I don't know what the internal situation is with WU, but I have to wonder if they're not really a going concern within TWC anymore. Between their terrible site redesign a few years back that just sort of stopped once it got to about 75% functional, and the way they seem to disable features whenever they start requiring a little more effort to support, it feels like there's maybe one or two people working on the site in their spare time trying to keep the lights on.
It's a shame because their forecast graph is quite possibly the best weather visualization ever - at a glance, I can see what the whole day is likely to look like. I will miss it when they inevitably discontinue it.
Edit: In fairness, I just visited the site for the first time in a while and it looks like they finally fixed their weather map - for at least several years after the redesign was "complete", the default view for the weather map would load so many weather stations it would grind pretty much any browser on any machine to a halt. Looks like that's no longer the case, so maybe something is still going on over there.
Their 10-day weather graph is one of my favorite interfaces. There is no better way to grasp the weather, and it transformed my conception of daily weather patterns. Try it:
I use the similar (but a bit wonkier) weather.gov graphs. I have it bookmarked on my phone, for checking the weather before I ride. After a bit you can figure it out at a glance. It also tells me sunrise and sunset times.
Unfortunately I don't think so. You use the "forward 2 days" button" to get more data, but they don't fit it on one page. You can get about 6 days total
They have a regular 7 day forecast[1] with text, but I prefer the graph for my commute.
I completely agree, as a long term WU user I am continually frustrated by their newer site design. The 10-day view is my favorite one but their site reliability and speed are atrocious. Like others in this thread I've been wondering what's going on over there and if they are just going to shut down some day. I used to even have my weather station send data to WU but I stopped doing that because I don't see any value in doing it anymore because their site sucks. Wundermap is slow as well and I've had issues with their iOS app lately too.
I would love to hear what people are using as alternatives. I already use Dark Sky on iOS as well, so that's one. I just want a good 10 day view :(
wunderground is my favorite, and I have a PWS that contributes to it, and I love the 10 day graph, but damn I was pissed when it was no longer the default view.
It seems spotty as to whether it remembers to use my PWS or not.
I just paid for Carrot (iOS weather app) plus the uberpremium subscription version so I can use WU data in it.
I used to use one of their weather maps. North Georgia gets a lot of random, unpredictable storms from the gulf. I used one of their maps to see if one of the thousand tiny half mile wide storms would pass over me. If it did, I could be in for a heck of a storm and need to unplug stuff. If not, all that effort would be wasted.
They had a map. It would show you all the little cells that rose to the level of concern along with speed, direction, and predicted location by time. I could match those cells to a violent-looking cloud sliding by in the distance. I could never find that map after the redesign. That was the only reason I went there.
I have to assume all their designers are from places where storm clouds come in big, predictable blobs instead of tiny terrors that miss you 90% of the time.
If you haven't try windy.com. Switched a while back (Central/West Georgia with same micro climate issue south of Carrolton). It works well for me ...phone app on the tractor is handy. They just switched their forecast model, a week ago?, so I don't have experience with the new model.
TWC is owned by IBM as well. With TWC absorbing Weather Underground you would think they would continue to support the great work from wunderground. In reality, seems that is not the case.
It's IBM. I'm not sure what's keeping this shell of a company functioning, but they are not a company I'd want to work at, if you could find a non-outsourced job, with regular layoffs and insane demands on their people. None of the people I know who've worked there have spoken well of it.
They still have some random dropouts and errors. Occasionally they'll send malformed json to the mobile app and it'll crash. Often times data from my weather station is rejected for no reason. Sometimes the 10 day graph just doesn't work or other data goes missing.
And of course they have no real support. It's a shame because having all the nearby weather stations aggregating data makes for some pretty accurate weather.
Thanks for sharing this. This is great! It even has an interactive shell. Unfortunately US only and likely soon to be gone... I normally use wttr.in, but telnet is way more old-school cool :)
> We realized we needed to make changes to ensure the highest level of quality, performance and uptime for our API users. As a result, we’ve made the difficult decision to retire the Weather Underground API.
Wait what? They weren't satisfied with the API, so they decided to end it instead of fixing it?
I pay for an app that has the option to use WU's API (Carrot Weather), so I'm upset by this. I've found their current weather and predictions to be more accurate than Dark Sky.
>They weren't satisfied with the API, so they decided to end it instead of fixing it?
Yeah, basically. They want you to use their ad-infested or paid app instead, they don't want you using someone else's (or worse yet, your own) app or program to get their data.
A lot of people were upset (and rightly so) when they EOLd the awesome weather app called Storm. The replacement was (and still is) garbage, riddled with advertisements. Initially there was no ad-free option. The Weather Channel apps are garbage, and full of ads too. Same with the website.
I'm surprised that IBM has done such a terrible job with TWC.
The International Business Machines Corporation wants you to use Weather Company Data feed now. You'll need to call them for pricing, however. Current estimates on that pricing are $400/mo if you read the commenters.
In the end, you are not an International Business, so the International Business Machines Corporation has no more use for you. You should've expected this when they bought out The Weather Channel. Likely Carrot Weather cannot afford it either. International Business Machines Corporation likely doesn't want 3rd party applications to compete with its Weather Company's 1st party apps.
Carrot Weather's uber-premium subscription level is more expensive PRECISELY because of the access to WU data. I hope it includes the new pricing model, since I just paid for it.. yesterday.
That quote is bizarre corporate-speak. It sounds like “We would need to change our product to make it perfect. So instead we’re going to kill it.” It doesn’t make sense.
Surprisingly, I think it means exactly what it says: accessing WU through the API was flaky and unreliable, with lots of downtime. It sucks to maintain as-is, and they can't be bothered fixing it, so they're killing it off.
(Of course, it was a unique data source, and still better than actually using their actual site, so this is a nasty bit of corporate indifference.)
For the users of that data none. For them the one where 95% broken was a pain to maintain and wasn't bringing in much money. "We realized we needed ... for our API users" is meaningless pretty much every decision gets some justification where it's somehow better for the users but it's just a thin enamel over whatever real business justification there is 95% of the time.
For users, none at all. I'm sort of assuming the thing wasn't stable enough to stay up without ongoing effort, given that they seem to have stopped maintaining/supporting it a bit ago and only now killed it off. But for everyone except IBM, this is clearly negative.
So now I'm confused by this. On one hand they say they are retiring it, then they say
"For developers who use WU API data for non-commercial purposes, you will have access to a new plan for a personal use, low call volume API. Stay tuned for more details as we build this out."
Being a user this lack of clarity is annoying (a free one, but my previous company paid for the service.)
I learned how to the api it 5 years ago while working at a start up when they needed weather to normalize data. It was pretty easy to use and the price was decent.
I got a free developer account and get the weather at my location every hour with their free developer api..
The site compares the last 90 days vs last year so as a New Englander I can be sure when complaining about the weather that it was in fact better last year.[1]
Ouch, I loved weather underground. What's a good alternative then?
I loved that I could just use the weather station down the street and always had very accurate weather predictions. When it says it rains, it actually rains right where I am, unlike other weather services.
Also curious to see what CARROT Weather will replace it with.
I've personally been using darksky.net recently for my personal weather checking. It doesn't use personal weather stations, and there isn't a free programmatic API so it definitely isn't a replacement. I've been looking for a while and I haven't found something that fills the niche that was wunderground in its hay day.
This is what the developers I know have been doing as well.
Ambient Weather has been moving into the PWS data space. They have released their own smartphone apps and they have an API:
https://www.ambientweather.com/api.html
Ouch indeed. I have a RainMachine irrigation controller that hooks into WU forecasts to be smart about when to water. It can use other data sources (e.g., NOAA), but in our microclimate area, WU’s access to highly-localized personal weather station data has been pretty useful.
There’s some active discussion of alternatives in the RainMachine forums [1].
Carrot weather already has Dark Sky as the default. You need to pay for a subscription if you want to use it on the iPhone, and a higher-tier subscription for the Watch. I'm assuming they're just remove WU and have DS as the only option.
“On display? I eventually had to go down to the cellar to find them.”
“That’s the display department.”
“With a flashlight.”
“Ah, well, the lights had probably gone.”
“So had the stairs.”
“But look, you found the notice, didn’t you?”
“Yes,” said Arthur, “yes I did. It was on display in the bottom of a locked filing cabinet stuck in a disused lavatory with a sign on the door saying ‘Beware of the Leopard.”
We see less exposure of data because it's more profitable to push people to use proprietary apps that either cost money or are infested with annoying ads. This is why screen-scraping exists and is so popular, despite all the calls from nay-sayers who scream "don't screen-scrape! use the API instead!"
The Wunderground API was home to my first ever 'real' personal project, a little crawler to get some historical data I was interested in. To this day, it's still the best (multinational) weather source I've found.
I'm really sad to see it go, and I expect you're right that we'll see fewer things like this in the future.
What a shame. Carrot Weather is by far my favorite weather app ever, but here in Europe, its forecast is useless unless you pay extra to get the weather underground data. Then it’s in-line with what all other local weather forecasts produce.
I assume with the API going away so will carrot weather's access to weather underground.
I have a client who had me start a project 5 years ago to create a new product line; we integrated some weather API into it years ago and I couldn't remember who we used, but I just checked and it turns out that it is Wunderground.com--- but the lucky part is that the product is still unreleased due to scope creep, bad specifications, and horrendous delays. So I guess this is just one more delay to re-implement the weather functionality. You know there's trouble when your APIs start dropping off like this... We probably won't be finished for a long time yet anyways! I wonder what the next API to close will be...
I remember helping the WU team as a PWS operator since about 2007, and I found the existing XML feeds lacking back then. I gave them some new data models to implement for the XML feeds, and they were thoroughly open and awesome for the help from a PWS owner. Ended up a Lifetime member.
Definitely haven't been the same in the last 3-4 years, though. In the last 6 months, even the platform has become a bit touch and go.
Anybody knows a good, worldwide weather "niceness" rating API with a free tier? The only scoring system I found was the Hugo Poppe method[0], but seems like I'd have to implement it myself.
What does this mean really? I was just going to buy a PWS for my grandpa for christmas. Actually, I was going to buy it for myself but figured I was only going to view it via the 'wundermap' so might as well give the station and console to him. Are they killing off access from PWS's or only how you can retrieve the data?
I too am curious about this. A low priority project on my list at work was to set up a weather station and add a weather feed on our public website, and i'm now not sure where things are going.
another problem: there are now going to be SO MANY google results pointing to some API/documentation/forum posts that aren't going to work anymore.
I wouldn't count on NWS surviving the Trump administration, for that matter. There has been a strong push to privatize it going back to the Bush #43 era.
They do offer a free API but it is nowhere near as full-featured as WU. They have many tens of thousands fewer weather stations that report current conditions, for example. The variables returned are fewer when it comes to conditions and forecasts. Additionally, depending on your pay tier, WU had all kinds of historical info + specialized weather forecasts that NWS/NOAA does not offer. Even metrics like UV are rarely in NWS API products but could be expected from WU.
WU was also global, while NWS is US-only.
I love the NWS API btw. It's just nowhere near what the WU API was in terms of the promised data returned from API calls. NWS API has always been far superior when it comes to uptime!
They do offer a free API but it is nowhere near as full-featured as WU.
While I don't disagree with you that the NWS has thousands fewer weather stations than WU, it should be noted that the NWS has very strict standards for weather sampling and reporting.
NWS weather stations are set up in a particular way to ensure consistent, accurate data. WU, less so.
For example, last week I was in a part of the country where the only weather station for 80 miles was a PWS from WU. According to its data, there is a 40-degree (F) spike in temperature every morning around 7am. Clearly, this is a thermometer that is being affected by the sunrise. This doesn't happen to NWS equipment.
What does it mean? They will retire the api but will make another one? Are they planning to point PWS to other competitors? In what other ways was the wunderground api providing services?
Side note- their site has been suffering performance issues for a long time.
IBM did not handle this well. This official announcement is months late. The API was abandoned months or years ago but they were still taking people's money for it. When they finally cut off the payment form and removed their API plans without notice in the summer there was an uproar. It hurt me deeply, making 1-2 months of dedicated work directly against this API to be useless.
Very bad decision-making and handling at IBM and I will not be supporting any of their businesses in the future. It is not reliable and they shut down without notice.
Edit: A side anecdote. The API was so bad (in terms of service uptime) that I learned a lot about how to write retry logic for API calls on mobile. So many calls would fail for various reasons. But randomly. So for each API call you want to make, you should have 5-10 error cases to catch and then retry. Eventually you'd get a successful call!
(I switched to the free US NWS / NOAA API for All Clear Weather. It is better in performance but I miss a lot of the data available from WU, and the US-only thing is a downer too. All Clear: https://play.google.com/store/apps/details?id=com.allclearwe...)
I didn't either, and I was wondering why the system was rapidly degrading. It used to be my top go-to site/app, but it became so unreliable, crashing so frequently on Android I had to remove it from all devices, and rarely use it anymore on web/desktop.
I never 'looked under the hood', but it seemed to be straining & breaking the load of mountains of new cruft code. Somehow not surprised to learn that this happened on IBM's watch these days when they seem to have lost the plot in so many other ways.
For Android, FlowX [1] is a great app that is my first go-to now (no relation other than as a happy user). This Earth map [2] is a great map for web.
Anyone have other recommendations for web/desktop?
Was at a workbench type event at IBM's Germany headquarter about their Cloud infrastructure.
While discussing microservices, they mentioned that they own WU. It was actively suggested we could use data from WU for our own services.
This was last month(ish).
--
The end of WU was not handled well by IBM.
Similar story with the tech consultancy I work for. IBM tried to pitch the WU API to us during the summer for a large-ish project for a major client of ours. Thank god we ended up going with Dark Sky instead or that could've been quite a few man hours down the drain, not to mention us looking bad to the client.
I know someone mentioned their app as a competitor in your other thread, but I’m not sure if anyone pointed it that they also have an API: https://darksky.net/dev
They combine multiple sources of data. Even if you don’t want to use their API, I’ve found their list of sources handy for European data (though finding what you need in Europe is still confusing as heck).
Unfortunately I don't have specific code examples. I was operating in stealth mode at the time and the code was closed source. I could dig some examples probably later today?
From memory, data might return normal json but with "" empty string instead of metric forecast but imperial forecast is there. Or the other way around. Or both empty. Or 0 values for variables that shouldn't be 0. Or error state json with not much info. Things like that.
There was no way of knowing it was abandoned. I am taking a guess at the length of time ago it happened; my guess is informed by lack of quality of service and public-facing updates. It's reasonable to assume that the API degraded in quality over time as fewer and fewer / maybe no people were working on it.
I think it happened around May/June/July ish of this year. I remember creating a free acct using the free tier API for a coding project I did for an interview around April. Shortly after they killed the free access and I guess now it is totally dead :(
It's a shame because their forecast graph is quite possibly the best weather visualization ever - at a glance, I can see what the whole day is likely to look like. I will miss it when they inevitably discontinue it.
Edit: In fairness, I just visited the site for the first time in a while and it looks like they finally fixed their weather map - for at least several years after the redesign was "complete", the default view for the weather map would load so many weather stations it would grind pretty much any browser on any machine to a halt. Looks like that's no longer the case, so maybe something is still going on over there.