A SIMPLE ALGORITHM FOR SEQUENTIALLY INCORPORATING GRAVITY TO ENHANCE SEISMIC TRAVELTIME TOMOGRAPHY: APPLICATION TO THE UPPER-CRUSTAL STRUCTURE IN PUGET LOWLAND, WASHINGTON

The earth's three-dimensional upper-crustal structure can be revealed by modeling variations in seismic first-arrival traveltimes and in potential field measurements, which is particularly useful in regions having sparse borehole-control and limited outcrops. In the Puget Lowland, Washington State, USA, we demonstrate a simple method, designed to improve our resolution of the 3-D geometry of Cenozoic basins, for sequentially satisfying seismic first-arrival traveltime and gravity residuals in an iterative 3-D inversion. The algorithm is portable to any seismic analysis method that uses a gridded representation of velocity structure. Our technique calculates the gravity anomaly resulting from the velocity model by converting to density with Gardner’s rule. The residual between calculated and observed gravity is minimized by weighted adjustments to the model velocity-depth gradient where the gradient is steepest and where seismic coverage is least. The adjustments are scaled by the sign and magnitude of the gravity residuals, and a smoothing step is performed to minimize vertical streaking. The adjusted model is then used as a starting model in the next seismic traveltime iteration. The process is repeated until one velocity model can simultaneously satisfy both the gravity anomaly and seismic traveltime observations within user-defined acceptable misfits. We test our algorithm with data gathered in the Seismic Hazards Investigation in Puget Sound (SHIPS) experiment. We perform resolution tests with synthetic traveltime and gravity observations calculated with a checkerboard velocity model using the SHIPS experiment geometry and show that the addition of gravity significantly enhances resolution. We calculate a new velocity model for the region using SHIPS traveltimes and observed gravity, and we show examples where correlation between surface geology and modeled subsurface velocity structure is enhanced. These improvements include enhanced resolution of the Everett basin, of the Olympic accretionary core complex, and of northwestern end of the Southern Whidbey Island fault. This technique can be readily applied to other areas having both tomography and gravity data, including the Georgia basin, the San Francisco Bay area, and the Los Angeles region in the Cordilleran forearc.