watts per nanometer. And these are always normalized to 1,000 lumen output.

So in other words, watts per nanometer emitted per 1,000 lumens of output from the source, OK. Second function scotopically weights its output. So as to better represent visibility impacts to the human eye, the human retina's more sensitive to wavelengths in the shorter range.

So by weighting the results according to the scotopic function, it gives a little better representation of the relative ability of the light source to obscure visibility to an unaided eye. So let's say someone standing in their backyard, trying to do some star gazing.

And this person right now, is actually under the light dome produced by the city. OK, so these are scenario parameters are common to all of the sources being compared. Now we get to a little bit more specifics.

The baseline light source characteristics — of course, these are characteristics that just apply to the baseline. And I'll point out that the baseline here shown in red in both the chart and the table. And then we also have the comparison light source characteristics, which then are shared by all of the other products shown in blue here that are being compared against that baseline.

So the first here is, actually, it's pretty obvious that that is the baseline is just the one that you're going to select among those that you're comparing — pretty straightforward. The percent uplight pertains to and is really determined by the type of fixture that's represented by the baseline.

So there are four options in here. And I'm not going to spend time right now describing when you might select a particular value. But there's some of that explanation in the instructions tab. You'll find it in there. So let's leave that at zero for now.

Same thing applies to the percent uplight characteristic that you might select for the comparison sources. I will say that in most cases, particularly, with LEDs, particularly, with most conversions are taking place today, you'd expect this now to be 0%. But that isn't necessarily always going to be the case.

Let's say if these are decorative post-top replacements where you might have a little bit of uplight. Or even if you have 0% uplight luminares that have been installed in a tilted orientation or something, you may have something that has changed the characteristic.

So you basically have the same four characteristics here. We'll leave that again at zero. Then finally, there's the relative lumen output of the replacement system compared to the baseline. Most frequently, we're seeing substantial reductions in the lumen package of LED products that are replacing any kind of glass lamp product.

But again, that isn't necessarily always going to be the case. For example, you might be expanding the area covered like adding some parking lots or a new housing addition or something at the same time. You're replacing the lights. Or you might have already replaced the lights with LEDs, along with a dimming system.

And you want to know what the contributions are at different levels of dimming. So in order to provide maximum flexibility here, you have the option of really dialing this in that the lumen package of the comparison source compared to the baseline, anywhere from 1% of the output all the way to 200%.

So that, hopefully, can represent pretty much any scenario that you might be encountering. And that's just a straight scalar multiplier, which makes sense if you think about it. If you cut the amount of light output by 10%, you reduce your contribution to sky glow by that same amount, 10%. It's pretty straightforward.

All right, so I can pick up the pace here a little bit now. The first thing that you might notice — and again, this is a very sort of apples-to-apples comparison. Everything is equal here. So all we're seeing here in this chart right now is basically the influence of the actual spectral content and the spectral content only.

The first thing you might see is even, then, there's not a whole lot of difference here between the different white light sources. Metal halide and the mercury vapor actually make the largest contribution to sky glow at this near position when everything is equal.

And we're looking at this in unweighted terms. So this is just kind of the raw sky glow score. And this is per 1,000 lumen output, et cetera. You might also have already noticed — but you certainly will after I point it out — is that there's really very little correlation here between the CCTs of these products and their resulting influence on sky glow.

The mercury vapor, if you recall, this was about 3,400. It has a higher contribution than all of the subsequent LED products, including this all the way down here at the end. That's 5,200. This was 3,400, and this one's 5,200. So this really illustrates the problem.

We're trying to use CCT as a control measure. It's really simply just not accurate enough. Again, as I said earlier, I think this is becoming a little better known now, I think, throughout the whole lighting community.

Next, let's go topically weight these results. So, again, this will better reflect the relative abilities of each of these sources to obscure the view of the night sky that this person might have standing in their backyard inside the city. OK, so we'll do that, change the scotopic, hit Calculate.

It doesn't go automatically. You have to hit the button. And now you do see some better correlation between certainly between CCT and this effect when you scotopically weighted these results. Now, note that we haven't changed anything about the actual sky glow here.

We're just waiting at — we're emphasizing different components of those wavelengths differently. And so we're seeing this effect. I really want to caution anybody about jumping to any conclusions at this point or any time based on just a limited set of comparisons, because I also tend to refrain from talking about these values too much, because they jump all over the place.

They change very readily if you change any of these other options or even when you change the baseline, the actual numerical values change. Their relative positions are the same, or their relative multipliers are the same. But the numbers themselves change a lot.

So you really want to be sure of what it is you're looking at before you draw any conclusions from it. So like, for example, this violet pump 2,700K product used to have a higher influence when these were not scotopically weighted than the mercury vapor. Now, that position has reversed.

But these are the kinds of things that you're looking for when you change some of these different input conditions. What are the effects in terms of the contribution to sky glow or to that effect that you're interested in.

So let me let me, next, demonstrate — something I mentioned earlier, how we can clean up these graphics. You can see this one is getting a little bit messy here. And I can also show you something interesting here at the same time. So if I click anywhere within this graphic, you'll see those little buttons just came up.

Here's the one that looks like a funnel. This is the chart filters. If I click on this, it sets up a list there of all the things that are displayed. And, actually, this by itself is a cool feature. You can just sort of scroll down. I don't know how well. I guess you could see the darker ones.

Anyway, as I move down through here, it highlights the particular wavelength. So it's a way to visualize — look at each of these in the context of all of the others — very neat little feature. But let's say I just want to get rid of all those the fastest way to eliminate all of them at once, click Select All.

And then let's just take a look at, for example, the metal halide. And you also have to go down here and remember to hit Apply, or your changes won't take hold. So if I say Apply, immediately, you would notice that this has been re-scaled again. We got rid of the other higher peaks.

These are now the higher peaks, so this whole graphic has been scaled to that. OK, that's what the source looks like in terms of its spectral content. Now, let's go down here and do the same thing. Now, first, let me change so we can see these a little bit easier.

Let me — whoops — change the view here to — I just want to make the screen a little bit smaller so we have a little bit better view, throw all of these together. So 90%. That gives us a little bit better ability to see both those at the same time.

So let's go down and do the same thing here. I adjust that, and we hit apply. I think this one is very interesting to look at, because what you're seeing here is demonstrating the interaction of those shorter wavelengths. There's a lot of short wavelength content in the metal halide.

You're seeing what the interaction looks like in terms of scattering by the components of the atmosphere. You see it's actually a very nice looking curve. Let's investigate the effects there a little bit more of those atmospheric conditions.

So here again, this is clear, low particulate. Let's change that to clearer with a higher particulate. So again, a dusty or smoky atmosphere would have you keep your eye on the ball down here on that graphic.

When I hit that, not a huge change but some. It seems to be leveling out those results a little bit. My guess is I'm not an atmospheric scientist. But my guess is when you have higher concentrations of particulate and larger sizes and so on, a lot of that scattered light is basically being absorbed by those particles.

So you see a little bit of a flattening of that curve — wasn't huge. This is still a clear atmosphere and so on. But there was some effect there. Now, let's go up to a complete cloud cover. And what I want to do before I hit Calculate is I want you to note the scale here.

We're looking at the peaks or somewhere 10 to the minus 9. Somewhere over here, it gets to 10 to the minus 10 — up here — 10 to the minus 9. When I hit Calculate — now this thing looks remarkably like the source SPD.

And I think this probably coincides with most people's experience you have, which if you have low cloud cover, most of the light is directly reflected back into the city. And so it looks very similar in terms of what the original SPD was is what it looks like when it's being reflected back into the city.

And if you recall, I said, most of the magnitude over here was 10 to the minus 9. Now we've gone to 10 to the minus 8. In other words, we've increased the amount of sky glow at this nearest location by a factor of 10, which is also, I think, probably coincides with most people's experiences — how much brighter it gets when there's complete cloud cover over the city.

I want to throw out another little caution here. We keep saying — and I will continue to say that this tool is really intended for looking at relative kinds of changes. In order to calculate the relative impacts, however, the tool has to match our original model, did calculate absolute impacts.

And that's what you're actually looking at here. This is a depiction of an absolute impact. We wouldn't recommend that you try picking any sort of values off here for your site. The applicability of these specific values to any specific site, i.e. yours, is only as good as the assumptions that went into creating this little contained universe.

This is how it looks in a little contained universe. So as well as those assumptions fit your site, which is probably not very well in most cases. As they depart from that, then you could expect these results to depart from your site.

But then when we're talking about relative changes, like I just did, I said, OK, going to complete cloud cover, increases this influence by a factor of 10 that we think those are actually pretty sound.

So let's move next to a little more realistic scenario where we're going to replace — let's say we're replacing an incumbent high-pressure sodium light source with LEDs. Let's say this is a city that doesn't deploy any mercury vapor anymore or a metal halide on their streets. So we don't really care about those.

I'm going to clean this up a little bit. I'm just going to take those, remove those from the comparison. So I'm going to clear the metal halide and clear the mercury vapor. OK, so we just got the HPS. And now we're just looking at the different LEDs that might be being considered to replace this product.

And, of course, I've got a can setting here. You can put in anything from any source that you've obtained your own SPDs. So let's go back to calculate. You'll notice here that this list has been updated. But none of the numbers or the other effects here have been updated until we hit Calculate.

Before we do that, I want to go back to change a couple other things here. So let's go back and include all of the SPDs — same thing here. And that includes, for example, you notice that the metal halide and the mercury vapor are still showing up in this list. But those will disappear as soon as I hit Calculate.

Let's also take a look at changes to the distant observer. And this doesn't really matter. But let's go back to clear low particulate. We'll leave scotopic in. That's good to go with for now. So I hit Calculate.

And here we are, again. These are scotopically weighted. We are seeing some influence here — does appear that the higher CCT products do have a greater ability to obscure the view of the night sky. Again, same thing, particularly, in relation to this high-pressure sodium baseline.

But, again, now we're back to looking at really just the impacts of the different spectral contents. OK, and we're throwing some extra emphasis on the short wavelength content according to the scotopic function.

I think probably a lot of the people on the phone here know that by now, anyway, that you're almost never going to pursue a lumen-for-lumen replacement in an LED conversion. This is the exact same number of lumens coming out of an LED product than any HPS product that was replaced.

You just typically don't need as large of a lumen package from the LED product. So a value that I've seen that I'd say was common in my experience is a 50% reduction in the replacement lumen package. And I've seen some conversions with much higher percentages than that. That's not a stretch of the imagination.

But if you don't like 50%, obviously, you can put in anything. And you should absolutely put in the numbers if you actually do an AGI calculation and so on. And you're looking at different products. You should put in whatever number applies to your specific situation.

I also should point out — this doesn't mean — I'm not talking about a 50% reduction in illuminance. But 50% left light output from the LEDs that's required to satisfactorily meet the illumination requirement in a given street or area lighting application.

OK, so let's scroll this up and pick 50% just for purposes of illustration here. As I mentioned, this has a scalar impact. So when I hit that button, that simply reduces the contribution by 50%. And, now, actually, you notice that the level of sky glow here are getting pretty close to the baseline.

If you're furthermore a site like a city, say, like Cambridge, Massachusetts that is putting in a dimming system, and they are dimming their lights, they've achieved this kind of a change. But now, they're also dimming their lights after midnight from midnight to morning, 12:00 to 5:00 or 12:00 to 6:00 AM or whatever it is.

They're cutting those lights by another 50%. So all of these columns would be cut again in half at which point, you've reduced the sky glow contribution from the streetlighting system below that of the high-pressure sodium system that's been replaced across the board independent of the CCT.

And I've actually done this calculation for Cambridge. And this is in fact what they've achieved in that location. OK, finally, let's investigate the impact of even just a little bit of uplight in the incumbent. OK, so this would be the case — and here's a case I'd say in many if not most ongoing conversions in the United States.

There are still tons of drop-lens overhead streetlights out there. You can find these in almost every city. And in some locations, they might even still be a majority. I'd say the city where I live, Vancouver, Washington, they are the majority. They're all over the place.

So anyway, very common. Let's assume, give them the benefit of the doubt and assume these fixtures had been replaced in the last 10 or 20 years. They also, they have only a very modest uplight characteristic of maybe 2%. OK.

Now, because all of their replacement products are 0%, we're about to see the impact of, we're introducing uplight to the baseline and then removing it in the comparison products. So you're going to see what that impact of that uplight was, OK? Now, keep your eye on the middle chart there, just because I don't want you to miss anything.

OK, so, clearly, what this shows is basically the overwhelming impact that uplight has in terms of contribution to sky glow. And if these results are scotopically weighted, even, I think this is, overall, is a realistic scenario. But even if you wanted to go back to 100%, of course, that just returns those to the previous level.

You could you see, basically, compared to getting rid of uplight, all of these other actions, including dimming and changing, substituting different SPDs, different CCTs, basically, are really operating on the margins, at least, as long as you're talking about remaining with a broad spectrum or a white light type of source.

I forgot to hit Apply there earlier. So one thing I think that the tool is very useful for is demonstrating general principles like this one — really getting rid of uplight should always be the first component of addressing sky glow issues.

But I want to point out, again, also, that this tool is only showing the impact of removing the uplight from this particular system that you're investigating here. So that could be the entire streetlighting system, or it might just be an individual parking lot if that's what somebody was using this tool for or what have you. There are still likely to be lots of other sources of uplight, such as buildings and signage.

And unless you're looking at some kind of policy affecting all of those others, as well, their impacts are still going to be very visible. And, in fact, are likely to become dominant in the overall context, because what you've done is you've moved the streetlights from something that looked like this down to something that looks like this.

Any of those other sources that are putting light into the night sky, emitting light above a horizontal orientation is going to largely have — there's some other details here that I'm skipping over. But they're largely going to have this kind of contribution to the sky glow generated by a populated area. So it's just something to keep in mind here when you're looking at these results.

OK, so let's return to the slide presentation. And I just have a couple more slides to finish things up. What are some limitations of this comparison tool? I've mentioned some of these. As I've mentioned, we have described a very simplified universe here.

So, for example, there are only three atmospheric settings that are representing something that's virtually infinite in nature. You can see it over there in this photograph that I actually took out the window of an airplane in August when I was flying back from Boston when we were having all the fires out west. You can just see the sky is full of smoke particles.

We have scattered clouds. There's probably moisture in the air and everything else. This is virtually impossible to model something like this. And so we simply just had to come through with some simplifying assumptions — a number of them in order to make this thing manageable.

Similarly, we've only modeled at two observer locations. So we have the one right on the edge of the city underneath the light dome. And we have someone 40 kilometers distant from the city center. But for example, we know that the change in sky glow contribution is not linear between those two points.

And I can't tell you, or we can't tell you what the difference is, say, at 10 kilometers from the city center or 20 kilometers from the city center or 100 kilometers from the city center. We only have these two points. Those others would be great to know. But it requires literally tens of thousands of model runs at each of those locations in order to generate those results.

So we think this is still very useful to give some sort of general insight on these things. But again, this is what I was saying earlier. It's not necessarily going to apply specifically to your site. Similarly, we only have these four options, very generic options for uplight.

I guarantee any products that you have are not going to match exactly any one of these values. But, again, I think this is still valuable for providing insight on the general trends. One that I didn't show today is that the biggest change comes from when you start with nothing to something.

So when you go from 0% to 2%, has the largest incremental impact. You still have another growth in the impact when you go from 2 to 5 and from 5 to 10. But those increments, those are smaller, even though this overall increment, this percentage change is smaller, it has a larger increment because it went from nothing to something.

OK, and I promise for the last time, it's not a model. It's suitable for looking at relative impacts. But if your needs involve looking at absolute impacts, if that's what you need, you're going to need to engage an actual model that's able to take in all the other specific details of your situation. And it isn't making these kind of gross simplifications that we've had to make here, but also, to create more of a generic easy-to-use tool.

OK, and I've also mentioned this one multiple times. We're only looking at the contribution from the system you're investigating. And there are likely to be many others, like you see here, lots of other sources that you're not influencing, unless you are modeling some sort of city lighting ordinance, or something that might be affecting a much larger, maybe, all of them or something.

And then finally, Miro Kocifaj has continued to refine his model in the time since our investigation. And so these results were all generated. They're based on modeling results based on the state of the model in early 2017. I'm not really sure how much he's continued to refine this. I'm not sure how much the results would change if we ran everything again today, probably not a lot, but there might be some differences.

OK, so there you have it. At the moment, we're asking users to request a copy of the tools so that we could keep track of the ones out there in case someone finds a bug that we want to address and correct in all the copies in circulation. We might also be coming out at some point, it's possible we'll come out with a version 2 if we get some good suggestions for updating it.

I've already gotten a very good one from someone in the astronomical community. I don't know if we can accommodate it. But I'm going to check into that to see if we can. It's a very good suggestion. It would be a very nice feature if we can offer it.

If nothing big turns up in the next few months, we'll probably just post the tool itself at this address, so everybody can download the tool directly. This is also where the FAQ generated from any questions submitted during this webinar will be posted, as well as the webinar recording itself.

We've also set up an email address here for submitting questions and bug reports. This is also contained in the Instructions tab as I noted, but I just included it here for reference. That mailbox will be monitored for some period into the future. I don't know for how long but for some time.

I hope this has been useful in teaching you about the tool and how to use it, and what it's best used for. And our team sincerely hopes that the tool itself is something that you find useful as well and fun even. I found it actually to be quite a nice little diversion from my normal work day.