Thanks to modern-day satellite technology, weather-predicting supercomputers, and the now-ubiquitous smartphone, it's easier than ever to find out what the weather's going to do in the next hour or day. But just how far into the future can a forecast actually predict?
Just last week, AccuWeather introduced 90-day forecasts, much to the dismay of meteorologists who assert that predicting any weather patterns more than 10 days in advance is pretty much pointless. Before launching their completely unscientific 90-day forecasts, the longest forecast offered by AccuWeather was 45 days - the predictions of which were dismissed as "not rooted in any science currently available to meteorologists" by The Washington Post back in 2013.
It's true to say that improvements in technology have made weather forecasts more accurate. A 5-day forecast from today is as accurate as a 3-day forecast was back in 1995, Ars Technica reports, and we can now trust 10-day forecasts to some extent whereas the outer limit was a week back in the 1980s.
But trying to predict something three months away is a whole different matter.
It's the sheer number of variables and contributing factors that make weather forecasting so difficult, and the more time that passes, the lower the chance of being able to make predictions that are accurate.
It's the butterfly effect in action - one small change tomorrow can ultimately lead to a massive difference after 90 days.
"People should not use long-range forecasts as a strict guide, but rather look at how the weather patterns evolve," says AccuWeather, admitting that its 90-day forecasts can't be relied upon to be completely accurate.
Even with that disclaimer, the forecasts can't be trusted at all, according to US meteorologist Dan Satterfield. "We are in the realm of palm reading and horoscopes here, not science," he said in a blog post. "This kind of thing should be condemned and if you have an AccuWeather app on your smart phone, my advice is to stand up for science and replace it."
In building up a weather forecast, meteorologists combine information about current conditions with what we know about atmospheric laws, detailed data about what the weather has done in the past, and advanced computer simulations (which are getting more complex and intelligent all the time).
The Met Office in the UK - widely recognised as one of the world's most accurate forecasting organisations - says its predictions are pretty reliable up to four days, based on an error-checking measurement it's developed itself. That's up from a single day in 1980, which shows how far we've come, but it's still nowhere near 90.
So if you want an answer to the question we started this article with, it depends on what margin of error you're willing to allow, but it seems that 3-4 days are just about the limit if you want a good chance of knowing what's going to happen.
Up to 10 days, you can get a less accurate - but occasionally helpful - forecast, but anything after that, and you might as well have a guess yourself.