Monte Carlo Simulation of Seismic Location Errors for Moving Vehicles

Abstract : A method was developed to predict the distribution of seismic source mislocation errors for tracked vehicles using bearing and range (r) location statistics obtained from field data. The method is of use for 1) guiding the design and deployment of seismic sensor networks and 2) statistically modeling seismic network tracking performance. In the present work, a seismic array is composed of one or multiple seismic subarrays each of which makes estimates of bearing and range. The bearing and range estimates are taken to be normally distributed around the true target location with errors sigma(sup b) and sigma(sup r) respectively. When these sigma's are used in our Monte Carlo simulation, we chose a true source point, and network geometry (array locations). Using these source and sensor positions, a set of true bearing and rare computed. The Monte Carlo simulation is done with 100 trials, in each trial a random error is added to the true bearing and r for each array according to the observed sigma. For each of the 100 bearing and r Monte Carlo measurements, a weighted least squares solution for the East and North value of the source is found. A statistical fit for all 100 trials is used to find the 9007) confidence location ellipse surrounding the true source point. Two types of displays are presented. The first fixes sigma(sup b)and sigma(sup r) and makes a map that displays the error ellipse axes and orientation as a function of map position. The second uses a fixed map position and displays the error ellipse axes and orientation as a function of sigma(sup b) and sigma(sup r). This allows convenient analysis of how the location error changes with changes of the array measurement accuracy. The later format is particularly suited for evaluating the impact of seismic senor configuration on network tracking performance. Results are presented for a number of different network configurations representative of battlefield situations.