I'm the author of the video you link to in your question.
I'm sorry that I didn't reply to this when you asked three months ago but I've only just seen your message on Reddit. Your account seems to have been deleted so I'm copying my reply here.
In terms of wind direction, this was only a simple model so I cheated and picked an angle randomly. A simulation that "evolves" with changing wind and weather is something I'm working on but the biggest problem is making it fast enough to use.
For wind and rain, what I did was simulate a square of particles, all one pixel apart and implemented as an array of floats for wind elevation and an array of floats for wind moisture. The wind moisture at the start is set quite high to represent the fact that it is moist air carried over the sea.
I then move the cloud along the wind direction over the map. When the elevation (height) of the map is greater than the wind height, then the wind height is moved up to this elevation. Otherwise, the height of the wind decreases slowly. This simple model can be used as a very simple approximation of air pressure when the elevation of the world is much less than the elevation of the wind. My video colours the raw wind elevation blue but you could colour the difference between wine elevation and land elevation instead for a more interesting effect.
Each point in the cloud loses some moisture at each step depending on temperature (cold areas losing more rain i.e. particularly over mountains). Once it has lost all of its moisture it is out of rain, so a "rain shadow" develops, represented by a desert if the temperature is hot. You could improve this model by having wind gain more moisture over lakes or sea.
Finding the best constants for how quickly wind particles lose moisture, or how much moisture they have initially, was just trial and error until I found a combination that looked good and gave me a full range of terrain types on a typical map.
Hope that helps -- good luck with your project! Let me know how you get on.