Winter Storm Jonas brought our area of the country to a standstill for a few days. Leading up to the storm, I was watching closely as the two weather apps I use predicted what would happen and how much snow we would get. As the storm grew closer, the apps diverged more and more on the amount of snow, and I though it would be interesting to record their predictions and how they compared to reality.
There are two weather apps that I typically use to get an idea about the hours and days ahead. Each app performs distinct jobs. I use Dark Sky for the “hyper local” forecast and hourly predictions, and Accuweather for long-term forecasts and a general overview of the coming days.
Dark Sky has based its reputation on the “hyper local” forecast, that is, not only when it generally going to rain in your area, but being able to pinpoint within about 10 minutes when rain will start and stop. It has a distinct and attractive UI and gets regular updates. It’s a universal iOS app (meaning it’s one app for both phone and pad) has an Apple Watch component (which I have never used, since I don’t have an Apple Watch). On both my iPhone and iPad, Dark Sky gets a prominent space on the first home screen.
Because it specializes in long-term forecasts and has outlooks for things like outdoor activities (which take into account things other than just temperature and precipitation, like wind and humidity), the Accuweather app also has a place on my iPad for weather forecasting. Where Dark Sky only shows the next week, Accuweather goes for fifteen days. It is a free app; however, there’s a glaring banner ad on the top, and the look of the app is severely outdated. It doesn’t receive updates very often. There is an Accuweather Platinum app that is for-pay, but I have not tested it.
As we got closer to the storm, it was shaping up like this:
- Snow would begin sometime Friday (22 January 2016) afternoon, and start to fall at a rate of about 1 inch per hour
- Snow would continue throughout the night and all day Saturday (23 January 2016) at a similar rate
- Snow would continue into the night on Saturday and end sometime in the early morning Sunday (24 January 2016)
For this comparison, I recorded what each app predicted for the storm on 19:15 on Thursday (21 January 2016) night, as well as what the National Weather Service predicted:
Thursday night, 7:15pm Dark sky: Friday: 6-9" Saturday: 14-23" AccuWeather: Friday day: 0.1" Friday night: 5" Saturday day: 5.9" Saturday night: 3.1" National Weather Service 18-30" total
So, Dark Sky was confident that we were going to get a lot of snow. At the high end, it was predicting even more than the NWS was. Accuweather, on the other hand, was way on the low end, less than 15″ total. Accuweather also broke it down into AM and PM totals. Dark Sky separated it into days, whereas the official NWS forecast just gave a total for the whole event.
Instead of starting sometime Friday afternoon, snow began to fall at my house around 11:30 AM. What began as a light snow soon intensified into the inch per hour that was forecast. At nightfall on Friday, I made my first measurement.
Nightfall total here in New Creek: 6.5″. We got an inch an hour since it started about 11:30 this morning.
Already, it looks like the Accuweather app severely underestimated the severity of what was coming.
I took another measurement at daybreak on Saturday morning.
Daybreak total in New Creek: 19.5″. The rate held steady at 1″/1hr throughout the night.
We’re already over the Accuweather predicted total, and we still had the whole of Saturday to go.
One more at nightfall on Saturday:
Nightfall total at New Creek: 22.5″.
With only three more inches throughout the day, the rate slowed considerably. It was still snowing when I made that reading. All told, we had about 25″ of snow from that storm. By daybreak on Sunday, drifting snow made it pretty tough to get a measurement that I trusted. But it appeared that it finally stopped snowing around 11pm Saturday.
So, how’d we do?
National Weather Service
|Total||18-30″ (24″ midpoint)||25″|
The National Weather Service did not give out daily totals, but hit the mark for the total storm accumulation pretty closely.
|Total||20-31″ (25.5″ midpoint)||25″|
Dark Sky did poorly on the daily totals, but like the NWS was accurate on the storm total.
Accuweather was a huge miss. Not only did they miss the daily accumulations by a wide margin, the total was off by 10 inches, which is a very significant amount.
This very unscientific, one-data-point test confirms (to the extent it can) that Dark Sky belongs in a prominent place on my iOS devices. The test does not, though, show enough to make me delete the Accuweather app, because I like some of the capabilities that it has that Dark Sky does not. I’d like to do some more testing to see if the app is always this wrong, or if this was an anomaly. If it consistently gets predictions wrong, there’s no sense in keeping it around. For example, I love the 15-day forecast. I’ve never stopped to look at how accurate it is. If it’s filling me with hope (or dread) and is consistently inaccurate, there’s no need to open it up to see what the next two weeks look like. On the other hand, if it’s just bad at snow, or all precipitation, I can live with that because of the other features it provides.
So, as always, the result is that more testing needs to be done. Neither app was able to predict the daily totals with any accuracy, though Dark Sky had the event total right, whereas Accuweather was not right about any of the snow accumulations.
And the National Weather Service was vague, but hit the total very well.
Monday January 25th, 2016