search

NOAA-PMEL / LAS

Live Access Server
https://ferret.pmel.noaa.gov/LAS/
The Unlicense
13 stars 5 forks source link

How should cruises be chopped of for rectangular regions #1604

Open karlmsmith opened 6 years ago

karlmsmith commented 6 years ago

Reported by @AnsleyManke on 29 Jul 2014 21:19 UTC (For SOCAT 4)

If cruises come round a landmass or otherwise have a curved shape, and we set a region that chops off the middle of the curve, then the data contains the start of the curve, then what should be a gap, not represented by any data, and then the end of the curve.

When drawn as straight lines, this results in colored lines drawn across the edge of the plot where the line was chopped off. Do we want to draw this differently? How to find such locations?

Migrated-From: http://dunkel.pmel.noaa.gov/trac/las/ticket/1597

karlmsmith commented 6 years ago

Attachment from @AnsleyManke on 29 Jul 2014 21:19 UTC several cruises with a curved shape coming_round_greenland

karlmsmith commented 6 years ago

Attachment from @AnsleyManke on 29 Jul 2014 21:20 UTC A subregion cuts off cruises: lines are drawn across the bottom edge chop_off_greenland

karlmsmith commented 6 years ago

Attachment from @AnsleyManke on 19 Jun 2015 17:18 UTC

greenland_larger_request

karlmsmith commented 6 years ago

Comment by @AnsleyManke on 19 Jun 2015 17:18 UTC If the data request was slightly extended, so that it includes points beyond the requested region, then the plot would in effect draw all of that larger set of data, windowed to the requested region. Then that line isn't visible.

As long as the data is reasonably dense, asking for even another fraction of a degree of latitude would be plenty.

Attaching a plot where I've made a data request a bit bigger than the region in the above plots, then slightly shrinking the plot region, to see the effect.

karlmsmith commented 6 years ago

Comment by @AnsleyManke on 23 Jun 2015 16:42 UTC Here's another good solution:

We could use the sample number to draw the correct plot. The line segment would stop on either missing data or a gap in the sequence numbers. For this purpose, the decimated datasets need this to be the sequence after decimation, not the decimated version of the sample number from the full dataset.

This would also solve ticket 1696, where a constraint has been applied and we draw a line connecting the valid points with the constraint but would like to skip the ones where the constraint is not met.