Neuromorphic cameras as event-driven scintillometers for boundary-layer turbulence characterization.

Abstract

Image-plane scintillation concentrates at high-contrast boundaries, where turbulence-induced wavefront distortions are amplified by local intensity gradients-precisely where neuromorphic event cameras are most sensitive. We exploit this physical correspondence to estimate the refractive-index structure constant Cn2 over near-ground horizontal paths (∼200-300m) without an active transmitter, a co-located scintillometer, or a trained regression model. Relative irradiance sequences reconstructed by event-only temporal integration yield pixel-wise scintillation indices that are inverted through a spherical-wave model. Four field campaigns spanning two seasons and daytime through nighttime regimes, validated against a sonic anemometer, confirm hourly agreement within a factor of 1.2-4× under well-illuminated conditions. Mechanical vibration from wind loading and insufficient scene contrast are the limiting conditions, pointing toward mount stabilization and active target illumination for continuous 24 h operation.