Hacker News new | comments | show | ask | jobs | submit login
Windy.com (windy.com)
580 points by davesque on Aug 24, 2017 | hide | past | web | favorite | 103 comments



We are happy that you like windy.com. If you want to help us with this project, then report all issues to community.windy.com We love bug reports from programmers, with all screenshots etc (Ivo)


Windy was coded by billionaire founder and owner of Seznam, which is czech search engine (and media company), one of only three other search engines in the world that still beats Google in local market.


Has he actually coded it himself? Do you have more information?

Seznam is a remarkable company. Their maps (http://mapy.cz) are second to none, my favourite feature are touristic maps which work at least within Europe.


Yeah, it was a side project he made for himself as he is a pilot and windsurfer. [1]

Though for windsurfing I find windguru.cz is more useful (also Czech(!)) and for flying its hard to beat ForeFlight.

It should also be clarified that Seznam is one of the few companies to beat Google in searches with the Latin alphabet. Local search engines are still beating Google in Chinese, Korean, Japanese and Cyrillic alphabets.

[1] https://www.forbes.com/sites/forbesinternational/2017/02/06/...


Wow. Mapy.cz is amazing. Just checked my German town. Not only more information than Google, it's way faster. Everything feels instant.


In addition they did a pretty good scan (somehow) of the center of Brno to produce a remarkably nice 3D model of the city you could scroll around in, Google seems to also have this now. Here's roughly the same area from both services:

https://mapy.cz/zakladni?x=16.6075148&y=49.1942864&z=19&m3d=...

https://www.google.cz/maps/@49.1926496,16.6076641,133a,35y,2...


It seems to be based on OpenStreetMap (I recognise some changes I made), which is indeed a lot more detailed than Google Maps in some areas.


It is, see the copyright in the bottom-left corner.


Yeah, OMS is consistently better than Google in Germany (or at least where I looked), but this site still feels better.


This site was originally based on https://github.com/cambecc/earth, the open source version of https://earth.nullschool.net


As far what I know (I read an article about him and this project in Forbes Czech), he works on this project personally with only a few other developers. Originally he made it for himself because as a hobby pilot he needs an accurate weather forecast.


Why only 3? I can list: Baidu, Naver, Yandex, Yahoo Japan, Seznam


Well, it seems there more, I dont remember by which measure it was three. I had Seznam, Baidu and Yandex in mind.


In case others are wondering: Speed is shown in kt which means "knot," equal to one nautical mile per hour, approximately 1.151 mph.


The other two being Yahoo Japan and baidu?


Also Yandex (Russia)


That is pretty cool! Although hijacking the back button is a bit annoying.


They hijack a lot of standard buttons. I couldn't command-L to focus into the address bar. I have a special disdain for this trend.


We check it, this seems like bug


Huh, Cmd-L works for me (Safari, latest version of Sierra)


It's called history API, and it's not annoying, they should use more .replaceState though


Looks like that isn't purposeful, it adds coordinates as a parameter after you load the page and hitting back takes you "back" to the plain homepage.


They should be using replaceState instead of pushState there. An all-too-common mistake.


It works really well on mobile! Including the hijacking the back swipe feature.


It's the JavaScript framework that does this. Like the commenter said, you have to replace path for every view instead of push


He we would like to fix this bug. Can you get in touch with us on community.windy.com > Feedback


Interesting note, I play a lot of golf competitively, and they've basically recently allowed players in tournaments to use phones (obviously players don't do that much or if at all, concentration and all that).

But the one specific rule is that players can't use their phones to check the weather, and even more specifically the wind direction. Wind makes a huge difference on the course, and being able to know the exact direction of the wind where the ball is flying would be really helpful. Other part is being able to know if the wind shifts during the round. Before you start you can check the wind direction, but if that changes, you could be out of luck. This seems like a perfect golf aide, so much to the point where it's a penalty in a tournament.


Even the best models commonly available will only operate at a 3km resolution, which realistically isn't that much more useful than a 10km or 20km grid for ageostrophic winds (wind not affected by the ground). A model might show you have 8kt winds out of the SSE, but things like water, treelines, hills, etc. could have a large impact on a golf ball.

It'd be easy enough to look at the forecast for a 3km model like the HRRR, but the challenging part would be guessing at how the various features on a course interact with that forecast.


Right, I talked about this below, but really the main thing is just knowing the which of the 16 labeled directions the wind is coming from, and then checking back every few holes if possible, and then make arrows on the course map with that overall direction.

And yes, trees and hills totally have an impact, but the main impact is the difficulty of having a very in tune knowledge of where the wind above the trees and hills is blowing. If you're hitting a normal full shot, the ball flight will be (usually) only affected by the standard wind direction.

Again of course, there are gusts that make that difficult to time a shot, and ball flight and shot type make a difference as well, but knowing and trusting the overall direction would be really helpful.


I race sailboats competitively, and obviously wind direction and speed affects the outcome even more.

Electronically, in most classes of boats you're allowed to get any information that's available to anyone (i.e. you can hit any public website, but you can't call your local weatherman to ask his opinion). But really, there's not that much information available for day races (long distance races are a different animal)

One of the things that's become common over the last 10 years or so it to use a forecasting service. Basically you talk with a meterologist each morning or perhaps a few days in advance and they let you know what they expect to happen. You end up with statements like, "If you see big puffy clouds coming in, you should expect the wind to shift to the right of about 30 degrees shortly. But if the clouds don't come in, or you see flat ones, you won't get the shift until the puffy clouds fill in."

They're pretty good at forecasting what will happen over a large area or a day, but for really small areas (a few square miles) or really short periods, they can't be that accurate because the movement of the systems varies in speed and direction too much over short periods.

But try a forecasting service, it really can be helpful to know what's happening around you. Around $100 a day is reasonable last time I did this.


It does not seem precise enough to give you localized condition on that scale. I tried zooming in to see how it was reporting wind and waves I can see from my current location, and there was simply no data once you got down to a range of a few miles.


A range of few miles doesn't make that big a difference on the golf scale. Really, the only deep down information I'd want to have is like down to the 16 directions on a compass. These courses have maps of the holes and you're able to draw arrows on the map of the general wind.

The key part of this site seems like I'd be able to check back every few holes to see if it's shifted enough to change the arrow. When checking for a local course, I can see enough to where I'd trust it, but then again, having more correct wind information would be better.


> exact direction of the wind where the ball is flying would be really helpful

When I did tournaments we would look at the flagstick, trees and also release blades of grass.

I could see near exact speeds helping (could make the difference between putting for birdie vs chipping from off the green) but not any more of an advantage than a rangefinder. Do they let you use a rangefinder in your tournaments or no?


What I can't get over is the speed.

Ever since tile maps have become the norm, most of the weather radar services are unbearably slow on my DS(Hel)L connection. This loads fast.

I wonder what they are doing right.


Yes, the hurricane looks pretty:

https://www.windy.com/?29.770,-95.363,5

I noticed Windy.com took me straight to this spot due to my location near the approaching storm, so in case you're in a part of the world with a little less going on right now, check out the winds coming to Texas.


That storm is looking serious. Not only is it a powerful storm but it looks like it is going to hang over Corpus Christi for maybe 2 days and then slowly head up the Gulf coast for 3 days after that. They are talking 3 feet of rain and that is a fairly flat area so I can only hope people can get away from there.


Or, conversely, what are the tile maps doing wrong?

I have a 100 Mbps connection and tile weather radar maps are some of the poorest experiences on the web with that speed, too. Why can high-resolution, high-framerate video load fast, but not a map?


I think part of it is that Google Maps and Bing Maps do not have a sense of urgency. So for one thing, the base maps are poorly performing.


This looks incredibly similar to https://www.ventusky.com/ doesn't it?



A bit of post hint.fm streamline history:

It can be sourced back to Cameron Beccario's project getting some exposure in late 2013 after launch as open source. https://github.com/cambecc/earth

It's cited as direct motivation by windy: https://community.windy.com/topic/4/about-windy

---

For the record: windy, launched as windity in late 2014, this was 1 1/2 years before ventusky:

"ventusky is owned by InMeteo, built to compete with windyty, which was originally a copy of my site's open source repo"

"if you're curious: http://hint.fm/wind launched Mar? 2012, e.n.n Dec 2013, windyty Nov 2014, and ventusky Jul 2016"

https://twitter.com/cambecc/status/784448346471530496

---

I also recommend his talks at "The Graphical Web" 2014: https://www.youtube.com/watch?v=QXNODLWhSbw

and openvis 2015: https://www.youtube.com/watch?v=OLSmNZm1e0k


Except this one immediately runs

Does that thing where you hit back and it cancels the loading scripts and site actually runs


eerily so



This site just went straight to my bookmark list, brilliant.

Small criticism: Every time you move it creates new page entry in the back button list, so once you have moved around a bit you can't use browser back button to easily go back to previous website.


That was my same thought. This is one of the few rare times I don't think using the browser history makes sense for an SPA (at least, in its current implementation).


Great project! I recently wrote a detailed technical post on how to implement a similar visualization with WebGL — check it out: https://blog.mapbox.com/how-i-built-a-wind-map-with-webgl-b6...


We use a lot of @mourners node libs on backend


Very similar to: https://www.ventusky.com


I was just going to say that. Because I found that on HN. I had also found this here https://earth.nullschool.net/#current/wind/surface/level/ (before ventusky)


I wrote in their forums suggesting they add air quality/pollution info and greenhouse gas emissions to their maps and it was done in about three days. I was impressed.

Btw I think they use Riot.js on their front end?


They do use Riot, there was a blog post somewhere about it a bit ago.



Yeah, I'm a little up and to the left of that (a couple hundred miles inland.)

While the big spinning mass offshore is the headline, the subtitle is that everything east of I-35 is going to get somewhere between a foot and two feet of rain. Not snow, water.


Do you know what they grow around Corpus Christi[0]? Looks like a lot of rich, dark soil that is about to be washed out to sea...

[0]https://www.google.com/maps/@27.6570039,-97.6556344,44285a,3...


It reminded me of this awesome project: https://earth.nullschool.net/


well that would be the first major hurricane to strike the US in over ten years. hopefully people aren't too complacent and still take them seriously.

I would hope FEMA does better this time but recent track record of that agency still is not good.


Agree. I saw many projects like "Windy" based this.


They also provide free API (http://api.windytv.com/)

Which is very cool to track ocean sailing.

I have made several trackers to follow around the world sailing races/adventures.

https://gis.ee/lb/

https://tracker.ee/


Are there ways to use them outside JS? E.g. to get georeferenced video?


Looks like they are using http://leafletjs.com/


You know what strikes me? Look at the overland place where the winds move quickly. Those are our cities. We're living off whiffs in the aerodynamic backwaters on a world of windy metropoli.


That's funny, as some friends of mine made the exact same thing years ago, and it even had a similar name (Weendy):

https://techcrunch.com/2013/02/14/weendy-an-extreme-sports-a...

They have since pivoted to something similar, as AFAIK they didn't get enough traction.


Is there a name for the hotspot wind system south of Africa? That looks intense.


"Roaring Forties" and "Furious Fifties".

https://en.wikipedia.org/wiki/Roaring_Forties


As someone who works with mapping data and web based maps regularly, this website is excellent in terms of usability. The ease of switching overlays, adding symbols, saving selection, adjusting the map are all excellent and intuitive. The ability to drill down on symbols added in a smooth and sensible way is excellent. This is how you make web maps for specialist data!


This has been my go-to for sailing conditions for a while. I used other sites before it (Predict Wind, Surf Watch), but Windy is fast and responsive and usable on a phone. The data for sites like these all ultimately come from the same sources (the big weather models) so the differentiators in this space are mostly in the user experience.


Holy hell please give me my back button back.


Yeah browser push state for days


Other than the whizbang interface there's nothing really innovative going on here as far as actual science... Same with all the other me-too sites that use that same streamline animation code. Some of the visualization is downright misleading, but whatever. The ventusky wave animation is awful and physically incorrect.


Great website, has been around for some years as windity and windytv. I guess windy will be its final name. I usually find windguru.cz more easy to read, but windy offers a cool visualization that I think gives more context. It's really cool to check it during hurricanes.


Not sure where the data is coming from but my area is showing arrows to the north west. We nearly never get wind from the south east and looking outside this map isn't accurate.


I have no source for this, but it might be possible that the wind it's reporting is from a higher altitude. 4k to 5k meters above the ground the wind could be moving in the opposite direction from the ground.

Source: http://www.eniscuola.net/en/argomento/air-in-motion/winds/hi...


There's an altitude slider bottom right.


It reminds me of hint.fm/wind


Looking at this made me realize how insanely huge storms are in the southern ocean. Hurricanes and typhoons up north have nothing on that.


That's the so-called "Roaring Forties":

"The Roaring Forties are strong westerly winds found in the Southern Hemisphere, generally between the latitudes of 40 and 50 degrees."


Which also raises the question about how the map projection / scaling affecs the percieved size and data layout.


Interesting. Does anybody know why it happens?


There is a continuous ring of ocean the whole way around Antarctica. That water can build up a lot of momentum as it is pulled around the Earth by tidal forces. At least, that's what the captain told me, I'm sure there are meteorologists around here with a much deeper understanding.


Just now we have launched weather radar layer: https://www.windy.com/?radar,27.863,-96.237,8


It's going to be hot in SF today...

https://imgur.com/gallery/NiMJk


240° is wind direction


The UI is very small on a big screen. The entire right hand legend and menu is taking up around 10% of my screen width, so the button are tiny!


FWIW, I found hang gliders really enjoying this site (alongside some other obscure wind estimation sites) for planning whether to go flying.


Thanks to this website I discovered that procrastination can reach new levels... there's no limit to it.


Is it just me or is this website really slow?

(Especially after pressing the "play" button in the lower left).


It was fast for me on my iPhone


This is pretty, but I don't think it is going to be 35 C in the North Atlantic any time soon.


Why do you only upload the latest three atmospheric models only one time (0000 UTC) per day?


and if you want to see the windiest city in the world, it's in the South Pacific:

https://www.windy.com/-41.332/174.793?-41.801,174.793,8


Trying to back out of this website after zooming in is really frustrating.


My friend that works in the aerospace industry uses Windy all the time.


Oh shit what's going on by Texas


Hurricane Harvey


Crazy, people's lives being ripped out by strong winds and I just see some swirls on a screen.


What model is this pulling from?


In the menu it has a link explaining it: https://community.windy.com/topic/12/what-source-of-weather-...


Not sure, but there's a "forecast model" radio button in the bottom-right of the interface which you can play around with.


This site/app is my favorite because you can compare three models side by side: gfs, ecmwf and nems.


Not sure why they even bother pulling gfs, its miles behind the euro.


Thanks for correctly saying "these data" in the menu


I wanted to build something similar for surfing forecasts a couple years ago, but couldn't find any decent data source... Where is the data coming from? Is this publicly available?




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: