Determination of Geothermal Anomalies Through Multivariate Regression of Background Variables at Yellowstone National Park Using Landsat 5 TM Thermal Band Data

Abstract

Geothermal anomalies of Yellowstone National Park (YNP) are identified and quantified using Landsat 5 TM thermal band data. Multivariate regression of independent background variables that effect thermal emissivity, including elevation, slope, aspect, insolation, vegetation, water, soil moisture, and exposed land, were utilized in this study to create a comprehensive background filter for the raw imagery. Subtracting the multivariate background model from raw Landsat 5 TM data accentuates large geothermal anomalies such as Grand Prismatic and less thermally evident features such as the Old Faithful Geyser while removing significant false anomalies from the imagery. Geothermal anomaly emittances within YNP were calculated with a range of 40-120 W/m2. False positives for geothermal activity were reduced in the scene, with remaining ones focused on bare earth slope, consistent with other studies. A differencing between known geothermal pool temperatures and model residual temperatures at 25 sites indicates an average difference of 347 K (stdev 12 K), suggesting scalability from residual output to corrected temperature detection. The methodology employed for detecting known geothermal anomalies in YNP could be utilized to detect unknown geothermal potential in underexplored geothermal regions.