“Modeling Sensitivity of GPS Networks to Ground Deformation in Southern California”

Larger imageNathan brainstorming, organizing, and writing pseudo-codes to determine how to conduct tests.

Throughout much of Southern California, high-precision Global Positioning System (GPS) stations have been erected to monitor ground motion resulting from elastic deformation around the plate boundary. These stations come in two flavors: continuous and campaign. Continuous GPS stations automatically collect data and lots of it, so their errors are relatively small. Campaign GPS stations, on the other hand, require manual resurveying, but cover a much broader area. The costs associated with placing new continuous stations and resurveying old campaign stations are rather high – thousands of dollars per site.

Larger imageNathan using computer programming to analyze data.

The GPS network in Southern California constitutes about 200 continuous GPS stations and several hundred more campaign sites, each with its own data and time series. We explored varying levels of spatial smoothing to determine shear strain and dilation, and performed Monte Carlo tests to determine the expected variance on these values. Extending the temporal span of GPS observations for individual stations will decrease model errors and result in better estimates on strain or other model parameters. Other factors, such as distance from faults, also affect stations’ contributions to models. Using computer programs, such as Matlab, to artificially extend stations’ time series data, we can determine the effect a resurveyed site has on the model and infer which sites would most benefit the model if resurveyed. This in turn shows the optimal distribution of resources to improve model estimates.

Larger imageNathan looking at Devonian strata in the gorge next to the Earth and Atmospheric Sciences building.

The other challenge is then to present improvement data in an efficient manner. For this purpose, we created a public website and Google EarthTM tour. The combination of these tools allow for easy distribution, and visualization in the interactive environment of Google Earth. The next step will be to compare the models resulting from this research to updated models from Crustal Motion Map version 4 to determine the accuracy of these estimated error improvements.

View field notes from other SURE Interns:   Select an intern... Alex Wilk Alexandra Millar Christina Velasquez Hannah Waterhouse Joshua Soble Lauren Steinberg Melissa Nunley Patrick Dooling Trevor Thomas Tyanna Schlom