Event Date
Abstract
Soil moisture has a heavy influence on the stability of shallow colluvial hillslopes. Because of this influence, it is often necessary to monitor seasonal variations of soil moisture at various individual sites. Soil moisture measurements over large areas can be obtained via low orbit satellites. In particular, the NASA Soil Moisture Active Passive (SMAP) Level 4 Root Zone Soil Moisture (L4_SM) product provides near real-time measurements of soil moisture on a gridded spatial resolution of 9 km. Unfortunately, a resolution of 9 km is too coarse for use in soil moisture monitoring at a single site. Therefore, satellite-based soil moisture data must be downscaled to a resolution conducive for an individual site-scale assessment. This study presents the results of efforts to downscale the SMAP Level 4 Root Zone Soil Moisture product through machine learning (ML) and a proposed bias reduction technique. Specifically, this study investigated Random Forest and Soil Evaporative Efficiency (SEE) downscaling methods to obtain site-specific moisture data. The proposed bias reduction calibrated the locally downscaled data to that of ground-based data through a linear regression approach. The proposed bias reduction technique incorporates a multivariate regression analysis, which characterized relationships between site-specific soil texture data and SMAP L4_SM root zone soil moisture correction factors. The downscaling approach was performed over various in-situ sites across the Commonwealth of Kentucky to yield site-specific L4_SM soil moisture estimates downscaled to 1 km. Once downscaled, the site-specific SMAP data were then used to investigate incipient failure conditions using infinite-slope stability models. These models were verified using known landslide occurrences in Kentucky. Further, the study herein presents a novel workflow for the creation of dynamic landslide susceptibility (i.e., susceptibility that change with changing soil moisture) map and soil moisture forecasting over a domain within Eastern Kentucky. Forecasting of soil moisture was conducted through the development of a Long Short-Term Memory (LSTM) forecasting machine learning approach. Forecasts of soil moisture were then assimilated into an infinite slope stability equation to provide forecasts of hazard analyses. These forecasted hazard analyses were investigated over known landslides with satisfactory results obtained.
Bio
Dr. L. Sebastian Bryson is the current Hardin-Drnevich-Huang Professor of Civil Engineering in the Department of Civil Engineering with a joint appointment in the Department of Earth and Environmental Sciences at the University of Kentucky. Dr. Bryson is also the current Department Chair for the Department of Civil Engineering. Dr. Bryson’s research interest focuses on applied geotechnics and includes field instrumentation and monitoring of constructed facilities, in-situ response of earth structures, performance prediction of supported excavations, and in-situ and laboratory testing of soils. Current research projects include: landslide geohazard assessment and prediction, deformation-based design methodology, and multi-hazard response of infrastructure systems. Dr Bryson is a Fellow in the American Society of Civil Engineers (ASCE). Dr. Bryson is also certified as a Diplomate Geotechnical Engineer (D.GE) by the Academy of Geo-Professionals (AGP) of ASCE in recognition for his specialized knowledge and skills, professional ethics, commitment to lifelong learning, and continued professional development in the field of geotechnical engineering. Dr Bryson received his Ph.D. from Northwestern University, his Master of Engineering degree from Howard University, and his Bachelor of Science degree from the Florida A and M University/Florida State University College of Engineering. Dr. Bryson teaches undergraduate courses in Geotechnical Engineering and Introduction to Research. He also teaches graduate courses in Advanced Soil Mechanics, Advanced Geotechnical Engineering, and Stability of Earth Slopes. Dr. Bryson is a registered professional engineer in the states of Kentucky, Wisconsin, Illinois, Ohio, Michigan, Indiana.
Light refreshments will be provided after the seminar.