New Sensors to Discriminate Between Nuclear Explosions and Chemical Explosions or Natural Events

Period of Performance: 07/09/2002 - 01/08/2003


Phase 1 SBIR

Recipient Firm

Information Systems Laboratories, Inc.
10070 Barnes Canyon Road Array
San Diego, CA 92121
Principal Investigator


The fusion of seismic and magnetotelluric data holds promise in discriminating between underground nuclear blasts and earthquakes. MT measurements can sense several different effects of nuclear blasts: the acoustic wave which travels slowly away from the blast center, the upward propagating wave which generates Alfvén waves, and the direct electromagnetic pulse (EMP) which propagates through the earth's atmosphere after traveling from the buried location to the surface. These signals are mixed in with noise signals which will mask their presence: plane wave ionospheric resonances, and worldwide lightning which can mimic EMPs from explosions. This proposed work will reduce this noise to increase the SNR to detectable levels, and rule out lightning as the source of the EMP. We will explore the aspect of sensor arrays, beam-forming, and multivariate noise reduction. We will draw on our experience with multivariate algorithms for coherent noise mitigation, and in the location of global lightning. A demonstration experiment will validate the concept of using a localized array of sensors in California to derive directional EMP location information, and the identification of plane-wave signals which constitute noise. This experiment will demonstrate the concept of fusing magnetic with seismic data to differentiate between nuclear explosions and earthquakes. This work will demonstrate that important observable electromagnetic phenomena produced by explosions are observable after noise mitigation operating on multi-sensor arrays is executed. It will also demonstrate that spurious signals such as lightning can be differentiated from the sought-after explosion-induced EM signals, using geographic location techniques. It will also demonstrate the utility of multi-sensor data fusion of standard geophysical data streams, with magnetotelluric data. New algorithms developed in this program to maximumally identify and remove plane-wave ionospheric signals from magnetometer arrays will be written using the most modern numerical algorithms and languages (most probably in Matlab). This software will have as a baseline for comparison, the pre-existing MT multivariate array processing code written by Egbert (1997), which is used in multi-station MT data collected for petroleum exploration. Improvements over the Egbert algorithm could be sold to the petroleum exploration businesses who actively use MT methods for reconnaissance exploration of geologic structures not easily imaged with seismic techniques. If the Phase I study identifies significant improvements attainable using new software in comparison to the Egbert algorithm, then this will provide impetus in Phase II to further include these algorithms into a fully develop software product, which will have applications to petroleum time-series magnetotelluric analysis.