Off-campus Michigan Tech users: To download campus access theses or dissertations, please use the following button to log in with your Michigan Tech ID and password: log in to proxy server

Non-Michigan Tech users: Please talk to your librarian about requesting this thesis or dissertation through interlibrary loan.

Date of Award

2020

Document Type

Campus Access Master's Thesis

Degree Name

Master of Science in Geological Engineering (MS)

Administrative Home Department

Department of Geological and Mining Engineering and Sciences

Advisor 1

Snehamoy Chatterjee

Committee Member 1

Thomas Ooommen

Committee Member 2

Xin Xi

Abstract

Precipitation is a key triggering factor for landslides. In Colorado, precipitation triggered 75% of the landslides between 2007 and 2017, as recorded in NASA’s Cooperative Open Online Landslide Repository (COOLR). Often, landslide hazard characterization at regional or statewide scales uses sparse rain gauge measurements or assumes constant precipitation. This study explores the need for improved precipitation measurements for landslide hazard characterization. Identifying spatial and temporal precipitation patterns allow for improved understanding of their influence on geohazards, like landslides. Rain gauges collect in situ point data, while satellites gather areal precipitation data remotely with large spatial coverage. Integrating point rain gauge observations with downscaled areal precipitation data can improve the overall spatial resolution of precipitation maps. The areal precipitation product being downscaled is the monthly Integrated Multi-satellitE Retrievals for Global Precipitation Measurement (GPM) data (IMERG) for Colorado for January, April, July, and October 2018. The spatial resolution of the product is improved tenfold from 0.1° x 0.1° to 0.01° x 0.01°, using area-to-area kriging (ATAK) based on an areal variogram deconvolution algorithm. The rain gauge monthly precipitation is interpolated with the downscaled satellite precipitation products using a simple cokriging algorithm, and the results are assessed for conditional bias with leave-one-out cross-validation (LOOCV). Error statistics are calculated to assess the relationship between rain gauge observations and the LOOCV results. The LOOCV error statistics are compared to error statistics between the coarse-scale GPM data and rain gauge observations to demonstrate the models improve precipitation estimates. Applying the models to Colorado landslides, it is found that Spring and Summer 2017 GPM average accumulated precipitation within seasonal landslide zones is statistically different from outside the seasonal landslide zones in Colorado’s mountains with a 99% confidence limit. When integrated with rain gauge measurements, the average accumulated precipitation at improved spatial resolution is also significantly different with a 99% confidence limit from the coarse-scale GPM data for April and July 2018 in Colorado’s mountains. This research suggests improved precipitation maps could influence landslide hazard characterization.

Share

COinS