Improving the Amplitude and Frequency of Wind Gusts in Live Weather

[WxMarc]

Over the past several months of flying in MSFS, the wind gusts in live weather seem to have a similar frequency everywhere – regardless of terrain, altitude, etc. Additionally, the amplitude of each gust seems to be nearly identical. The wind speed value appears to oscillate back and forth between a minimum and maximum value several times per minute. I suspect this variation is too extreme.

In the United States (which I am most familiar with), the standard timeframe for observing wind gusts is 10 minutes. As an example, if a METAR reports winds of 28012G26KT, it is quite common for the 26 knot gust to be an anomalously strong gust that occurred several minutes before the observation. When I fly in MSFS, I get the sense that live weather is creating maximum gusts several times per minute. If the METAR reports gusts to 26 knots, those peak gusts will occur in the sim very frequently and rapidly. In the real world, the strongest gusts generally occur once every several minutes, with smaller amplitude gusts on shorter time scales. Gusts can also last for multiple seconds in the real world, while in MSFS, they always seem to last for less than a second.

I would suggest adjusting the gust model with much more varied behavior – many small variations in speed on a second-by-second basis, a few large peaks occurring every several minutes, and several intermediate peaks on the sub-minute time scale. Ideally, this should be randomized, rather than following a predictable pattern.

[LKFJP]

I think there’s already a wishlist item for this?

Edit: Here

[WxMarc]

As I understand it, the bug report you linked is a different issue. That user is reporting that the wind gust graph in custom weather is not reflecting the parameters input on the custom weather dialog box. Specifically, it is not allowing the user to display a relatively steady wind with small gusts.

My post is specifically about the frequency and amplitude of gusts in live weather. It’s possible that the issues are related somewhere in the core code, but I’m focused here on the need for improved variety of gusts in live weather – less frequent maximum gusts, more variety of gust amplitudes and durations, etc. I just edited the topic title to clarify that I’m referring to live weather.

[LKFJP]

I obviously don’t have access to the code, but I’d speculate that they’re one and the same. The game’s weather “engine” has capabilities that can either be set manually or set based on current real world data. The game’s lack of fidelity with real world weather is a well known thing, especially w.r.t. cloud types, so it doesn’t surprise me at all that the wind gust model is also… lackluster.

On the plus side, that other thread is tagged “feedback logged”, so maybe they will do something about it?

[KezMaefele]

For clarity my issue which you linked is about user configured weather. It is not necessarily related to live weather, unless the metar numbers are interpreted by the faulty mechanism that draws the graph. This is likely to be true, but not guaranteed to be true, so a distinct issue is reasonable. Honestly the whole wind gust situation looks a lot like some non-pilot programmer misunderstood the requirement, and non-pilot QA said, yup, it varies, ship it!

I focused on the graph because it is trivial to demonstrate and if it’s all the same algorithm, fixing that will hopefully fix everything else as well.

[hobanagerik]

I hadn’t looked at this for a while, so I found a place with high winds, and also gusty.

KDFW
KDFW 180953Z 19017G28KT 10SM FEW250 04/M09 A2985 RMK AO2 PK WND 20028/0953 SLP109 T00441094

I then let an external tool plot the ambient wind at ground level for a few minutes.

Seems bang on as far as weather gusting from 17kts up to 28kts.

I’ll now look for somewhere else in the world that has similar weather, and see what the pattern is there.

LYTV
LYTV 181030Z 15017G28KT 9999 SCT013 BKN040 16/14 Q1004 NOSIG

And for the plot:

And just in case there was something affecting the results with teleporting around, somewhere with very little wind in the same session.

ZMBH

This place has no reported gusts, but there is always very slight changes in the wind velocity. I left the chart on auto scaling so it looks more energetic than it is, as you can see from the scale on the left, ranging from about 1.4kts to less than 3kts.

What I would argue here is that although the wind strengths are clearly different, and therefore the amplitude changes in the chart, that the tempo or frequency of these changes is roughly the same across all three charts i.e. there doesn’t seem to be much difference in how often the wind changes speed, only the amplitude changes, and that a non-gusting wind area is changing just as fast, and often, as a gusting area, it’s just not as noticeable.

Look across the three charts, and look at these peaks in particular, and see how they correspond across all three of them. That pattern can be seen in all three.

[LKFJP]

@hobanagerik

Look across the three charts, and look at these peaks in particular, and see how they correspond across all three of them. That pattern can be seen in all three.

I wonder if there’s an easter egg hidden in there? How long is the repeating pattern? Maybe it’s something completely non sequiter like Ken Lum’s Melly Shum hates her job?

[WxMarc]

This is really great data! Thank you for collecting it and sharing the results. I would make a couple of observations based on this better quality information:

1. The sim is not correctly interpreting METAR winds. If winds are reported as 19017G28KT, the plots show that the sim is creating a varying wind between a minimum of 17 knots and a maximum of 28 knots. That is incorrect. 17 knots is the average wind speed (over a 2 minute interval in the US METAR standard), and 28 knots is the maximum wind gust (recorded over a 10 minute interval in the US). In order for 17 knots to be the average, there would need to be lulls where the wind speed drops below 17 knots.

2. Gusts to near the maximum reported value seem excessively frequent. Over the course of three minutes, the charts show several gusts to at least 27 knots (within 1 knot of the 10 minute maximum).

I spent a little while searching for high-resolution real-world wind data to compare, and I found a couple of interesting academic papers. The following plot shows an example of wind speed measured over a 10-minute interval with the peak gust highlighted:

This comes from a 2022 paper in Applied Sciences entitled “Characterization of Wind Gusts: A Study Based on Meteorological Tower Observations” by Yan, Chan, et al. In this case, the 10-minute maximum gust was a clear outlier, with other gusts of widely varying magnitudes throughout the 10 minute period.

The same paper also includes the following histograms showing the relative frequency of a variety of wind speed parameters (including gusts and gust amplitudes) in observed data. The histograms show that gusts tend to follow a distribution with small gusts more common and large gusts less common:

[LKFJP]

We could dream that they’d read an academic paper or two when they implemented the wind gust model, but from @hobanagerik’s data, it looks like they obtained a profile from somewhere and are just scaling that and playing it back on repeat: “number 9, number 9, number 9”. That’s pretty cheap. I’d give them some credit if there was at least an easter egg hidden in that pattern somehow, but I doubt it.

[Elethes]

Has this been in the dev’s team radar? I am surprised that this topic has so few replies.

I’ve played with airlines for a long time, where live weather wind speeds feels good enough. But in the last months I’ve started playing a lot with smaller aircraft and even considering the difference in weight and size it is just not enough to justify the bull riding it is.

It is almost certain to happen in any session in any place, from Canada to Patagonia. The aircraft is agressivelly pushed from one side, specially during take-offs and landings. And yesterday, during a flight from Quebec City to Halifax area, I’ve experienced two moments where for several minutes the aircraft was pushed hard just from one side. Just a really strong wind.

It is clear it happens just in live weather. Custom winds and gusts as high as 10 knots are a breeze when compared to live weather wind speeds. And I concour, it is following the METAR values in its max. Maybe that explains why this behaviour is so clear when flying close to ground layers.

It feels the physics is tailored to big aircraft. Maybe the model for some reason was concepted in way that you can favor only one side. I don’t expect a solution that model fluid dynamics correctly, that would be overkill and ask for a lot of processing resources. But maybe tailor the behaviour through a parameter in the formula that takes into account the weight and form factor?

Or better yet, a way for us to change the proportion of max and minimum speeds of live weather or its mean velocity through sliders? Because sadly turbulence options don’t solve this, be it medium or low.