This paper reviews recent progress toward understanding the dynamics of the middle atmosphere in the framework of the Atmospheric Dynamics Research InfraStructure in Europe (ARISE) initiative. The middle atmosphere, integrating the stratosphere and mesosphere, is a crucial region which influences tropospheric weather and climate. Enhancing the understanding of middle atmosphere dynamics requires improved measurement of the propagation and breaking of planetary and gravity waves originating in the lowest levels of the atmosphere. Inter-comparison studies have shown large discrepancies between observations and models, especially during unresolved disturbances such as sudden stratospheric warmings for which model accuracy is poorer due to a lack of observational constraints. Correctly predicting the variability of the middle atmosphere can lead to improvements in tropospheric weather forecasts on timescales of weeks to season. The ARISE project integrates different station networks providing observations from ground to the lower thermosphere, including the infrasound system developed for the Comprehensive Nuclear-Test-Ban Treaty verification, the Lidar Network for the Detection of Atmospheric Composition Change, complementary meteor radars, wind radiometers, ionospheric sounders and satellites. This paper presents several examples which show how multi-instrument observations can provide a better description of the vertical dynamics structure of the middle atmosphere, especially during large disturbances such as gravity waves activity and stratospheric warming events. The paper then demonstrates the interest of ARISE data in data assimilation for weather forecasting and re-analyzes the determination of dynamics evolution with climate change and the monitoring of atmospheric extreme events which have an atmospheric signature, such as thunderstorms or volcanic eruptions.

In this study we analyze infrasound signals from three earthquakes in central Italy. The M_w 6.0 Amatrice, M_w 5.9 Visso, and M_w 6.5 Norcia earthquakes generated significant epicentral ground motions that couple to the atmosphere and produce infrasonic waves. Epicentral seismic and infrasonic signals are detected at I26DE; however, a third type of signal, which arrives after the seismic wave train and before the epicentral infrasound signal, is also detected. This peculiar signal propagates across the array at acoustic wave speeds, but the celerity associated with it is 3 times the speed of sound. Atmosphere-independent backprojections and full 3-D ray tracing using atmospheric conditions of the European Centre for Medium-Range Weather Forecasts are used to demonstrate that this apparently fast-arriving infrasound signal originates from ground motions more than 400 km away from the epicenter. The location of the secondary infrasound patch coincides with the closest bounce point to I26DE as depicted by ray tracing backprojections.

The 2017 North Korean nuclear test gave rise to seismic and low-frequency acoustic signals, that is, infrasound. The infrasonic signals are due to seismo-acoustic coupling and have been detected on microbarometer array I45RU in the Russian Federation at 401 km from the test site. I45RU is part of the International Monitoring System for the verification of the Comprehensive Nuclear-Test-Ban Treaty. We analyze the seismo-acoustic coupling by making use of array-processing and backprojection techniques. The backprojections show that infrasound radiation is not confined to the epicentral region. More distant regions are found to be consistent with locations of topography, sedimentary basins, and underwater evanescent sources. The backprojections can be used to estimate the average infrasonic propagation speed through the atmosphere. We discuss these findings in the context of infrasound propagation conditions during the sixth nuclear test. It is suggested that propagation from the test site to I45RU may have occurred along unexpected paths instead of typical stratospheric propagation. We present several scenarios that could be considered in the interpretation of the observations. Electronic Supplement: Details on signal characterization and infrasound propagation conditions.

The underground nuclear tests by the Democratic People's Republic of Korea (DPRK) generated atmospheric infrasound both in 2013 and 2016. Clear detections were made in the Russian Federation (I45RU) and Japan (I30JP) in 2013 at stations from the International Monitoring System. Both tropospheric and stratospheric refractions arrived at the stations. In 2016, only a weak return was potentially observed at I45RU. Data analysis and propagation modeling show that the noise level at the stations and the stratospheric circumpolar vortex were different in 2016 compared to 2013. As the seismic magnitude of the 2013 and 2016 nuclear test explosions was comparable, we hypothesize that the 2016 test occurred at least 1.5 times deeper. In such a case, less seismic energy would couple through the lithosphere-atmosphere interface, leading to less observable infrasound. Since explosion depth is difficult to estimate from seismic data alone, this motivates a synergy between seismics and infrasonics.

Accurate prediction of Sudden Stratospheric Warming (SSW) events is important for the performance of numerical weather prediction due to significant stratosphere-Troposphere coupling. In this study, for the first time middle atmospheric numerical weather forecasts are evaluated using infrasound. A year of near-continuous infrasound from the volcano Mount Tolbachik (Kamchatka, Russian Federation) is compared with simulations using high-resolution deterministic forecasts of the European Centre for Medium-Range Weather Forecasts (ECMWF). For the entire time span the nowcast generally performs best, indicated by a higher continuity of the predicted wavefront characteristics with a minimal back azimuth difference. Best performance for all forecasts is obtained in summer. The difference between the infrasound observations and the predictions based on the forecasts is significantly larger during the 2013 SSW period for all forecasts. Simulations show that the SSW onset is better captured by the 10 day forecast while the recovery is better captured by the nowcast.