User not logged in - login - register
Home Calendar Books School Tool Photo Gallery Message Boards Users Statistics Advertise Site Info
go to bottom | |
 Message Boards » » Fuck you apple. Page 1 2 3 4 5 [6], Prev  
HaLo
All American
13376 Posts
user info
edit post

^Good. Android users don’t deserve nice things

7/6/2020 6:32:27 PM

synapse
play so hard
60188 Posts
user info
edit post

Fuck apparently not now I have to dredge around for a new weather app

7/6/2020 8:52:59 PM

TreeTwista10
Laugh, Think, Cry
143235 Posts
user info
edit post

there are only hundreds to choose from

such as MyRadar

[Edited on July 6, 2020 at 9:00 PM. Reason : .]

7/6/2020 8:55:15 PM

synapse
play so hard
60188 Posts
user info
edit post

There are hundreds but I've never seen anything as good as Dark Sky

(Hence the characterization of dredging)

7/6/2020 10:33:56 PM

wdprice3
BinaryBuffonary
45903 Posts
user info
edit post

^3. this.

7/8/2020 1:30:53 PM

darkone
(\/) (;,,,;) (\/)
11482 Posts
user info
edit post

What do you want from a weather app?

DarkSky's key features are that they downscaled certain weather variables based on terrain and land use and that they used motion interpolation on weather radar for prediction of near-term precipitation onset. I'm sure they also have some secret sauce about how they weight forecasts from different numeric models to arrive at a deterministic forecast. Other than that, they use the same numeric models as everyone else. There's not a lot of proprietary weather tools and information out there and of the ones that do exist, few are actually good.

7/8/2020 3:04:25 PM

synapse
play so hard
60188 Posts
user info
edit post

English holmes

7/8/2020 10:41:25 PM

StTexan
All American
2129 Posts
user info
edit post

^^is it seriously that fucking good?

7/9/2020 12:29:47 AM

darkone
(\/) (;,,,;) (\/)
11482 Posts
user info
edit post

From my perspective, DarkSky impressed people for two reasons:

First, the developers realized that below a certain size scale, the biggest impact on the temperature at a given location was elevation and land-use. So they used that to take weather observations and predictions made on of a spatial scale of 1 - 10 km and interpolated those data to the scale of 10s of meters with good success. Basically they just used your location elevation information to make the temperature forecast more precise. That doesn't mean it was more accurate however.

The really cool thing they didn't that no one had done for a phone app before was motion projection for radar data. Basically, you see blobs of precipitation on radar. You analyze their movement over successive observations to calculate a motion vector. Then you use that vector to translate the blob forward in time. As such, their app could say, "It's going to rain in 22 minutes." and they would usually be pretty close. It's not a novel approach in the weather research community, but in was for a phone app. Over short periods of time, it's usually not a horrible assumption that a storm is going to maintain it's size, shape, intensity, speed, and direction.

DarkSky used to have blog posts which talked about this but they've been memory-holed.

As for weather predictions, all the models are run by governments and are either free (NOAA) or available to anyone willing to pay the licensing fees (ECMWF). Everyone works with the same data. Private models do exist, but there typically targets for specialist industries like wind power who like to know about specific parameters not captured by the public models. There are a variety of companies who try to sell their own forecasts based on some statistically informed weighting of the various public models. I don't think I've see anyone offer rigorous proof that they're making better predictions. There's very little "secret sauce" when it comes to weather forecasts. Human forecasters still exist because they're still pretty good an identifying when a model has missed something and gone off the rails.

DarkSky also had a good API which made them popular with developers.

7/10/2020 11:23:19 AM

wdprice3
BinaryBuffonary
45903 Posts
user info
edit post

I liked DS because of the interface and radar. The predictions were nice, but probably had <75% accuracy.

7/10/2020 12:01:58 PM

darkone
(\/) (;,,,;) (\/)
11482 Posts
user info
edit post

The best radar app is RadarScope, but it's targeted at professionals.

7/11/2020 4:35:52 PM

StTexan
All American
2129 Posts
user info
edit post

^^^So when the Fish would be in front of a map during a snow storm and be “this is what the model is showing, but i think this could be less than map shows” etc?

Give a firm example of accuracy and precision. Is it guessing rain and 80 when it ends up being mostly cloudy and 93?

7/11/2020 11:38:26 PM

darkone
(\/) (;,,,;) (\/)
11482 Posts
user info
edit post

One could teach a whole semester-long class on forecast verification and assessment.

7/12/2020 1:39:18 PM

colangus
All American
742 Posts
user info
edit post

I just got fucked out of $300. I sent back a pair of AirPod Pros back in May (2nd pair I returned as the 1st pair didn't fit correctly). Both times they gave me a label w/ tracking.... both tracking numbers were wrong.

Both times they sent me - for no apparent reason- the shitty free pair of plugin earbuds you get w/ the old iphone/ipod (audio plug, not power).. I was like WTF?!?!

I've reached out via chat and phone.... every time they tell me a SR support rep will call me in 5 days... never hear back.

7/17/2020 4:40:37 PM

wdprice3
BinaryBuffonary
45903 Posts
user info
edit post

synapse


you find a new app yet?
I've been using the free version of NOAA weather radar & live alerts. it's ok I guess

7/20/2020 9:09:07 AM

GrimReap3r
All American
2732 Posts
user info
edit post

^for radar, RadarScope is the way to go for sure.

7/20/2020 9:19:48 AM

 Message Boards » Tech Talk » Fuck you apple. Page 1 2 3 4 5 [6], Prev  
go to top | |
Admin Options : move topic | lock topic

© 2021 by The Wolf Web - All Rights Reserved.
The material located at this site is not endorsed, sponsored or provided by or on behalf of North Carolina State University.
Powered by CrazyWeb v2.38 - our disclaimer.