|
|
05-06-2014, 20:28
|
#406
|
Registered User
Join Date: Aug 2009
Location: oriental
Boat: crowther trimaran 33
Posts: 4,414
|
Re: New Grib Presentation
Quote:
Originally Posted by Baikal
Join wishes
+1
|
It must be interpolating between data points. It does this over time (for interpolate mode in playback) but not yet over space. Is this correct? We must make it clear to the user, because our interpolation could potentially display less accurate data if it is just a simple linear interpolation.
|
|
|
06-06-2014, 00:27
|
#407
|
Registered User
Join Date: Sep 2012
Location: Baikal
Posts: 580
|
Re: New Grib Presentation
Quote:
Originally Posted by boat_alexandra
It must be interpolating between data points. It does this over time (for interpolate mode in playback) but not yet over space. Is this correct? We must make it clear to the user, because our interpolation could potentially display less accurate data if it is just a simple linear interpolation.
|
Yes, I understand that interpolation affects the accuracy of the forecast. For example do interpolation only for wind arrows, but not for OverlayMap. Allow the user to select and warn of possible deterioration of the accuracy of the forecast. You can also do a partial interpolation up to a certain scale.
|
|
|
06-06-2014, 00:48
|
#408
|
Registered User
Join Date: Aug 2009
Location: oriental
Boat: crowther trimaran 33
Posts: 4,414
|
Re: New Grib Presentation
The overlay map already interpolates (with the use gradual colors option in preferences) as does the cursor data.. so implementing higher density arrows would be trivial. The question then is how? What ui interface, a simple check box with a spinbox to specify density (like the number display) ?
I think the separate settings/preferences is a bit cumbersome. Also the interpolation over time does not work with moving gribs (which means they cannot be used properly with weather routing either). I wonder if that should be added as well.
|
|
|
06-06-2014, 03:46
|
#409
|
Registered User
Join Date: Sep 2009
Location: Angers - France
Boat: Beneteau First 29 Ptizef
Posts: 844
|
Re: New Grib Presentation
As it is, wind arrows are drawn each time and if we increase the concentration, we have performance problems in dragging and zooming (I already tried some times ago)
So I think there is a little more work to do (may be draw only one time for a value and put it in a cache)
JP
|
|
|
06-06-2014, 04:26
|
#410
|
Registered User
Join Date: Sep 2012
Location: Baikal
Posts: 580
|
Re: New Grib Presentation
Maybe increase the existing pavement arrow twice as and no more
|
|
|
06-06-2014, 05:03
|
#411
|
Registered User
Join Date: Jul 2010
Location: Hannover - Germany
Boat: Amel Sharki
Posts: 2,537
|
Re: New Grib Presentation
Quote:
Originally Posted by Ptizef
As it is, wind arrows are drawn each time and if we increase the concentration, we have performance problems in dragging and zooming (I already tried some times ago)
So I think there is a little more work to do (may be draw only one time for a value and put it in a cache)
JP
|
JeanPierre,
the zyGrib code is open source, you may have a look how it was done there.
However zyGrib uses Qt and this might be the better solution than wxWidgets...
Gerhard
|
|
|
17-06-2014, 03:04
|
#412
|
Registered User
Join Date: Aug 2009
Location: oriental
Boat: crowther trimaran 33
Posts: 4,414
|
Re: New Grib Presentation
Quote:
Originally Posted by Ptizef
As it is, wind arrows are drawn each time and if we increase the concentration, we have performance problems in dragging and zooming (I already tried some times ago)
So I think there is a little more work to do (may be draw only one time for a value and put it in a cache)
JP
|
Do you have an example of a grib that is particularly slow? I can optimize the wind-barb rendering quite a lot, and this is probably needed if we begin to interpolate between them.
I also implemented two more improvements to the grib plugin. One is to allow interpolation over time for moving gribs. It only works if all of the slices use the same grid size, but I think this is always the case? Otherwise it must interpolate over space to make the interpolation over time (which I didn't implement)
The other improvement is to use polar interpolation rather than vector interpolation for wind and currents over both time and space. I believe this is a big improvement as a change in wind direction but not wind strength before would change the wind strength to something lower between the two time slices. Of course if we try hard enough we can always find a counter example (when wind strength also changes greatly) so as noted before, there is also error in the interpolated data. For this reason, it may be possible to use a better interpolation than polar (averaging speed and direction independently) but vector interpolation is obviously faulty.
If it were possible to get the rate-of-change and other parameters from the weather model we could do a lot better, but these have no field in gribs and are of little use to a casual user viewing each frame of a grib.
I have pushed these changes to my fork of opencpn:
Code:
https://github.com/seandepagnier/OpenCPN
I will wait a few days for feedback before submitting a pull request. I only had one moving grib to try.
|
|
|
17-06-2014, 14:55
|
#413
|
Registered User
Join Date: Sep 2009
Location: Angers - France
Boat: Beneteau First 29 Ptizef
Posts: 844
|
Re: New Grib Presentation
Quote:
Originally Posted by boat_alexandra
Do you have an example of a grib that is particularly slow? I can optimize the wind-barb rendering quite a lot, and this is probably needed if we begin to interpolate between them.
|
Yes I have a file with a resolution of about 6'x12' which is very slow. The most impact seems to come from pressure isobars then from wind barbed arrows.
you can get it at
Værdata i GRIB-format - yr.no
Load the file named "metno-neurope.grb" (size more than 10Mb)
Furthermore, to display barbed arrows at interval less than the resolution, we face the problem of the endless increase of the overlay to be read as zooming in( as for the colored map overlays). May be there is another way to do this ?
Quote:
Originally Posted by boat_alexandra
I also implemented two more improvements to the grib plugin. One is to allow interpolation over time for moving gribs. It only works if all of the slices use the same grid size, but I think this is always the case? Otherwise it must interpolate over space to make the interpolation over time (which I didn't implement)
|
Generally yes, moving files have the same grid size.
But there are the files with a changing resolution over time ( 0.5°x0.5° for a first period then 2.5°x2.5° etc...) . we have an example with the files got from saildocs for more than 8 days horizon.
When the change of resolution happens, it implies a change of the size. So at this moment, interpolation do not work .
So , to be complete,we'll need interpolation over space
Quote:
Originally Posted by boat_alexandra
The other improvement is to use polar interpolation rather than vector interpolation for wind and currents over both time and space. I believe this is a big improvement as a change in wind direction but not wind strength before would change the wind strength to something lower between the two time slices. Of course if we try hard enough we can always find a counter example (when wind strength also changes greatly) so as noted before, there is also error in the interpolated data. For this reason, it may be possible to use a better interpolation than polar (averaging speed and direction independently) but vector interpolation is obviously faulty.
If it were possible to get the rate-of-change and other parameters from the weather model we could do a lot better, but these have no field in gribs and are of little use to a casual user viewing each frame of a grib.
I have pushed these changes to my fork of opencpn:
Code:
https://github.com/seandepagnier/OpenCPN
I will wait a few days for feedback before submitting a pull request. I only had one moving grib to try.
|
I have poor internet possibilities and very little time at this moment but I will very pleased to test it as soon as possible when it will be in the opencpn git
many thanks
Jean Pierre
|
|
|
17-06-2014, 23:32
|
#414
|
Registered User
Join Date: Aug 2009
Location: oriental
Boat: crowther trimaran 33
Posts: 4,414
|
Re: New Grib Presentation
Quote:
Originally Posted by Ptizef
Yes I have a file with a resolution of about 6'x12' which is very slow. The most impact seems to come from pressure isobars then from wind barbed arrows.
you can get it at
Værdata i GRIB-format - yr.no
Load the file named "metno-neurope.grb" (size more than 10Mb)
|
Ok, I will try this one.
The isobars aree something completely different, and doesn't really benefit from interpolating data points over space (except the lines might look smoother) One possibility is to use the isobar routine from climatology (which is also the same routine in wmm) which is significantly more efficient than what is used in the grib plugin.
Quote:
Furthermore, to display barbed arrows at interval less than the resolution, we face the problem of the endless increase of the overlay to be read as zooming in( as for the colored map overlays). May be there is another way to do this ?
|
Yes, the overlay itself would not be changed in any way. In opengl mode it already linearly interpolates. The only change proposed is to add an space interpolation option for wind and current arrows, along with a spacing parameter (same as "numbers")
Quote:
Generally yes, moving files have the same grid size.
But there are the files with a changing resolution over time ( 0.5°x0.5° for a first period then 2.5°x2.5° etc...) . we have an example with the files got from saildocs for more than 8 days horizon.
When the change of resolution happens, it implies a change of the size. So at this moment, interpolation do not work .
So , to be complete,we'll need interpolation over space
|
I think since .5 is a multiple of 2.5, we can get away without interpolation over space because we can just use the larger spacing. I need to adapt the program to handle this properly, so I need an example file.
Are there cases where you won't have an exact multiple (like 1.5 and 2.5 spacings) I think this case was already broken before for non-moving gribs.
Quote:
I have poor internet possibilities and very little time at this moment but I will very pleased to test it as soon as possible when it will be in the opencpn git
many thanks
Jean Pierre
|
Sure, it's up to you, but anyone who wants to test this can just add a remote branch of my fork which should take little internet traffic to pull.
|
|
|
27-06-2014, 04:07
|
#415
|
Registered User
Join Date: Mar 2012
Location: Boston, MA
Boat: 1981 Bristol 32 Sloop
Posts: 17,523
|
Re: New Grib Presentation
This is a little off track, but Gribfiles.com has some good information summarized for different grib types/sources. Could be useful.
https://www.gribfiles.com/#/Weather-models
|
|
|
08-12-2014, 08:23
|
#416
|
Registered User
Join Date: Mar 2012
Location: Boston, MA
Boat: 1981 Bristol 32 Sloop
Posts: 17,523
|
Re: New Grib Presentation
Add manual mail request area selection option (typing or drawing) (#FS 1235)
https://github.com/OpenCPN/OpenCPN/c...89e0a9d9d6e54f
FS#1235 : UI Grib Request Area Selection -options
I've been trying to test the new Manual and Area box selection functions and have a few questions about how they work.
When the manual box is checked two sets of coordinates appear
Max Lat, Max Long. Min Lat Min Long
are these set by the current Chart Screen Area (yes) or by the Current Grib being displayed (No).
When "Start Graphic Selection" button is pressed the plugin expects the user to drag out an Area Selection by diagonal corners. (Click corner, hold down, drag diagonal, release on corner)
When this is started a red box appears.
If "Stop Graphic selection" is picked the process is stopped.
After completion of Area box selection the red square remains until "Continue" is selected and the red area box disappears, integral lat long values calculated and placed in the Mail "send ... request area, ready to send the request by email.
Ptzif, Very nice implementation. Thank you. My only suggestion might be to somehow prompt the user to Drag diagonally an area box!
|
|
|
14-12-2014, 03:38
|
#417
|
Registered User
Join Date: Feb 2010
Posts: 619
|
Re: New Grib Presentation
I'd like to say a BIG "Thank You" to all the contributors to the new features!
I find most useful the Weather Table and - of course - the Particles view. Thanks a lot, Sean. The particle view is a welcome relax to all-discrete viewing. It also shows some local phenomena in a more apparent way, I think. Some I just start learning to see...
Moreover, it is great fun to watch!
Thanks again,
Piotr
|
|
|
14-12-2014, 16:19
|
#418
|
Registered User
Join Date: Aug 2009
Location: oriental
Boat: crowther trimaran 33
Posts: 4,414
|
Re: New Grib Presentation
Does anyone know of a web access to grib files we can support in the plugin? It would be faster and more automatic than email.
|
|
|
15-12-2014, 02:49
|
#419
|
Registered User
Join Date: Jul 2010
Location: Hannover - Germany
Boat: Amel Sharki
Posts: 2,537
|
Re: New Grib Presentation
zyGrib
|
|
|
16-12-2014, 17:02
|
#420
|
Registered User
Join Date: May 2011
Posts: 847
|
Re: New Grib Presentation
Quote:
Originally Posted by boat_alexandra
Does anyone know of a web access to grib files we can support in the plugin? It would be faster and more automatic than email.
|
But please let's not drop, of stop optimising, email; those of use that travel in less civilised areas rely on email, and over very thin routes such as sailmail via SSB
Chris
Sent from my iPhone using Cruisers Sailing Forum
|
|
|
|
|
Thread Tools |
Search this Thread |
|
|
Display Modes |
Rate This Thread |
Linear Mode
|
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
|
Advertise Here
Recent Discussions |
|
|
|
|
|
|
|
|
|
|
|
|
Vendor Spotlight |
|
|
|
|