GRIB2 PlugIn

[Gilletarom]

Dave

There are two years I expect that. It is magic.

Best regards. Gilletarom.

[rgleason]

FYI wgrib2 does not list Windows.

[Ptizef]

Hi All

I extracted grib2 files from French Site :
Model AROME : 0.025 resolution, 0h-6h and 07h-12h ( about 2 x 32 MB)
tested only with Linux:
Grib_pi 4.0 loaded both in one time (concatenation) and displayed apparently correctly the parameters :
wind - gust - Pressure - Cloud cover - Air temperature
Globally it's far slower than grib1 :
-refreshing screen when parameters are changed is slow
-loading is long, panning and zooming is slow and it is worse when isobars are displayed. in this case, the zone is saturated with the pressure values and moving screen become very slow
I tested also an ARPEGE file with 0,1 resolution and SP1 ( about 39MB)
same result as above
see screen shots

Jean Pierre

[Rasbats]

A mistake writing copy.bat.

datatime in the grib_set commandline should read 00 (not 60);

The title bar of the Grib2 plugin now shows the correct base date/time (2011-03-31 00:00:00).

But .... the first record is 1300utc and still not 1200.

Also the current boxes have N/A N/A for direction and force.

Mike

[Ptizef]

Mike
Could you post the files you are reading?
thanks
JP

[Rasbats]

JP,

Post #45 of this thread has the files.

If this does not help please let me know.

Mike

[bdbcat]

Ahoy GRIB2 testers...

I have posted a new github commit correcting several problems noted in this thread. It is available on github as branch "GRIB2".

I have also built a Windows installer, and uploaded it as before. Here is the link.
http://opencpn.navnux.org/beta/opencpn_4.2.1_setup.exe

The fixes involved mostly the BSH (Current) GRIBv2 files.

Technical digression...
For the technically interested, the BSH current files are the first GRIBv2 files I have seen that use JPEG2000 for coverage map encoding. Our Jasper library needed a little touchup. Also, oddly, I found that the grid embedded in the GRIB file itself was what might be called "sparse". Alternate points in the grid, both x and y, were empty, i.e. GRIB_NOTDEF. This scrambled our spacial interpolation logic. So I implemented code to "fill out" a sparse grid when encountered, using simple linear averaging logic. This makes me a little nervous...

I cannot understand why the GRIB was produced by BSH with essentially twice the resolution required for the available data. This method is not prohibited by GRIB2, but seems very inefficient. I note that Meteo France and NOAA GFS GRIBv2 files seem to always contain a full grid.

All comments and feedback appreciated.

Dave

[maitai]

FYI in qtVlm we have decided that these gribs are just plain wrong, after quite some work some years ago. I actually don't know of any routing software able to exploit them, including Adrena.

Just my 2cts.

[rgleason]

Translated french link below
https://translate.google.com/transla...3D51&sandbox=1

Translated
https://translate.google.com/transla...-text=&act=url

Leads to
Index of /ensemble/naefs/grib2/raw

Picked one path
Index of /ensemble/naefs/grib2/raw/00/000

Downloaded each type and tested. Not many work but some do

Ones in the last link are not working at all.
Hung up and have to restart. Need a better way to escape bad files rather than restarting whole program.

[bdbcat]

Rick....

Which one(s) of the CMC_naefs files did you try?

With latest MSW O4.2.1, I just tried this one, with no trouble.
http://dd.meteo.gc.ca/ensemble/naefs..._allmbrs.grib2

This is PRMSL (pressure reduced to Sea Level) for the entire world. Nice isobars.

Limited WiFi bandwidth here, so cannot check more just now....

Dave

[Ptizef]

Dave

Tested new git with Linux
Current files seems now displayed fine . Data at cursor position are here and
overlay map's OK

But grib2 files from French "arome" model no longer work
I get a segment fault. here the call stack :
""
#0 0x7fffd3bcd26e GribRecord::getValue(j=<optimized out>, i=246, this=<optimized out>) (/home/jeanpierre/Documents/OpenCPN/plugins/grib_pi/src/GribRecord.h:178)
#1 ?? GribRecord::getInterpolatedValue (this=<optimized out>, px=<optimized out>, py=<optimized out>, numericalInterpolation=numericalInterpolation@entr y=true, dir=dir@entry=false) (/home/jeanpierre/Documents/OpenCPN/plugins/grib_pi/src/GribRecord.cpp:497)
#2 0x7fffd3b9de99 CursorData::UpdateTrackingControls(this=0x4df5000) (/home/jeanpierre/Documents/OpenCPN/plugins/grib_pi/src/CursorData.cpp:310)
#3 0x7ffff5cc02de wxAppConsoleBase::CallEventHandler(wxEvtHandler*, wxEventFunctor&, wxEvent&) const() (/usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0:??)
#4 0x7ffff5e46537 wxEvtHandler::ProcessEventIfMatchesId(wxEventTable EntryBase const&, wxEvtHandler*, wxEvent&) () (/usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0:??)
#5 0x7ffff5e4691e wxEvtHandler::SearchDynamicEventTable(wxEvent&) () (/usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0:??)
#6 0x7ffff5e469ae wxEvtHandler::TryHereOnly(wxEvent&) () (/usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0:??)
#7 0x7ffff5e46a73 wxEvtHandler::ProcessEventLocally(wxEvent&) () (/usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0:??)
#8 0x7ffff5e46ad5 wxEvtHandler::ProcessEvent(wxEvent&) () (/usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0:??)
#9 0x7ffff5e46837 wxEvtHandler::SafelyProcessEvent(wxEvent&) () (/usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0:??)
#10 0x7ffff5dac751 wxTimerImpl::SendEvent() () (/usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0:??)
#11 0x7ffff6377fdb ??() (/usr/lib/x86_64-linux-gnu/libwx_gtk2u_core-3.0.so.0:??)
#12 0x7ffff113e913 ??() (/lib/x86_64-linux-gnu/libglib-2.0.so.0:??)
#13 0x7ffff113deaa g_main_context_dispatch() (/lib/x86_64-linux-gnu/libglib-2.0.so.0:??)
#14 0x7ffff113e250 ??() (/lib/x86_64-linux-gnu/libglib-2.0.so.0:??)
#15 0x7ffff113e572 g_main_loop_run() (/lib/x86_64-linux-gnu/libglib-2.0.so.0:??)
#16 0x7ffff1a0aa07 gtk_main() (/usr/lib/x86_64-linux-gnu/libgtk-x11-2.0.so.0:??)
#17 0x7ffff6359055 wxGUIEventLoop:oRun() () (/usr/lib/x86_64-linux-gnu/libwx_gtk2u_core-3.0.so.0:??)
#18 0x7ffff5d022b3 wxEventLoopBase::Run() () (/usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0:??)
#19 0x7ffff5cc7746 wxAppConsoleBase::MainLoop() () (/usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0:??)
#20 0x7ffff5d53e80 wxEntry(int&, wchar_t**) () (/usr/lib/x86_64-linux-gnu/libwx_baseu-3.0.so.0:??)
#21 0x504d88 main(argc=1, argv=<optimized out>) (/home/jeanpierre/Documents/OpenCPN/src/chart1.cpp:827)
""
These files worked with the previous git (Yesterday) See my post #48

will test with Windows as soon as possible

Jean Pierre

[Ptizef]

Mike

Originally Posted by Rasbats

JP,
Post #45 of this thread has the files.
If this does not help please let me know.
Mike

Ok thanks
I tested with linux and the last git
Cursor at data position are here and overlay map is OK now
but for me too dates and times are very curious:
For IBI_ucurr.grb the starting date is 31/03/2011 00:00 but data
are from 08/03/2016 14:00 to 08/03/2016 14:00 ???
For current_no_20160308.grb2 starting date is 08/03/2016 00:00
but data start at 08/03/2016 00:30
I have not the IBI_vcurr.grb

Jean Pierre

[rgleason]

Very large Arome grib2 eventually works but it takes an endless time loading. Endless. I thought it was frozen. It seemed to help to close a second hodden opencpn window via the taskbar that always is opened. Then the Arome grib appeared.

[bdbcat]

jp...

On the linux Arome GRIB2 crash...

I see no such crash on small GRIBs.

Is this with a full METEO France (large) GRIBv2 file? Or a small one from did-g's server? If small, please post, or show download URL, please.

Thanks
Dave

[Ptizef]

Dave

I get grib2 files from MeteoFrance at this URL:
https://donneespubliques.meteofrance...id_rubrique=51
My selection is in the shot attached

The result is a file of about 32MB
These files was read normally (although slowly ) in Linux with the previous git until Yesterday. I was even able to read two files at a time thus 12 hours of predictions
I had not tested with that git level in Windows, but I tested now both the new installer and a local vs2010 new git build , and I get also a crash.
You can find debug mode call stack in shot attached

Jean Pierre