NavList:

Some parts of the celestial sphere have relatively few bright stars and a poor selection of "official" navigation stars. These are "star deserts". Where are they? What are the worst parts of the sky for finding navigation stars?

I started thinking about this tonight because I realized we're entering the "Null Milky Way" season. For the next few months in mid-northern latitudes there's no visible Milky Way in the evening sky, at some point during the evening. The North Galactic Pole passes high overhead, and the "ring" of the star clouds of the Milky Way runs around the local horizon, rendering them invisible. Really this only applies at one specific latitude (north Florida -ish), but it's more or less true across a broad band near there.

Many of the brighter stars are near the belt of the Milky Way, though it's certainly not an overwhelming feature of their distribution. With the Milky Way down, we're looking into one prominent star desert in the evening this month and next. Among the brightest stars, Arcturus is close to the North Galactic Pole. If it wasn't there, the entire region around the NGP would be a huge star desert.

A couple of the star deserts are familiar and relatively obvious. There's a big hole without bright stars generally between Altair and Fomalhaut, near the South Galactic Pole. This includes the zodiac constellations Capricornus, Aquarius, and Pisces, which are famously faint. There's another big empty area centered on Hydra after its head is cut off, including the small, very faint (and yet ancient) constellations Crater and Corvus. This star desert would be even more prominent were it not for the arbitrary inclusion of Gienah in Corvus among the 57 nav stars --one of the faintest of the "official" navigation stars.

I was pondering writing some code to generate a KML (a map file) that I could drop into Google Earth displaying the positions of the navigation stars at some arbitrary date/time with zones around them indicating their relative brightnesses (magnitudes). Then I thought... ChatGPT. So I submitted my list of the 57+1 navigation stars with radii for my zones already settled. I let it generate the KML file, which it happily did using simple plain english prompts on my end (meaning that I said, "make me a KML with these lat/lon points surrounded by octagons of these dimensions"). I started with just five stars as an experiment. Why octagons? Becuase the KML data format doesn't support circles natively, and if you try to draw them you get rather large numbers of points. ChatGPT's first version was correct, but it treated my "radii" as sizes in meters making the octagons invisible where I was imagining five or ten thousand times bigger, so I had to ask again. It then produced the KML file, as requested, in about fifteen seconds and also displayed the code it used to generate the files. I did not need the code, but it was nice that it was included. I requested changes, and it complied. I was not surprised that ChatGPT had trouble drawing an octagon for Polaris. With a simple algorithm, that's a special case.

I'm attaching the KML files created by ChatGPT. There are two. I asked it to scale up the radii again, which it did, and as an example of its random "creativity", it also changed the style of the octagons. On the first pass they were transparent and blue. On the second, solid white. If you have Google Earth installed, just download these KMLs and open them. They should by default open in Google Earth. You can then zoom out to a planet-wide view and roll the sphere around. Ignore the Earth. Ignore the continents and countries. It's just a virtual celestial sphere that you can roll about to see the star deserts. It's a fun toy.

Good luck identifying the stars visually in these KML maps! Each octagon represents a star, but they don't exactly jump out at you. The coordinates are your friend here. For example, there's an octagon centered on Kansas at longitude -99.2° and latitude 38.8°. That's Vega. The latitude is just the star's dec as usual. The longitude is 180°-RA, where RA is the right ascension of the star or equivalently the longitude is SHA-180°. Note that this places the stars in a reverse celestial sphere view where the constellations are shaped and oriented as they are in the sky instead of the inside-out view often considered traditionally as "correct" since the traditional view is from "outside" the imaginary celestial sphere.

I'm including a couple of images. These are screen caps from Google Earth to illustrate how the KMLs should look. Load the first. Then load the second over it. Both were generated by ChatGPT. I have left them in their weird glory, just as produced by the robot. But the thing is, they do just what I wanted them to do, and they were produced in perhaps a tenth of the time that it would have taken me to generate them in my own code. In fact, writing this NavList message has taken me far more time than generating the maps of the star deserts.

After I was essentially done, I asked ChatGPT if it recognized the coordinates. It did not. I then explained I was observing its evolution, and it replied like it was running for office. Maybe that's how A.I. will take over the world... the old-fashioned way. See its flattery in the last image below.

Frank Reed

PS: I just peeked inside the KML files, and really, the only thing I object to, at a quick glance, is something that I have seen many, many times in human-generated mapping files. Many of the latitudes and longitudes are provided with a precision that implies subatomic positioning. We don't need that many digits past the decimal point. :)

File:

File:

File:

File: