AUTHOR=Martinez S. N. , Schaefer L. N. , Allstadt K. E. , Thompson E. M. 

TITLE=Evaluation of Remote Mapping Techniques for Earthquake-Triggered Landslide Inventories in an Urban Subarctic Environment: A Case Study of the 2018 Anchorage, Alaska Earthquake

JOURNAL=Frontiers in Earth Science

VOLUME=Volume 9 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/earth-science/articles/10.3389/feart.2021.673137

DOI=10.3389/feart.2021.673137

ISSN=2296-6463

ABSTRACT=Earthquake induced landslide inventories can be created using field observations, but doing so can be challenging if the affected landscape is large or inaccessible after an earthquake. Remote sensing data can be used to help overcome these limitations. The effectiveness of remotely sensed data to create landslide inventories, however, is dependent on a variety of factors, such as the extent of coverage, timing, and quality of the imagery, as well as environmental factors such as atmospheric interference (e.g. clouds, water vapor) or snow and vegetation cover. With these challenges in mind, we use a combination of field observations and remote sensing data from multispectral, light detection and ranging (lidar), and synthetic aperture radar (SAR) sensors to create a ground failure inventory for the urban areas affected by the 2018 M 7.1 Anchorage, Alaska earthquake. The earthquake occurred during late November at high latitude (~61° N), and the lack of sunlight, persistent cloud cover, and snow cover that occurred after the earthquake made remote mapping challenging for this event. Despite this, 43 landslides were manually mapped and classified using a combination of the datasets mentioned previously. Using this manually compiled inventory, we investigate the individual performance and reliability of three remote sensing techniques in this environment not typically hospitable to remotely sensed mapping. We found that differencing pre- and post-event Normalized Difference Vegetation Index (NDVI) maps and lidar data worked best for identifying soil slumps and rapid soil flows but not as well for the small soil block slides and rock falls that occurred. The SAR-based methods did not work well for identifying any landslide types because of high noise levels likely related to snow. Some landslides, especially those that were more subtle or didn’t leave lasting marks on the landscape, were identifiable only from the field observations. This work highlights the importance of the rapid collection of field observations and provides guidance for future mappers on which techniques, or combination of techniques, will be most effective at mapping landslides in a subarctic and urban environment.