Infrasound data inversion for atmospheric sounding

Abstract

International audienceThe International Monitoring System (IMS) of the Comprehensive Nuclear-Test-Ban Treaty (CTBT) continuously records acoustic waves in the 0.01-10 Hz frequency band, known as infrasound. These waves propagate through the layered structure of the atmosphere. Coherent infrasonic waves are produced by a variety of anthropogenic and natural sources and their propagation is controlled by spatiotemporal variations of temperature and wind velocity. Natural stratification of atmospheric properties (e.g. temperature, density and winds) forms waveguides, allowing long-range propagation of infrasound waves. However, atmospheric specifications used in infrasound propagation modelling suffer from lack and sparsity of available data above an altitude of 50 km. As infrasound can propagate in the upper atmosphere up to 120 km, we assume that infrasonic data could be used for sounding the atmosphere, analogous to the use of seismic data to infer solid Earth structure and the use of hydroacoustic data to infer oceanic structure. We therefore develop an inversion scheme for vertical atmospheric wind profiles in the framework of an iterative linear inversion. The forward problem is treated in the high-frequency approximation using a Hamiltonian formulation and complete first-order ray perturbation theory is developed to construct the Fréchet derivatives matrix. We introduce a specific parametrization for the unknown model parameters based on Principal Component Analysis. Finally, our algorithm is tested on synthetic data cases spanning different seasonal periods and network configurations. The results show that our approach is suitable for infrasound atmospheric sounding on a regional scale

Evaluation of wind and temperature profiles from ECMWF analysis on two hemispheres using volcanic infrasound

“…In earlier infrasound inversion studies [e.g., Lalande et al, 2012;Assink et al, 2013], it has been assumed that adiabatic sound speed profiles are essentially correct, which leaves any misfit between modeling results and observations to errors in the wind profiles. This assumption seems justified considering the typical magnitude of errors in temperature and wind profiles (i.e., from comparisons between ECMWF analysis and lidar/wind radiometer data ).…”

“…Recently, studies have focused on the development of inverse methods to estimate upper atmospheric wind updates from infrasound data [Le Pichon et al, 2005b;Drob et al, 2010;Lalande et al, 2012;Assink et al, 2013;Arrowsmith et al, 2013]. Inversions for upper atmospheric data using infrasound data are not only useful in further reducing the uncertainty of these parameters but are also beneficial for applications that require a precise modeling, such as for estimation of explosion magnitude [Assink et al, 2013].…”

“…Instead of a more rigorous inversion procedure that involves propagation paths, travel time, and bearing deviations Lalande et al, 2012;Assink et al, 2013], we simply seek the ensemble of effective sound speed updates that would likely explain the observed spread in trace velocities. As such, we present these results to provide a first order of magnitude estimate of updates.…”

Comparison of co‐located independent ground‐based middle atmospheric wind and temperature measurements with numerical weather prediction models

“…This has motivated the development of atmospheric remote sensing methods [e.g., Drob et al, 2010;Lalande et al, 2012;Landès et al, 2014;Assink et al, 2014a]. In particular, the main characteristics of SSW events have been successfully derived from directional microbarom amplitude variations resulting from changes in stratospheric propagation conditions [Evers and Siegmund, 2009;Smets and Evers, 2014].…”

Infrasonic interferometry of stratospherically refracted microbaroms—A numerical study

“…For such an estimation, an inversion of the model of infrasound propagation through the atmosphere is required. Numerical studies for building such an inverse model of the atmosphere are done by Drob et al (2010) and Lalande et al (2012).…”