Let me share a couple quick examples of things a meteorologist does to improve upon model output when weather forecasting.
A meteorologist is trained to know how the atmosphere works in the real world, so when the computer models show things which do not make sense in terms of atmospheric physics (which they often do) then the forecaster knows to adjust and correct the model and ignore things it shows that are likely to be wrong.
In the ECMWF model forecast below the area circled in green on the 500mb jet stream chart is favorable for short-wave disturbances aloft which spark thunderstorm clusters, sometimes squall lines, mesoscale convective complexes or even a Derecho. However, being imperfect the model rainfall forecast (right-hand panels) is for drier than normal weather. So a meteorologist crosses that out (X) and makes a forecast for normal to above-normal rainfall ignoring the models dry forecast.
But automated weather WEB sites and phone APPS would just show you the wrong model forecast so you’d be misled. The same thing happens with model forecasts of temperatures and snow/ice etc. A trained meteorologist does not just say what the model shows must be the forecast. People look at an APP and wrongly think... that must be what every meteorologist is forecasting. NOPE, nope and NOPE.
You can see in the verification of actual rainfall below that the model was out to lunch, it was not dry in the X area:
So when the website or APP you use keeps changing or gives you a bogus forecast don’t blame the meteorologists unless you checked with one, we have nothing to do with WEB or APP forecasts.
The same goes with if you get a busted forecast from a meteorologist, that does NOT MEAN every meteorologist had the exact same forecast. Back before I retired my forecast was routinely different some or a lot from other forecasters. I even had people give me a hard time about a bad forecast when they had not heard or seen MY forecast, they just assumed mine was like the one they saw. The public can be clueless on how this works lol smh.
ANOTHER EXAMPLE:
In the model output below there is a GIANT difference in rainfall forecast between the two major global model equations.
Where the two models place the wettest weather and the driest weather is very different in large parts of the country:
Wet or dry?
A good meteorologist does not just blindly accept model output as the forecast.
How do you know which one is right? Or if both are wrong?
Instead I would use research, my training in weather physics, synoptic weather pattern recognition, known model biases and shortcomings, and years of experience to fix and adjust the models to make my own forecast.
Your WEB weather page or APP does not and CAN NOT not do that!
And do you even know which model your website or APP uses? Do you know when to trust them and when not to do so?
In easy patterns APPs will be OK, and many use radar and just move the rain or snow forward for some number of hours and that can work but only AFTER precipitation has already formed and is moving through the area. But thats extrapolation and not a true forecast of the future when radar is still clear.
So now you know much more than most of your friends and family about weather forecasts.
For more insights follow me on Twitter @MellishMeterWSB.