I’ll add my findings that this is not so much a matter of over/under-done gust effects. It’s a missing element, see:
The turbulence right now feels very similar to the experience of flying through a cloud, where the plane shudders and shakes like a wet rat in rapid tossing jolts. Until it comes out clear on the other side…
Unfortunately, in the sim, this behavior is continuous and monotone. In clear blue sky, at FL380, a plane now shakes constantly, like it’s driving over cobblestones. As if inside an invisible cloud of infinite uniformity and size.
Truth is: There is a missing element in the data pipeline that abstracts RW weather data into the sim. In technical terms, this is known as “Amplitude Modulation”. In noise function design, the concept is more often known as “octaves”. I’m sure the capable programmers at Asobo are familiar with these notions, yet they seem to have been left out of the current implementation of gusts so far.
Currently, we have wind that does this:
That is a single octave of randomness, shifting the winds up or down to a given fixed amplitude.
Ideally we’d like to see turbulence that feels like this:
While that might appear altogether different, there is but a single small difference between those two functions: A Single Extra Layer of Random Noise.
Setup another layer of noise at a much broader period than the one governing the wind changes, then simply multiply the output of that first signal with this extra layer. The result will be as you see in the second example.
Ideally, you’ll also want to clamp the output of the second layer above zero, but leave it’s noise amplitude to range between -1 and 1. The result then will be that whenever the function commands a value below zero, you will have a pause in the turbulence. A simple, yet VERY convincing depiction of the well known lulls and smooth patches often found in between bursts of roughness.
That much is a critical lack that currently reduces the existing gusts simulation to a crude caricature of turbulent flight conditions.
At present, the system does not appear either under or over done. It is just too simple to pass muster, yielding far too regular, unauthentic perturbations and requires the simple addition recommended here for it to become truly believable.