Attached is spectra of the EndX Anemometers during a wind event over the weekend.  The start time is 2100 on 26 May when the average speed hits 15mps and stays up for ~5 hours with averages touching 30mps.  These channels come in as meter per second so I only applied a calibration of 1 so I think the units displayed are correct.

Check the mapping in the third attachment of 49387 to see the location of the Fence wind sensors.  Verbally, SPEED_4 is ~collocated with the WIND_ROOF sensor.  Speed_3 is west and high, Speed_2 is closer to the fence and only 12' up.  During this windy period the direction of the wind was pretty 'northerly' coming almost straight down the Xarm toward the corner

Generally, the WIND_ROOF sensor has a higher ASD than the FENCE sensors.  Does this indicate it is more sensitive or more noisy?  Based on the time series though, the PEM weather sensor is more responsive than the PCE Fence sensors, even the #4 up on the roof.  The other two sensors are understandably appear less sensitive as they are at less than 1/2 the altitude as the roof top sensors.