Fernando A. Martínez-Torres
Years participated in RESESS:
Improving Slab 1.0 subduction zone models using regional constraints from the Eastern Pacific
Academic Affiliation: Senior, University of Puerto Rico, Mayaqüez, Geology
Science Research Mentors: Gavin Hayes, USGS and Synergetics Inc.
Writing Mentor: Fran Boler, UNAVCO
Fernando A. Martínez Torres was a 2009 and 2010 RESESS intern. He was a star recruiter of new RESESS interns for the summer of 2010 in his home department at the University of Puerto Rico, Mayaguez. Fernando plans to attend graduate school in the field of geophysics. Fernando loves to travel and see new places, and has an upbeat outlook on life. He got the prize for visiting the most National Parks in one summer as a RESESS intern.
Knowledge of the geometry of subduction zones is essential for understanding the rupture processes of great (magnitude > 8) earthquakes, and is a key constraint for many other modeling efforts, including the study of subduction zone mechanics and interpretation of seismic structure and mantle geodynamics. Inferences of plate geometry are usually made based on seismicity; the hypocenters of subduction zone earthquakes tell us the locations of their fault rupture planes—our best estimate of the location of the slab—so it is important to obtain well-resolved hypocentral locations. In this study, we work with Slab1.0, a global three-dimensional model of subduction zone geometries. These models are compiled primarily using teleseismic data, which can often be less accurate than regional data because they use global observations rather than using local seismic networks. In this study, we compare Slab1.0 models with models compiled from regional studies, digitized from geo-science literature. After digitization, these models can be quantitatively compared to Slab1.0 models covering the same region, generating difference maps of their vertical offset. We also analyze the methodology and data sets used by each study to verify the accuracy of the regional models. By quantifying the differences between models and by assessing the reasons for those differences we can identify areas where Slab1.0 can be improved.