Fast level-set methods for large-scale geospatial-information based inverse gravimetry problems and applications to threats detection

点击翻译按钮获取中文摘要

The principal investigator and his colleagues will develop fast level-set methods for large-scale geospatial-information based inverse gravimetry problems and apply these tools for threats detection. They aim at addressing some challenging inverse problems arising from geospatial-intelligence applications and remote sensing from both theoretical and computational perspectives: fast stable numerical continuation of gravitational potentials and fast high-resolution reconstruction of subsurface attracting-source inclusions. First, their proposed fast algorithms for numerical continuation of gravity fields will overcome significant, long-standing difficulties in inverse gravimetry problems because the new methods can be applied to gravitational and magnetic potentials in arbitrary computational domain in any dimension. Second, their proposed fast level-set methods for large-scale geospatial-information based inverse gravimetry problems will be the first of its kind which is able to recover both a domain and its potential density from geospatial-information based gravitational measurements. Third, the new fast method will be developed into efficient algorithms and codes on streaming-architecture based high performance computing platforms. Fourth, the resulting codes will be tested on synthetic data sets and GRACE satellite data sets. Consequently, this new set of powerful computational tools will be developed for the first time for large-scale geospatial-information based inverse gravimetry problems. The project will address some fundamental issues in geospatial sciences, remote sensing, geosciences and scientific computing. The US government has been developing remote sensing technology by deploying advanced radars and satellites to collect geospatial information for many years. Solving the inverse gravimetry problem is one of the many essential tasks to analyze and classify the collected surveillance data. This research endeavor will enable further advances and breakthroughs in the Defense and Homeland Security sector in the form of fast and efficient algorithms for the detection of underground tunnels and caves, including man-made underground structures. This is an example of what this work concerns. Such underground tunnels can exist along the US border. The problem is to determine the existence and location of such underground tunnels. The research done here is relevant to analyzing geospatial information collected by all such applications in threats detection. Students from the PI's institution are involved in this innovative interdisciplinary research project.