Light Pollution and Skyglow

[mark.m]

I've been studying astro cameras the past few weeks, trying to decide what I will get to replace my SBIG ST-9 camera. And as I've analyzed camera specifications, I've come to the realization that once a camera is good enough that background skyglow dominates image noise, no additional dollars spent on a better camera (lower noise, better pixel size match, ...) will gain anything in performance.
And so I set out to measure skyglow here in Portsmouth. I went into my image archive and pulled out 9,030 images that I've taken since 12/5/2019, and measured the amount of skyglow in each. I learned a few things:

• There are two distinctly different sources of skyglow, and they behave differently: moonlight and light pollution.
• Light pollution has a distinct color to it -- it seems to be something in the yellow/orange family.
• Skyglow due to moonlight is strongly blue. (Probably for the same reason that the sky is blue when the sun is out, the sky is also blue when the moon is out.)
• Skyglow due to moonlight can be as much as 5X to 10X as bright as light pollution (here in Portsmouth).
• A typical amount of light pollution here in Portsmouth is about 1 photon per second per square arcsecond of sky
• (This one surprised me.) The amount of light pollution doesn't seem to correlate with the time of the night. (I really expected background skyglow to drop around the time of "last call" in Newport.)

[Pete]

To quote Lord Kelvin, “When you can measure what you are speaking about, and express it in numbers, you know something about it, when you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind; it may be the beginning of knowledge, but you have scarcely, in your thoughts advanced to the stage of science.”

Very nice Mark. Maybe it's one of my "being an engineer" things.

Can't argue with you regarding camera noise but while I've got some uncertainty about the Nyquest theorem (that 2 arc seconds per pixel is optimum sizing for balance between sensitivity and resolution) it makes sense for local seeing. But I'm not sure how it applies to local noise in the form of light pollution.

Those of us who do scientific imaging find ourselves in a very anxious position right now. CMOS detectors have problems with linearity, random pixel noise, and amp glow and just aren't suitable for quantification. But Kodak has ceased CCD production and Sony is terminating CCD production next year. Any of us with an older CCD camera have to be worried about what we'll be doing when the camera dies. To the point where, even though it's probably not going to fit into the budget, I'm searching for a replacement to my SBIG ST-8XME. Only with a CCD larger than the current 9.18mm X 13.77mm chip.

Your Godzilla mount makes your situation unique as you needn't integrate the image with guiding. But please keep us posted as you investigate your next (and possibly last) CCD camera.

[mark.m]

Pete:
I don't know where the quote comes from, but "follow the money" seems to be quite apropos here. When Sony and Kodak (ON Semi) saw the handwriting on the wall about the future of CCD imaging, the first thing they did was cut way back on research and development investment in CCD technology. That seems to have happened maybe 5-10 years ago.
Meanwhile, they (particularly Sony) realized that CMOS was the future and they have invested heavily in recent years. As a result, the most recent generation of CMOS sensors (Sony IMX455, GPixel's GSense 2020 and GSense 400) are really impressive components. Dark noise is no longer an issue at all, read noise is several times smaller than the best that's available in CCD, and dynamic range is better than what most CCD chips can achieve. Further, the camera manufacturers have found ways to deal with amp glow; they're now using USB 3.0 and large in-camera DRAM memory to read out the sensor so fast that amplifier glow has become quite small and relatively easy to remove during image calibration. Linearity of these three devices is better than the linearity of my SBIG camera. Several AAVSO people using these most recent CMOS cameras have found them better for science than the vintage CCD devices.
There certainly are some remaining issues -- the Chinese camera manufacturers (ZWO, QHY) are using their in-house engineering teams differently than SBIG or FLI did, and it's sometimes hard to get reliable (and well-translated) performance numbers and design info.
I've been assessing cameras using 4 scenarios ("use cases"), each of which covers one of my most challenging observation needs. Each scenario has two stars of interest (a variable one and a reference star), although in 3 of the 4 scenarios the reference star is trivial and can be ignored. Scenario #1 is a variable star that is very bright (mag 6) in near-infrared, being compared to a star 6 magnitudes fainter. (This scenario checks dynamic range.) Scenario #2 is an exoplanet transit and checks a camera's ability to support high-precision photometry. Scenario #3 measures the brightness of a red giant at blue wavelengths, checking a camera's blue sensitivity. Scenario #4 just looks at a faint star at visual wavelengths.
For each scenario, I find the best exposure strategy for each camera (CMOS cameras often have mode and gain settings). Then I measure the resulting photometric accuracy. That's how I found that once skyglow becomes limiting, camera optimization is way different than what you should do if you're sky is really dark (at least if you're trying to image individual stars, not nebula).

[Pete]

Yup, no new CCD sensors for some time, and all CCD cameras are a bit dated as they use these old chips. Chip size seems to be stalled in the mid-range too. My 12 year old SBIG ST-8xme has a 16mm diagonal, and that's exactly the size all of the midrange priced CCDs are at.

You get a much larger chip for your money with CMOS for sure. If you can use CMOS for photometry then I've no worries about using them for astrometry.

The July issue of S&T has a 4-page review of QHY latest camera. $5000 for a HUGE 24mm X 36mm chip!!! The article goes into detail comparing CCD vs. CMOS and overall Dennis DiCicco believes the transition is inevitable. Educational.

Well, I'm out tonight experimenting with supernova detection. Something new, eh?

[menardre]

Mark

Thanks for the very interesting and informative article.

The astrophotography camera market is really changing rapidly with many new cameras. It is hard to keep track of what is going on.

Your article really helps.

Roger

[mark.m]

It is hard to keep track of what is going on.

Very true -- one of the things that this new technology is uncovering is where our prior understanding might have been a little bit sloppy.

Case in point relates to Pete's earlier comment about Nyquist sampling and finding the right pixel size. There's been a recent thread on one of the AAVSO forums that has clarified the old rule-of-thumb about wanting about 3-5 pixels per full-width-half-maximum (FWHM) of star profiles. When the dust settled, we had pretty much come to a consensus that the 3-5 pixels/FWHM is a minimum, not a target (at least for science imaging). When looking at the SNR associated with stellar brightness measurements, there is no penalty for having "extra small" pixels, as long as the camera internal noise is pretty small so that performance is limited by skyglow. (And low noise is one of the things that's becoming more common in astro cameras these days.)

The camera in the S&T review (QHY600M) is definitely one that's on my short-list. But I'm also realizing that once the camera reaches a certain point, the most economical way to improve image quality is to improve the optical quality of the system. I'm using a 20+ year old classic Meade SCT, and never get better than 4 arcsecond FWHM. A better-corrected optical system will improve my photometry more than spending twice as much on a super-duper camera.

[Galactus]

I don't know where the quote comes from, but "follow the money" seems to be quite apropos here. When Sony and Kodak (ON Semi) saw the handwriting on the wall about the future of CCD imaging, the first thing they did was cut way back on research and development investment in CCD technology. That seems to have happened maybe 5-10 years ago.

Mark, while not exactly germane to the topic, the line, "follow the money" is spoken by Deep Throat (Hal Holbrook) to Bob Woodward (Robert Redford) during a midnight meeting in a parking garage in the film All the President's Men. Oddly enough, Woodward has reported that Deep Throat never said the line, but that this was the screenwriter's device.

[mark.m]

Oh, neat! It never occurred to me that the origin was that recent. I figured it would date back to the mid 1800's and the great railroad barons or maybe even to the infamous Charles Ponzi (early 1900's).
- Mark