Good vibrations: Seismological inference of elephant behaviour

Monday, 12. November 2018
14.00 – 15.00

Room Z816, the University of Konstanz

Iain Couzin

Prof. Tarje Nissen-Meyer, University of Oxford

Good vibrations: Seismological inference of elephant behaviour"

Tarje Nissen-Meyer, Beth Mortimer, Kuangdai Leng

Seismology has come a long way from constraining seismicity and Earth's interior. Owing to recent developments in instrumentation on land, sea, and above, numerical techniques and supercomputing, we now quantify and understand information from those complex vibrations that continuously excite our planet and its surface at scales from nanometer-scale cracks to global oscillations, and thereby illuminate their underlying physical processes. In particular, methodologies to discriminate vibration sources have been applied to meteorite impacts, nuclear explosions, glaciers, landslides, hurricanes, trains, ocean-shelf interactions, and animals. 

In this talk, I will first introduce modern seismological inquiry with large, diverse datasets, state-of-the art computing, inverse problems, with applications such as getting at Earth's dark matter and extraterrestrial life. This  sets the stage for our brand of mammal seismology. As a first step, we recorded seismic vibrations from elephants in Samburu Park, Kenya. Aided by numerical seismic modeling, we suggest that their various seismic signals (rumbling, running) may propagate as clearly identifiably signals many kilometers afar with favorable terrains and noise environments. This provides an effective means of potential information transfer in addition to or instead of acoustic transmission. Moreover, we extract "source parameters" such as elephant location and discriminated their time-dependent behavior. There are a number of advantages in using seismic methods for wildlife: The non-invasive, passive and relatively cheap deployment of seismometers buried in the ground suggests to attempt seismic monitoring for complementary data on wildlife behavior and whereabouts. Furthermore, the instantaneous nature of our analysis lends itself well to applying these ideas to combat the increasingly worrisome poaching problem. 

We will embark on field work in Mpala Research Centre, Kenya, to conduct source-function playback experiments, and deploying a large seismic broadband array  together with cheaper accelerometers, videotraps, and drones. This will establish a novel testbed of seismic monitoring of wildlife, discriminating between different species, and applying machine-learning techniques to allow for quick, possibly in-situ discrimination and analysis of continuously recorded vibration wavefields in such ecosystems. May elephants sense the faint first blibs of distant earthquakes prior to tsunamies and other impending dangers, or even apply seismic imaging? Could seismic cues traversing between different species affect their behavior? I will argue that a mix of local geology, geophysics and seismology, ecology, wildlife behavior, mammal physiology, blended by multidisciplinary data acquisition and mining with novel deep-learning techniques shall help us in further unraveling the connection between life and our dynamic planet.