New Respect for Navico 4G Radar

[Dockhead]

I had one of my hairier harbor entrances the other night.


After 350 miles of sailing from Helsinki across the main part of the Baltic Sea to Kalmar in SE Sweden. In the Kalmar Sund after passing the tip of Oland, we encountered a downpour just as the sun disappeared, and had to sail about 50 miles down the sound in the dark and pouring rain.


The last 15 miles or so approaching Kalmar bridge is complicated with a lot of unlit buoys and rocks and doglegs in the channel. Just at the start of this the rain tapered off and we got -- a dense fog.


Visibility was such that I could hardly see even the end of my own boat.


So it was pure instrument flying.


I have not greatly appreciated my 4G radar, hating the poor bearing discrimination and useless MARPA, but in these conditions, it shined.


This radar has the best signal processing of any I have ever used, something which gives it supremely useful guard zones, a big plus.


But here also -- the very high signal to noise and very few false returns is exactly what is needed in conditions like this. Bearing discrimination is not important; you're not running radar plots.



Using that radar, I picked my way through those last 15 miles with confidence, and got safely into port without a false step. At times set to 100 meter range, something my old Pathfinder radar didn't even have -- would have all been within the main bang.



The only tense moment was when a strong, unknown return appeared almost dead ahead in a narrow part of one channel. I was about to throw on the brakes when we heard a screech of bird cries, and the target return broke up into a bunch of small spots flying in every direction.

[Pelagic]

Exactly what radars SHOULD do

Does this radar have manual tuning?

[Dockhead]

Originally Posted by Pelagic

Exactly what radars SHOULD do

Does this radar have manual tuning?

Yes, and rarely used other than sea clutter and rain clutter.

[Nicholson58]

We had a similar fog experience on Lake Michigan. Couldn’t figure out the crazy signals moving on the screen. Our 2g unit was showing individual seagulls both flying and floating. 4g upgrade this year. SIMRAD.

[NahanniV]

Originally Posted by Dockhead

.......
I have not greatly appreciated my 4G radar, hating the poor bearing discrimination and useless MARPA, but in these conditions, it shined.
.......

Have you tried the OpenCPN implementation of MARPA ?

Much better than on a plotter.

[Dockhead]

Originally Posted by NahanniV

Have you tried the OpenCPN implementation of MARPA ?

Much better than on a plotter.

I have tried it, admittedly not extensively, and didn't find any difference. Doesn't the radar do the computations in any case?


And I'm not sure that the computational method is the issue in any case -- I think you just don't have good enough data to do MARPA well with such poor bearing discrimination.



That's my theory, anyway, and I would be glad to be corrected if I don't understand something.

[mvmojo]

Love my 4G Radar. Here's a panoramic photo off the stern of my boat in dense fog at Block Harbor. Only a handful of boats visible. But on the 4G radar - holy moly! Hard to count the number of boats within 1/8 nm...

Panoramic looking aft. Notice the sailboat with 2 dinghy's behind it...
deck.jpg Views: 250 Size: 121.2 KB ID: 200679" style="margin: 2px" />

Radar set at 1/8 mi offset looking aft. The dinghy's on the closest sailboat show up as a separate target - great target discrimination!:

[NahanniV]

Originally Posted by Dockhead

I have tried it, admittedly not extensively, and didn't find any difference. Doesn't the radar do the computations in any case?


And I'm not sure that the computational method is the issue in any case -- I think you just don't have good enough data to do MARPA well with such poor bearing discrimination.



That's my theory, anyway, and I would be glad to be corrected if I don't understand something.

In the case of Navico Displays, the RADAR is doing the computations.

In the case of OpenCPN your computer is doing the computations.
If you have a good compass connected to the 4G, it can work very well.
Perhaps your compass or the way in which it's connected could be the reason for your problem with bearing discrimination ?

[Pelagic]

Originally Posted by Dockhead

Yes, and rarely used other than sea clutter and rain clutter.

Interesting, looking st MV mojo screenshot the discrimination is amazing.
Far better than the very expensive S-Band radars I use commercially for heavy rain and sea conditions.
In those conditions, I rely heavily on manual tuning to get best performance

[Dockhead]

Airmar H2183, supposedly the best non-satelite compass which is feasible for a recreational vessel.

Are you sure O does MARPA itself?

Originally Posted by NahanniV

In the case of Navico Displays, the RADAR is doing the computations.

In the case of OpenCPN your computer is doing the computations.
If you have a good compass connected to the 4G, it can work very well.
Perhaps your compass or the way in which it's connected could be the reason for your problem with bearing discrimination ?

[Nicholson58]

Originally Posted by Pelagic

Interesting, looking st MV mojo screenshot the discrimination is amazing.
Far better than the very expensive S-Band radars I use commercially for heavy rain and sea conditions.
In those conditions, I rely heavily on manual tuning to get best performance

Broadband is easy and power consumption is very low. Running 6 mph one is not so concerned about 25 miles out. Our 2g sees dinghies, nav marks, seagulls. I’m looking forward to the new 4g

[NahanniV]

Originally Posted by Dockhead

Airmar H2183, supposedly the best non-satelite compass which is feasible for a recreational vessel.

Are you sure O does MARPA itself?

Yes I’m sure, the in dome MARPA is proprietary, so, it was implemented in the RADAR plugin.

The most accurate way to get correct bearings is to supply the 4G directly with heading data, depends what interface box you have. When the 4G dome has heading data it attaches it to each spoke as it is received.

[Dockhead]

OK, but why would heading data be needed at all for MARPA? Surely the most accurate calculation will use RELATIVE bearings like we do in a normal hand radar plot.

Originally Posted by NahanniV

Yes I’m sure, the in dome MARPA is proprietary, so, it was implemented in the RADAR plugin.

The most accurate way to get correct bearings is to supply the 4G directly with heading data, depends what interface box you have. When the 4G dome has heading data it attaches it to each spoke as it is received.

[Paul L]

Originally Posted by Dockhead

OK, but why would heading data be needed at all for MARPA? Surely the most accurate calculation will use RELATIVE bearings like we do in a normal hand radar plot.

Of course you need solid heading data. You have to remove the radars bouncing around horizontal movements from the radar reported target movement.

[Dockhead]

Originally Posted by Paul L

Of course you need solid heading data. You have to remove the radars bouncing around horizontal movements from the radar reported target movement.

OK, so you mean it's used to stabilize the relative bearing? How? I would be interested to know the exact algorithm involved -- it's interesting.


Edit: Sorry, brain fart. Of course we reduce relative bearings to true in a classic radar plot. That way minor changes of our course don't create the appearance of a changing bearing when it's still a collision course.


But I don't see any reason why ARPA needs to work that way. It can instantaneously calculate a more accurate plot using relative, not true bearings, and our instantaneous COG and SOG. This takes the compass out of the equation and should be much more accurate on a small vessel lacking a $50,000 gyrocompass.



Interesting question -- how does ARPA do it? True bearings, or relative?


If true, then maybe a small boat MARPA could be developed doing it the other way, and maybe it would work much better.


I note that Furuno ARPA for some reason works 100x better than anyone else's MARPA, and not just the ARPA features. Maybe they have a different computational approach?


How is it done in OpenCPN?