Spatiotemporal information preservation in turbulent vapor plumes

Monitoring of airborne chemicals is essential for public safety and environmental applications, but the complex nature of stochastic airflow limits the accuracy of data acquired from chemical sensors, including the identity, concentration, and location of a chemical source. Here, we explore the limits on information retrieval in a turbulent environment using temporally modulated vapor plumes. By exploiting the physical and statistical properties of turbulent flow, we demonstrate higher classification rates (≥20 Hz) at lower chemical error rates (≤0.1%) than prior reports. Physical constraints on vapor sensing are explored, including finite spatial sampling, large path distances (≥3 m), dilute concentrations (≤1 ppb), dynamic wind velocities (V = 0–5 m/s), cross winds, and occluded airflows. Our results highlight the fact that as turbulence increases, smaller structures can be embedded in an air flow, which increases measurement variance, but also makes each plume more quickly and accurately identifiable. For a given wind velocity, there exists a modulation rate which optimizes the rate of preserved information. This raises interesting possibilities for the identification of optimally preserved patterns in a turbulent environment.