Big Brother - thy name is Google.
I can’t, actually. Take a day like today - today’s forecast is basically ‘rain all day’, but it’s not raining right this minute. If I go outside, I can tell that it’s not raining right now (my head is dry, whee!), and I can tell that it “feels like rain”. But my weather app can tell me “Light rain starting in 25 minutes”, which means I know I can probably go across the street to get a cup of tea without my umbrella. It’s pretty cool.
Apparently forecasting has improved in just a matter of months.
My weather app used to predict rain in 10% increments. Now the percentages are 20%, 12%, 11%, 3%, 2%, 1%, 0%. Such detail!
Yes, and I worry about this some. If I lived a more “interesting” life, I would probably opt out. But then I think that 1) you never know who might want to stalk you or why, and 2) where the next big data leak/breach will come from (Google? probably only a matter of time), and it becomes clear that this stuff is quite dangerous even for mundane people. Or something happens and I speak out and all of a sudden become an important activist for Cause du Jour, and somebody potentially has access to what flights I have tickets for, when my bills are due, when I’m expecting a package at my house, where my kid goes to daycare.
It is terrifying.
But it’s also kinda nifty to be told if I need to leave early to make it on time. So it goes.
Most very short term personal rain forecasts are actually radar interpretations. It simply looks at NWS radar, sees which way the wind is blowing, and looks 5, 10, 20, 30 minutes upwind of your location. If it sees rain there, you get a “forecast” of rain in 5, 10, 20, or 30 minutes.
It works pretty darn well for the very short term. But in no sense is that weather forecasting as in predicting the flow of atmospheric temperature, pressure, and humidity to predict when and where clouds will form or dissipate or when and where rain will begin falling from an otherwise dry sky or when and where a raining parcel of air will dry up again.
IOW, the technique is 100% valid for what it does. But it’s not extendable to longer timeframes or weather factors other than precipitation.
Already answered, but here is an analogy I use that people seem to resonate with:
If I flip a coin, I really have no clue whether it will come up heads or tails.
If I flip 1,000 coins, I can predict with pretty good accuracy that approximately 500 of them will be heads and 500 will be tails.
If I weight the coins to slightly favor one side over the other, I still can’t predict the outcome of a single flip, but I can predict how the long term trends will change.
Even though we have a high short term variability, we can still reliably identify long term trends.