f13.net

There's been some discussion in a couple of different threads here over the past decade, but just bits and pieces, so here's a thread for the topic.

I used to use Intellicast for their radar, which featured maps showing enough detail down to county lines and major roads that you could tell exactly where what was, with images usually between 10-30 minutes old and an animation loop of the recent past so you could see exactly what was coming and how fast.  Intellicast has been assimilated by Weather Underground and it's gone to shit basically overnight. Roughly one day in three something breaks and they stop updating the radar completely for hours at a time. like lots of hours. 5, 10, half a day, more. And it's not my browser cache, I've tried it on completely different devices after it stopped updating and still see the same hours old results. IBM is not doing themselves any favors with their branding all over that crap site. NWS has good high-rez, usually up-to-date radar with helpful warning overlays and minimal clutter, but doesn't do the fancy work of differentiating between rain and snow, or screening out non-precipitation haze or whatever. And no loops.  So, what's a good source for better radar than NWS now that Intellicast is no more?  I *really* miss almost live hir-rez radar loops. :(

As for forecasts, I've found NWS to be by far the most accurate. Weather.com has been crap around here, and WU is hit and miss and wading through their cluttered maps and ads to get to the actual data is a migraine-inducing nightmare. Google's hourly forecast actually seems better than those two, which is odd as I think they get their data from weather.com? So are there any better options I'm overlooking?

Oh, and I'm in central Kentucky but travel around the southeast so I'm looking for national coverage, not a local TV station.

NWS absolutely has loops. I've been using them for years.

https://radar.weather.gov/radar.php?rid=OKX&product=NCR&overlay=11101111&loop=yes

https://radar.weather.gov/Conus/northeast_loop.php

NWS for me too, with all the locations I go to, it's the overall most accurate. Radar isn't the most finely grained, but works.

We owned Intellicast the whole time. The internet response is amusing.

Not sure if it's what you are looking for, but www.windy.com is very nice.

Not ours. :awesome_for_real:

Been using this since you mentioned it the last time. Only use it at home and when I hear about some big weather thing going on.

On my iPhone I use MyRadar Pro, but mostly when I'm trying to tell if that TStorm over there is going to hit us on the lake.

I thought I remembered you saying that before. My gripe is that since the changeover, wtf has happened anyway? Why has it been down 4-7 hours at a time at least 3 times in what, 10 days?

Have you tried a 'metar and taf' app to grab your local airfield forecasts?

Usefulness depends a lot on how many airfields you have nearby, but for a simple 24 hour forecast I've never found anything more accurate or with less extraneous bullshit.

My only response here is that the WU site is a relative shambles, but I think this is generally known. If you think the Intellicast moaning was bad, wait until we turn down WU.

Note that the backend data is the same in all cases. Unfortunately there is processing being done along the way which can cause wonkadoo in any given medium. For example, there have been bugs in iOS flagship that were not in Android, but the data delivered is the same.

Pulling METAR data is great, but what do you do when you are n kilometers from one? This question is True Meteorology.  :oh_i_see:

Who is "We" here Yegolev?

Short answer is IBM.

Longer answer is the collection of long-running previously-extant companies as follows:
Weather Services International, descended from other companies which were more like a WX version of Wozniak in his garage (Andover, MA), merged with Weather Channel in some year I forget.
Weather Channel minus the TV studio and cable stuff. Byron Allen owns that now.
Weather Central (Madison, WI), sold to Weather Channel or WSI depending on who you ask in 2012.
Weather Underground (SanFran, CA), probably bought in 2012?.

Of that collection, I work most closely with the WSI people even though I live in the ATL. WSI delivered PilotBrief in 1980, and some of those people are still around. Lots of wacky shit in here. MIT grads. Spare PDPs on people's desks. Live ASCII weather forecast maps. Largest so-called IoT implementation that I'm aware of. Lots of smart people.

IBM bought the shit out of us in 2016 but "we" are still here, and Ginni likes to mention us by name. I've managed to learn surprisingly little about weather and media in 3.75 years, other than how much of it I don't understand.

For that, I use an app called AeroWeather which is great - you see all your favorite locations at a glance and then drill into forecasts. (Since we go into the mountains a lot, we can get conditions and forecasts for passes and airports in the small towns we visit).

How do you rate the forecasting for mountain passes? Passable(haha)?   I find the interpolation of forecasts across a large Z to be very hard.

Usually the passes don't have a forecast, just current conditions (because they have weather stations at the top of them). It would be nice if they did some kinda forecast or at least trending (is it snowing more or less than an hour ago?).

And here we go again. another storm and another outage, 3 hours and counting so far. wtf?  This would be EVERY time a storm has come through since the changeover the damn radar stops working. That's about as useful as a flashlight that only works in the sunshine.

Wait, what the fuck? Where are you posting from, BBN? PARC? Is it 1978 where you are?

In some places, yes. The thing which puts out the ASCII forecast is running on a OOS emulator and is contractually obligated to exist until 2021. While there are scientific advances in many places, people who NEED wx forecasts are strongly attached to their delivery processes. Similar to how the gubment is strongly attached to how they deliver things like forecast data and alerts.

And another storm coming through and WU's radar loop for my region has stopped again. An hour and a half since the last update, normally updates every 15 minutes or so.  National is current (well, 25 minutes old), but Bowling Green is out. Again.

That's 4 snow/storm events in a row it's stopped working.

Do you have a URL?

Hawaii having an effect on wind:
https://earth.nullschool.net/#2019/02/10/1500Z/wind/surface/level/orthographic=-157.33,22.74,1979

Sorry I missed this query. Whatever issues were happening seem to have been resolved.  As of a few days after that Feb 7 gripe the radar has been relatively current (within half an hour) every time I've checked it since. 

But it hasn't really snowed more than once since! LOL  (can't imagine how snow would have an affect unless it was just driving more traffic, in which case tornado season should be even worse?)  But to answer your question, the link I use is https://www.wunderground.com/maps/radar/current/bwg (https://www.wunderground.com/maps/radar/current/bwg), but when the outages were happening at Bowling Green they were also happening at Charlotte and Dayton (the next two closest).  Which is all odd because the national map was still updating the affected region, only the regional maps were not updating.

So, for now WU 's Radar animation is back to being may favorite for seeing what's coming in the next few hours.  NWS still usually has the most up-to-date (updated every 15 minutes approx) snapshot plus tornado and other warning overlays.

Unlike some other wx datasets, radar does produce more digital data when there are more echoes.

There is a 5-minute radar product but I don't happen to know its origin. Chances are that it is data-only and NWS may not be generating visuals from it. Naturally, any processing system would need to work 3 times faster to generate maps off of that dataset.

Well, I've given up on WU. My regional radar has been down several more times since my last whine, including two of the last three storm fronts.  Again, the national loop showed current data but the local loop was up to 24 hours out of date in the worst case.  Seriously, if more than x number of raindrops breaks the damn thing it's pretty worthless for tracking storms and such, no?  I don't really have much use for radar when the weather is fair.

So I've held my nose and crossed my fingers and reinstalled Flash on my devices so I can see the NWS loops like it's 1999 all over again. I guess I shouldn't complain too much, for a government system it's not all that extraordinarily out of date. And at least it still works when I need it the most.

On a separate note, wouldn't it be cool if you could get a live weather radar overlay on Google maps while you're driving?  That and a live satellite feed showing the actual traffic jam up ahead of you on the maps satellite layer instead of two year old imagery. I know, I know. Ridiculously expensive to image, process and transmit the entire country at that resolution even once a day much less every 10 or 15 minutes.  Not to mention the whole clouds thing. But it would/will be cool when we get to that point, maybe integrating imagery from drones and traffic cams and such?  Maybe even add imagery from the steering cameras in autonomous cars.  Hm. that would be quite a bit of telemetry to broadcast, prohibitively expensive to process and probably not even physically possible to fit in the RF spectrum.

You are correct in that it would be a shit-load of data to schlep around.

Sorry about the WU site, there are X layers between the data processing systems I am familiar with and what you see on any endpoint. Not so with NWS but in that case you're just getting the radar imagery. That is probably what you want, anyway.

For the DIY WX nerds: https://www.kickstarter.com/projects/weatherflow/tempest-a-revolutionary-personal-weather-system/description

Public announcement of a huge amount of work on our backends:
https://www.cnbc.com/2019/11/14/ibm-hopes-to-change-weather-forecasting-around-the-globe-using-big-data-and-a-new-supercomputer.html