Climatology

[seandepagnier]

I improved the handling of overlays by using npot textures instead of texture rectangles if supported. This won't affect some drivers much, but on others it could be faster or avoid other problems. It fixes several issues with viewport rotation.

[seandepagnier]

I want to use this image for the download page.

[rgleason]

Rick:

With everything on in the Display Panel, everything checked, and if Isbars is checked on for each datatype, plus the overlay is on for each datatype, that is a lot of calcs, so it is slower and isbar regen is more of a problem.

If you turn off opengl, the overlay is turned off so if all isbars are on, it will be faster, but still slow.

Opencpn 3.3.1824 seems to be a little better about this due to the opengl and optimizing.

Sean:

I improved the handling of overlays by using npot textures instead of texture rectangles if supported. This won't affect some drivers much, but on others it could be faster or avoid other problems. It fixes several issues with viewport rotation.

I am talking about the situation when some idiot decides to check all the boxes to display all datatypes at once, with overlay textures checked on (opengl is on too), and requesting to have isobars drawn for every datatype!!!! -- (Later) You can't even understand all the overlays when they are displayed so there's a natural limit there too that most users will never reach. How many overlays displayed at one time is reasonable?

IMHO this is completely unreasonable, ridiculous, outrageous, extreme. We need a check on user control here I think, until everyone has even faster computers in 5 years. I am going to suggest:
First limit-
1. Isobars: No more than 5 datatypes with isobars "on" (Later, actually display).
2. Overlay (opengl on): When opengl is on, and more than 5 datatypes have overlay textures "on" (Later, ready to actually display and cause calcs.. needs some thought where to put the limit).
Second limit- If opengl is off, there is no overlay texture calcs, so the Isobar limit could be raised to something else like 7 datatypes with isobars "on".

When a user checks an additional isobar display, if the limit is reached a message comes up "Too many isbars checked, limit reached" and the box is not checked. (Later, maybe we should focus on the actual calcs, and display, so the user can set up his datatype to display as he wants, but he just cant check too many display boxes.....Your input would be very helpful here please.)

We would need to figure out what the limits should be and what is reasonable.

--In the image above for download page, I like the sustained wind colors, very very nice.

[seandepagnier]

I think something similar is done in grib, only a single overlay is allowed at once.

So I suppose it could simply uncheck any fields which already have an overlay map.

I think it's possible to use 2 or more in some cases, but not really very useful.

Sean

[rgleason]

Sean,
I git and compiled most recent climatology. It works with all isobars checked and with all overlays ON , I would never use it that way (I'd be an idiot) but the Isobar progress bar never came up and never got lost. I did not hit any limits where there was a message, so I am wondering what you did, but I think this is working now for all normal human beings...(----!!!!!

I think it should be released for general consumption. Unfortunately we have been changing VPS providers over the last 4 days (pretty seamless) but my file limits for Opencpn Beta File thingie got changed and I don't have access to change it back yet, so I cant upload the 6mb exe . It is at 2mb limit now.

[rgleason]

When we zoom out too far the opengl texture overlays get out of scale horizontally.
It is a minor annoyance really, but should be fixed if possible.
http://www.cruisersforum.com/forums/...ml#post1626927

Have posted in flyspray.
FS#1487 : Texture Overlay widens out of scale when zoom out (Opengl)

[seandepagnier]

I have made a new release of climatology in git:
https://github.com/seandepagnier/climatology_pi

The required climatology data is rebuilt and can be obtained:
opencpnplugins - Browse /climatology_pi at SourceForge.net

The difference is only for wind and currents:
- The gales and calms are now computed more correctly
- The spacial resolution is half degree world grid instead of one degree
- The resolution for wind speed and directions is improved
- The compression of the wind data is greatly improved.
- Areas with insufficient data are omitted
- Current data for 0.2 knots and above instead of 0.3 knots

The package is 15% larger, but I didn't want to sacrifice the quality of the wind atlas.

My plan next is to produce an optional high resolution version with .25 degree spacial accuracy for wind, even more accuracy for directions and speeds (but this will greatly increase the database, up to around 15MB)

Next, it is possible to produce the following:
- Day/Night slider to show the changes in wind over the course of 24 hours at each time in the year
- Derivative terms or weekly averages to improve daily interpolation
- Wind atlas with speed buckets so you can see percentage of time wind is at various speeds for each direction.
- El Nino slider to show how wind and currents change.

[rgleason]

Climatology_pi updated by Sean with a new more accurate DB downloaded from Sourceforge! See Sean's notes above.

Opencpn Beta File Thingie rguser rgpass
Directory: 3.2.2 PI Win
File: climatology_pi-0.13-ov3.2.2-d-db-new-win32.exe

I have used this plugin with Wind and Current and "Grib_pi" and "Weather_Routing_pi". The Gulf current eddys are represented better and the 10-12 wx-routes all calculated fine.

Windows users, please test it out. Thanks.

PS: Sean, your next versions sound enticing. Would you mind making the "Day" area in "Month" "Day" just a little wider? - maybe take some space from the month field? It is difficult to see the second numeral in the day field.

Thanks so much!

[sbfreddie]

Originally Posted by boat_alexandra

I have made a new release of climatology in git:
https://github.com/seandepagnier/climatology_pi

The required climatology data is rebuilt and can be obtained:
opencpnplugins - Browse /climatology_pi at SourceForge.net

The difference is only for wind and currents:
- The gales and calms are now computed more correctly
- The spacial resolution is half degree world grid instead of one degree
- The resolution for wind speed and directions is improved
- The compression of the wind data is greatly improved.
- Areas with insufficient data are omitted
- Current data for 0.2 knots and above instead of 0.3 knots

The package is 15% larger, but I didn't want to sacrifice the quality of the wind atlas.

My plan next is to produce an optional high resolution version with .25 degree spacial accuracy for wind, even more accuracy for directions and speeds (but this will greatly increase the database, up to around 15MB)

Next, it is possible to produce the following:
- Day/Night slider to show the changes in wind over the course of 24 hours at each time in the year
- Derivative terms or weekly averages to improve daily interpolation
- Wind atlas with speed buckets so you can see percentage of time wind is at various speeds for each direction.
- El Nino slider to show how wind and currents change.

Sean:
Ed here (Freddie). Will this Plugin run on the CubieTruck (ARMv7)? Does it need full OpenGL?
I would like to try getting all of your plugins working on the CubieTruck, but am not sure which if any require full OpenGL as opposed to OpenGL-ES2.

Thanks,
Freddie

[seandepagnier]

Originally Posted by sbfreddie

Sean:
Ed here (Freddie). Will this Plugin run on the CubieTruck (ARMv7)? Does it need full OpenGL?
I would like to try getting all of your plugins working on the CubieTruck, but am not sure which if any require full OpenGL as opposed to OpenGL-ES2.

Thanks,
Freddie

All of my plugins work without opengl, however some features do not always work without opengl. The overlay maps for climatology need opengl. In the case with opengl-es2, it will work correctly (I use climatology on the cubie truck) As far as I know all the plugins work with opengl-es2, if not I would like to know.



I did a small experiment to build wind atlas maps for each time of day since the data is every 6 hours. I found the differences a lot more than I expected, especially on coastlines near tall mountains (like andes mountains) the daily land/sea wind effect can be seen far out to sea in these areas. In other areas, it is clear the wind direction shifts on a daily cycle 20-40 degrees is common in many areas. I expect this knowledge is useful when beating against the prevailing winds to time tacking appropriately. Unfortunately including this data would increase the database size to around 12MB.

Next experiment is to compare wind and current data built for different el-nino indicies.

[rgleason]

did a small experiment to build wind atlas maps for each time of day since the data is every 6 hours. I found the differences a lot more than I expected, especially on coastlines near tall mountains (like andes mountains) the daily land/sea wind effect can be seen far out to sea in these areas. In other areas, it is clear the wind direction shifts on a daily cycle 20-40 degrees is common in many areas. I expect this knowledge is useful when beating against the prevailing winds to time tacking appropriately. Unfortunately including this data would increase the database size to around 12MB.

-Sean would that 12mb increase be from only areas where the differences are significant? ...Or from increasing the data frequency over the entire world? Can the plugin process data that is not normalized spatially and over time? - I would not mind the 12mb increase for better routing!

-I wonder if there are places where the 6 hour data would be significant improvement?

-What would you use to define "a significant improvement"? --Where the wind changes more than 90 degrees and/or more than 20% of speed during the subject period?

Next experiment is to compare wind and current data built for different el-nino indicies.

PS: Very interesting results.

[rgleason]

Sean has updated Climatology
- improve el-nino test for cyclones - fix possible warning

Also the most recent CL-DATA-1.0 Database download from Sourceforge is included.

Note that at the moment Weather_Routing will not recognize this new Climatology data, there is a popup note from Weather_Routing.

This version of climatology is about 6042kb rather than 5400kb due to the .5 degree wind data.

Opecpn Beta File Thingie rguser rgpass
In folder 3.2.2 PI Win
climatology_pi-1.0-ov3.2.2-f-db-new-win32.exe

[rgleason]

Sean,
With Opencpn 3.3.2424 running Climatology, it appears that the Overlays or the Background Land outline is fogged or fuzzy and the overlays seem to be a little weird.
Is this being caused by something in Opencpn 3.3.2424?

[rgleason]

Sean
See this post about the Climatology overlays at Zoom out scales.

http://www.cruisersforum.com/forums/...ml#post1708979

I have updated the Climatology writeup and images, and Thomas has added many menu items.

[rgleason]

See these screenshots.
http://www.cruisersforum.com/forums/...-138517-6.html
Is there any chance for improvement?