The GPS Analysis Package for Exploration and Understanding of Geodetic Sensor Web Time Series Data

We introduce the GPS Analysis Package (GAP), a Matlab toolbox for GPS data exploration and understanding. The toolbox is designed to support scientists and engineers studying the motion of the solid Earth both in an academic environment and in the course of NASA missions such as UAVSAR and future InSAR satellite missions. It includes an ensemble of low-level routines to perform basic signal processing operations, such as removal of secular motion, de-noising, and removal of seasonal signals. It also includes a suite of more sophisticated statistical pattern recognition techniques, including hidden Markov models and Bayes nets, to detect changes, identify transient signals, understand regional motion, and uncover relationships between geographically removed nodes in the GPS network. Finally, it provides an assortment of methods for estimating missing observations in the network. We provide usage examples of the package applied to particular scenarios, including the 2010 El Mayor-Cucapah earthquake, the 2011 Tohoku-Oki earthquake, and ongoing slow slip events in the Cascadia region. We also demonstrate the utility of the package within a web portal and web services environment by showcasing its use in the QuakeSim web portal. The QuakeSim portal allows easy access to GPS data sources provided by multiple institutions as well as a map and plotting interface to quickly assess analysis results. Finally, we show the extensibility of the package to other problem domains and sensor network data sources, demonstrating the analysis tools as applied to seismic network data, autonomous robotic navigation, and fault detection in engineering data streams from the International Space Station.