Physical constraints on visual anemometry using vegetation displacement statistics

TL;DR

This study investigates the physical limits of visual anemometry using vegetation displacement, revealing regime-dependent correlations and proposing parametric models to improve wind speed inference across different wind conditions.

Contribution

It identifies fundamental constraints on vegetation-based wind measurement and introduces parametric relationships to enhance generalization of visual anemometry methods.

Findings

Wind and vegetation speeds follow Weibull distributions.

A sigmoid function describes the relationship between wind and vegetation scale factors.

Correlation between wind and vegetation is regime-dependent, enabling VA in intermediate winds.

Abstract

Visual anemometry (VA) leverages observations of fluid-structure interactions to infer incident flow characteristics. Recent work has demonstrated the concept of VA using both data-driven and physical modelling approaches applied to natural vegetation. These methods have not yet achieved generalization across plant species and require site-specific calibration. We conducted a laboratory study in an open circuit wind tunnel using overhead imagery of three vegetation species to assess the utility of vegetation displacement fields for wind speed inference. Both the wind and vegetation speeds exhibited a two-parameter Weibull distribution. The relationship between the scale factor (one of the two parameters) of the wind and vegetation was found to be well described by a sigmoid function, indicating three regions of distinct structural response to the wind loading at low, intermediate, and high wind speeds. Within an intermediate range of wind speeds, the wind and vegetation scale factors are nearly linearly proportional, thereby facilitating VA. Importantly, the wind and vegetation scale factors were found to be uncorrelated in low and high wind regimes, revealing a fundamental constraint on VA using structure response data. We discuss the physical basis for these regimes and present additional parametric relationships that can be exploited in the intermediate wind regime to potentially generalize inference of the wind speed and direction.