Odor source location can be predicted from a time-history of odor statistics for a large-scale outdoor plume

Abstract

AbstractOdor plumes in turbulent environments are intermittent and sparse, yet many animals effectively navigate to odor sources to find food [1]. Computational and lab-scaled experiments [2, 3, 4] have suggested that information about the source distance may be encoded in odor signal statistics, yet it is unclear whether useful and continuous distance estimates can be made under real-world flow conditions. Here we analyze odor signals from outdoor experiments with a sensor moving across large spatial scales in desert and forest environments to show that odor signal statistics can yield useful estimates of distance. The probability distribution of statistics from individual whiffs (contiguous odor sequences) are correlated with distance, but their correlation coefficient is poor. However, we show that a useful estimate of distance can be found by incorporating whiff statistics from time history of ∼10-seconds, resulting in a strong correlation ofR2∼0.70. We identified whiff concentration and duration to be the most informative features. By combining distance estimates from a linear model with wind-relative motion dynamics, we were able to estimate source distance in a 60×60 m2area with median errors of 3-8 meters, a distance at which point odor sources are within visual range for animals such as mosquitoes [5, 6]. Finally, we show that such estimates are only feasible if odor signal information is recorded at temporal resolutions of≥20 Hz. Together, our results provide a compelling case for how odor signal statistics could be used by animals, or man-made machines, to optimize plume tracking decisions at large spatial scales with natural wind.