The latest version of the Weather Channel app offers both past and future radar as options whe viewing the radar map. How does it know the future radar picture? My WAG: An extrapolation of current conditions? But, it seems there must be more. The composite picture does not seem strictly a linear model. Can anyone explain?
what needs to be explained? They extrapolate weather based on continuous worldwide data of temperature, wind speed, humidity, sunlight, cloud cover, cloud height, particulate matter, etc… It’s fed into computer models that visually display the estimated outcome of the information disseminated.
What don’t I get? I think they’re really pushing it. First, I notice there is always a disconnect between the “Past” loop (which comes up to present), and the “Future” loop. Why?
Next, can it predict if a line of thunderstorms is going to fizzle out before it reaches me? Obviously, extrapolating the instantaneous data alone would predict the line continuing in the same direction, same strength, same speed, ad infinitum. So somehow, statistics must get factored in to figure out when this line of storms will fizzle…or strengthen! I guess it doesn’t matter if they’re wrong, so what have they got to lose by projecting the “Future” radar?
You can usually plot a “simradar” picture from a weather model to simulate what the radar picture may look like in the future. I’d say that’s what TWC is doing. There’s really no other alternative. If you took the current radar picture and tried to simulate the future, you’d have to take into account humidity, pressure, temperature and a million other data points to simulate it. That’s a weather model.
So the present radar loop is what was/is actually happening, then when it hits the present, it switches over to a model. The model will be taking current conditions and plotting what may develop over the next hour or so. I’m no expert (although I do run my own model 3 times a day at home 'cause I’m a nerd) but I think the model resolution is going to be the main reasons for the difference.
The resolution of the model is going to be a lot less than the resolution of the radar. The best models have a 3km resolution, but the computation on such a detailed model is so hard that it takes way too long to run unless it’s for a small area. TWC is most definitely not doing that, they’re using something lower resolution, most likely the National Weather Service’s 13km Rapid Refresh (RAP) model. The output isn’t going to match the real radar exactly. Here is an image of precipitation in the current RAP model of the northeast. Notice how low resolution the precipitation is.
And, you’re right, it’s TWC and they’re trying to make money. The future radar looks cool and people probably like it even if it’s not very accurate. Beware the big corporate weather names. It’s pretty much impossible to predict any weather out to 10 days yet Accuweather has a 25 day forecast where they predict the wind speed to tenths of miles per hour. Why? So they can advertise that they’re the only place with a 25 day forecast.
Extrapolation need not be linear. If we have a general theory of how a system behaves, we can extrapolate future states based on that theory.
I don’t get the question. This is what weathermen have always been doing, their reason for existence. It’s the very reason you pay attention to the weather report in the first place. All that’s changed is that they’re now delivering their forecasts in the form of pretty pictures. But thirty years ago, when the weatherman said that there was a 40% chance of thunderstorms later in the evening, he was making that statement based on the same sort of models that are currently being used to make that map (of course, the current models are much more sophisticated and detailed than they were 30 years ago, but qualitatively the same).
The app is much more specific. It purports to show the exact areas where rain will fall and those areas where it won’t. My experience is that the future radar is highly inaccurate, and I would not rely upon them. They are based on current meteorological data, including humidity, dew point, present rains and their movements. Often the rain predicted will dissipate before it reaches your area, or a shower will develop unexpectedly.
Ummm, it’s called weather forecasting for a reason.
In the current implementation, they simply record the weather as it happens. It’s then stored in an underground vault until 2081, when time travel is invented. The data is then sent backwards in time to the appropriate date, and looped in as usual.
Username and post combo for the win!
Yeah, and they try to predict those, too. You’re only just now discovering that weather forecasters sometimes don’t get it right?
It’s not really a matter of getting it right or wrong, it’s really impossible to predict exactly how a storm will evolve, at least to the point where you can make a fake radar image for it. You can often predict general characteristics like how much rain it will produce, whether it will intensify or dissipate, or if it will produce any severe weather, but a lot of how a storm behaves is due to either small local weather conditions that are practically impossible to detect or due to the internals of the storm itself.
Think of a radar image as being like a fire. You may be able to predict whether or not it has enough fuel to keep burning for a while (and you still may be wrong because there’s a lot of little details you don’t know) but you can’t predict it’s exact shape.