Charts II: NGA - 2700 Charts

[Surveyor]

Originally Posted by nohal

Thanks for the report...
All of them were uploaded with wrong names. Fixed.

Pavel

Oops! I grabbed and stitched over 800 charts! I was bound to make a boo boo or two.

[HappySeagull]

Originally Posted by recri

First off, I didn't leave them blank I ended up filling them in, just like I tried to resolve the location of Puerto Morelos. But now I'm repenting, because some of them were harder to do, and some of the harder ones didn't turn out to be very convincing.

So on 61270 you could just read off the grid along the edge, and that would be fine. But on 22082 you have transfer a measurement across the chart to a marked grid.

Better to use other labeled points on the chart for georeferencing than to get into transferring measurements. If you supply a value that requires some work to come up with, then who ever wants to check it needs to do the same work.

yes,I know.Just don't do the hard ones.Skip along,be happy.don't be unhappy,or we'll lose calibrators!If you DO take the time to figure what errors are,that's great too.That chart 22082 (and I see your difficulty!)would be easier to play with as a jpeg downloaded and done in MapCal.I marked it so.It's not locked though.Someone else can go in and root away to their heart's content.

[bcn]

Calibrators!

What about to identify all skewed (90/180/270) charts in your region to make those available at short term?

Another fast scan might give us all the charts with multiple subcharts (rectangular/not rectangular)!

Region 9 seems to be quite an orphan. Any godfather?

Hubert

[recri]

The stats page shows only 1330 charts untouched, so we have touched half the charts and passed the halfway mark sometime this weekend.

[bcn]

Yes,

and screening the charts as proposed will bring us more quickly to usable charts - and knowing where we have to invest more efforts.

[wdkester]

Originally Posted by bcn

Calibrators!

What about to identify all skewed (90/180/270) charts in your region to make those available at short term?

Another fast scan might give us all the charts with multiple subcharts (rectangular/not rectangular)!

Region 9 seems to be quite an orphan. Any godfather?

Hubert

Hubert, I (long with others) are doing the Caribbean part of the 20000 series. I got my info from the stats page NGA Charts project stats

26148 is 180 ddeg skew, 24463, 25001, 25484, 25609, 25613, 26320, 26341, 26342, 27080, 28006 are 90 or 270 deg skew.

Is this what you wanted?

[bcn]

Yes,

exactly. The other type of charts are the grouped ones: rectangular/not rectangular. You can identify them quite quickly just having a look on the miniature charts and marking their status correspondingly.

That will help Nohal to know what we are dealing with. And once the skewed charts identified those can be rotated and we have more charts to work with - in case we are running out of work.

Hubert

[recri]

Here are the morning's crop of problem charts:

processed: 1054 charts
accepted: 992 charts, mean ratio 1.00469 stddev 0.00191

problem: chart 23115 ratio 0.01032 - 4d error in corner coordinate
problem: chart 27044 ratio -0.49682 - lat/lng swapped in corner coordinate
problem: chart 27211 ratio 0.67606 - 10' error in corner coordinate
problem: chart 44400 ratio 0.00806 - 10d error in corner coordinate
problem: chart 54167 ratio 0.28474 - corners from different charts
problem: chart 54180 ratio 0.28575 - corner coordinate of inset entered
problem: chart 54279 ratio 1.36566 - 27' error in corner coordinate
problem: chart 54360 ratio 0.55304 - 3d error in corner coordinate
problem: chart 54418 ratio -3.74932 - 2d error in corner coordinate
problem: chart 55110 ratio 0.07020 - corner coordinate W instead of E

missingref: 54344 54369 - marked "need special PLY" when "corners not visible on cutouts"

One way to incorporate this computation into the web calibration interface, without getting too intrusive, would be to add a "Summary" panel which displayed a few easily computed metrics on the chart and how they measure up to the values found in previously accepted charts. It would save a lot of time if people could fix problems without entering data into the database and waiting for a KAP to download and check in opencpn.

The summary panel could display:

Is the SW corner SW of the NE corner?
Is ((N edge length)/(S edge length)) approximately 1?
Is ((E edge length)/(W edge length)) approximately 1?
Is (distance(SW,NE)/distance(SE,NW)) approximately 1?
Is (aspect(projected corners)/aspect(REF points)) approximately 1?

[recri]

Here's another interesting corner coordinate example for the manual.

In chart 22263 the NE corner longitude disagrees by 1" with the longitude label immediately above it. That the corner longitude is 1" wrong is apparent when you view the KAP in opencpn.

[HappySeagull]

Originally Posted by recri

One way to incorporate this computation into the web calibration interface, without getting too intrusive, would be to add a "Summary" panel which displayed a few easily computed metrics on the chart and how they measure up to the values found in previously accepted charts. It would save a lot of time if people could fix problems without entering data into the database and waiting for a KAP to download and check in opencpn.

The summary panel could display:

Is the SW corner SW of the NE corner?
Is ((N edge length)/(S edge length)) approximately 1?
Is ((E edge length)/(W edge length)) approximately 1?
Is (distance(SW,NE)/distance(SE,NW)) approximately 1?
Is (aspect(projected corners)/aspect(REF points)) approximately 1?

Wow! can this be done?and
Recri,will your ChartCrawler work when the complicated charts begin too?super tool
editf course I guess you can't know that yet

By the way,who's leaping in to fix these?

[bcn]

Originally Posted by recri

Here are the morning's crop of problem charts:

processed: 1054 charts
accepted: 992 charts, mean ratio 1.00469 stddev 0.00191

..........
problem: chart 54167 ratio 0.28474 - corners from different charts
problem: chart 54180 ratio 0.28575 - corner coordinate of inset entered
problem: chart 54279 ratio 1.36566 - 27' error in corner coordinate
problem: chart 54360 ratio 0.55304 - 3d error in corner coordinate
problem: chart 54418 ratio -3.74932 - 2d error in corner coordinate
problem: chart 55110 ratio 0.07020 - corner coordinate W instead of E

missingref: 54344 54369 - marked "need special PLY" when "corners not visible on cutouts"

These are corrected - great that we are some hours ahead to produce errors
-the checker is really a big, big help

The missingrefs: that happens when one forgets to change the status when in the end the corners do not appear.

Will try to finish the first run through 5xxxx series soon

Hubert

[recri]

Originally Posted by HappySeagull

Wow! can this be done?and
Recri,will your ChartCrawler work when the complicated charts begin too?super tool
editf course I guess you can't know that yet

I expect it will, the plans/insets should show up as sub-chart's in the database, then they'll have the same data available.

By the way,who's leaping in to fix these?

Most of the errors just seem to go away without fuss, there are only two in this set that were in the last set.

[GDD]

Hi,

I've been looking through this thread trying to work out just is happening.
Are people manually calibrating every map on the NGA website?
How can I get involved?
Is there a depository somewhere of maps that have been calibrated?
Is there some sort of sign-up sheet for who is calibrating which maps?
It looks like an excellent project!

cheers,

george

[recri]

Originally Posted by GDD

Hi,

I've been looking through this thread trying to work out just is happening.
Are people manually calibrating every map on the NGA website?
How can I get involved?
Is there a depository somewhere of maps that have been calibrated?
Is there some sort of sign-up sheet for who is calibrating which maps?
It looks like an excellent project!

cheers,

george

Welcome George --

You've nailed it. Register your self at

NGA Charts Status at OpenCPN.info (English)

and start digitizing. The region 9 charts looked sort of lonely last night, but feel free to jump in anywhere. A link to the most recent release of the evolving manual of digitization practices should appear in the last few days of posting.

Roger

[paulr]

Originally Posted by recri

Here are the morning's crop of problem charts:

problem: chart 23115 ratio 0.01032 - 4d error in corner coordinate
problem: chart 27044 ratio -0.49682 - lat/lng swapped in corner coordinate
problem: chart 27211 ratio 0.67606 - 10' error in corner coordinate
problem: chart 44400 ratio 0.00806 - 10d error in corner coordinate
problem: chart 54167 ratio 0.28474 - corners from different charts
problem: chart 54180 ratio 0.28575 - corner coordinate of inset entered
problem: chart 54279 ratio 1.36566 - 27' error in corner coordinate
problem: chart 54360 ratio 0.55304 - 3d error in corner coordinate
problem: chart 54418 ratio -3.74932 - 2d error in corner coordinate
problem: chart 55110 ratio 0.07020 - corner coordinate W instead of E

missingref: 54344 54369 - marked "need special PLY" when "corners not visible on cutouts"

Absolutely no doubt that this is a useful way of checking, but I would like to sound a note of caution. All the charts need checking manually too, and, I suggest, by at least one person other than the original calibrator.

I have found quite a few errors that cannot be picked up by computation, but which could have a serious impact on the credibility of the project.
54418 mentioned above is a case in point. The soundings datum is set to MHWS, not MLWS. Just by random checking, I've found and corrected half a dozen or more incorrectly specified soundings datums.

Whatever warnings we attach to these charts, people are still going to use them for navigation. Specifying fathoms instead of metres, for example, could have very serious consequences.

Paul