09/19/2018 (Wed) 10:59:02
Weather is a chaotic system; errors grow exponentially with time. There is no workaround around this unfortunately, even just thermodynamic fluctuations will ruin the situation, it's against the laws of physics to deterministically predict it.
But one can predict the randomisation time of the system (Lyapunov exponents), and for weather in general it's a couple of days in the tropics, and about two weeks in temperate regions. But again, it depends on what kind of system you're observing, smaller systems randomise faster. So if it's impossible to predict where exactly storms will develop next day (only that they will in the region), you can predict how a storm will develop for a couple hours ahead, but meanwhile large systems like weather fronts can be propagated with quite the accuracy week or more ahead. Rossby waves in jet stream can be predicted for a whole month ahead.
Recent development; this grew over night. Storms started coalescing into one single mass around midnight.