Abstracts Category : Other

Multispectral and Hyperspectral Remote Sensing of QuaternarySediments in Tule and Snake Valleys, Lake Bonneville, Utah

by Kianoosh Hassani

Lake Bonneville was the largest water body thataccumulated in the Great Basin during the late Pleistocene. Itslatest major lacustral cycle lasted from 30 ka to 12 ka and muchevidence of the lake remains are still evident in the landscapetoday. This thesis investigates the use of Landsat-8 multispectralimagery for mapping the Quaternary deposits in the Tule Valleyportion, and Hyperion (EO-1) hyperspectral data for mapping part ofthe adjacent Snake Valley of Lake Bonneville. Maximum likelihoodclassification was applied for Landsat 8 data, and the two spectralanalysis approaches of linear spectral unmixing and spectral anglemapper (SAM) were applied to the Hyperion dataset. Furthermore,X-ray diffraction (XRD) analysis of a Lake Bonneville marl sedimentsample characterized the dominant minerals in that sample. Thisinvestigation relied on Sack's (1990) Quaternary geologic map ofTule Valley as the reference for the remote sensing analysis. Thisstudy investigates if those sources of information can approach inquality and detail the traditional map that relies on fieldwork andair photo interpretation. Results illustrate that hyperspectral andmultispectral data have potential value for Quaternary geologicalmapping. Maximum likelihood classification yielded overall accuracyof 51% with successful discrimination of Qlf, Qeg, Qes, Qlm, Qac,and bedrock. However, complete separation between severallacustrine and alluvial classes was not achieved. In general, theHyperion spectral angle mapper (SAM) and spectral unmixing resultsdiscriminated relatively well among the three endmembers ofcalcite, gypsum, and quartz across portions of the Snake Valleystudy area. The high fraction abundance values on the fractionalimages reliably represented pixels dominated by calcite, gypsum, orquartz. Some confusion between classifications are attributeded tothe local mixing of classes at the pixel scale, overlap inmineralogy, similarities in the nature of surface weathering, andthe limited spectral resolution of the Landsat image. Resultsindicate that these methods have value for mapping extensive areasof Lake Bonneville and other desert lake basins faster and moreefficiently than has previously been possible.Advisors/Committee Members: Sack, Dorothy (Advisor).