INTRODUCTION Seismic Refraction Tomography Study of the Potrillo CHECKERBOARD TESTING On May 16-18, 2003, the geophysics group at the University of Texas at El Checkerboard testing was carried out by adding 0.5 percent Paso conducted a wide-angle seismic refraction/reflection experiment across a sinusoidal perturbations to our smoothed velocity models with x and z portion of the southern Rio Grande centered on the Potrillo Volcanic Field in the Southern : Opportunity for Collaboration dimensions of 10x5 km and 16x8 km. (PVF). The PVF is perhaps best known for its xenolith localities at two volcanoes within the field, and Potrillo Maar. The experiment is Between Geosciences and Computer Science Checkers are well resolved within the upper 5 km for the 10x5 km velocity part of a joint project to study magmatic contributions to crustal evolution in an perturbations and down to 10 km for 16x8 km perturbations. Deeper than intracontinental setting using the southern Rio Grande Rift as a natural Matthew G. Averill, Roberto Araiza 10 km we lose both ray coverage density and resolution of features less laboratory. A pool of seismic instruments that were designed and built with than 16x8 km. funds provided by the Texas Higher Education Coordinating Board and the National Science Fopundation were deployed by a team of about 65 student Faculty advisors: Vladik Kreinovich, Kate C. Miller, and G. Randy Keller volunteers. The field work and data analysis was funded primarily by a grant from the Texas Higher Education Coordinating Board to UTEP and Texas Tech University of Texas at El Paso PVF2003 10x5 km checkerboard test

University. Additional funding came from the National Science Foundation EAR 0 --SP 1 --SP 2 --SP 3 --SP0 4 --SP 5 --SP 6 0 --SP 7 --SP 8 (Continental Dynamics Group) and The El Paso Water Utilities. 0 0 0 INITIAL RESULTS 0 PVF Experiment Layout and 0

) -10 -10 ) The velocity model below was derived from 5188 first arrival times input to a 3D 0 k m tomography (Hole, 1982). The inversion was carried out in a 230x25x68 km model -20 -20 (

Record Sections t h

space with a grid interval of 1 km. The RMS misfit for the model is 0.13 s. p

Profile Length: 205 km Source: 7 2000 lb shots D e p t h ( k m -30 -30 D e BACKGROUND For purpose of display, the 3D result has been averaged in the y-direction to produce Instrumentation: 795 Texans at variable 2D model plots. The velocity is masked by the ray coveraged shown at the bottom. -40 -40 The Rio Grande Rift is a major Tertiary tectonic feature that profoundly spacing (100, 200, 600 m) modifies the lithospheric structure of the southern Rocky Mountains of North The velocity model shows the signature of Basin and Range structure with low 0 50 100 150 200 America. Patterns of magmatism and extensional structures of the rift both velocity basins between higher vlocity uplifts within the upper 5 km. A high velocity Distance (km ) crosscut and reoccupy older structures including those associated with zone (> 6.2 km/s) between 5 and 10 km underlies the and may Precambrian continental assembly, late Paleozoic Ancestral Rockies and late represent the basement cored Rio Grande Laramide deformation. A broad elevated -2.1 -1.5 -0.9 -0.3 0.3 0.9 1.5 2.4 Percent Mesozoic to Early Tertiary Laramide tectonism. The region is an excellent place region of velocities greater than 6.5 km/s is located below the Potrillo volcanic field. to study magmatic modification because it is a location where comprehensive Moho depth across the model is at approximately 32 to 36 km. geophysical data sets and a rich suite of basement samples can be obtained. In fact, all crustal levels are accessible, with (1) shallow basement exposed in nearby mountain ranges, (2) a remarkably diverse deep- and middle-crustal Plot of Observed (green) vs. Calculated (red) Travel Times for each Shot within the Model Space PVF2003 16x8 km checkerboard test xenolith suite in two Quaternary maar volcanoes, and (3) a middle crustal along with Calculated Travel Time Residuals (blue) 34 --SP 1 --SP 2 0 --SP 3 --SP 4 --SP 5 --SP 6 --SP 7 --SP 8 0 xenolith suite in Eocene intrusions near El Paso. Thus, these data sets provides 32 0 0 0.01 0 0 30 0 28 0 0 0.01 a fantastic opportunity to establish the relative importance of magmatic input to 26 0.01 -0.01 24 0 0.01

22 ) -10 -10 ) formation of the crust by integrating the geophysical structure of the region with 20 0 0 18 0.01 0.01

0 k m 16 0 0 detailed geobarometry, geothermometry, geochronology, and chemistry of ( 14 12 -20 0 0.01 0 -20

T i m e ( s ) 0 surface and xenolith samples. The modeling of these data with current 10 0 t h

0.02 p 8 0 6 0.01 0.01 tomographic inversion algorithms is an opportunity for further interdisciplinary 0 0.01 4 D e p t h ( k m -30 0.01 -30 D e 2 collaboration between our geophysics and computer science groups, because 0 2-D Initial Velocity Model 0.8 the software is difficult to use, models are hard to edit, and the inversion is 0 00.01 -40 0.01 -40 0 5 5 5 5 5 5 5 5 5 5 5 0

Vp (km/s) -0.8 R e s i d u a l t ( ) inherently unstable. 9 6 6 6 6 6 6 6 6 6 6 6 100 200 10 y=25 km 8 Distance (km) 20 3-D Model 0 50 100 150 200 7 7 7 7 7 7 7 7 7 7 7 7 30 Distance (km ) Z ( k m ) Space 6 40 8 8 8 8 8 8 8 8 8 8 8 -2.0 -1.4 -0.8 -0.2 0.4 1.0 1.6 2.5 50 5 PVF2003 R1DV4 Masked Velocity Model Y=15km x=230 km t ? Percent if 60 ange l 4 y R 0 100 200 ande alle r illo Mtns . X (km) V io G otr lin Mtns R amide) up edar Mtn. olumbus ar rank z=68 km Hacita C C (L East P F

--SP 1 --SP 2 --SP 3 --SP 4 --SP 5 --SP 6 4 --SP 7 --SP 8 4.5 4 4.5 4 4.5 0 5 5 4.5 4.5 4 5 4 45.5 4.5 55. 0 5.5 5 5 5.5 5.5 6 5 5 COLLABORATIVE WORK 5.5 5.5 5.5 6 6 Near-Vertical Reflection Data 6 6 6

PVF Near Vertical Deployment RECORD SECTIONS ) -10 -10 ) Incorporate expert knowledge: 253˚ 6.5 k m -20 6.5 -20 ( t h In order to improve the usability and expand the usefullness of the p Shot Point 1 7 D e p t h ( k m -30 -30 D e tomographic inversion code we are combining the knowledge and Potrillo Volcanic 7.5 32˚ 32˚ 7.5 experience of our geoscience department with the resources and Field Kilbourne Hole -40 8 -40 programming experience of the our computer science department. Our goals include development of user interface software designed 1125 1150 1250 1700 1450 1650 1675 for both ease of data input, creation of the initial model, and displaying 1350 1550 0 50 100 150 200 Distance (km ) SP5 results adapted to needs of the geoscience community, parallelization km of the inversion algorithms to better utilize computing capabilities and 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 possible improvement of inversion algorithms to incorporate additional 500 1230 1400 1800 2600 0 50 Velocity (km/s) available data and stabilization of the calculations.

-107˚ Shot Point 5 Some of the solutions computed by the current methods are 0 mathematically sound, but geophysically unsound. The geophysicist Pg Pg PVF2003 Ray Coverage Density Model then realizes the source of the error and repeats the process. We

5 plan to incorporate interval uncertainty and other types of Intracrustal --SP 1 --SP 2 --SP 3 --SP 4 --SP 5 --SP 6 --SP 7 --SP 8 Reflectors Shot Point 6 0 0 uncertainty to use a geophysicist's knowledge and intuition in the solution process, to produce models which also agree with geophysics. PmP PmP 10

) -10 -10 ) T i m e ( s ) k m PmS ? PmS ? -20 -20 ( 15 t h p Use advanced methods:

D e p t h ( k m -30 -30 D e Fast Marching Method (Sethian, 1996): Unlike typical methods, this 20 -20 -15 -10 -5 0 5 10 15 20 25 30 -40 -40 method avoids iteration by, among other optimizations, adapting Offset (km) Dijkstra's shortest-path algorithm to the continuous case. It runs in 0 50 100 150 200 O(n log n), where n is the number of points in the domain. Distance (km ) Level Set Method (Osher and Sethian, 1988): This method, although 0.0 6.5 45.0 83.5 124.6 172.8 239.6 2582.0 Coverage (Rays/cell) slower than the Fast Marching Method, is more general. It efficiently takes into account discontinuities which are important in geophysical applications