Why are weather forecasts sometimes so different from each other? Don’t they all use the same data and science?


I’m talking about information displayed on different weather websites. One will say that there’s a very high chance of rain in a place at a certain hour, for multiple hours, and another will say that there will be just some light rain in the first hour, then it will be dry/sunny.

In: 22

In some countries they do have different sources and so e companies own measure equipment across country. In other countries there is only one groverment maintained source but they have different type of subscription which has some “delay” according to plan. So.e do pure gussing or settelite image analysis only

When you get into the details of it, it’s an incredibly complex and difficult job. Weather is chaotic, everything affects everything else, and we have a series of best guesses based on how well we can predict the future based on past patterns in a world which is heating up and changing (weakening our predictive power).

Anything that involves predicting the future is risky, you’ll note that all traders get the same information about the stock market yet make different investment decisions, same with weather.

There are a couple of main models that almost all forecasting sites get their information from. The US model is GFS and the European model is ECMWF.

Some websites will just pass on Unmodified forecasts direct from one or other of these models.
Some websites will combine the data in some way and present a likely average.
Some websites apply local knowledge of particular regions and geography to make more precise localised predictions based on one or both of the models.

No. Not everyone uses the same data, although data is shared across agencies. No, not everyone uses the same science, although the science is shared across agencies. But, there are two other things to consider, as well: 1) even with the same data and the same science, 2 different models will give 2 different results, and any given website/page/app *might* be using its own model, or they might just be passing along gov-issued data, but even in government agencies, different models are used, so are they sourcing the raw data or the post-modeling data when they pass it to you? And 2) predicting the future is very, very hard. The further in the future you’re trying to predict, the harder it becomes. That’s why forecasts are fairly precise an hour from now, but very vague 2 weeks out. And even making predictions for a month from now or a year from now just isn’t done in meteorology. Anything beyond about 3-4 weeks is _usually_ the province of climatology, a completely separate science of attempting to predict the future.

Say you draw a whole bunch of squares on the ground, then you hold a bouncy ball up in the air above them. You have a whole bunch of people watching. Everyone can see the squares, can see you, they know everything about the bouncy ball you’re holding…they have complete knowledge of how everything looks *right now*.

You ask them to make predictions about where the bouncy ball will go when you drop it. You end up getting several different predictions from different people, because even though they know everything about the setup *right now*, they can only make predictions about what *might* happen next. Then you drop the bouncy ball and see if any of their predictions are correct.

If it’s someone’s *entire job* to make predictions about where the bouncy ball will land, they’ll probably make a pretty good guess…but they still can’t be sure, and they may still come to different conclusions than someone else.