Hacker News Reader

Comment by hosh | Hacker News Reader

That works the best close to noon, but I don’t think that works so well at around dawn or dusk.

Latitude also matters. The closer it is to the arctic or antarctic circle, the iffier that gets.

This isn’t a trick so much as just how the sun and the earth defines the plane of the ecliptic and how we experience it because of Earth’s rotation.

There are some interesting things that can be done with starlight navigation. The Polynesans, for example, might navigate by latitude based on how far up Sirius is in the sky, or using the little dipper (and Polaris) in the Northern Hemisphere as a clock.

View on HN

Starlight navigation is incredible. I was in the marines with a helicopter pilot who had a second set of wings on his uniform, an odd pair I'd never seen. Turns out he was a starlight navigator on C-130s before becoming an officer and getting his pilot's wings. He used to wear the starlight navigator ABOVE the pilot wings, which I always thought was incredible. The training and skills you need to get this set of wings is quite an undertaking.

I can navigate with a map and compass, but you're using terrain to help you locate yourself on the map. Being able to do it in the expanse of the ocean or the air, with only the stars is pretty amazing.

The concept of celestial navigation is actually pretty simple. Actually doing it requires physical skill.

The general idea is this. Take for example the star Altair, at any given instant it is directly above exactly one spot on Earth. If you were at that spot and measured it's elevation above the horizon, it would be 90 degrees. That point is called the "Geographic Position (GP)". So, if you were at some unknown location and measured Altair to be 45 degrees, you would know you were 45 degrees from Altair's GP, and all of the points satisfying that can be plotted as a circle on a map. Repeat that for another star and you end up with two circles that intersects at two points. You could repeat with a third star, but in practice you have a rough idea of where you should be, and you choose stars that produce a large enough circle that you can disregard one of the two points. And taking that "in practice" one step further, the diameters of the circles are so large, you really just need to draw part of the circle as a straight line, and that lets you plot in a small enough area that map distortions are of no significance (unless your very close to the poles). There's a standard form called a "VP-OS" which makes the process a lot easier. A video showing how to do it is [1].

To determine the GP of a star (or planet) you need an ephemeris of some sort, there's software to do it, but a copy of the Nautical Almanac provides enough information to make the process simple enough to do with paper and pencil.

Actually measuring the elevation of a star on a moving, bobbing ship is where the real skill comes in. At the equator, being off by one degree means your result will be off by about 60 miles. Skilled navigators can fix their position within a few hundred feet!

[1] https://www.youtube.com/watch?v=bVIPorb_aMM

Leave it to hacker news comments to summarize something I thought was black magic into paragraph that I completely understand. Not sure I could replicate it but it makes sense to me now.

So, this has been around a lot longer than we had accurate star charts that actually understood the motion of planets and stars, right? Do you have any summary of how these charts that describe where a certain star is above on the earth on a specific day came from before we understood the motions of the stars?

I'm not too familiar with the ancient history of it, but it requires an accurate clock. There's a good book called "Longitude" that goes over the development of the clock that could be used at sea. But, before then, it was well known that you could measure the altitude of Polaris to get your latitude, so ships would sail to the latitude they wanted, then turn East or West.

We've had accurate star charts for a long time, Hipparcus had made accurate enough measurements to detect the Earth's slow 26,000 year wobble called precession. The key piece missing for navigation was a clock.

I assume that doing this technique with a computer and a camera could achieve really good accuracy - you really just need a zoomed image showing the horizon and some stars in a few directions, and by averaging across hundreds of stars you ought to be able to achieve perhaps 1/100th of a pixel of precision, which works out to 1.8 meters of position accuracy. Modern tech can see stars even in daylight too.

I wonder why that technique isn't used instead of GPS - back in the 70's before GPS was deployed, it surely must have been an attractive option.

Even today, in warzones with GPS blocked, it seems like a good option.

On land, it can be hard to very accurately measure the horizon (mountains etc).

However you can use a similar technique by using a weight on a string to make a perfectly vertical line, and seeing the alignment between stars and that.

https://en.wikipedia.org/wiki/Lockheed_SR-71_Blackbird#Astro...

Uncertainty in atmospheric refraction limits exactly how accurate you can get. Celestial Navigation books say it's better to make several measurements of the same star than to measure a bunch of different ones.

From what I understand, some land surveyors often still use a sun sighting to determine true North.

GPS has the advantage of working when it's cloudy, or when the sky is blocked by trees or buildings. And a GPS receiver is pretty cheap, durable, compact, and foolproof compared to a telescope and camera system which will have to be recalibrated regularly.

Something I only learned recently: Boeing 747's had a sextant port, for use in navigation. This is only one of several aircraft variants (including military aircraft) which used sextant navigation.

<https://virginflightdeck.blogspot.com/2010_09_01_archive.htm...>

Mentioned at least once on HN: <https://news.ycombinator.com/item?id=5719703>

OK, rather more than once: <https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...>

IIRC, ICBMs also use starlight navigation.

They do, you may also find this [1] interesting, it doesn't cover the starlight navigation system but the 19,000 piece Advanced Inertial Reference Sphere (AIRS)

https://www.thedrive.com/the-war-zone/30254/this-isnt-a-sci-...

Why does it work worse in higher latitudes and closer to dawn and dusk?

The shadow of the tip of the stick will always move in a straight line from east to west right? At high latitudes I could imagine the shadow gets so long that finding the tip might become hard. Around dawn and dusk, what is the problem? Does the shadow become too faint? Does the shadow move to slow to give a big difference in position?

At dawn and dusk, aka twilight, the sun is below the horizon. So you only get indirect illumination. Contrast is terrible, and shadows will be diffuse. The direction should still be mostly right, but you will be hard pressed to define the end of the shadow.

The position of the sun can be a bit weird depending on the season the further away from the equator you are. Especially in the periods where the sun never rises above the horizon. Other than that I don't know exactly what problems you might have. The midnight sun perhaps flips the direction?

Reply on news.ycombinator.com