Celestial Navigation with OpenCPN

[seandepagnier]

I added some very basic support for celestial navigation. Sun Moon, and planets are supported. You can enter an elevation, and various error parameters, and it draws an "area of position" for it. Kind of like a line of position, but it is fat if you tell it the time accuracy is 100 seconds, or the sextant is only good to a degree.

I have here a screenshot example with sun, moon, and mars shots


Improvements would:
1. include stars (which ones?)
2. finish implementing azimuth shots, and compensate for global magnetic variations (so you can use your compass instead of sextant, or use both for even more accuracy)
4. compensate for refraction
5. compensate for parallax (moon)
6. what else?

Unfortunately due to the high-precision planet database, the patch is 1.5 megs and I cannot post it easily on the forum.

[cagney]

Looks great! Do you have a download link?

As for which stars to include, a good starting point would be the stars used by H.O 249 / Selected Stars.

6. what else?

I'll tell you when I have had a chance to test your program :-)

Thomas

[cdennyb]

OMG that is so baddd azzzz. I can hardly wait to give it a try. Do you have a place to enter stuff like height of eye of observer, etc.?
Me too, where's the download link to try it out?
nice work. You ALL rock.

[seandepagnier]

I added a bunch of stars, I got azimuth shots working.. they are cool, and they work! It is a line of position that is perpendicular to the azimuth lop at the point. I did some test sights. If you give both elevation and azimuth, you get location from only a single shot. Otherwise 2 azimuth shots could intersect.

Also compensating for observer height, refraction, and parallax (for the moon) as well as edge shots.

Unfortunately I don't own a sextant, so I cannot improve accuracy more than it is. Anyone have a spare?

I need to make azimuth shots work for magnetic north now as well as true north which should result in some interesting (and hopefully weird) areas of position, since the declination changes based on geographic location (screenshots to come)

I'll post a patch soon, but I would rather just push into a branch on the opencpn git.

[seandepagnier]

Here shows two shots where the sun was claimed to be at 270 degrees. The orange is magnetic compensated, cyan is not. Both shots claim up to a degree of error in azimuth (determining their thickness), and incline unknown (no sextant)

You can also see how the lines of position end up at their respective poles (geographic and magnetic)

I have seen cases where the line of position for a magnetic compensated shot is S shaped, or even inflexes 5-6 times and some unusual shapes!

[seandepagnier]

here is a nice shot I took of the moon giving its azimuth relative to magnetic north.

[seandepagnier]

Download the patch here, and let me know some feedback!

http://www.dacust.com/inlandwaters/p...vigation.patch

[cagney]

Thank you very much for the excellent celestial plugin.
As someone who has a long way to go before the miles done with a gps on-board catches up with the miles done with only a sextant and a chronomete(not counting a primitive RDF), perhaps some comments could help.

I love the feature where the celestial line of position(LOP) is plotted directly on the chart. With a paper-chart this method is only available with very high sextant altitudes in tropical regions.
For centuries mariners has been forced to approximate this LOP with a tangent to the LOP, in their vicinity, with the help of ugly trigonometry. Many mariners lost sight of what celestial navigation really is and thought it was the forced workaround solution itself!
On a "pure" theoretical level a sextant height gives you the distance to a known point on the globe, combine this with a bearing of this place at the same time, and you have a fix, as everyone who has done coastal navigation(without a gps) knows. The present state of the celestial module is a great illustration to this.

When it comes to using the plugin for practical celestial navigation a few changes are necessary.

A compass bearing of a celestial body at sea can never create a useful LOP, the accuracy is way to poor.
I think that the plugin assumes an accuracy of 1degree. Even on a big vessel with access to a pelorus, this kind of accuracy will be a struggle. It is definitely not obtainable with a hand bearing compass on a small boat with any kind of sea running. Even an accuracy of 5 degrees will be a struggle, and then only of objects with a low altitude.
But let's assume an accuracy of 1 degree and an altitude of the body of 30 degrees. The geografical position is then (90-30)x60 miles away (3600 M ).A rough estimation of the accuracy, using the rule of thumb that a 1 degree error after 60 M gives a cross error of 1 M, we have a possible error of 60 M. Not usable for practical navigation, and no point, as two normal sights gives way better accuracy.
If someone could invent a device where bearings could be taken to within 0.1 degrees, at sea, and used for an altitude of less than say 60 -70 degrees, the situation would change.

Suggestions for the plugin to make it a practical tool underway.

Don't use the computer for time. If you take celestial serious you have to have an independent timepiece, that you carefully track for performance. There is no point in doing the same for the computer time. You cannot have the computer next to the sextant and enter the time instantly. A round of 3 - 4 star sights takes time and the calculations may happen quite some time later.
This will also make it possible for the user to build confidence in the plugin by comparing results to examples in text books.
*The user must be able to enter the time independent of the computer.
*The azimuth calculation in present shape must go or clearly be labled "for illustration only".
*Calculate and display the azimuth for every sight, can be used for checking the compass under certain circumstances.
*State clearly how altitude corrections are done. Obviously not all cases are covered(low altitudes), as it is not possible for the user to enter temperature or barometric pressure.
*Change "height" to "height of eye". Generally use standard terminology, "altitude", not "elevation", see Bowditch.
*Use lower "Limb" as default, as this is what is normally used, if available, with a sextant.
* Let user enter sextant altitude using degrees minutes and decimal minutes, as this is what you get with most sextants, not degrees and decimal degrees.
*Include calculations for running fixes, two sights of the Sun for example. User enters course and distance between sights.
*Include a star identification feature. User enter time, sextant altitude and DR position (or better). The plugin names the star and ask if the user accept, before producing a LOP.

Thomas

[seandepagnier]

Originally Posted by cagney

Thank you very much for the excellent celestial plugin.
As someone who has a long way to go before the miles done with a gps on-board catches up with the miles done with only a sextant and a chronomete(not counting a primitive RDF), perhaps some comments could help.

I love the feature where the celestial line of position(LOP) is plotted directly on the chart. With a paper-chart this method is only available with very high sextant altitudes in tropical regions.
For centuries mariners has been forced to approximate this LOP with a tangent to the LOP, in their vicinity, with the help of ugly trigonometry. Many mariners lost sight of what celestial navigation really is and thought it was the forced workaround solution itself!
On a "pure" theoretical level a sextant height gives you the distance to a known point on the globe, combine this with a bearing of this place at the same time, and you have a fix, as everyone who has done coastal navigation(without a gps) knows. The present state of the celestial module is a great illustration to this.

When it comes to using the plugin for practical celestial navigation a few changes are necessary.

A compass bearing of a celestial body at sea can never create a useful LOP, the accuracy is way to poor.
I think that the plugin assumes an accuracy of 1degree. Even on a big vessel with access to a pelorus, this kind of accuracy will be a struggle. It is definitely not obtainable with a hand bearing compass on a small boat with any kind of sea running. Even an accuracy of 5 degrees will be a struggle, and then only of objects with a low altitude.
But let's assume an accuracy of 1 degree and an altitude of the body of 30 degrees. The geografical position is then (90-30)x60 miles away (3600 M ).A rough estimation of the accuracy, using the rule of thumb that a 1 degree error after 60 M gives a cross error of 1 M, we have a possible error of 60 M. Not usable for practical navigation, and no point, as two normal sights gives way better accuracy.
If someone could invent a device where bearings could be taken to within 0.1 degrees, at sea, and used for an altitude of less than say 60 -70 degrees, the situation would change.

The actual accuracy can be specified. It is true that 1 degree error is nearly impossible to achieve under typical circumstances, but if you only have one body, and you have elevation, and 5 degrees error on azimuth, it does narrow down the area quite a bit.

Also, until I ran out, i was selling digital magnetometers (digitalsurveyinstruments.com) which could actually measure azimuth to better than 1 degree accuracy. On a moving boat.. I'm not there yet!

Unless this feature causes problems, or should be changed, I think we should leave it, and it is interesting to use.

Suggestions for the plugin to make it a practical tool underway.

Don't use the computer for time. If you take celestial serious you have to have an independent timepiece, that you carefully track for performance. There is no point in doing the same for the computer time. You cannot have the computer next to the sextant and enter the time instantly. A round of 3 - 4 star sights takes time and the calculations may happen quite some time later.

It should be possible to manually modify the time, I didn't add this yet.

This will also make it possible for the user to build confidence in the plugin by comparing results to examples in text books.
*The user must be able to enter the time independent of the computer.
*The azimuth calculation in present shape must go or clearly be labled "for illustration only".
*Calculate and display the azimuth for every sight, can be used for checking the compass under certain circumstances.

How do I calculate the azimuth without knowing position? Or am I allowed to use gps? Then I could just give you incline and azimuth. Do you want to support gps crosschecking?

*State clearly how altitude corrections are done. Obviously not all cases are covered(low altitudes), as it is not possible for the user to enter temperature or barometric pressure.

Do you think support for temperature and pressure should be added?

*Change "height" to "height of eye". Generally use standard terminology, "altitude", not "elevation", see Bowditch.
*Use lower "Limb" as default, as this is what is normally used, if available, with a sextant.
* Let user enter sextant altitude using degrees minutes and decimal minutes, as this is what you get with most sextants, not degrees and decimal degrees.

This is all very easy, I will try to do the above soon.

*Include calculations for running fixes, two sights of the Sun for example. User enters course and distance between sights.

Great idea, but it significantly complicates the user interface. Would it make sense to just allow a course and distance to be entered for every sight (default is 0 distance), then shift the plot by that amount?

*Include a star identification feature. User enter time, sextant altitude and DR position (or better). The plugin names the star and ask if the user accept, before producing a LOP.

Thomas

A good idea also, suggestions for actual user interface appreciated.

I would also like to support lunar angles.

The only other thing is... this plugin needs much more testing to determine if it is functioning properly, I suspect small numerical errors that could be improved (feel free to look at the source)

[svpresent]

Hi,

First of all my compliments for this awesome feature

My first question: I can't enter the exact time of my sightings.
Is there a reason this field is fixed to the computer time of entering?

Thanks!
Len.

[svpresent]

Well, never mind my previous post
Guess I was too enthousiastic to read first and act later.

About the user interface...
Exact timing is done (well, by me) by starting a stopwatch at the moment of the sighting or yelling at my wife to start it.
So you go inside with your sextant and with a running stopwatch.

It would be handy to have :
1) the pc synched with gps-time (for the diehards a synch function with a radiobeacon would be nice)
2) An entry-field where I can put in any value, I mean the amount of minutes and seconds the stopwatch can show.
3) I hit enter at the exact moment the stopwatch shows this value.
4) The pc subtracts these passed minutes/seconds from the pc time registrated.

Cheers, Len.

[addboy]

I'm new to OpenCPN and would like to know how to install the plugin?
THX Dave

[svpresent]

When you've downloaded the v2.2 Build 809 and you have no special functions running, you right-click on the map. In the following menu there is an option "show celestial plugin".

Personally I'd prefer a button but I guess the team is being thrifty with the available buttons-space or want to reserve it for stronger navigation-related functions.

But hey, however you start this baby, it's super!

Cheers, Len.

[nohal]

Originally Posted by addboy

I'm new to OpenCPN and would like to know how to install the plugin?
THX Dave

Dave,
you have to use the 2.2 beta to use the plugins.

Pavel

[cagney]

geckosenator

but if you only have one body, and you have elevation, and 5 degrees error on azimuth, it does narrow down the area quite a bit.

It doesn't "narrow" down the area enough and is useless for practical navigation. It could possibly qualify as an "emergency navigation" method.

From the wiki

For the measurements (Azimuth and Elevation), only either is required, but both may be specified. This allows navigation with sextant and/or compass. If both are provided, the geographic location can be pinpointed with a single sight (with errors of course)

Totally misleading. One cannot get a position from one single sight with any kind of acceptable accuracy, and using bearings of heavenly bodies for fixing a position is definitely not recommended.

Unless this feature causes problems, or should be changed, I think we should leave it, and it is interesting to use.

This feature can possibly cause problems, if someone takes the present writing in the wiki seriously, and try to follow the instructions!
I don't mind having the feature in the plugin, as long as it is stated "for illustration only, not to be used for practical navigation".
Perhaps the position line(area) in this illustration should be based on 5 degrees accuracy, instead of 1 degree, to give the user a more realistic idea of the (in)accuracy.

How do I calculate the azimuth without knowing position? Or am I allowed to use gps? Then I could just give you incline and azimuth. Do you want to support gps crosschecking?

Good question! I was stuck in the "assumed-position-syndrome"
The same problem arises with star identification.
No, I don't want gps crosschecking. It may be a good idea to allow a user defined position though, dead reckoning position for example, or a position derived from two sights.

Do you think support for temperature and pressure should be added?

No, just state in the instructions "Not to be used for sextant heights less than 10 degrees", or something similar. This argument was meant to contradict the wiki statement

Each sight .... takes all possible sight errors into account.

...

Great idea, but it significantly complicates the user interface. Would it make sense to just allow a course and distance to be entered for every sight (default is 0 distance), then shift the plot by that amount?

Yes that makes sense.

The only other thing is... this plugin needs much more testing to determine if it is functioning properly, I suspect small numerical errors that could be improved (feel free to look at the source)

I'll try to do some testing....
If time & date could be entered by the user, it would be possible to test against published examples in various books, and I could dig out my old sights notebook........

Thomas