Contour detection

I just finished big improvements in the contour detection of the GRIBs. The performances are now correct (a few milliseconds per GRIB layer, for a full Pacific GRIB 2 degrees by 2 degrees).
Now the contour detection can be part of a GRIB Animation.
In addition, there are now new check boxes to show and hide already computed contour lines, just like for the faxes.
The picture shows PRMSL and 500HGT contours detected in a South Pacific GRIB. Notice that there is now an option to show only the detected contours (with no wind or other GRIB data).

User Exit

Rendering of faxes and GRIB data seems to be an endless topic..., the soft this blog is about provides some options, but there is way more to do.

That's why we came up with this user-exit feature, that allows anyone who knows a bit of Java to do whatever he wants, with the faxes - reworked or not (color and transparency) - and the GRIB data. This creates an entry in the popup menu of the chart panel, which you can select to trigger external code. The only restriction is to implement a given Java Interface, and you can access the context of the chart panel. 

Above is an example of such a user-exit, smoothing and animating the GRIB data currently displayed in the chart pane. 500mb charts look a bit like a big slug drooling across the ocean...

We will soon post on the site a repository of such user exits (along with explanations about the way to write yours), the possibilities are endless.

The rendering provided by Processing is gorgious, I still need to look into it, for now I'm struggling with thread issues. They manage their own pool apparently, and I need to somehow sneak in... We'll see. 

The video above is a preview of the animated GRIB slug. It's smaller than the real one, but the picture on top will give you an idea of the actual size.

Smoothing the GRIB

The smoothing algorhythm mentioned in an earlier post can be adapted to the GRIB Data.
It allows you to turn this:

into that:

This smoothing is spatial, in other words, it fills the gap between the points.
This can reduce substantially the size of the GRIB data to download, as it gives the possibility to rebuild the data between the actual points. This will need some validation, but that sounds promising. SailMail users might be concerned; on the example above, the size of the GRIB file is ~15 times smaller.
The next step will be to add a time dimension, smoothing the wind across several frames.
Keep posted!

Smoothing GRIB Data

In order to minimize the size of the data that come back from the GRIB requests, I am working on some smoothing feature for the GRIB viewer. This intends to render a smoothed view of the GRIB files, even if the cells are bigger than 1 degree by 1 degree. This smoothing can also happen along the time dimension, from one GRIB layer to another, which would allow a better routing.
The idea is to determine the four closest GRIB points, and smooth the wind speed and direction, geographically, and along time, in case the GRIB has several frames, and you request the wind data between two frames. On the right is a first test of such a smoothing. The red are the GRIB data, the blue are calculated by the smoothing algorithm.