Fine Calibration of Compass using Radar

[Dockhead]

Originally Posted by stinkpots

HI again Dockhead
Here is a video which shows it well. If you do the manouver quickly wind and tide will not affect your result.

Thanks for this.


This would work despite a current provided you have an accurate way to determine where the ship's head is. Maybe some people have a couple of objects along the centerline they can sight down. If you don't, I suppose you could make a transit along your deck if you had points of reference to measure to.


This is a real condundrum as all these methods are +/- probably a few degrees and nowadays we have compasses accurate to fractions of a degree.



And Adelie's point is well taken about bearing resolution of our little radars.


What's a sailor to do?

[Dockhead]

Originally Posted by requiem

For aligning the heading, what comes to mind is to find a tall mountain or other object with a well-surveyed position, then get as far from it as you can whilst still being able to see it. Point your bow at the object and then adjust your heading sensor to match the object's bearing.

Assuming you can get a few miles away, and your boat's position error is no more than a few meters, the error should be quite small. An exercise for the reader...

That seems right to me too.


But again the same problem -- how do you know your bow is "pointed at" the object? With any accuracy?

[Dockhead]

Originally Posted by Dockhead

. . .

And Adelie's point is well taken about bearing resolution of our little radars.

What's a sailor to do?

But actually maybe not. It's true our little toy radars have poor bearing resolution, but that just means an object becomes an out of scale blob.

But you can CENTER that blob on a chart object with what I think is a great deal of accuracy.

So maybe this is the way forward.

1. Get the compass and radar accurately lined up using the chart overlay.

2. Then check to see if the compass seems to be aligned with the ship's head.

3. If not then the radar needs to be aligned.

4. Rinse and repeat.

If radar and compass are closely aligned, then MARPA will work right and radar bearings will be correct, no matter how out of line the radar and compass are with ship's head.

And one more method of aligning the compass has occurred to me:

1. Wait for dead slack tide on a calm day.

2. Motor slowly in one direction. Note any deviation between COG and HDG.

3. Motor slowly on the reciprocal course. Note COG/HDG deviation

4. Split the difference and change compass alignment.

5. Repeat until the deviation is equal on both courses.

This will produce error to the extent that the boat doesn't motor quite straight through the water, and with slightly offset shafts on most of our boats, there might be some of this. But motoring slowly should minimize that.

Of course we need this alignment to be correct so that deviation between COG and HDG can be interpreted as current.

The B&G Zeus plotters have this lovely SailSteer screen which even has an arrow with numbers for set and drift calculated from these deviations. Which is gibberish unless you have accurate STW and accurate HDG but which is extremely useful if you do have.

[Paul L]

I don't think you need a perfect compass radar alignment for MARPA to work well. The addition of the Sat compass should radically improve the stability of the radar image and the ability of the MARPA algorithms to reasonably acurrately track targets, Small boat MARPA needs a reasonably stable image to function acceptably.

The Sat compass only needs a single point correction. This is to correct any install error of the compass antennas to the centerline. You should be able to do this single point correction easily by running toward or away from a charted range in calm conditions.

You should be able to adjust the radar in simlar way with a less accuracy, as you do not have a digital readout to 1 degree.
There is an MAIB report that faulted a slightly misaligned radar as a partial cause of a fog based English Channel yacht ship collision.

[requiem]

Originally Posted by Dockhead

Of course we need this alignment to be correct so that deviation between COG and HDG can be interpreted as current.

Ah, so that gets to the root of the issue; just what do we really mean by heading anyway?

Your COG and HDG approach here treats heading as an abstract value derived from COG corrected for current. Intuitively we want to consider heading as having objective meaning (related to the ship's centerline), but is that the truthiest approach?

Suppose you can get a perfect alignment down the centerline, but manufacturing imperfections in the keel mean that you always have 3° of leeway on one tack and 7° on the other? Or, suppose you have a pelorus mounted with an error of 20° to starboard: what would be the practical impact if all your instruments where then exactly indexed to that?

Once deviation has been removed from the picture, a misaligned heading would seem to be similar to compass variation (only globally fixed). It would only be relevant when needing to communicate with an external party using a different variation value, or when working with charts. My initial thought is that it should have effectively no impact to ARPA performance as it would still accurately track changes in heading, and the absolute value of the heading is not relevant.

For charts, the video posted earlier made a good point: how precise are the measurements someone makes if the compass rose is marked every 2 degrees, especially if someone isn't using triangles? Also, what is the printed accuracy? I once recall a small error comparing the rose to the meridian lines: was it in transferring the measure to the rose, or a print error?

[Bycrick]

A GPS compass is designed to read Heading, ie the direction that the centerline of the boat is pointing. Heading is completely unnecessary to "stabilize" a radar display which normally is Heading Up. It’s when one tries to get fancy and display a chart in Heading Up mode or display the radar in North Up mode, that "stabilization" becomes necessary. But both of those are relative to the boat. MARPA just adds an extra complication because there you don’t want to use Heading data, you want COG.

[Paul L]

Originally Posted by Bycrick

A GPS compass is designed to read Heading, ie the direction that the centerline of the boat is pointing. Heading is completely unnecessary to "stabilize" a radar display which normally is Heading Up. It’s when one tries to get fancy and display a chart in Heading Up mode or display the radar in North Up mode, that "stabilization" becomes necessary. But both of those are relative to the boat. MARPA just adds an extra complication because there you don’t want to use Heading data, you want COG.

For Marpa to work well you need to stabilize the radar relative to the continually changing heading. Marpa is working on a target that's bearing relative to the heading (not COG). As the heading quickly varies the radar is seeing a different bearing to the target, this needs to be compensated for. The Sat compass should do an excellent job at this.

[requiem]

Originally Posted by Paul L

There is an MAIB report that faulted a slightly misaligned radar as a partial cause of a fog based English Channel yacht ship collision.

I believe this may be the case: Gas Monarch and Whispa. It's rather interesting (LPG carrier vs yacht), and so I may start a separate thread on it should time permit.

Of relevance to this thread, the yacht's radar was offset to starboard by at least 2°, the radar beam width was almost 5°, and thus the skipper altered to starboard not realizing the other ship, approaching head on, had already crossed her bow.

[Bycrick]

I do understand how MARPA works. My point was that it doesn’t work very well on small boats. Fighting to make it work as well as AIS does is sewing silk purses from sows’ ears. Collision avoidance is based on current position, SOG and COG. Readily and simply available from a GPS. Trying to derive them, in real time, from a moving platform is going to be inherently inaccurate by comparison. Before AIS, MARPA was better than hand-plotting. But that’s no longer the case. It reminds me of the various attempts foe electronic celestial navigation systems that were bandied about in the 1980s.

[Paul L]

Originally Posted by Bycrick

I do understand how MARPA works. My point was that it doesn’t work very well on small boats. Fighting to make it work as well as AIS does is sewing silk purses from sows’ ears. Collision avoidance is based on current position, SOG and COG. Readily and simply available from a GPS. Trying to derive them, in real time, from a moving platform is going to be inherently inaccurate by comparison. Before AIS, MARPA was better than hand-plotting. But that’s no longer the case. It reminds me of the various attempts foe electronic celestial navigation systems that were bandied about in the 1980s.

The MARPA functions on my Furuno radar work very well offshore. I use them often on the many non- AIS targets.

[Bycrick]

Dockhead — I’ve read your threads about MARPA. I’ve never seen the Furuno implementation, but agree that the Simrad version is marginal. It’s certainly nice to have. My only question is whether or not it’s worth the effort to polish it up. To me, it’s a secondary system, only useful for tracking smaller, non-AIS targets. You’re sailing under entirely different conditions than I am.

[Dockhead]

Originally Posted by Bycrick

A GPS compass is designed to read Heading, ie the direction that the centerline of the boat is pointing. Heading is completely unnecessary to "stabilize" a radar display which normally is Heading Up. It’s when one tries to get fancy and display a chart in Heading Up mode or display the radar in North Up mode, that "stabilization" becomes necessary. But both of those are relative to the boat. MARPA just adds an extra complication because there you don’t want to use Heading data, you want COG.

It is true that simple relative bearings displayed on a radar in head up mode don't require stabilization, but everything else does.


MARPA requires good HDG data (as PaulL explained), and radar overlay requires good HDG data. On my boat, radar overlay is a standard operating mode, not "getting fancy". YMMV.



North Up radar display (preferred by pros for a great many situations) requires good heading data.

[Dockhead]

Originally Posted by Bycrick

Dockhead — I’ve read your threads about MARPA. I’ve never seen the Furuno implementation, but agree that the Simrad version is marginal. It’s certainly nice to have. My only question is whether or not it’s worth the effort to polish it up. To me, it’s a secondary system, only useful for tracking smaller, non-AIS targets. You’re sailing under entirely different conditions than I am.

Well, it's true that there are relatively few targets which you can't track by AIS, and MARPA is not the only way to do collision avoidance by radar. I doubt if I would be investing money solely to improve MARPA, but if it's a side effect of improving other things I'm not against at all. MARPA could be really useful in those few cases where you can't use AIS.


I sailed thousands of miles this summer, literally from the Southernmost point in the Baltic to the Northernmost, and within 80 miles of the Easternmost, and within 50 miles of the Westernmost. All without AIS as my AIS set blew up and customs problems prevented me from taking delivery of a new one.



It was a really good exercise in polishing up alternative methods of collision avoidance. MARPA was working somewhat better but was not my main tool. Mostly I was doing hand plots, and using the VRM/EBL on the radar like in olden times. It works pretty well.


Big problem with the Navico MARPA implementation is that targets don't stay caught, so you have to acquire all over again, and again and again. I'm thinking good HDG data at a high data rate might help with this. It can't be the radar itself -- a 4G unit which has really good signal processing. Vanishingly small incidence of false alarms with radar guard zones, for example, which makes those supremely useful.

[Dockhead]

Originally Posted by Paul L

The MARPA functions on my Furuno radar work very well offshore. I use them often on the many non- AIS targets.

You actually have ARPA, not MARPA (the difference is automatic acquisition of targets), and the Furuno implementation has a fantastic reputation.


I would consider Furuno nav gear next time.

[Dockhead]

Originally Posted by requiem

Ah, so that gets to the root of the issue; just what do we really mean by heading anyway?

Your COG and HDG approach here treats heading as an abstract value derived from COG corrected for current. Intuitively we want to consider heading as having objective meaning (related to the ship's centerline), but is that the truthiest approach?

Suppose you can get a perfect alignment down the centerline, but manufacturing imperfections in the keel mean that you always have 3° of leeway on one tack and 7° on the other? Or, suppose you have a pelorus mounted with an error of 20° to starboard: what would be the practical impact if all your instruments where then exactly indexed to that?

Once deviation has been removed from the picture, a misaligned heading would seem to be similar to compass variation (only globally fixed). It would only be relevant when needing to communicate with an external party using a different variation value, or when working with charts. My initial thought is that it should have effectively no impact to ARPA performance as it would still accurately track changes in heading, and the absolute value of the heading is not relevant.

For charts, the video posted earlier made a good point: how precise are the measurements someone makes if the compass rose is marked every 2 degrees, especially if someone isn't using triangles? Also, what is the printed accuracy? I once recall a small error comparing the rose to the meridian lines: was it in transferring the measure to the rose, or a print error?

This is a deeper dive, and interesting.


Yes, my idea for aligning the compass DOES correct for any misligned keel -- it aligns the compass to the direction the boat moves through the water in the absence of current and solves the problem of the misaligned keel.


I think you may be right about MARPA (which it seems needs high consistency but perhaps the absolute value is not important), but as we have been discussing, accurate HDG data is needed for radar overlay, true bearings, etc., and accurate HDG data makes the comparison to COG more valuable.