I have these observations concerning smoothing in FSUIPC 4.5.3.
I have selected to allow changes to FS's own weather.
Flights in FSX with wind, temperature and altitude smoothing enabled under Jeppeson weather control (Note- just using Jeppeson as an example here, I've also used and tested all other weather emulation programs):
1) Wind smoothing works wonderfully as it has through many FSUIPC versions and the last two FS versions.
2) Pressure smoothing has an unexpected behavior-
It appears only the altimeter is getting "smoothed" (display of the altitude) and not the pressure the aircraft is flying in. I've determined this by observing the following:
a) Although the altimeter is getting smoothed nicely (indicator is responding to altitude changes more slowly over time) while observing (A:INDICATED ALTITUDE,feet) the actual altitude the model is responding to while observing the GPS/Radar altimeter (A:PLANE ALT ABOVE GROUND, feet) is not getting smoothed. Even while flying over large expanses of water where terrain should not be influencing the GPS display of the altitude, large, uncontrolled jumps of the plane's actual altitude are being observed. Modulating the plane's altimeter display DOES have the positive attribute of fooling the Autopilot into thinking an altitude change has been modulated, thus the AP is not driven to large compensation maneuvers. A well configured AP in the aircraft.cfg is able to smooth pitch compensation without this assistance however.
b) As a result of un-modulated altitude changes, the performance of the aircraft in response to the large air density changes observed by large speed changes in the IAS display also supports the conclusion the actual altitude of the aircraft is un-modulated.
3) Very much like the Altimeter smoothing characteristic, while observing SAT Temperature (A:TOTAL AIR TEMPERATURE, celsius) the actual air temperature of the environment the aircraft is flying in is un-modulated (jumps up and down in large steps- as it always has in FS.) This is also validated while observing the large airspeed changes and required compensation of the AutoThrottle mechanism in response to large efficiency changes in jet engine performance at high speed and altitude.
Wind smoothing enabled turbulence characteristic-
My impression of the turbulence emulation this feature re-introduces as a result of smoothing removing the otherwise intended results of turbulence set by cloud and wind layers is it's a bit over-done. I appreciate being able to suppress this feature. With all FSUIPC4 smoothing disabled I created a static weather condition using the FS weather dialog where I fly a turning, descending pattern into the runway through several cloud layers with the top being very turbulent thunderstorm with only slight turbulence at the bottom. The aircraft under test is my own flight model with very well known and predicable flight characteristics. The plane behaves as expected with a rich expression of bouncing and rolling, but always remaining in reasonable AP control.
When using FSUIPC4 wind smoothing in Jeppeson flying straight and level through clear air at high altitude and high speed in an area known and proven to be relatively calm based on the metars for the area, the plane is driven to flip over by the turbulence. Checking the suppression box reduces this to a manageable level although my impression while making long flights of thousands of miles, at high altitude and speed (thousands of feet higher than the highest clouds) the turbulence seems too pervasive where I'm flying in fairly rough bouncing and weaving 90% of the time.
Rant against MS FS:
I've been flying in FS since version 1 for nearly 25 years. Ever since dynamic weather was introduced, the developers have always got it wrong as far as the philosophy of the data is always right, we just fly the data (with little interpretation.) This digital approach to interpreting digital data provided by metars just doesn't work using the assumption a large number of station data near each other will "mostly" reflect an average condition, thus the stations will "modulate" each other over time, slowly transitioning over distance to new, different conditions. In the datasets we have access to over many years, it's quite evident this ideal perspective is not obtainable without a radical change in the datasets we've had access to (metars.) A good example is the Puget Sound area of Western Washington in the US. Many of you reading this also live in richly diverse terrain where no two stations even a few miles apart have the same data due to islands, water, farmland plains, mountains many thousands of feet all within a few miles of each other. Add to this problem that of flying over large expanses of water where the last and next station data encountered are thousands of miles apart with weather as different from one another as is possible in the digital world. Sorry FS devs, the response of the sim between widely separated data by reverting to clear and calm between stations does not work. I mean "work" by saying "realistic." This is want I want. Real world realistic, not strict "the data is god" realistic. An top of all this, the large portions of the dataset is indeed very faulty. Between uncalibrated instruments, data corruption, balloons getting lost, errors in hand-entered data, etc, the dataset is at best a distant representation. Fine. I can accept the dataset with all its faults if a rule of expected-normal interpretation is imposed over it to transition through the errors elegantly. Right now and in all my FS experience, the errors are allowed too much influence on the interpretation.
The valiant attempts representing tens of thousands of man hours by many talented and dedicated teams of third party developers over the years should be ample evidence to the FS developers that users are NOT happy with this philosophy of real world digital data interpretation (no coded transitioning effects of temperature, wind, or pressure.) It has been left up to third party developers without the required support needed to solve this horrendous problem and has taken years to even get close while some major issues are still unresolved. I'm giving notice to those dev members who have moved on to other flight simulator projects that I'm not satisfied with continuing this attitude of digital accuracy at the cost of real-world emulation. Our planet modulates EVERYTHING. Transitions EVERYTHING. The Metar Dataset left to itself does not transition enough. This is what I expect. This "Attack Radar" philosophy of weather simulation just HAS to go.