Improving nuclear explosion detection using seismic and geomorphic data sets

Cleat Philip Zeiler, University of Texas at El Paso


The ability to detect and locate nuclear explosions relies on the collection of seismic data, picking seismic phases, comparing amplitudes of seismic phases, and using velocity models to invert for location. The current monitoring system provides reliable results for large-yield sources detected at regional to teleseismic distances and at known test sites. With the increased availability of local datasets and the possibility of evasion scenarios that mask the yield, we improve the current methods of nuclear explosion detection by understanding local phase phenomena.^ The first chapter of this dissertation addresses the characterization of local/regional phase phenomena and source discrimination associated with near surface testing. I develop a local source discrimination technique and characterize the phases produced by near-surface explosions. We use the Source Phenomenology Experiment (SPE) broadband data set collected across the Colorado Plateau during the summer of 2003, which recorded explosions in hard and soft rock mines. We optimized a local surface wave magnitude scale derived from a stable, regional surface wave magnitude (Russell, 2006) for the explosions in each lithology. Magnitude scales for the local phases (Pg, Lg/Sg and Rg) were also tested and compared. The regional and teleseismic discrimination techniques employed were optimized for local distances to distinguish between source and rock type. We found that the magnitude and amplitude ratios were able to discriminate between small earthquakes and explosions at local distances, with each performing the best in the hardrock lithology. However, we believe that the ratio techniques would perform equally in both lithologies if multiple stations were used to establish the ratio values. We also determine that the source lithology and large-scale geologic features control most of the variability in the amplitude measurements. While several misclassifications are noted in the model, we designed the model to prevent false negatives for the simulated nuclear explosions. In this scenario no nuclear explosions will go undetected, but additional investigation needs to be performed to understand the false positive matches.^ The second and third chapter of the dissertation explores the errors associated with picking seismic phases in the process of identifying and locating seismic sources. The study is broken into three parts, which each use distinct data sets to isolate the factors needed to determine an appropriate error model for measuring picks on seismograms. The first data set describes the reliability of a single analyst, the second data set demonstrates the picking culture among institutions, and the third data set defines an error model derived by multiple analysts picking a common data set.^ The fourth chapter is an integrative method that characterizes a tectonic region using traditional methods combined with a moving-window Z/R ratio technique. The final chapter establishes a geoinformatics database, which is used to correlate wave phenomena to the local geology.^

Subject Area


Recommended Citation

Zeiler, Cleat Philip, "Improving nuclear explosion detection using seismic and geomorphic data sets" (2008). ETD Collection for University of Texas, El Paso. AAI3310688.