Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

SQM Limitations

  • Please log in to reply
49 replies to this topic

#26 Redbetter

Redbetter

    Fly Me to the Moon

  • *****
  • Posts: 7318
  • Joined: 16 Feb 2016
  • Loc: Central Valley, CA

Posted 11 May 2019 - 01:42 AM

According to the Unihedron website the SQM are calibrated to a NIST standard.  But if you dig a bit further you'll find the "NIST standard" is basically a commercially made photographic light meter, designed to measure in tens to hundreds and thousands of lumens.  That is -several orders of magnitude brighter than anything the SQM is intended for.  I don't see how a sensor calibrated to those levels so far off can be called 'calibrated' for very tiny levels of light. The sensor and electronics would have to be extremely linear in the response over many magnitudes of brightness. Also, the sensor is not sensitive to blue wavelengths, which is where the vast majority of lumen output of LED streetlights falls, along with increased scattering of bluer light, the SQM will 'not see' much of contribution as more and more LEDs contribute to light pollution and so underestimate the amount of light pollution.  It may even lead to the wrong conclusion that light pollution is decreasing over time.  An example of bad data is worse than no data.  

 

 I see many discussions on CNs of observers 'splitting hairs' in arguments about dark skies and tenths of readings completely unaware of the difference between precision and accuracy v. full scale calibration range of instrument readings.  A SQM precisely fits the old engineering & science adage of "measure it with a micrometer, mark it with a chalk, cut it with an axe."  

And yet it works.  So the assumptions above should be evaluated. 

  • I don't know that the sensor and electronics have to be that linear.  Afterall, the magnitude scale is not linear, but logarithmic, and this serves to smooth the errors.  In a 2005 Italian evaluation (link) the resultant magnitude standard error was +/- 0.028 mag over a range of about 8 to 20 MPSAS...and it actually had lower deviation at both ends of that range than in the middle. 
  • The spectral response of the meter appears to be pretty strong throughout the relevant visual scotopic wavelengths, including the blue, still about 75% of peak at 450nm, and somewhere in the 90's at 510nm.  So it is going to be able to detect the LED emissions with only somewhat reduced efficiency rather than "not see much." 
  • Your argument is perhaps backwards...the SQM is also sensitive to the red that we don't see well at night, so the lack of red contribution makes it harder to correlate to some existing photometry systems that are heavily weight toward the yellow or red when the sources change to a bluer type.  The V band and CIE photopic band have the same problem, even more so.  The CIE scotopic should be somewhat better in that regard (being insensitive to red), but ironically it doesn't capture the blue as well as an SQM will.
  • The closer conditions are to pristine, the smaller any error introduced by LED or other manmade lighting sources.  That is because the fraction of the manmade sources shrinks so much.  So counter to the argument you have made above, darker conditions will show the least departure for visual.  The actual concern would be for light polluted conditions.

When it comes to light polluted conditions I have seen some interesting things scotopic-vision wise.  In an area of rapid suburban sprawl around me, I noticed the sky becoming slightly darker with conversion to LED streetlamps, despite being the less desirable 4K variety.  Being "full cut off" they throw quite a bit less light into the sky, and I can see that.  I can also see how they throw an order of magnitude less light into my bedroom at night (light replaced this week.)  The HPS lamps stuck out and threw a great deal of light to the side and up.  I first noticed the improvement at the eyepiece when the neighborhood next door was converted about 2 years ago.  And the SQM-L indeed showed some improvement as well.  I expect this to be short lived with the new homes and schools being built, etc., but it ran counter to the conventional "wisdom" about LED street lights.  And if these were warmer LED's the concern would largely disappear.

 

No single meter can be trusted for anything if one does not have ways to cross check it.  The SQM is no different than many other meters in the garage or elsewhere in this regard. 

 

The ones discussing these things tend to know a lot more about calibrating and measuring things than you give us credit for.  I will take a meter that gives reproducible results (and can be tied to various calibrations) over somebody's +/- half a magnitude approximation with uncalibrated eyes any day night of the week.  Even if the meter is shown to have a significant departure from actual, that can be corrected for accuracy and regardless it still provides an orders of magnitude more precise value for comparison on its own.   


  • Jon Isaacs likes this

#27 ngc7319_20

ngc7319_20

    Viking 1

  • *****
  • Posts: 657
  • Joined: 25 Oct 2015
  • Loc: MD

Posted 12 May 2019 - 10:41 AM

In a recent thread (sorry I can't find it!) someone else reported inter-unit variations as high as 0.4 magnitude.

 

I'm the fellow reporting 0.4 magnitude discrepancy between SQM meters.  I've been studying the situation and can provide some details.  Here is a plot comparing readings on four SQM meters since 2015.  I am plotting differences between meter #X  (X=1, 3, 4) and meter #2 which I think is the most stable one.  As I will discuss, meters #1 and #3 have visible detector problems, and meters #2 and #4 look OK.

 

SQM comparo.JPG

 

My measurement technique is to discard the first reading, and then take the average of the next 5 readings on each meter.  Then immediately do the same with the next meter, etc., eventually rotating back to the first meter.  Readings from any one meter over a few minute session should agree to ~0.02 mag (often the agreement is 0.01 mag or better).  All readings are on clear moonless nights aiming at the zenith.  My typical readings are around 20.20 mag.

 

Meter #1 (red squares) was bought around 2005.  By 2015 it was showing my local sky brightness getting DARKER by 0.1 mag.  To confirm the trend I purchased meter #2 in 2015, which seemed to show meter #1 was in error.  Since 2015 meter #1 has continued to drift towards darker readings and now reads ~0.4 mags too dark.  Looking carefully at the detector in this meter with a magnifier I can see two problems with it: (1) the detector chip has physically moved towards the edge of the hole in the red filter, and is no longer aimed exactly outward. How this could happen?  The meter circuit board appears to be held in place by a piece of foam rubber, and perhaps it has migrated or deteriorated. The meter has never been disassembled, etc.  (2) The other problem is that there is a fine particulate on the inside surface of the blue-green IR-block filter.  It looks a little like an eyepiece that has developed fungus or something.  The meter has been stored in a dry, clean environment most of its life, but obviously is used on dewy nights now and then.  

 

The blue circles are meter #3 which I purchased to check the other two.  It seems fairly stable, but reads about 0.1 mag brighter than meter #2.  Looking at the detector in this meter I can see it also has a problem.  There is some yellow glue-like stuff covering the detector lens.  This probably explains its different response -- either due to spectral issues, or optical field-of-view differences from the other meters.

 

The green triangles are meter #4 which was bought new about 3 months ago.   Meters #2 and #4 are in excellent agreement (so far).  Their values are generally within 0.02 magnitude of each other.  Looking at their detectors with a magnifier shows the detector is well centered in the hole in the red filter, no contaminants, and no stray glue problems.

 

What can I conclude from this?  

 

(a.) The meters can be extremely accurate.  Meters #2 and #4 typically agree to 0.02 magnitude or better.  Meter #2 is at least 4 years old, so there is some evidence of long-term stability. The detectors in these two meters pass inspection with a magnifying glass.

 

(b.) The meters are not necessarily stable over many years.   Meter #1 which is 14 years old, reads 0.4 mags too dark, and the drift rate is currently ~0.1 mag per year. This detector now appears anomalous when inspected with a magnifier (de-center, aiming problem, contamination).

 

(c.) Manufacturing errors are possible and can affect results: Meter #3 with yellow glue on the detector lens consistently reads too bright by ~0.1 mags.

 

(d.) Like any precision scientific instrument, the calibration should be checked on a regular basis, probably every 6 months for these.

 

(e.) Two of my four meters have obvious problems that can be identified by inspecting the detector with a magnifying glass.  Four is not a statistical sample, but a 50% problem rate is worrisome.

 

(f.) A casual user with a single meter cannot be guaranteed of accurate results -- the detector anomaly rate is high and changes are possible over time.

 

Suggestions:  Get several meters (minimum of 3, preferably 4 or 5) and compare them every 6 months or so.  With >3 meters you hopefully weed out ones which are bad. Also check the detector with a magnifying glass for anomalies (de-centering relative to the hole in the red filter, foreign matter, glue on detector lens, etc).


Edited by ngc7319_20, 12 May 2019 - 11:39 AM.

  • havasman and Enkidu like this

#28 ngc7319_20

ngc7319_20

    Viking 1

  • *****
  • Posts: 657
  • Joined: 25 Oct 2015
  • Loc: MD

Posted 12 May 2019 - 11:05 AM

Heres what the detectors look like in these meters.  Meter #4 is the newest and I believe this is how they should appear.

 

 

Meter 4.jpg



#29 ngc7319_20

ngc7319_20

    Viking 1

  • *****
  • Posts: 657
  • Joined: 25 Oct 2015
  • Loc: MD

Posted 12 May 2019 - 11:08 AM

This is meter #1 which is 14 years old.  The detector inside has migrated over time and appears de-centered and tilted.  There is also some kind of contamination in the meter.  It currently reads 0.4 mag too dark, and drifts darker at about 0.1 mag  / year.  What causes the migration over time?  I note the circuit board appears held in place by a piece of foam rubber - maybe it has migrated or deteriorated, but I am not sure.

 

Meter 1.jpg


Edited by ngc7319_20, 12 May 2019 - 11:15 AM.


#30 ngc7319_20

ngc7319_20

    Viking 1

  • *****
  • Posts: 657
  • Joined: 25 Oct 2015
  • Loc: MD

Posted 12 May 2019 - 11:11 AM

This is meter #3.  There appears to be stray yellow glue covering the detector lens.  This probably introduces some color effect, and well as changing the optical properties of the lens (i.e. field of view will be altered).  It seems to be the same yellow glue that is used to hold the blue-green IR-blocking filter over the detector.

 

Meter 3.jpg


Edited by ngc7319_20, 12 May 2019 - 11:12 AM.


#31 Tony Flanders

Tony Flanders

    Hubble

  • *****
  • topic starter
  • Posts: 16756
  • Joined: 18 May 2006
  • Loc: Cambridge, MA, USA

Posted 12 May 2019 - 02:32 PM

Here are a few thoughts. First, there is a great deal of useful information on the SQM website, including the specifications of the underlying detector. Note that the response is inherently linear; the device is basically counting the amount of time required to accumulate a certain number of photons. This is obvious in use; the amount of time required to obtain a reading is inversely proportional to the intensity of the light. Linear, not logarithmic.

 

Second, Unihedron was more than happy to replace my SQM that had drifted, when I informed them of the problem. They confirmed that the sensor had become de-centered. It might be worth bringing them into this conversation; they are, after all, scientists and engineers with a passion for astronomy, not businessmen motivated by making money.

 

Third, if each device is indeed individually calibrated, then it would be possible for a device's readings to drift toward the bright side if the sensor shifts so that it's better centered than it was originally. But it's much more likely to drift toward darker readings, since there are an infinite number of ways for the sensor to be placed wrong, but only one way for it to be placed correctly.

 

Finally, it's interesting to speculate on how an inherently superior device could be made. In particular, the CCD and CMOS sensors used in cameras can accumulate enough data to determine sky brightness quite accurately in a shorter time than the photodiode underlying the SQM can. And that's with the handicap of exquisitely fine spatial resolution, acquiring data for literally millions of individual points in the sky, compared with the single average reading obtained by the SQM.

 

And somehow the camera makers have managed to keep temporal drift and unit-to-unit variation under pretty tight control. They do, of course, have the advantage of being able to amortize their development and manufacturing costs over millions of units.



#32 Jon Isaacs

Jon Isaacs

    ISS

  • *****
  • Posts: 76136
  • Joined: 16 Jun 2004
  • Loc: San Diego and Boulevard, CA

Posted 12 May 2019 - 03:57 PM

If every user had to go through your calibration routine and create an individual table every year or few months, the instrument is useless for general practical use.   Users only need is to understand its not a precision laboratory instrument and to take the readings with a big grain or several big grains, of salt as described in my earlier post.  

 

Ron:

 

I am not suggesting you do it.  That's up to you.

 

I am proposing a method that could be used to investigate the drift and stability as well an caibrate it against a known standard.   That in itself would be valuable.

 

Laboratory instruments also require calibration and understanding if one wants to make accurate measurements. It's part of the game.

 

Jon



#33 Ron359

Ron359

    Apollo

  • *****
  • Posts: 1112
  • Joined: 21 Apr 2008
  • Loc: -105 +39

Posted 13 May 2019 - 09:48 PM

Here are a few thoughts. First, there is a great deal of useful information on the SQM website, including the specifications of the underlying detector. Note that the response is inherently linear; the device is basically counting the amount of time required to accumulate a certain number of photons. This is obvious in use; the amount of time required to obtain a reading is inversely proportional to the intensity of the light. Linear, not logarithmic.

Good to see at least Tony is looking into the details of the instrument on the website, as I've been suggesting people do.  It would be nice and much easier to critically evaluate the SQM if all the assumptions and 'variables' made by the designers were put in one document instead of scattered through several papers and even a thread of emails in one of their links.  I would note to Tony's comment,  that linearity is not perfect, but the photodiode frequency output is -+ 1% of the full scale, i.e. 2%.  Nothing is perfect so thats just one variable.   And there are probably others in the diode specs such as temp. etc.  All that adds up to their stated precision of -+ of 0.1 mag/arc sec or precision of 0.2 mag/arc sec of a stellar mag.  So as I've tried to point out, if you're getting readings that vary a couple tenths it may not be your sky -although it could be - you just won't be able to tell with a $100 instrument you can't calibrate to some standard 'candle'.   The meter readings documented over several years in prior posts show this amount variation in each of the instruments.  And also show that the instruments do vary in accuracy over time or due to variations in the build.  There are more 'variables' in the assumptions in the SQM documents of how bright a star should be in the FOV, the Milky way etc.  This may confuse those who assume that any number to the right of the 0 are both a statement of precision and are also accurate.  

 

One issue I've previously pointed out about the reduced sensitivity of the SQM to LED street light pollution is easily found by 'connecting' two dots in the SQM documentation.  The SQM uses the photodiode as the sensor to turn photons into a electrical frequency output.   BUT,  they also put a filter in front of the sensor - which i think basically acts as a UV-IR blocking filter as used my many film or ccd imaging cameras for many years.  In fact, its a HOYA filter CM-500.  This makes sense since you don't want the photodiode output influenced by non-visible wavelengths since the purpose to measure visible wavelengths of light to human eyes.  

 

LED streetlight wavelength output is well documented and has a maximum peak at about 450 nanometers (nm) A DOE study gives a variation of peak output from 445-460nm.  I've included one spectrograph of several LED streetlights v. High press. Sodium (HPS) output.  You can find the spectral curve sensitivity of both the photodiode and the HOYA filter on the SQM website.   I've done a couple quick screen grabs of the relevant spectragraphs and upload them here.   As you can see at 450nm the photodiode (by itself) response @ 450nm is right about 60%.   If you look at the spectragraph of the HOYA filter which covers the photodiode, it passes 88% at 450nm.  There is a steep falloff at shorter wavelengths but the max readings are easy to see.  So 60% of 88% of the 450nm LED transmission comes out to SQM 52.8% sensitivity and thats at the peak of LED streetlight output.  

 

As prior post noted, yeah, it works.  As the saying goes, even a broken clock is right twice a day. The  (undocumented) assertion of 75% sensitivity to LED output, just doesn't fit the SQM own facts.  Their spectrographs show the SQM will not 'see' ~50% of the contribution from LED streetlights.  If you have no LED streetlights in your area, you wont' have a problem.   But if LEDs are being installed or will eventually be 100% of your light pollution its easy to see your SQM readings will be dropping over time.  What your eyes sense about increasing or decreasing light pollution in the blue and near V LED output wavelengths is a whole nother subject and still subject to study.  The AMA has issued papers and guidelines, the DOE has a decent summary of the subject.  https://www.energy.g...ight FAQs_1.pdf

 

  LED_vs_HPS_70W.jpeg

 

The SQM was designed when LED street lights and light pollution was barely a thought of future city planners. A way to simply adjust the calibration of the SQM to correct for all the other things like drift or changing sensitivity would be nice,  but just isn't happening any time soon.  So as previously noted, for $100 bucks take your readings with a few grains of salt with understanding of the limitations and avoid the temptation to divide the decimals into little piles of decimal dust.   Clear and 22nd mag./arc sec. dark skies...  

Attached Thumbnails

  • Screen Shot 2019-05-13 at 10.20.01 AM.png
  • Screen Shot 2019-05-13 at 10.25.40 AM.png

Edited by Ron359, 13 May 2019 - 10:00 PM.


#34 Jon Isaacs

Jon Isaacs

    ISS

  • *****
  • Posts: 76136
  • Joined: 16 Jun 2004
  • Loc: San Diego and Boulevard, CA

Posted 13 May 2019 - 11:13 PM

A prime example of the difference between precision and accuracy.   

 

Actually it's an example of the potential for errors in the map used for comparison.

 

There was no way to measure the accuracy...

 

Jon


  • mountain monk and Redbetter like this

#35 ngc7319_20

ngc7319_20

    Viking 1

  • *****
  • Posts: 657
  • Joined: 25 Oct 2015
  • Loc: MD

Posted 14 May 2019 - 12:23 AM

 

LED streetlight wavelength output is well documented and has a maximum peak at about 450 nanometers (nm) A DOE study gives a variation of peak output from 445-460nm.  I've included one spectrograph of several LED streetlights v. High press. Sodium (HPS) output.  You can find the spectral curve sensitivity of both the photodiode and the HOYA filter on the SQM website.   I've done a couple quick screen grabs of the relevant spectragraphs and upload them here.   As you can see at 450nm the photodiode (by itself) response @ 450nm is right about 60%.   If you look at the spectragraph of the HOYA filter which covers the photodiode, it passes 88% at 450nm.  There is a steep falloff at shorter wavelengths but the max readings are easy to see.  So 60% of 88% of the 450nm LED transmission comes out to SQM 52.8% sensitivity and thats at the peak of LED streetlight output.  

 

I'm not sure I understand your concern about the SQM spectral sensitivity and LED street lights.  Yes, if you multiply the HOYA filter and photodiode at 450nm you get 53%.  But the same calculation at 500nm, where the sensitivity of the dark adapted eye is maximum, gives 75% x 90% = 68%.  So the ratio of the SQM 450nm / 500nm sensitivity is 53% / 68% = 78%.  The same ratio for the human eye is only 48%.  So if anything, one could argue that the SQM is *too* sensitive at 450nm.   So are you concerned that the SQM is too sensitive relative to the human eye at 450nm?  Or not sensitive enough since LEDs have more output at 450nm?

 

eye_spectral_response.PNG

 

From

https://www.telescop...al_response.htm


Edited by ngc7319_20, 14 May 2019 - 12:25 AM.

  • Redbetter likes this

#36 Redbetter

Redbetter

    Fly Me to the Moon

  • *****
  • Posts: 7318
  • Joined: 16 Feb 2016
  • Loc: Central Valley, CA

Posted 14 May 2019 - 12:25 AM

Good to see at least Tony is looking into the details of the instrument on the website, as I've been suggesting people do.  

I have read up on this, which is why I can tell that your claims are incorrect.   For example, the SQM is temperature compensated, and using it I can verify that I don't see drift with large temperature swings in the instrument. 

 

The SQM actually captures much of the peak of LED's, which is well off the scotopic peak as well...  See Fig. 13 from Cinzano 2005 to see how far off.  From this and some other data I gathered the scotopic response is only ~45-46% at 450nm.  The SQM exceeds that!  And the difference grows wider at even shorter wavelengths where the SQM response is far greater than visual.  So I will ask the other readers here:  will the SQM undercount or overcount LED input compared to our scotopic vision?   

 

I am going to assume that these apparent errors of fact and interpretation are unintentional.  I don't disagree that 4K LED's have the potential for greatly altering the nature of light pollution and the way it is quantified, but in some regards the SQM will do a better job of capturing that then our eyes can based on CIE scotopic response curves. 

 

The real solution is to move to warmer white LED's, at which point this becomes largely irrelevant.  There is no good reason to be using such dazzling LED's for night time use.  I suspect that 4K LED's will prove to be more of a fad/generational thing that is phased out.  I would still rather have these 4K's in the lamp outside than the unshielded HPS that was there before creating so much glare.  I can see the difference and that is what matters.


  • Jon Isaacs likes this

#37 Ron359

Ron359

    Apollo

  • *****
  • Posts: 1112
  • Joined: 21 Apr 2008
  • Loc: -105 +39

Posted 15 May 2019 - 01:31 AM

I have read up on this, which is why I can tell that your claims are incorrect.   For example, the SQM is temperature compensated, and using it I can verify that I don't see drift with large temperature swings in the instrument. 

 

The SQM actually captures much of the peak of LED's, which is well off the scotopic peak as well...  See Fig. 13 from Cinzano 2005 to see how far off.  From this and some other data I gathered the scotopic response is only ~45-46% at 450nm.  The SQM exceeds that!  And the difference grows wider at even shorter wavelengths where the SQM response is far greater than visual.  So I will ask the other readers here:  will the SQM undercount or overcount LED input compared to our scotopic vision?   

 

I am going to assume that these apparent errors of fact and interpretation are unintentional.  I don't disagree that 4K LED's have the potential for greatly altering the nature of light pollution and the way it is quantified, but in some regards the SQM will do a better job of capturing that then our eyes can based on CIE scotopic response curves. 

 

The real solution is to move to warmer white LED's, at which point this becomes largely irrelevant.  There is no good reason to be using such dazzling LED's for night time use.  I suspect that 4K LED's will prove to be more of a fad/generational thing that is phased out.  I would still rather have these 4K's in the lamp outside than the unshielded HPS that was there before creating so much glare.  I can see the difference and that is what matters.

Because its obvious you haven't actually understood or looked at Elvidge's LED spectrograph and understood the wavelength v. Radiance (energy) output that I presented to support my conclusion that the SQM will underestimate the LED contribution to LP,  and you just want to assert I'm incorrect, or just trash talking the SQM I was going to respond with a detailed discussion of wavelength v. output Radiance of LEDs.  

 

But since you already reject what I've presented from simple observations of published spectra, by just a CNs member with decades of  doing astro and earth sciences under my belt, it will waste much less of my time, and everyone else, to just recommend you read a recently published in-depth science field study of the SQM accuracy with regards to changing light pollution sources.   Since you probably won't read it, here is one conclusion presented after their extensive field measurements and data analysis:   Enjoy.  

 

"It is interesting to note that the massive replacement of street illumination with white LEDs will result in darker SQM readings and for the photopic vision due to the important blue component in their spectra. However the resulting sky is brighter for the human eye scotopic vision for the same radiance....As final conclusion, the SQM is a good instrument but is not good enough to trace the evolution of a change of sky brightness in a color changing world. "

 

Sky Quality Meter measurements in a colour changing world

A. S ́anchez de Miguel1,2,5,+⋆, M. Aub ́e1,+†, J. Zamorano2, M. Kocifaj3,4, J. Roby1,

and C. Tapia2.  Mon. Not. R. Astron. Soc. 000, 1–15 (2014) Printed 19 January 2017

 

https://pdfs.semanti...db88451d23c.pdf


Edited by Ron359, 15 May 2019 - 01:40 AM.


#38 ngc7319_20

ngc7319_20

    Viking 1

  • *****
  • Posts: 657
  • Joined: 25 Oct 2015
  • Loc: MD

Posted 15 May 2019 - 01:13 PM

Because its obvious you haven't actually understood or looked at Elvidge's LED spectrograph and understood the wavelength v. Radiance (energy) output that I presented to support my conclusion that the SQM will underestimate the LED contribution to LP,  and you just want to assert I'm incorrect, or just trash talking the SQM I was going to respond with a detailed discussion of wavelength v. output Radiance of LEDs.  

 

But since you already reject what I've presented from simple observations of published spectra, by just a CNs member with decades of  doing astro and earth sciences under my belt, it will waste much less of my time, and everyone else, to just recommend you read a recently published in-depth science field study of the SQM accuracy with regards to changing light pollution sources.   Since you probably won't read it, here is one conclusion presented after their extensive field measurements and data analysis:   Enjoy.  

 

"It is interesting to note that the massive replacement of street illumination with white LEDs will result in darker SQM readings and for the photopic vision due to the important blue component in their spectra. However the resulting sky is brighter for the human eye scotopic vision for the same radiance....As final conclusion, the SQM is a good instrument but is not good enough to trace the evolution of a change of sky brightness in a color changing world. "

 

Sky Quality Meter measurements in a colour changing world

A. S ́anchez de Miguel1,2,5,+⋆, M. Aub ́e1,+†, J. Zamorano2, M. Kocifaj3,4, J. Roby1,

and C. Tapia2.  Mon. Not. R. Astron. Soc. 000, 1–15 (2014) Printed 19 January 2017

 

https://pdfs.semanti...db88451d23c.pdf

Is this just a debate over the purpose of the SQM?  If the purpose is to measure the human visual impact of light pollution, then it seems fine.  If anything, the blue-ward response of the SQM seems too high to me.  On the other hand, if the purpose is to measure light pollution in some absolute sense, then yes, it will probably underestimate light in the far blue (which is increasing due to LEDs).  Is that the underlying question here?

 

The blue light from LEDs is especially damaging since light scattering goes something like the wavelength to the -4 power.  (It is why the sky is blue.)  Blue light will scatter much more than red or green, and scatter into geographical areas that are currently dark.


  • Redbetter likes this

#39 Redbetter

Redbetter

    Fly Me to the Moon

  • *****
  • Posts: 7318
  • Joined: 16 Feb 2016
  • Loc: Central Valley, CA

Posted 15 May 2019 - 02:15 PM

Is this just a debate over the purpose of the SQM?  If the purpose is to measure the human visual impact of light pollution, then it seems fine.  If anything, the blue-ward response of the SQM seems too high to me.  On the other hand, if the purpose is to measure light pollution in some absolute sense, then yes, it will probably underestimate light in the far blue (which is increasing due to LEDs).  Is that the underlying question here?

 

The blue light from LEDs is especially damaging since light scattering goes something like the wavelength to the -4 power.  (It is why the sky is blue.)  Blue light will scatter much more than red or green, and scatter into geographical areas that are currently dark.

That seems to be where Ron is having trouble with this.  The curves and the article he linked to above actually confirm what I said earlier:

 

"Your argument is perhaps backwards...the SQM is also sensitive to the red that we don't see well at night, so the lack of red contribution makes it harder to correlate to some existing photometry systems that are heavily weight toward the yellow or red when the sources change to a bluer type.  The V band and CIE photopic band have the same problem, even more so.  The CIE scotopic should be somewhat better in that regard (being insensitive to red), but ironically it doesn't capture the blue as well as an SQM will."

 

As the lighting source being tested moves from longer to shorter (LED blue) wavelengths the accuracy of the SQM with respect to scotopic (night time adapted vision) actually improves.  This can clearly be seen in the figures in the article with errors that center around 0.0 for LED's.    The one thing the article noticed from bench measurement of the device was that there was more of a response in the red tail all the way into the infrared for the SQM than the filter and sensor were supposed to impart.  So resultant SQM and scotopic under bluer lighting are closer to each other than I would have predicted based on the other known curves.   

 

The impact is that the SQM has historically seen the sky as somewhat brighter than it should (more toward the way the photopic curve would render it).  This didn't matter much in darker sky conditions, but is more of an issue in bright/urban areas, which are in the mesopic range anyway, muddling the comparison.  However, the SQM is now historical baseline and that historic baseline includes yellow/red/orange that are disappearing from the lighting and being replaced by blue. 

 

Probably the best way to summarize it is like this:

  • Historically the SQM has seen the more yellow/orange/red sources as brighter than it should have if one was trying to capture scotopic response alone.
  • For more light polluted environments, historically this has resulted in pessimistic (low MPSAS) readings for the sky. 
  • The shift toward LED's actually improves the accuracy of the SQM relative to scotopic (night adapted vision) because the redder source of error is minimized/removed. 
  • Unfortunately and ironically, improving accuracy in the scotopic range is not helpful in analyzing the changing sources of light pollution because the baseline has been skewed toward the photopic. 

But the one thing that is a glaring deficiency in the paper and most of the studies I have read is a failure to fully appreciate the impact of light thrown out at very shallow angles.  It isn't just the source of the light that matters, but the direction.  That shallow angle light passes through many times more atmosphere and is scattered, increasing local sky glow.  (It also creates a great amount of ambient glare, but that is somewhat of a different impact.)   I remember reading a summary of a study a few years ago confirming that this was more of an issue for light pollution than actual uplighting which, while being utterly useless, was scattered far less.  The LED conversions I have seen have all been full cut off replacement of unshielded HPS fixtures that are not adequately accounted for in the study.



#40 Ron359

Ron359

    Apollo

  • *****
  • Posts: 1112
  • Joined: 21 Apr 2008
  • Loc: -105 +39

Posted 15 May 2019 - 02:19 PM

Is this just a debate over the purpose of the SQM?  If the purpose is to measure the human visual impact of light pollution, then it seems fine.  If anything, the blue-ward response of the SQM seems too high to me.  On the other hand, if the purpose is to measure light pollution in some absolute sense, then yes, it will probably underestimate light in the far blue (which is increasing due to LEDs).  Is that the underlying question here?

 

The blue light from LEDs is especially damaging since light scattering goes something like the wavelength to the -4 power.  (It is why the sky is blue.)  Blue light will scatter much more than red or green, and scatter into geographical areas that are currently dark.

To answer your question all you have to do is read the OP and subject line by Tony Flanders..  What are some of the limitations of the SQM?  Of course a limitation might depend on your use or application of the data.  As Tony and the science paper writers noted there are many using SQMs for far more than just backyard astronomy.  The primary use is monitoring light pollution for many reasons.  But some have decided that an obvious big limitation for the primary purpose of monitoring changing light pollution is just not so, and have turned this into some sort of debate based on personal attack instead of the data and real limitations.  As noted in the conclusion of the science paper I quoted, the limitation of lack of sensitivity to the blue LED output also applies to human night vision assessments of increasing LP.  Myself and other astronomers I know have known about this problem for a few years.  I didn't even get into the Rayleigh blue light scattering issue which as you note makes LEDs far worse as light polluters, shielded or not.  

 

As I said in a previous post, bad data is worse than no data at all because it will lead you down the wrong paths of conclusion and harmful for decisions if action is needed and what action is needed.   Thats a standard principle of doing science or engineering.  It might even prove frustrating for backyard astronomers who are trying to fight LP or educate about good sources or ways to use lights in hopes of improving their backyard viewing or how many gallons of gas they need to buy to get to a real and not imagined dark sky site.  


Edited by Ron359, 15 May 2019 - 02:36 PM.


#41 Redbetter

Redbetter

    Fly Me to the Moon

  • *****
  • Posts: 7318
  • Joined: 16 Feb 2016
  • Loc: Central Valley, CA

Posted 15 May 2019 - 05:17 PM

 But some have decided that an obvious big limitation for the primary purpose of monitoring changing light pollution is just not so, and have turned this into some sort of debate based on personal attack instead of the data and real limitations. 

The personal attacks have come from you.  I have been focusing on analyzing the SQM, while you have been making "hat band" comments and belittling everyone's ability to understand precision, accuracy, meter calibration, or read a graph.  The people you are conversing with tend to be scientists, engineers, etc. so the condescending tone is not helping your argument.  We are astronomers as well and have made our own observations.  Mine happen to be at odds with yours...and yes, I have had decades of science/engineering and observing.

 

Turning back to the actual subject

As noted in the conclusion of the science paper I quoted, the limitation of lack of sensitivity to the blue LED output also applies to human night vision assessments of increasing LP.

 

The part in italics is incorrect, and an inaccurate summary of the paper or the figures within.  The paper demonstrates in Figure 7 that the error between the SQM reading and the actual scotopic spectra is 0.0.  The error is greater for warmer sources, and the meter sees them as brighter than they really are to scotopic vision.  I guess I will just have to continue repeating this until it is understood.  It is not insensitivity of the meter to blue relative to scotopic human vision that is the concern.  It is the greater sensitivity of the meter to the red that results in it reading brighter for warmer lighting sources.  This has been the historical norm.  As those sources are removed and replaced, it is the reduction in red/orange/yellow that is reducing the response of the meter when the replacement source is primarily blue. 

 

The reading with LED sources is actually more accurate and very closely correlates with scotopic vision.  The problem is the baseline with longer wavelength was less accurate, reading brighter than it was designed to be, behaving more like a photopic vision meter.  The irony is that you are rejecting it because it is now likely to show close to zero departure from scotopic.   

 

Furthermore the quote in a previous post is also somewhat out of context.  It was referring to the change from one state to another, not the direction of the accuracy of the underlying reading.  Granted, this requires some deeper reading/examination of the graphs to appreciate what is actually being said.

 

What the conclusion actually says and is relevant is:  "As final conclusion, the SQM is a good instrument but not good enough to trace the evolution of a change of sky brightness in a color changing world."  That is spot on.  We have the historical baseline SQM data, but we have to put it in context when comparing SQM values as LED conversions are made. 

 

As I said in a previous post, bad data is worse than no data at all because it will lead you down the wrong paths of conclusion and harmful for decisions if action is needed and what action is needed.

 

But it is not bad data.  The data is sound, but limited.  It is of a single photometric system, SQM, rather than a spectra.  That has always been known, and relying on a single band when the source spectra changes dramatically has always been a weakness/an approximation. 

 

The difficulty is the lack of supporting data as to the light source.  As the paper demonstrates, with an understanding of the relative source emissions, it is possible to account for this, but it means reconstructing representative spectra.  That is tricky and very much subject to interpretation. 

 

With regard to quality of data:

I have frequently found it necessary to adjust historical data when something was found to be different with regards to the basis of the data.  Some change occurred (intentional or not) and the new data was effectively on a different or even drifting scale, yet unknown to those measuring it at the time.  I have seen that in the lab, I have seen that in pilot plants, I have seen it in full scale operation.  I have seen it at home and most recently on a project with one of my son's teams at school.  I have had to adjust calibration factors, create compensation schemes for them, alter probes and sensors, or simply recalibrate.  The old data was still valuable/good, and after adjusting for known factors, it was often accurate as well.   And when such adjustments were made, other things suddenly began to "fit" that had not previously matched. 

 

"Bad data" to me is data that cannot be made useful because there is no way to compensate for them.  It cannot be characterized because of a lack of information.  These could be unexplained outliers.  They could be from upsets that put the equipment in a transient regime that could not be adequately characterized.  Often I discovered sampling errors, sometimes it was due to inadequate purging, not catching a live stream flow, samples that were sensitive to oxidation, sunlight, time in general, precipitation, degassing and/or liquid/liquid phase separation.  I might no the direction of the error, but have no way to quantify it sufficiently.
 



#42 Ron359

Ron359

    Apollo

  • *****
  • Posts: 1112
  • Joined: 21 Apr 2008
  • Loc: -105 +39

Posted 15 May 2019 - 06:27 PM

 

"Bad data" to me is data that cannot be made useful because there is no way to compensate for them.  It cannot be characterized because of a lack of information.  These could be unexplained outliers.  They could be from upsets that put the equipment in a transient regime that could not be adequately characterized.  Often I discovered sampling errors, sometimes it was due to inadequate purging, not catching a live stream flow, samples that were sensitive to oxidation, sunlight, time in general, precipitation, degassing and/or liquid/liquid phase separation.  I might no the direction of the error, but have no way to quantify it sufficiently.
 

Maybe some night when you are miles from nowhere under 21.3 mag/arc sec skies with your scope and your SUV or truck won't start at 4am,  because you are actually out of gas, even though your gas gauge reads 1/4 or more full, you'll figure out what bad data are.  And, duh!, cause you didn't believe your gas gauge was giving false readings, you made a bad decision to pass up that last gas station 10 miles before where you turned off the main road to your dark site. There is always hope your cell phone still works despite having 0 bars on your service reception.  


Edited by Ron359, 15 May 2019 - 06:33 PM.


#43 Enkidu

Enkidu

    Sputnik

  • *****
  • Posts: 47
  • Joined: 16 Oct 2017
  • Loc: EU

Posted 16 May 2019 - 04:18 PM

At the social level, the SQM (and SQM-L) make it possible to calibrate other people's observations against one's own experience.

I agree these are social measurements; very useful as a general indication between observers. In addition, regular observing with a friend who also has a meter is a good way to calibrate : )

 

 

Not only are the dark end differences probably more important but they're also more difficult.

Absolutely agree on this, esp. for light pollution maps (as you mentioned).

 

 

There is a company called Astromechanics that now produces an SQM monitor which reads a 10 degree fov.

I'll be receiving this Astromechanics meter and look forward to testing locally (~18.4) and at nearby sites up to ~21.4. The narrow FOV seems more practical. I'll calibrate the Dark Sky Meter app to it, for a second reading.

 

 

There is a smartphone app, dark sky meter, that uses the phones camera to measure sky brightness. Based on the comments in this thread I wonder how accurate the measurements are.

The app's creator seems passionate about getting the best results, and optimistic about getting close to a dedicated device. He mentions moving to raw data in order to avoid the jpeg compression that makes dark readings difficult.

 

My phone has a 60˚ FOV and "hardware drift" is unlikely. If the developer's goals are met, more widespread use would generate a lot of data, and could help create awesome light pollution maps.

 

For now, I'd like to use the Astromechanics meter with Glenn LeDrew's visibility chart to check the contrast of extended objects. I think that's the only direct, personal impact this device may have (for me) during an observing session.



#44 Ron359

Ron359

    Apollo

  • *****
  • Posts: 1112
  • Joined: 21 Apr 2008
  • Loc: -105 +39

Posted 16 May 2019 - 07:13 PM

This latest bit of news for health effects concerns of excessive blue light in LED light pollution just further highlights the need for a SQM type monitoring device that accurately measures the huge amount of short blue wavelengths peak output of most LED streetlight and other sources of LED LP.  

 

https://www.cnn.com/...trnd/index.html

 

Only a few give much more than a fleas butt for backyard astronomers' faint fuzzies, but once the environmental and medical communities come together they will have 'the clout' with potential to really reduce the unnecessary increases and perhaps roll-back the huge loss of the night sky to LED light pollution that is underway.  Those doing the studies to scientifically document LP need an instrument with indisputable accuracy that will stand up in a court if needed.  

 

I bet the builders never imagined this need when it was first built for backyard astronomers.  But IMO if Unihedron wants to remain a leader as a maker of these devices, its far past time to update and upgrade their SQM sensor sensitivity to shorter wavelengths of peak LED output and fix apparent QC problems in their build that have been shown in this thread and other professional study I cited earlier.  A problem for backyard astronomers is such an improved instrument might cost well above a $120 bucks.  



#45 ngc7319_20

ngc7319_20

    Viking 1

  • *****
  • Posts: 657
  • Joined: 25 Oct 2015
  • Loc: MD

Posted 16 May 2019 - 09:08 PM

 

I bet the builders never imagined this need when it was first built for backyard astronomers.  But IMO if Unihedron wants to remain a leader as a maker of these devices, its far past time to update and upgrade their SQM sensor sensitivity to shorter wavelengths of peak LED output and fix apparent QC problems in their build that have been shown in this thread and other professional study I cited earlier.  A problem for backyard astronomers is such an improved instrument might cost well above a $120 bucks.  

Well, there is probably much we can do ourselves.  We can cross-check our meters, cross check with friend's meters.  Look at the detectors with a magnifier, and see if they look OK.  Send the bad meters back.  Send the sick ones in for service or whatever.

 

How about if we find a blue filter we could stick over the SQM to measure the 450nm light?  Or maybe Unihedron could make a run of altered meters with enhanced blue sensitivity? 

 

I personally would not want the current meter design to go away -- I want to continue adding to my 15+ years of data on different observing sites with a uniform set of readings.



#46 Redbetter

Redbetter

    Fly Me to the Moon

  • *****
  • Posts: 7318
  • Joined: 16 Feb 2016
  • Loc: Central Valley, CA

Posted 16 May 2019 - 09:44 PM

 

I bet the builders never imagined this need when it was first built for backyard astronomers.  But IMO if Unihedron wants to remain a leader as a maker of these devices, its far past time to update and upgrade their SQM sensor sensitivity to shorter wavelengths of peak LED output and fix apparent QC problems in their build that have been shown in this thread and other professional study I cited earlier.

 

In troubleshooting, it is crucial to identify the root cause of a problem, rather than misdiagnosing and treating only a symptom.  Misidentify the cause and you are likely to implement a "solution" that will not work.  The above is an example of that.

 

As has been pointed and is documented in the paper linked to in the form of figures, the SQM sensitivity to shorter wavelengths of LED's is not the issue.  As the paper cited demonstrates the error is effectively zero with respect to the short wavelengths.   As lighting shifts to LED's the meter more closely tracks scotopic vision than before.  The problem is not with the blue end of the response, it is with the red.  This might well seem counterintuitive, but when you examine the paper, that is what one sees. 

 

The meter has been somewhat too sensitive to the longer wavelengths (again as shown in the paper).  Therefore as those longer wavelengths disappear that bias relative to scotopic vision also disappears.  The blue emissions are correctly counted, but the reading doesn't brighten with respect to scotopic as expected because of the reduction in red.  The former bias to the red is part of the historical measurements with the SQM.    

 

As for what could be done going forward, that is trickier.  A meter that more effectively mimicked scotopic with redder sources could be made, by more effectively filtering the red.  But it would show a step change up in MPSAS for the current lighting mix when compared to the present meter.  So light polluted areas will suddenly read darker than the present meter.  Ironically, the blue end is the area where the response would least need adjustment (and might require adjustment in the opposite direction if one tried to balance with the current mix.) 

 

Again it is a matter of reading the graphs and understanding what they mean when it comes to the hardware.



#47 havasman

havasman

    Cosmos

  • ****-
  • Posts: 9094
  • Joined: 04 Aug 2013
  • Loc: Dallas, Texas

Posted 18 May 2019 - 03:22 PM

This may be interesting to an old objective data aficionado. The Astromechanics device arrived this morning.

 

SQM's, sm.JPG


  • Roragi likes this

#48 Roragi

Roragi

    Messenger

  • -----
  • Posts: 421
  • Joined: 24 Oct 2013
  • Loc: Madrid, Spain

Posted 19 May 2019 - 10:13 AM

I am happy with my unit, if any deviation is between them, but I think some take it very seriously, it is still a device to have a reference of how the sky is at night, mine is the classic 80 degrees and very happy with it.

 

 

Roberto.



#49 WoodyEnd

WoodyEnd

    Vanguard

  • -----
  • Posts: 2118
  • Joined: 21 Mar 2016

Posted 19 May 2019 - 05:53 PM

From a non experts point of view this is starting to sound like a "how many angels can dance on the head of a pin" argument.  To be truly scientific you would need multiple SQM's. One that would accurately cover our eyes spectral response and others that measure photographic light frequencies.  



#50 ngc7319_20

ngc7319_20

    Viking 1

  • *****
  • Posts: 657
  • Joined: 25 Oct 2015
  • Loc: MD

Posted 19 May 2019 - 08:56 PM

I am happy with my unit, if any deviation is between them, but I think some take it very seriously, it is still a device to have a reference of how the sky is at night, mine is the classic 80 degrees and very happy with it.

 

 

From a non experts point of view this is starting to sound like a "how many angels can dance on the head of a pin" argument.  To be truly scientific you would need multiple SQM's. One that would accurately cover our eyes spectral response and others that measure photographic light frequencies.  

 

I guess it depends on what you want from the SQM.   I was hoping to track changes in sky brightness over the years at my location.  I was pretty happy with it... until it showed my skies getting darker and darker over 10+ years, and then it turns out to be a problem with the meter.  It probably works well if you want to compare relative brightness at different sites over a short time period (hence avoiding effects of long-term changes in the meter).  Or just want to know if tonight is darker than last week, etc.  I think it is a hugely useful device... so long you understand the limitations.


Edited by ngc7319_20, 19 May 2019 - 08:57 PM.



CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics