Infrasound-Based Wind Profiling

Wind blowing over a stationary infrasound sensor located on the ground produces wind noise. According to Taylor’s Frozen Turbulence Hypothesis, this wind noise is caused by turbulent eddies of essentially fixed size being advected past the infrasound sensor at the speed of the wind local. Based on the idea that turbulent eddies have a range of sizes, with the size of the eddy giving information about the winds at a height proportional to its size, it has been suggested that turbulence induced wind noise can be used to infer vertical profiles of wind speeds (Murray, 2012; Priestly, 1965). Such vertical profiles of wind speed are particularly useful in a number of sectors including aviation and wind energy. While there are active ways to obtain wind profiles involving laser (LIDAR), radio-waves (RADAR Wind Profilers), and sound (SODAR), a passive infrasound-based approach would have advantages in terms of emissions, size, cost, and all weather capability.

In a 2018 AMS Annual Meeting presentation (Pepyne, 2018) we showed that an infrasound sensor with pressure inlet located at 1 meter above the ground was able to infer wind speeds at 10 meters above the ground. In our on-going work, we’ve teamed with NRG Systems who have given us access to data from a 60-meter instrumented tower located on the grounds of their Vermont facility. By recording the infrasound wind noise at the base of the tower and using truth data from anemometers at 10, 20, 30, 40, and 60 meters above the infrasound sensor, we will use supervised learning methods to seek to demonstrate the potential of infrasound-based wind profiling.

Murray, 2012: Passive Acoustic Detection of Wind-Turbine In-Flow Conditions for Active Control and Optimization. Univ. of Mississippi, National Center for Physical Acoustics Final Report to DOE (DOE Award No. DE-EE0003269).

Priestly, 1965: Correlation Studies of Pressure Fluctuations on the Ground Beneath a Turbulent Boundary Layer. National Bureau of Standards, Report 8942 (also M.S. Thesis, Univ. of Maryland, 1965).

Pepyne et al., 2018: A Machine Learning System for Pressure-Based Wind-Profiling: Proof-of-Concept Results, AMS 19th Symposium on Meteorological Observation and Instrumentation, Austin, TX.