Developing remote sensing methods for bedrock mapping of the Front Range mountains, Colorado

Abstract

The Colorado Front Range Mountains have a history of significant debris flow hazards capable of causing losses to both property and life. The flash floods in Larimer, Boulder, and Jefferson Counties exhibited this when a storm event on September 9-13, 2013 triggered over 1,138 debris flows in the Colorado Front Range leading to eight fatalities and causing damage to buildings, highways, railroads, and infrastructure. Following this event, the U.S. Geological Survey (USGS) studied the debris flows that were triggered by the rainstorm with the intention of modeling debris flow susceptibility in this region. The objective of this project is to assist in constraining the susceptibility modeling by creating and executing a methodology for using existing remote sensing technology to map bedrock outcrops. Calibrating against six smaller study areas that span different geologic formations and ecoregions of the Front Range Mountains, the goal was to produce a map of exposed bedrock outcrops over nine, 7 ½ minute quadrangles that encompass portions of the St. Vrain and Big Thompson watersheds. The benefit of using remote sensing is the ability to map the bedrock exposures in a time-efficient and cost-effective manner for a significantly sized area of interest. The primary purpose of this thesis project is to: (1) develop a land cover classification methodology capable of discriminating bedrock and colluvium and (2) compare the classification accuracy of each individual remote sensing method to select the best land cover classification method. Through the use of imagery data sets from the multispectral Landsat-8 Operational Land Imager (OLI) and Thermal Infrared Sensor (TIRS); the high resolution hyperspectral imagery from the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) sensor; and the Advanced Land Observing Satellite’s (ALOS) Phased Array type L-band Synthetic Aperture Radar (PALSAR) sensor, four methods were employed to attempt mapping bedrock outcrops. These methods included unsupervised classification, supervised classification, supervised classification with iterative unsupervised classifications, and sequential land cover classification. The sequential land cover classification method yielded the best overall results producing the highest sensitivity (user accuracy), precision (producer accuracy), and F_1 measure for the classification of Bedrock and Colluvium. The highest observed agreement (observed accuracy) an overall method was generated by the sequential classification scheme (77.72%). Considerable difference between the performance metric values for the bedrock and colluvium land cover classes versus all other land cover classes remains quite significant. Further research should be conducted to examine combining existing passive remote sensing methods with active remote sensing methods to map bedrock outcrops.