I completely agree - it’s not going to win any Apple Design Awards, but putting a screen icon on my iPhone home page makes a lot of sense for this - it’s the best local forecast you can get.
1 Like
I wanted to comment on that, since in the podcast “data is so expensive” was often mentioned. As far as I understand, NOAA data is freely available. In my country, and most EU countries, weather data is also available for free. It’s already been paid for…by taxes. So they are just paying for someone to aggregate date for them.
And if I want really accurate weather forecasts: I go to the homepage of our national weather service.
3 Likes
I’m not that interested in weather apps so this episode offered a different win for me.
Because I tried two of the recommended apps, I also started their respective trial subscriptions.
This was the trigger for finding a way to cancel subscriptions easier.
So, after some google magic, I came up with a Shortcut to “apps.apple.com/account/subscriptions” which I then put as a monthly recurrent reminder as “shortcuts://run-shortcut?name=NameOfYourShortcut”
So thanks for this episode which made me a better automation newbie 
3 Likes
On its 2011 Kickstarter Dark Sky openly noted where it got its data but its focus was on upcoming hyperlocal forecasts, and superior radar animations (“Our servers pull in this data, process it, clean it, extract velocity data, and convert it to the animations you see”). Over time it has become a more general weather app with extended forecasts, but its hyperlocal reports remain better than most apps (I’ve compared in NYC on a regular basis, for years), which helps explain why so many other weather app developers license its data via an API for which they pay $0.0001 per poll.)
I consider it useful. You can read about the guidelines they use in this article:
Thanks for an excellent and informative post.
I’ve said for years that I don’t really put a lot of trust in any forecast that talks about more than 48-hours away, but I’m generally talking about snowfall amounts. It seems like that can vary greatly.
I’m still curious about two things, which no one except the DarkSky developer(s) may know:
-
when DarkSky first came out, it seemed to be almost spooky-good about alerting me to when precipitation was going to happen when I was. Were they just doing something new with the data?
-
in more recent years, it seems like DarkSky isn’t nearly as good at that anymore. I don’t know if that’s an issue of scale, or what, but it’s a real bummer.
Dark Sky has long used neural nets to learn/train/make its predictions. IBM raced to match them with The Weather Channel and I’ve found its hyperlocal predictions to be different, and often better than Dark Sky in the NYC area.
@tjluoma - Snowfall is one of the hardest things to predict, and a 48 hour out forecast that verifies is really tremendously good. The atmosphere is dynamic vertically rather than just horizontally, and forecasting snowfall type (e.g. rain/snow/sleet) and location is incredibly complicated. Snowfall track is also notoriously hard to predict, since being ‘off’ by 5-10 miles can mean the difference between rain and snow. This article https://www.thoughtco.com/winter-weather-difficult-to-forecast-3444527 explains the difficulties reasonably well.
That’s a good point. There’s often a lot of change in that 48 hour period, but it’s enough that I know whether it’s going to be enough to pay attention to or not.
@bowline The problem with ‘neural nets’ or AI in forecasting is that you’re not really forecasting (identifying causation) when you’re using them, instead you are identifying correlations. While this sometimes can be useful, it can also lead to major ‘busts’ in forecasts. Weather models used by meteorologists are based on complex models based on advanced physics and then run on some of the most powerful super-computers on the planet. They’re imperfect, as a perfectly accurate weather model would need to model the interactions of each molecule of the atmosphere perfectly (which is both scientifically and computationally impossible at the current time).
I’d love to hear what Dark Sky’s developers are doing, but my guess is that they’re interpolating radar data - e.g. if they see precipitation 20 miles to the west of you moving at 40 miles an hour, in your direction, it’s not too hard to assume that there will be precipitation in 30 minutes where you are.
I also expect that there is a bit of bias in how we all remember the past. If, e.g., Dark Sky predicts that it will start raining in 10 minutes, and it does, we all think “gee, that’s cool!” But if it predicts rain in 10 minutes and it doesn’t start raining, we either don’t notice the fact (easier to note that it is raining than it isn’t) or we chalk it up to imperfect weather forecasting.
This seems to be true in my case as well–TWC app seems more accurate more often than Dark Sky.
Thanks. I’m just trying to determine which is the most accurate app (UI is far less important to me) for my area. For now, I’ve defaulted to TWC app.
I expected more from the “& Gear”.
Remember this one: https://512pixels.net/2015/02/on-the-burton-amp-jacket/ ?
What about “touch-screen” gloves?
Bags/backbags that can stand real “weather”?
@matt.gall et al, thanks for making a great episode even more interesting with this discussion!
For basic weather info I use the default app on the iPhone. It gives me enough for most days, and when I want more (usually during snow storms or potential severe weather in the summer… New England gets the best of both sometimes) I go to the NWS.
I’ve generally accessed the NWS forecast via the Facebook postings from my local office (NWS Boston) but I’ve added a link to my home screen for easier access.
1 Like
For my needs, the stock weather app plus TWC app are the perfect combination. The stock app provides most of what I need most of the time and the design is attractive. When I need more information I click on the TWC icon on the bottom left. Based on my non-professional research, the TWC app is one of the more accurate for my area.
1 Like
Happy to help - this discussion combines two of my loves, weather and Apple products
Honestly the stock weather app should serve about 90% of people well 90% of the time. NWS mobile website should solve for most of the rest. It’s just the geeks (weather and tech) who really need or want more.
2 Likes
Don’t forget your local TV station for a weather app. Here in Florida we are very weather focused. In NE Florida my favorite is N4J Weather - WJXT Weather. Ton of options and free! It will also give you other locations. Plus its hurricane tracker app - WJXT Hurricane Tracker (with a Watch app).
After this show, I got all excited when I saw a “renewed” version of the Netatmo Weather Station for $50 off.
Unfortunately, when it arrived, I realized that I couldn’t get HomeKit to work.
Long story short, I eventually figured out that it was version 2 of the Netatmo Weather Station, and HomeKit was added in version 3. The box actually showed a 2013 copyright date, and referred to the iPhone 5 as the latest version, which was from 2012.
Yikes.
It’s on its way back to the seller tomorrow.
Seriously- no mention of WhatTheForecast? Totally fun. Not for serious weather but fun.
I also endorse Radar Scope for storm/tornado avoidance and Dark Sky and Apple’ Weather for day to day use.
Living in tornado alley I’ve been grateful for the granularity Radar Scope provides when under tornado warning. The many radar products it makes available allows for me to evaluate the risk and not have to rely on the more general warnings standard radar products offer. It also permits me to speed up or slow down in order to avoid a nasty storm that will cross the road ahead. Tools like distance measurement and storm track direction and velocity makes this child’s play.
I’ve become the go to person for friends ranging from one in S. Florida who wants to know if he can get his morning walk done before it rains to someone wanting to know if a storm on a track to go over a game several states away is going to stop the game.
I don’t know what my subscription for the 1st of 2 Radar Scope tiers costs but it’s worth every penny. A single purchase covers all iOS devices including AppleTV.