Rebecca Bussard, postdoctoral scholar in the Department of Geosciences at Penn State, will give the talk "Mitigating Atmospheric Noise in InSAR Time-series Over Volcanic Targets Using a Convolutional Neural Network," at 2 p.m. on Friday, Sept. 4, in 117 Earth and Engineering Sciences Building on the University Park campus and via Zoom.
The event is part of a fall seminar series hosted by TRAnscendIng Natural hazards forecasting (TRAIN). The talk is free and open to the public.
Remote sensing techniques, like satellite imaging, are useful to monitor volcanoes and other natural hazards. However, atmospheric artifacts commonly cause noise that can mask deformation signals related to volcanic activity.
"Several techniques have been developed over the past few decades to mitigate atmospheric noise," Bussard said. "The success of these techniques, however, is highly variable depending on the regional setting and in some cases can even introduce more noise."
Bussard used U-Net, a convolutional neural network (CNN), which was trained to predict deformation from input consecutive unwrapped InSAR displacement maps.
"The CNN framework runs the input data through a series of convolution and deconvolution layers to handle the dimensionality of such a large image dataset," Bussard said. "We then train the CNN to isolate deformation from the atmospheric noise and measure how well the CNN output matches the original deformation patterns and magnitude."
Bussard graduated with a bachelor of science in physics with honors in Earth sciences from Penn State. She also received a doctorate in Earth sciences from the University of Oregon, where her dissertation was focused on mapping both distributed and focused volcanism with a wide geospatial toolkit including statistical modeling, optical remote sensing and microwave remote sensing. She is now a postdoctoral scholar at Penn State working with Christelle Wauthier, professor of geosciences, on mitigating volcanic hazards with geodesy and machine learning.
About TRAIN
TRAIN is an initiative through the Earth and Environmental Systems Institute (EESI) that aims to transform complex geophysical and atmospheric data and bridge the gap between raw information and community resilience. TRAIN aims to build the community at Penn State to become nationally recognized for research innovation to respond proactively to funding opportunities and coordinate natural hazard responses.
