Windy Customized Graphics - Version 3

  • Hi @WXcycles

    Awsome work!
    Can this by added by CSS as I'm running a web-app version on my MagicMirror?

    Greetings from Norway!

  • @TheStigh If you mean API, you can change colors as well. Use testing library from this link and check the object windyAPI.colors. Please send any further API questions to API topics. Thanks.

  • Increased the visual contrast and details within the high-pressure areas, to match with the contrast and details in the low-pressure areas.

    Setting is here:

    Pressure [pressure]

  • This is deleted.

  • Changed cloud display intensity thresholds to better display details

    Clouds [clouds]
    Low clouds [lclouds]
    Medium clouds [mclouds]
    High clouds [hclouds]

    Setting here:

    This is a fully linear display of both clouds and rainfall

    The Rain, thunder [rainClouds] part of the display is here:

  • I've rebuilt the temperature overlay and doubled its size to maximize shading smoothness and visual contasts.

    Temperature [temp]

    Setting is here:


  • I made an entirely new thunderstorm overlay as the standard windy overlay is highly non-linear and introduces many scale, intensity and location distortions to the forecast depiction. The replacement linear overlay removes such distortion and keeps the display within the correct visual proportions and has a display data range about 9 times higher than the original. It displays the forecast intensities and locations much more clearly.

    Thunder storms [lightDensity]

    The setting is here:


  • I rebuilt the sulfur dioxide overlay with double the zoomed detail increment, and far better contour interval shading. It’s now my largest overlay.

    SO2 mass [so2]

    Setting is here:

    This replacement overlay is a fully linear display:

    I'll be reworking the carbon-monoxide (CO) overlay in a similar way, soon-ish.

  • The carbon monoxide overlay has been rebuilt with a clearer contoured shading to 9,915 ppbv and is able to show point source areas.

    CO concentration [cosc]

    Setting is here:


  • Created a further CO overlay that has both 4 times as much contrast sensitivity (as yesterdays) plus 4 times as many contours. The result is an ability to see both low and high emissions more clearly.

    CO concentration [cosc]

    Setting is here:

    New York


    I’ll be reworking the particle and gas overlays further as I refine this new contour shading approach to displaying more complex forecast outputs which normal shading methods are unable to display with sufficient contrasts and details.

  • Sailor

    How do you explain this moiré effect or interference fringes effect that these settings give to the overlay display?

  • @idefix37 said in Windy Customized Graphics - Version 3:

    moiré effect

    16-bit RGB.

    CO concentration [cosc]

    Setting is here:


  • Sailor

    I have checked and paste your settings for CO layer. I see that this moiré effect is in fact voluntary. You have introduced darker strips in your colour settings to obtain some kind of isolines in the colour layer?
    So that’s nothing to do with interference fringes

  • Mostly yes, I deliberately created contour shading around the otherwise unseen peak values (more like topographic contours) to illustrate where they are, and to give a sense of their relative gradients and the peak locations and their scale.

    Plus I add color to indicate a rise or fall in value across the contour interval. The moiré itself is a rendering artifact of 1980s to 1990s era 16-bit display processing of what’s a fairly coarse global forecast pseudo-‘dataset’.

    I’m pushing the display to show emission and plume detail that can’t be seen (at all) within the standard non-linear and quite limited dynamic range of the standard Windy CO overlay. It’s a process of finding optimal levels of contour increments, plus starting them at a useful initial value (color and brightness) and deciding whether the initial contour channel should rise, or else fall (between 0 and 256). Plus to decide how deep the contour shading should rise or fall (i.e. how visually prominent they are). The oscillating depth of the contour channel likewise affects its increment spacing. A subtle shallow oscillation places the contours closer together (and is very sensitive to detail changes) while a deeper contour shading channel oscillation is visually less subtle (dark) and places the contour increments further apart (and is less sensitive to detail changes).

    Basically, everything affects everything else when doing this, so it’s a lot of experimentation to find a direction in the settings that works best. I’ve only scratched the surface to see what this approach can do. For instance, if the contour increment goes too high (too many contours and too close) the lower-level display values go blocky and coarse when zoomed-out. Plus the moiré effect blooms in the higher-level peak areas when zoomed-in. But if the contour increment is too low the visual detail drops away sharply. So finding a setting that works for displaying both the highest-peak levels and also provides detail in dispersing plumes gets tedious fast.

    This is the only way I’ve identified to linearly display the detail that’s already present but which can’t be displayed using normal shading methods, as the global data range is much too high for normal shading approaches to work (especially if using a non-linear shading). This CO overlay has by far the highest range of data values of all the Windy overlays, and this current version uploaded goes up to 9,100 ppbv (>4 times higher than the standard overlay).

    The current ‘Dust-mass’ overlay also contains very high values (up to 3,500 µg/m³) and was the first overlay I tried this contouring approach with (and will be refining it further yet) but the Dust-mass overlay contains only 1/3 of the CO overlay’s range. Getting so much forecast information to display within a linear display requires messing about to find optimal settings for each individual gas and particle overlay. The ozone overlay doesn’t need contouring at all as its’ value-range and increment level is relatively small, so a standard linear overlay shading approach works well for that gas.

    But any gasses with very high local peak value and associated regional to global downstream plume flow requires contours plus a (complicated to implement) color shading on top of it, to display the full peak value range appropriately. Creating such displays generates large and difficult to edit overlays. There are 4,000 lines of text in this one and the CO overlay uses almost all of the available 96,000 characters that Windy’s blog software permits me to upload. That many characters slows down page loading times in this thread, as you may have noticed.

    So I’m hoping to only have to use this technique within three overlays, namely, Dust-mass, SO2 and CO. I’m currently constructing editing tools to simplify and speed up editing and testing such contoured overlays.

    So these visual moiré interference fringes are for me just another one of the necessary compromises to balance.

  • This is a far more subtle carbon-monoxide overlay, the contour shading is translucent at lowest CO values but becomes progressively darker as the CO value increases towards a peak.

    CO concentration [cosc]

    Setting is here:


    Note: At very high local peak CO levels the contour display can get funky if the emission is asymmetric. Though you're quite unlikely to be confused as to where the forecast peak can be expected to occur.



  • Sailor

    I’m very impressed. Brilliant job !

  • @idefix37

    Thank you, I think I'll leave this one as it is until I've learned more from applying this method to the SO2 and Dust-mass overlays.

  • Dust mass [dust] - Removed

  • This is a final version of the Dust-mass overlay, can’t optimize it further at this point.

    Dust mass [dust]

    Setting here:

    11.png13.pngx.jpg20061106 Al-Asad-Iraq.jpg11.png

  • This is a fully recreated Cloud base overlay. It uses a precise 100 ft increment to 18,000 ft. There’s full color shading from 1 foot (bright scarlet red) to 18,000 feet (bright green). I removed all color shading above 18,000 ft (to de-clutter the display) but all the higher-altitude cloud-bases above 18,000 ft are being displayed in light grey.

    Cloud base [cbase]

    Setting is here:



    Cloud tops will be re-done soon.

Log in to reply

Windyty, S.E. - all rights reserved. Powered by excellent NodeBB
NodeBB & contributors, OSM & contributors, HERE maps