Windy Community
    • Unread
    • Categories
    • Groups
    • Go to windy.com
    • Register
    • Login
    1. Home
    2. dub
    D
    • Profile
    • Following 0
    • Followers 0
    • Topics 1
    • Posts 8
    • Best 5
    • Controversial 0
    • Groups 0

    dub

    @dub

    6
    Reputation
    205
    Profile views
    8
    Posts
    0
    Followers
    0
    Following
    Joined Last Online

    dub Unfollow Follow

    Best posts made by dub

    • RE: Missing: Large hail forecast

      @Filip_K The warnings alone don't really give a good picture of what areas are at imminent risk. (For instance, the map warnings are by county, but here in Texas, counties are often large enough that the NWS warns only portions of a county, e.g., Southwestern X County.)

      I'd suggest a new map for localized projected near future (no more than an hour, probably) severe weather: Hail/Tornadoes/Severe T-storms.

      I'm no meteorologist, but it seems like nearly all the data required to drive a pretty good hail/tornado/high winds risk map is already in the system (you could probably do a decent first cut at this by just summing data from some of these layers and visualizing that, then playing with weights to tweak): Storm severity and coverage from Radar and Satellite, CAPE, Cloud Tops, Lightning, Icing, Winds at various altitudes, and Freezing Altitude. (The altitude-based ones may require some computation and flattening first.)

      I realize this is a non-trivial integration, and I'd be fine if it's a Premium-only feature. (And it might help drive some subscription revenue for y'all as well.) I've been glued to Windy and a couple of other products continually recently to decide whether or not we have to move cars under cover - Ranchers have the same problem with their livestock. You want to know the risk so you only go to the considerable trouble to get everything protected if you're pretty likely to get whacked.

      (Other, related thoughts: Actual storm tracks (and maybe fronts?) could be an input to this, too (if you add those), especially if they're not just the simplistic linear ones from most Doppler radar products, which often show confusing overlaps of paths in turbulent storm conditions. Does anything in Windy's analysis code look for hook signatures from tornadoes in radar? (Maybe not, since it's not obvious you have a Doppler radar source...) Precip Rates could be useful for this, too (as well as Flash Flood risk visualization), but I'm not sure you have that as readily available in your data as the metrics above...)

      Thanks to the Windy Team for a truly amazing product!

      posted in Your Feedback and Suggestions
      D
      dub
    • RE: Windy offers Sounding Forecast

      This is awesome, but what would be even better, and really, really valuable as a next feature would be hail and near-term hail forecasts and warnings. It seems like there may be some model development needed (especially going forward), but just integrating the already-existing lightning and radar data to display hail presence and near-term probability would be a huge win.

      posted in Announcements
      D
      dub
    • Hey! Where's my Goat??!!! I want my Goat!!!

      I rarely use the speed control on the map playback, but it's useful to see how things like the arctic blast is going to flatten us here in Texas over the course of a few days. (Similarly useful, I imagine for those under a greater threat of hurricanes...)

      Imagine my surprise when the Goat has vanished - and taken the Turtle, and the Rabbit with him! Seriously, these are useful controls - please bring them back! And sack whoever is responsible for removing it - or at least ensure that they are bitten by a moose! (Oblig. Monty Python ref...)

      posted in Bug Reports
      D
      dub
    • RE: Suggestion for radar option

      FWIW, I find the current behavior of this just fine the way it is, and it's very nice to have the option of visually extending the actual radar into the projected forecast, especially for fast-moving fronts (common in much of the US, especially here in Texas!) The current UI design for this seems well though-out to me.

      posted in Your Feedback and Suggestions
      D
      dub
    • RE: Hey! Where's my Goat??!!! I want my Goat!!!

      @idefix37 Ahh, OK, I guess I just thought it worked everywhere. That makes sense. To be fair, I don't try to use the timeline for anything other than radar all that often. Glad the goat's not gone...

      posted in Bug Reports
      D
      dub

    Latest posts made by dub

    • RE: Missing: Large hail forecast

      @Filip_K The warnings alone don't really give a good picture of what areas are at imminent risk. (For instance, the map warnings are by county, but here in Texas, counties are often large enough that the NWS warns only portions of a county, e.g., Southwestern X County.)

      I'd suggest a new map for localized projected near future (no more than an hour, probably) severe weather: Hail/Tornadoes/Severe T-storms.

      I'm no meteorologist, but it seems like nearly all the data required to drive a pretty good hail/tornado/high winds risk map is already in the system (you could probably do a decent first cut at this by just summing data from some of these layers and visualizing that, then playing with weights to tweak): Storm severity and coverage from Radar and Satellite, CAPE, Cloud Tops, Lightning, Icing, Winds at various altitudes, and Freezing Altitude. (The altitude-based ones may require some computation and flattening first.)

      I realize this is a non-trivial integration, and I'd be fine if it's a Premium-only feature. (And it might help drive some subscription revenue for y'all as well.) I've been glued to Windy and a couple of other products continually recently to decide whether or not we have to move cars under cover - Ranchers have the same problem with their livestock. You want to know the risk so you only go to the considerable trouble to get everything protected if you're pretty likely to get whacked.

      (Other, related thoughts: Actual storm tracks (and maybe fronts?) could be an input to this, too (if you add those), especially if they're not just the simplistic linear ones from most Doppler radar products, which often show confusing overlaps of paths in turbulent storm conditions. Does anything in Windy's analysis code look for hook signatures from tornadoes in radar? (Maybe not, since it's not obvious you have a Doppler radar source...) Precip Rates could be useful for this, too (as well as Flash Flood risk visualization), but I'm not sure you have that as readily available in your data as the metrics above...)

      Thanks to the Windy Team for a truly amazing product!

      posted in Your Feedback and Suggestions
      D
      dub
    • RE: Suggestion for radar option

      FWIW, I find the current behavior of this just fine the way it is, and it's very nice to have the option of visually extending the actual radar into the projected forecast, especially for fast-moving fronts (common in much of the US, especially here in Texas!) The current UI design for this seems well though-out to me.

      posted in Your Feedback and Suggestions
      D
      dub
    • RE: Hey! Where's my Goat??!!! I want my Goat!!!

      @idefix37 Ahh, OK, I guess I just thought it worked everywhere. That makes sense. To be fair, I don't try to use the timeline for anything other than radar all that often. Glad the goat's not gone...

      posted in Bug Reports
      D
      dub
    • Hey! Where's my Goat??!!! I want my Goat!!!

      I rarely use the speed control on the map playback, but it's useful to see how things like the arctic blast is going to flatten us here in Texas over the course of a few days. (Similarly useful, I imagine for those under a greater threat of hurricanes...)

      Imagine my surprise when the Goat has vanished - and taken the Turtle, and the Rabbit with him! Seriously, these are useful controls - please bring them back! And sack whoever is responsible for removing it - or at least ensure that they are bitten by a moose! (Oblig. Monty Python ref...)

      posted in Bug Reports
      D
      dub
    • RE: Windy offers Sounding Forecast

      This is awesome, but what would be even better, and really, really valuable as a next feature would be hail and near-term hail forecasts and warnings. It seems like there may be some model development needed (especially going forward), but just integrating the already-existing lightning and radar data to display hail presence and near-term probability would be a huge win.

      posted in Announcements
      D
      dub
    • RE: Windytv 10 Public Preview

      The right-hand menu is awkward - right now, it collapses, but leaves a tag hanging out over the map that takes considerably more space than just leaving the icon up (especially for things with longer names, like "Temperature". I'm not convinced that having it vanish automatically is good. After using it a bit, I think I preferred the old behavior, where the arrows slid it out of the way until you opened it up again. I love the performance improvements in the new version, but I think in general, the UI changes are a step backwards.

      posted in Your Feedback and Suggestions
      D
      dub
    • RE: Windytv 10 Public Preview

      @ivo Depends on your browser - there's some slight benefit to putting this in the app...

      posted in Your Feedback and Suggestions
      D
      dub
    • RE: Windytv 10 Public Preview

      Just started trying the Beta. I notice you're open to suggestions about color schemes for the various metrics. Here's some of what I've learned about that over the years: There is a real art to doing this well - I've built complex graphics for visual data analytics several times, using different methods. my favorite is a seven-color scale (where you can get away with that little resolution): Low Critical - Magenta, Low Warning - Blue, Nominal - Green, High Warning - Yellow, High Critical - Red, Old /Possibly unreliable data - Gray about 20% off BG color. You can get away with squeezing a few more colors in there, but not too many. One of the issues with this is that although it's relatively "stoplight intuitive", it's tough on colorblind users (nearly 10% of population), esp. since red-green is the most common type, so confusion there can be really dangerous. If you want a tool to help you play with and explore some more complicated schemes to convey different kinds of info and even check for colorblind-safeness, check out http://colorbrewer2.org. (Note that you may want the "diverging" option for many of your color schemes...) I wish there was a way to specify the baseline schemes to modify, but most of what's there is decent as a starting place, if not always the most attractive aesthetically. Hope this helps!

      posted in Your Feedback and Suggestions
      D
      dub