Using infrasound observations and data assimilation to detect atmospheric variability over short timescales

Atmospheric variability at short time-scales (seconds to minutes) is challenging to detect, quantify, and include in numerical models of atmospheric circulation. Infrasound can be generated by natural and anthropogenic sources, and due to the low frequency of the signal, it can travel relatively long distances (hundreds to thousands of kilometers) and be detected by acoustic arrays. When detected, the observed wavefront properties quantities (travel time, backazimuth angle, apparent velocity) contain integrated effects of the atmospheric slab through which the wave traveled. We use data assimilation, in particular an ensemble Kalman filter, to invert these observations to atmospheric quantities. As observations, we use three days of daily infrasonic signals originating from 52 explosions. The signals propagated through the stratospheric waveguide and were recorded at a distance of 256 km. The assimilation background field is provided by the 10-member ERA ensemble reanalysis product, which is valid every 3 hours. The departures with respect to the background shed light to the atmospheric variability in very short time-scales (minutes).