Applications of Remote Sensing Data in Geoscience (Alan Yong, organizer)

Kristy Tiampo (Department of Earth Science, University of Western Ontario)
S. Samsonov (Department of Earth Science, University of Western Ontario)
John Rundle (Center for Computational Sciences and Engineering, UC Davis)
Spatio-temporal Bayesian Analysis for Integration of GPS and DInSAR Data

Friday 8:30-8:50, San Rafael

Abstract:

Recent work in the field of hierarchical Bayesian modeling in the atmospheric sciences has resulted in development of methods for the successful integration of spatially sparse but temporally dense data with data collected on a spatially dense grid at intermittent times (Wikle et al., 1998). Here we present a method adapted from this theory for the derivation of three-dimensional surface motion maps from sparse GPS measurements and two DInSAR interferograms using Gibbs-Markov random fields equivalency within a Bayesian statistical framework (Gudmundsson and Sigmundsson, 2002; Li, 2001; Samsonov and Tiampo 2005). It can be shown that the Gibbs energy function can be optimized analytically in the absence of a neighboring relationship between sites of a regular lattice and because the problem is well posed, its solution is unique and stable. The results of inverse computer modeling are presented and show a drastic improvement in accuracy when both GPS and DInSAR data are used. Preliminary results are presented using Southern California Integrated GPS Network (SCIGN) data and DInSAR data from the Western North America Interferometric Synthetic Aperture Radar (WInSAR) archive.


Michael J. Rymer (United States Geological Survey)
Angela S. Jayko (United States Geological Survey)
Joel E. Robinson (United States Geological Survey)
Michael J. Abrams (Jet Propulsion Laboratory)
Use of Multispectral Imagery, Aerial Photographs, and Geologic Mapping to Locate Faults in the Eastern California Shear Zone: Examples from the Indio Hills and Owens Valley, California

Friday 8:50-9:10, San Rafael

Abstract:

The eastern California shear zone (ECSZ) is an approximately 80-km-wide zone of deformation that accommodates about 24% of the relative PacificÐNorth American plate motion. The ECSZ extends from the Coachella Valley, where it is partitioned northward off the San Andreas fault, to the east of the Sierra Nevada, and then farther north. Because the shear zone is broad and diffused, faults are poorly expressed in some places, especially in alluvial fan environments. We analyze remote sensing data by implementing a multispectral approach, coupled with medium- and large-scale aerial photographs, and geologic mapping, to highlight previously unrecognized faults in late Quaternary alluvial fan deposits in the ECSZ. The two areas of study are 1) immediately north of the Indio Hills and 2) west and north of the Alabama Hills, in the west-central Owens Valley. For both locations we use Landsat, ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer), and MASTER (MODIS (MOderate Resolution Imaging Spectroradiometer) and ASTER) imagery, along with medium-scale aerial photographs. Additional data, such as large-scale, (1:4,000), low-sun-angle aerial photographs and geologic mapping are used in the Indio Hills; in the Owens Valley, TIMS (Thermal Infrared Multispectral Scanner) imagery are added. In both study areas faint fault "ghosts" show up on Landsat, ASTER, and MASTER imagery, especially when viewed through enhanced contrast filters, and on medium-scale aerial photographs. In the Indio Hills area, low sun angle (approximately 15 degrees off the horizon) aerial photographs clearly show fault features, including graben and offset fan surfaces marked with desert varnish and desert pavement. These newly recognized faults extend northward from faults exposed in uplifted mid- to late Quaternary deposits and project northward into prominent, well-expressed faults mapped in Cretaceous crystalline bedrock in the Little San Bernardino Mountains. In the Owens Valley study area, fault features are visible on large-scale aerial photographs, but are most noticeable on TIMS imagery. Although short sections of faults were previously mapped on aerial photographs, the TIMS imagery shows a heretofore undiscovered, through going, anastomosing fault zone. In both the Indio Hills and Owens Valley areas, these newly recognized fault zones should be taken into account when considering the local tectonic setting and earthquake hazards. We use these two examples to illustrate additional work that could be undertaken in other parts of the world with similar geologic and desert settings. We conclude that the type of multispectral imagery, image scale, and sun angle should be considered to optimize visibility of subtle fault features.


Tracy A. Purdum (California State University, Northridge; Jet Propulsion Laboratory)
Kenneth J. Hurst (Jet Propulsion Laboratory)
Dominic Mazzoni (Jet Propulsion Laboratory)
Classification of Hyperspectral Imagery for the Delineation of Geologic Formations

Friday 9:10-9:30, San Rafael

Abstract:

Cutting-edge technology and remote sensing allow a new approach for future geologic mapping. The classification of geologic formations within hyperspectral imagery by support vector machine algorithms (SVMs) can aid in delineating the extent of various lithologic units in a region. The objective of this study is to examine the performance of SVMs in classifying geologic spectra from the Panamint Range, California provided by the Hyperion sensor onboard NASA's Earth Observing-1 satellite. SVMs performed class assignment for all pixels throughout an image based on user-provided training examples. Results show SVMs produced a classified image in which the extent and pattern of the lithologic units bore a strong visual resemblance to the geologic formations on the map used for reference, and therefore can be regarded as strong candidates for future geologic classification studies on Earth and Mars.


Alan Yong (United States Geological Survey)
Susan E. Hough (United States Geological Survey)
Helen Cox (California State University, Northridge)
Kristy Tiampo (University of Western Ontario)
Amy Braverman (Jet Propulsion Laboratory)
Janet Harvey (UCLA)
Simon Hook (Jet Propulsion Laboratory)
Ken Hudnut (United States Geological Survey)
Gerry Simila (California State University, Northridge)
Characterizing Predicted Ground Motions Using Remote Sensing Data

Friday 9:30-9:50, San Rafael

Abstract:

Local rock and soil conditions are critical for determining the amount of shaking that can be expected during an earthquake. These conditions are usually determined from traditional geologic maps, but the amount of precision possible on such maps is limited. To circumvent this and other limitations, we use remote sensing data to determine the predicted ground motions. We collect samples of selective spectral data covering our southern California test regions at different scales. Our primary sources of spectral data come from ASTER, MASTER, and Landsat 7 ETM+ instruments. To map site conditions, we implement simple, empirical, and theoretical approaches using programmatic methods that include iterative and heuristic segmentation techniques via ArcGIS, ERDAS Imagine and ENVI packages. In addition, we explore the feasibility of applying statistical data fusion techniques to these data sets. We compare our results to the latest 1:100,000 scale geologic maps and to available ground motion measurement maps.