Ais vs marpa

[TonyS]

We have had MARPA for 10 years and it has safely got us through the channel shipping lanes even in thick fog. At SIBS on an impulse I bought an AdvanSea AIS RX 100, which was on offer. Over the winter I put my old masthead aerial on the pushpit and ran the AIS on OpenCPN with CM93 and a Garmin GPS 60 on my Vista laptop. Crossing the channel was the first time I connected it all up. Here is a picture of the shipping lane off Alderney.

We didn't change course or speed and had the AIS bleeping for 4 collision possibilities at the same time.
This year there were more ships than ever!
I also used MARPA but found the AIS much more informative and reassuring especially as we had less than 4 miles visibility and the ships would be appear enormous as they gradually became visible.
We had the collision range set at 2.5 or 3 miles but after a bit I had to turn the computer volume down as the constant bleeping was annoying.
We have 10 year old Raymarine navigation equipment so I cannot IS on the chart plotter. I may get a dedicated computer for the nav table that uses less power but the laptop did well on this occasion.

 

[FWB]

Remember that Vessels AIS transmissions are time delayed depending upon class and speed. Only really an issue for Class B but worth mentioning.
Also the AIS display is the position of the vessels AIS GPS antenna, some vessels are very long !

 

[RobbieW]

iirc MARPA can be delivered in two flavours, ground stabilised or sea stabilised. One of these used not to be common on leisure radar sets and, again iirc, was a contributory factor in the loss of the Wakuna several years ago. That collision was considered to be 'radar assisted' as the yacht crew misinterpreted the information displayed.

As you pointed out, AIS provides a much clearer picture of the actuals (ground stabilised?) than RADAR. I'm guessing your snapshot has quite a large range for illustrative purposes. What it doesnt show is your yachts track. One refinement to Open CPNs defaults you might want to make is to change the length of the COG predictor vector for both you and the AIS targets to the same value. I find this makes it much easier to see whats going on with the relative tracks of vessels around. I also set the collision range to 1 mile but thats much more of a personal choice.

 

[TonyS]

iirc MARPA can be delivered in two flavours, ground stabilised or sea stabilised. One of these used not to be common on leisure radar sets and, again iirc, was a contributory factor in the loss of the Wakuna several years ago. That collision was considered to be 'radar assisted' as the yacht crew misinterpreted the information displayed.

As you pointed out, AIS provides a much clearer picture of the actuals (ground stabilised?) than RADAR. I'm guessing your snapshot has quite a large range for illustrative purposes. What it doesnt show is your yachts track. One refinement to Open CPNs defaults you might want to make is to change the length of the COG predictor vector for both you and the AIS targets to the same value. I find this makes it much easier to see whats going on with the relative tracks of vessels around. I also set the collision range to 1 mile but thats much more of a personal choice.

Thanks! That is interesting. I couldn't find the yacht track at the time but have found it now. I am interested that you set the collision range to 1 mile. I wasn't sure whether to make it larger or smaller. The nearest a ship came to us was 1 mile which is nothing compared to sailing up Southampton Water when 100 yds in quite common. It always scares me though to pass in front of a ship going much faster even if it is 2 miles away. I am very happy to see the length of the far side! I have those doubts about what happens if the engine cuts etc etc.

It is interesting that you mention the Wakuna. We were in the same spot a few hours later, on the same day, having been assured that the fog was only on the French coast. That was the first time I had used MARPA and my wife and I took watches of 30 mins with the spray hood down. I wrote an article about it in Yachting Monthly.

Can I modify my 10 year old MARPA to be ground stabilised? It would certainly make it much better.

 

[Dave99]

I would want both. Not all boats have AIS yet and you are relying on the data being broadcast which is not always correct. AIS is a very useful tool in busy traffic areas but I wouldn't set sail on a passage of any length without radar. I would not be happy sailing in fog with just AIS.

 

[RobbieW]

.... I am interested that you set the collision range to 1 mile. I wasn't sure whether to make it larger or smaller. The nearest a ship came to us was 1 mile which is nothing compared to sailing up Southampton Water when 100 yds in quite common. It always scares me though to pass in front of a ship going much faster even if it is 2 miles away. I am very happy to see the length of the far side! I have those doubts about what happens if the engine cuts etc etc.

.....

Can I modify my 10 year old MARPA to be ground stabilised? It would certainly make it much better.

Like you, I was a Solent sailor so my tolerance to close encounters has been similarly shaped Sorry, no idea if your MARPA could be ground stabilised - try a Google search ??

 

[ParaHandy]

I also used MARPA but found the AIS much more informative and reassuring especially as we had less than 4 miles visibility

The Marpa algorithm was that in every 10 radar sweeps, a target had to appear 6 or 7 times in the same box otherwise the target was ignored. In a yacht rolling in a seaway the chances are fairly high that you might receive a confusing Marpa picture. Makuna has been mentioned in this thread and I remember calculating how long she had at the approach speed of the container vessel, to make a decision, and it wasn't very long at a 6nm range - less than 10 minutes. AIS crystallises the decision process in a way that small boat radar/Marpa can not. I don't think detection up to 6nm is good enough and the 7" radar display fitted to many radar sets just isn't good enough; apart from any other factor, the ease of use is not good enough - think Raymarine's clunky attempts to implement GPS facilities on their radar sets.

One observation which might be of interest is that in fog, merchant shipping leaves a respectable distance from adjacent ships but, in clear visibility, they are almost on top of each other. So, I've crossed the channel in poor visibility and it has been easy plotting a course through them when the separation is several miles, often easier from that point of view than in clear visibility.

I use Shipplotter AIS software which has CPA software which accounts for GPS aerial position. I think it's pretty good.

 

[Piers]

iirc MARPA can be delivered in two flavours, ground stabilised or sea stabilised. One of these used not to be common on leisure radar sets and, again iirc, was a contributory factor in the loss of the Wakuna several years ago. That collision was considered to be 'radar assisted' as the yacht crew misinterpreted the information displayed.

As you pointed out, AIS provides a much clearer picture of the actuals (ground stabilised?) than RADAR. I'm guessing your snapshot has quite a large range for illustrative purposes. What it doesnt show is your yachts track. One refinement to Open CPNs defaults you might want to make is to change the length of the COG predictor vector for both you and the AIS targets to the same value. I find this makes it much easier to see whats going on with the relative tracks of vessels around. I also set the collision range to 1 mile but thats much more of a personal choice.

Hi RobbieW,

Quite correct. I have yet to find a leisure radar that is sea-stabilised which is so important for MARPA. Leisure radars are almost all (if not all) ground stabilised which has been the cause of many radar assisted collisions, as you pointed out.

There are also many issues with AIS. First, it's also ground stabilised and hence suffers the same issues as ground stabilised radars. Second, it's transmissions are delayed so the vessel's position is always inaccurate. Third, there is no guarantee that the data being transmitted is correct (an issue which is the basis of many a commercial forum discussion.

Another point to make is that the COLREGs completely change the moment you are out of sight of another vessel (i.e., restricted vis) and knowing what action to take is so vital.

If you'd like to read a report I've written on all this which also details a test you can do to see if your radar is ground stabilised or not, see: http://www.playdeau.com/radar-assisted-collisions-and-marpa/

For me, you can't beat radar. At least you aren't relying on another vessel's transmissions. If your set is ground stabilised the simple answer is not to use it and instead, use the electronic bearing line (the equivalent of the mark one eyeball when watching another vessel to see if it's on a collision course with you).

Safe travelling - Piers

 

[Marsupial]

For me, you can't beat radar. At least you aren't relying on another vessel's transmissions. If your set is ground stabilised the simple answer is not to use it and instead, use the electronic bearing line (the equivalent of the mark one eyeball when watching another vessel to see if it's on a collision course with you).

Safe travelling - Piers

+1

Well done Piers, the best bit of advice on this thorny subject. AIS has a way to go before it becomes a reliable replacement for a hand bearing compass or an EBL on a RADAR set.

 

[Tomahawk]

What do yu mean by ground stabalised as opposed to sea stabilised?

Is that north up true motion as opposed to course up relative motion?

We have the option of either..

 

[RobbieW]

Piers,

I read your article and will admit I still dont understand the benefit of sea stabilisation. I can certainly see that if two ships are using different datums to establish their relative positions it will lead to problems.

Perhaps my problem is that in learning to sail mostly within sight of land, a fixed reference point is normally available so one tends to assess course by reference to that point ie COG. It is much more difficult to sail a straight course without those references and one that I suspect many yachties delegate to the autopilot. In fog, of course, all references are gone. Ok, in writing this perhaps the glimmer of understanding is coming through because RADAR provides a virtual eye (and EBL) making an analogy with a hand bearing compass or using part of the ship to provide a bearing line - these should be the things we do naturally to avoid collision with either fixed or moving objects.

Taking into account the deficiencies of AIS it still feels a more natural way to present the data though. Movement over the ground is what is 'real' whereas movement relative to the water is 'virtual'?

 

[prv]

I read your article and will admit I still dont understand the benefit of sea stabilisation

I didn't get it the last time it came up, either.

Piers's article helped a bit, in that the problem with the left-hand display is clear. Target B's vector is not pointing at the centre of the screen, so you naively assume it will pass in front. In fact you're not driving up the vertical line, you're sliding to the left with the tide and so B's vector is pointing at where you will be.

Thing is, this only applies if you insist on using basic head-up mode. As soon as you switch to north-up, surely it makes no difference which way your bow is pointing, and hence no difference whether your movement over the ground is caused by your own way, the tide, a combination of the two, or a squadron of invisible fairies pulling you along sideways.

What am I misunderstanding?

Pete

 

[RobbieW]

...
Piers's article helped a bit, in that the problem with the left-hand display is clear. Target B's vector is not pointing at the centre of the screen, so you naively assume it will pass in front. In fact you're not driving up the vertical line, you're sliding to the left with the tide and so B's vector is pointing at where you will be.
...

I didnt really find those illustration helpful as I'm not clear what I'm looking at. My experience of RADAR sets is they will show the trail of targets, not the predicted track, there is also no indication of range. So none of the targets appear to be a problem but the ones I'd watch would be C and A. That said I've little experience of either an ARPA or a MARPA set, ARPA on a JST ship where I wasnt really supposed to be playing with the features and MARPA on one or two yachts.

It could also be that I've got used to the generally clear way that OCPN presents the data it recieves about my and my potential targets predicted tracks.

 

[Cruiser2B]

Thing is, this only applies if you insist on using basic head-up mode. As soon as you switch to north-up, surely it makes no difference which way your bow is pointing, and hence no difference whether your movement over the ground is caused by your own way, the tide, a combination of the two, or a squadron of invisible fairies pulling you along sideways.

What am I misunderstanding?

Pete

Radar gives you a relative plot - takes two subsequent ranges and bearings of a target, connects the dots and voila: relative track of the target. For the MARPA to be able to determine the target's Co and Sp, it needs to solve the relative vector triangle using the knowns (relative vector and own vector) to determine the target's vector. You know that in a current your GPS will show your heading as being quite a bit different from where the bow is pointing and the SMG/SOG will differ from log speed. If the MARPA bases your own vector on GPS-derived Co and Sp then the target's Co and Sp will also show a similar difference to reality. When sea-stabilised, with ownship heading from the radar and speed from the log, the derived target's vector will show its heading more accurately, so that you will be able to more accurately determine where you are relative to the target (her aspect). Hope that makes sense.

 

[prv]

Radar gives you a relative plot - takes two subsequent ranges and bearings of a target, connects the dots and voila: relative track of the target. For the MARPA to be able to determine the target's Co and Sp, it needs to solve the relative vector triangle using the knowns (relative vector and own vector) to determine the target's vector. You know that in a current your GPS will show your heading as being quite a bit different from where the bow is pointing and the SMG/SOG will differ from log speed. If the MARPA bases your own vector on GPS-derived Co and Sp then the target's Co and Sp will also show a similar difference to reality.

Understand and agree with all of that, up until "difference to reality". What is "unreal" about a GPS-derived course and speed? Unless you're referring to inaccuracies in the GPS data itself, which I don't think you are (and I would suggest that the inaccuracies in the other inputs to small-boat radar are generally larger). If anything, a track and speed over the fixed ground is more "real" than one which is affected by tides, currents, leeway, etc - or at least simpler to model. I am this point, moving with this absolute vector, the radar finds out the relative vector between us, and some simple automated maths produces that absolute vector for the target. We're dealing only in points on a plane in this model, it doesn't matter whether we're powering ahead in still water, being deflected by a cross-current, or revolving slowly while drifting with the tide. We know how both points are moving across the earth, which is what matters.

you will be able to more accurately determine where you are relative to the target (her aspect).

What is the importance of knowing the aspect? Why can't I just treat the target as a point, that I don't want to intersect with my point? After all, if we're in restricted visibility, it doesn't matter which light segment I'm in for give way / stand on purposes.

Hope that makes sense.

Not really - but I do appreciate the help towards understanding. I'm not asking questions in order to be confrontational, but because I'm trying to learn.

Pete

 

[Piers]

Hi Guys, I've been out all day so will try and respond tomorrow or over the weekend. Good qns, but all answerable. Piers

 

[TonyS]

Hi Guys, I've been out all day so will try and respond tomorrow or over the weekend. Good qns, but all answerable. Piers

This has become far more technical than I expected but the discussion certainly provides food for thought.
One other difference is that it is very clear from AIS picture whether a ship has taken avoiding action either by changing speed or course. I found that very difficult to determine by MARPA, unless the turn is huge.
You can clearly see that the lower ship in green that has passed ahead of me has changed his course by his track. Whether it was to give more clearance or for some other reason is not clear.
Another point is that I understood the guys on the Wakuna didn't know how MARPA worked. They saw the blob of a ship on the starboard bow using the radar and thought they would turn round and let it pass ahead of them. As they were turned round they hit the ship on its starboard bow, head on. This removed the front of the yacht up to the mast.
I might add that when you get a Raymarine navigation system, the manuals together occupy about 10" of shelf space - probably a few thousand pages. The RL70 manual is itself an inch thick, which deals with radar and comes to MARPA about a third of the way through. Most non technical people have given up long before then. In my opinion it should be on page one as the only tool you really need and the essentials can be described in a couple of paragraphs. I taught my wife in 5 mins. and she likes playing with it when we see ships about.

 

[Cruiser2B]

Understand and agree with all of that, up until "difference to reality". What is "unreal" about a GPS-derived course and speed? Unless you're referring to inaccuracies in the GPS data itself, which I don't think you are (and I would suggest that the inaccuracies in the other inputs to small-boat radar are generally larger). If anything, a track and speed over the fixed ground is more "real" than one which is affected by tides, currents, leeway, etc - or at least simpler to model. I am this point, moving with this absolute vector, the radar finds out the relative vector between us, and some simple automated maths produces that absolute vector for the target. We're dealing only in points on a plane in this model, it doesn't matter whether we're powering ahead in still water, being deflected by a cross-current, or revolving slowly while drifting with the tide. We know how both points are moving across the earth, which is what matters.
What is the importance of knowing the aspect? Why can't I just treat the target as a point, that I don't want to intersect with my point? After all, if we're in restricted visibility, it doesn't matter which light segment I'm in for give way / stand on purposes.

By "reality" I'm referring to Colregs reality. Colregs are all based on apparent headings, not course over the ground. You are absolutely correct that it matters very little in a rule 19 situation, but many people refer to the marpa when dealing with any anti-collision situation incl those in good vis. TBH in the big ships, I used ground-stabilised data all the time - but we didn't suffer a lot of leeway, and drift was generally a small concern when proceeding at 20 kts; the errors could be discounted. In a small, slow boat in a strong current the course over ground may be vastly different from the heading (only because a 1 kt current is a greater component of track to a boat going 4 kts, than to one going 20 kts), so the resultant calculations made with that vast delta will show a similar divergence. For anti-collision, the relative vector is all that really counts, and it shouldn't be affected whether you're sea or ground stabilised. The only reason to use the true vectors is to gain an appreciation of aspect - if ground stabilised and affected by current/leeway as described, the aspect calculated by the marpa might be worthless (differ from reality). Again I don't see it as a great issue, as I confirm with the Mk 1 eyeball, but clearly the MAIB sees a problem.

Not really - but I do appreciate the help towards understanding. I'm not asking questions in order to be confrontational, but because I'm trying to learn.

Pete

I quite understand and didn't see you as confrontational at all. I hope I made it clearer, but I fear it's just one of those things that's better explained in front of the equipment.

 

[Piers]

What do yu mean by ground stabalised as opposed to sea stabilised?

Is that north up true motion as opposed to course up relative motion?

We have the option of either..

Hi Tomahawk,

Sadly no, it's not as simple as that. North up is a display orientation: North up where north is always at the top of the display; Course up where the course you are steering is always at the top of the display; Head up where the heading your boat is on is always at the top of the display. A common way to orientate the screen is Head Up, although this has a drawback which I'll come to later. The best orientation by far is North Up. More on this and why, later.

Meanwhile.....

Motion can be either be True or Relative. If set to True Motion, the display shows the land as static with all targets and your boat moving around the screen. If set to Relative Motion, your boat stays stationary in the middle of the screen and everything else moves relative to you.

Bear in mind that when in good vis and driving to avoid other vessels, you are watching how targets are moving relative to you. Either getting closer or further away, and either staying on a constant relative bearing to you or not. A risk of collision exists when a target is getting closer to you and is staying on a constant relative bearing.

Transferring this concept to using radar, you need it set to Relative Motion. To judge if the target is on a constant bearing relative to you, the easiest way is to use the EBL (electronic bearing line) to see if the target stays on it. An alternative to the EBL is to turn trails on and see if the trail is angled at you. At the same time, you are watching to see if the target is getting closer in which case you have a potential risk of collision.

In good vis, you will see if a risk of collision exists, and because you can see the direction the target is pointing (its aspect), you'll take the appropriate COLGRGs action. In restricted vis where you can't see the target and you are watching the radar instead, you'll see if the target is staying on a constant bearing relative to you and you can see if it's closing on you. Two things come into play at this stage. First, because you are in restricted vis the COLREGs change with Rule 19 coming into play. You need to know this - it's really important. Second, because you physically can't see the target you have no ide of the direction it's pointing (its aspect) so taking action may be quite the wrong action (Rule 19).

Having said all this what would I recommend? (Bear in mind this is how I use my radar and not an instruction!) Radar set to Relative Motion, Trails on (and set to 3 minutes), and if I see a risk of collision, I turn on True Vectors which will show the target's true direction, its aspect. Given this, I will know the correct rule 19 action to take.

Orientation

Head Up. This is one of the easiest to understand. Everything to the right is on the right. Everything on the left is on the left. Everything in front is in front and everything behind is behind. 'Simples' as they say. But the issue is that if your boat's heading changes or wanders at all, everything on the screen does the same making it much harder to keep an eye on relative bearings and ranges.

Course Up. This is very similar to Head Up and suffers the same problems.

North Up. Once you become familiar with radar, you'll find this to be a great orientation. Why? For two main reasons. Rather than targets moving around the screen as you change heading, the picture stays the same. Great for target watching. It's also the same orientation as your chart plotter so the two pictures fit well. It just takes a bit of getting used to, but once you have, you'll never regret it.

MARPA / ARPA

Let me say straight away that the best collision avoidance system to use when using radar is the constant bearing (EBL) and a reducing range. This gets the best from the radar. However, there is a feature that's been introduced (manufacturers like features - it helps sell product)

As the post on my website says, MARPA is the leisure version of ARPA. Like ARPA, MARPA relies on interpreting the targets direction and speed relative to you and uses this to calculate if a risk of collision exists. So the radar spends time calculating the target's direction and speed so there'll always be a lag, especially if the sea is emotional and headings wander at all.

In addition, any level of accuracy will depend on the inputs used for your boat's heading and speed. As you know, COG and SOG can vary hugely from boat heading and speed through the water depending on tide and wind effect, and this is at the heart of the problem. Using the former makes your radar ground stabilised whilst using the latter makes it sea stabilised. And it's this that has been recognised as a cause of radar assisted collisions.

It's always recommended that when using radar for collision avoidance, the set should be sea stabilised which means feeding it with boat heading and speed through the water. Google sea stabilised and collisions at sea (and variants thereof) and see what comes up. You'll find there's great discussion and variable thoughts on the subject, but it remains that the IMO, MAIB etc all recommend sea stabilised for collision avoidance.

In conclusion

In good vis, use mark one eyeball to judge constant bearing and reducing range, and take the appropriate action. In restricted vis, use your radar to see if there's a constant bearing and range. Use true vectors to judge the target's aspect and take the appropriate action according to rule 19.

Know your radar, and only use MARPA if you know your set is sea stabilised. A great radar guru, the late Robert Avis, has written two key books on the subject. Using Radar, and Superyacht Master. Both have really good sections on using radar.

Finally, on Play d'eau, I have a commercial radar and although its ARPA is brilliant I still double check a suspected target by using the EBL....

 

[Tomahawk]

Piers..

Thank you for your explanation...
I have a Raymarine C series chart plotter and radar...

How can I find out wheth it is sea or ground stabalised? Or does anyone know?