S63 charts - overzoom

[jongough]

I used OCPN last year with CM93 charts and was reasonably happy with the zoom capability. Most of the time I could zoom in enough to distinguish my anchor marks and follow tracks I had made.

This year I installed and used the new S63 chart capability (first the beta and now the official release) using official charts from the Australian Hydrographic Office. However the zoom capability is now much reduced. I now find that I cannot distinguish my anchor marks, they appear to be all on top of each other, and trying to visually follow tracks made before is almost impossible.

There is an issue in Australia in that many of the official charts are not properly georeferenced and also some have not been updated since 1770, they are Captain Cooks charts! So, whilst the shipping lanes and commercial traffic areas have good charts with lots of detail and you can zoom in a long way before overzoom kicks in, in other areas there is little detail and the zoom level is very poor.

With the current overzoom capabilities for S63 charts in some areas the boat length, as displayed, is about 0.5nm long. However, in these areas you really need to be within 10-20m of your anchor point to know you are safe, particularly when in the Great Barrier Reef. The GPS signals generally are good enough to get you within 3-10m, i.e. within a boat length of where you really need to be.

Now there are at least two opinions on zooming in more than the embedded scale of the chart suggests. This is what overzoom is.

1. Charts used for navigation should not be zoomed in more than the official scale will allow as there is no more detail to be found and the overzoom suggests a greater level of accuracy than is supported by the data.

2. Charts should be zoomable to the level of accuracy of the supplied location information, i.e. the GPS signal. Whilst the information displayed by the official chart will/may not be accurate, the location, i.e. lat/lon, will be accurate to any other information on the chart which is also derived by the location information. This would allow a skipper to approach a GPS location visually on OCPN and arrive at a known location.

In 1. the chart is being used for navigation based on marked depths, dangers, land, lights, bouys, etc. In 2. the chart is not being used for navigation, rather the plotting in 2D of the location, tracks and waypoints.

In case 1. there really should be no overzoom. In case 2. overzoom should be allowed to at least the GPS accuracy at the time. There is a potential problem though. This is where someone is in the second case, overzoomed, but attempting navigation using the chart information being displayed. In Australia this would be highly dangerous as I suspect it would in the rest of the world.

The questions are:
a. Should overzoom be allowed at all?
b. If overzoom is allowed, how much should be allowed?
c. How should it be indicated that the chart information being displayed at this level of overzoom is not accurate and MUST NOT be used for navigation purposes?

I have attached a PDF with 3 screen shots taken of Lady Musgrave Island lagoon at the south end of the GBR showing the issue. This lagoon is safe, but does have some isolated coral heads that come straight up from the bottom, 7-8m, to within <1m of the surface. Between 10:00 & 15:00 it is easy to see them, outside of this, or with overcast or strong wind it is almost impossible. I have marked the safe anchorages, many of the bommies (good diving!) so know where I can go. I just need to be able to get there. So I would like to be able to zoom in quite a bit further than is shown. Note: the current beta, 2302, does allow greater overzoom for CM93 charts.

Any ideas?

[bcn]

Jon,

there is no reason why zooming should be handled differently in the case of S63 or CM vector charts (and raster neither).

Before debating overzoom strategies a point to clarify:
your screenshots for CM and S-63 show the same details (1 & 2).
So the problem is that zooming-in/quilting with S-63 charts (your #3) does not behave the same?
But selecting the S-63 chart via the "piano" and then zooming-in renders better?

Hubert

[jongough]

On the CM93 chart I can zoom in further than the S63 chart. The CM93 shown is from 2011, the S63 from Oct 2014. I should really be using the 2014, or current charts.

The screen shots show CM93 at max zoom, piano key switch to S63 and zoom remains the same. Do any zoom on S63 and it immediately zooms out the max overzoom allowed for that chart.

On hover over the largest scale S63 chart gives an embedded scale 1:350000 and a scale on the info bar of 39900, where as on the CM93 there is no embedded scale and it zooms into in info bar of 2200. There is currently a limit in OCPN on the amount of overzoom allowed for S63 charts. The current beta, 2302, allows far more overzoom for CM93 and the base level chart, but the same restriction is on the S63 charts.

[bcn]

Looking into the charts you are debating it looks to me that the S-63 chart and the CM are using exactly the same underlying data.
For instance all the soundings are identical. The renderings of the coastline/outline of the lagoon are slightly different but the hard points are the same.

So just due the decision of the two chart providers how to present the data we see differences.
- AHS in S-63 says 1:350.000 which results with an overzoom of 1:10 in that 1:39.700 (btw.: the adjacent cells to the West and South show a scale of 1:90.000)
- while CM allows to zoom in up to 1:2200, that's a 1:40 from the stated 1:100.000 scale.

In my opinion there is no reason why CM charts can be over zoomed 40x while S57/S63 "only" 10x.
In our case this would lead to a scale of aprox. 1:10.000, sufficient for your purposes.

The other difference I see between the treatment of CM and S57/S63 charts is that in the case of the CMs the overzoom factor is shown in the status bar, but not for the S57/S63 ones.

[bcn]

Originally Posted by jongough

The GPS signals generally are good enough to get you within 3-10m, i.e. within a boat length of where you really need to be.

With a modern consumer GPS you can expect a precision of 2m CEP, with SBAS (WAAS, EGNOS,....) enabled a bit better.
I would not expect to see errors >3m in OpenSky conditions. Conditions we have on a boat usually.

You might have spikes with bigger errors at a quay, a container terminal, mangroves or the Norwegian fjords though. Sites with less satellites in view or multi path reflections which will lower the precision.
So please no automatic docking just with GPS.

In case you want to spent 15.000€ and being in the range of RTK differential corrections you can get a precision of 2cm or better in realtime at 100Hz.
Plus heading within 0.1° and pitch/roll at 0.05°. Happy autopilot and perfect Radar overlay. Different league, I suspect.

Hubert

[jongough]

The only charts officially available for Australia are the AHO S63 charts. No other charts are recognised as being suitable for navigation by the Australian authorities. The CM93 chart I have is from 2011 when commercial organisations could get these from the AHO, however that is no longer the case. So I have to use the S63 ones if I want to be legal AND, supposedly, have all the current navigation marks shown.

However, the issue is that GPS systems are far more accurate than ANY chart that is available in Australia. You can just see in the screen shots I took that according to the chart I went right over the reef, I did not. This is only one example, there are many more. In one case the chart is over 100m off and the GPS track shows you going over dry land!

Anyway, just because the chart has a scale at which it could be considered accurate does not mean that the user shouldn't be allowed to zoom in considerably more than the chart accuracy. In my case I have anchorages that I need to navigate to visually, not on auto pilot, and it would be useful to be able to zoom in to the level of accuracy of the GPS so that I can see from the wheel where my anchorages are. In the current scheme I can just about make out that I am in the lagoon somewhere and that there are multiple anchorages/bommies almost under the boat.

[jongough]

You may get the accuracy in other parts of the world, but in Australia there is no augmentation system to use and the satellite constellations tend not to be good for real accuracy. We quite often have most satellites on the horizon with only one or two high in the sky. This is fine, but rigging, boat hardware, etc, tends to get in the way and reduce the signal strength a bit. So normally we get 4-10m accuracy, which is fine by me. I just want to be able to use this accuracy to find my old anchorages as I know they are safe and I will not hit anything like submerged rocks, coral, etc.

[bcn]

Originally Posted by jongough

The only charts officially available for Australia are the AHO S63 charts. No other charts are recognised as being suitable for navigation by the Australian authorities. The CM93 chart I have is from 2011 when commercial organisations could get these from the AHO, however that is no longer the case. So I have to use the S63 ones if I want to be legal AND, supposedly, have all the current navigation marks shown.

However, the issue is that GPS systems are far more accurate than ANY chart that is available in Australia. You can just see in the screen shots I took that according to the chart I went right over the reef, I did not. This is only one example, there are many more. In one case the chart is over 100m off and the GPS track shows you going over dry land!

Anyway, just because the chart has a scale at which it could be considered accurate does not mean that the user shouldn't be allowed to zoom in considerably more than the chart accuracy. In my case I have anchorages that I need to navigate to visually, not on auto pilot, and it would be useful to be able to zoom in to the level of accuracy of the GPS so that I can see from the wheel where my anchorages are. In the current scheme I can just about make out that I am in the lagoon somewhere and that there are multiple anchorages/bommies almost under the boat.

Completely understood.
And I second your idea, that it should be possible to zoom in further.

Feature request so that this does not get lost? Plus the zoom factor in the status bar....

[bdbcat]

Guys...

OK, time for my overzoom sermon.

1. We have no way to evaluate the absolute accuracy of any chart, at any scale. Charts are representations of reality. We can say that the reasonably expected accuracy of a chart is a millimeter or so (pencil width) at the compilation scale. Paper charts show compilation scale by region, but this info is not available to the ECS, typically. So, if we assume that the compilation scale is 4x the printed scale (maybe typical?), then the on-screen position of geographic features is good to maybe 4 pixels at 4x overzoom. Keep this in mind.

Nigel Calder does a much better job of explaining this aspect of charting in his book.
Nigel Calder, How to Read a Nautical Chart - Best Sailing Books
Recommended reading.

2. There is nothing wrong with using OCPN as a "plotting sheet" at super high zoom factors. Only you can evaluate the accuracy of your positioning system. If you determine that your position is accurate to meter level precision, then use it. But there should be no reference, explicit or implicit, to the underlying chart if it is overzoomed too far.

3. It is obvious when using a raster chart at high overzoom that there is reduced information available. See picture 1 below. It is much less obvious with vector charts at the same scale. The vector chart "looks" more accurate, because the lines are rendered sharply at all scales. Pix2 below. That's the nature of vector charts. I think its clear that navigators would assume that Pix1 is not suitable for navigation. Not so clear regarding pix2.

4. We can use OCPN as a plotting sheet without charts by making and using an "empty" chart group. Nothing on-screen except the low resolution world chart, and grid lines if selected.

So, a reasonable feature request might be to develop a way to reduce the implied onscreen visual accuracy of excessively overzoomed vector charts.

But I will resist requests to allow arbitrary overzoom of all possible charts.

Dave

[bcn]

Dave,

well aware about the possible erroneously perceived safety when overzooming.

Now - if we can stretch CM93 40fold, then we should do the same with S-57/S-63. Or none.

Having the factor in the status line is another hint what one is doing in this instant.

The big OVERZOOM is a statement already. What about a second one in the lower right corner from 10x on and one in every corner from 20x on? For the distracted navigator....
The program might blow a horn as well

Hubert

[bdbcat]

Hubert...

I agree that it is a little bit inconsistent to allow OZ to 40 for cm93, but less for S63/S57.

I might agree to a compromise, say 16X for all vector charts.

It is not really possible to state the overzoom factor on a general quilted chart, unless we declare that the OZ factor applies to the smallest scale chart patch currently visible. We don't do this now, but could.

And btw, on Flyspray, I do not agree that this issue is a bug in O. This is more accurately a Feature Request.

Thanks
Dave

[bcn]

Dave...

Originally Posted by bdbcat

Hubert...

I agree that it is a little bit inconsistent to allow OZ to 40 for cm93, but less for S63/S57.

I might agree to a compromise, say 16X for all vector charts.

The CM93 factor was the result of the complaints about the too restrictive solution, wasn't it?

It is not really possible to state the overzoom factor on a general quilted chart, unless we declare that the OZ factor applies to the largest scale chart patch currently visible. We don't do this now, but could.

Can you explain how the quilting of CM93 works? What does "general quilted" imply?

And btw, on Flyspray, I do not agree that this issue is a bug in O. This is more accurately a Feature Request.

Thanks
Dave

Requested closure of #1568 - noticed the wrong category once placed...
Please don't forget about the OZ factor in the status line for the rest of the bunch.

Thanks

Hubert

[cagney]

Originally Posted by jongough

......
.....
where as on the CM93 there is no embedded scale
....

Well.... there is actually.



Chart 01980456.D has a scale of 1:100.000 .
Switch to single chart mode and hover the cursor over the yellow bar.
So, for some reason, the CM93 chart is of much larger scale than the S63 chart.

Thomas

[bcn]

Thomas...

as discussed in this thread (#4): http://www.cruisersforum.com/forums/...ml#post1674993

What would you recommend with respect to the limits/handling of overzoom?

Hubert

[jongough]

Just to let you know there is a Flyspray feature request for overzoom, FS#1520.

I had coded a solution and submitted it, but I was asked to open raise this question for general discussion.