High-fidelity detection of rockfalls in Yosemite Valley by coupling historical imagery with Structure-from-Motion photogrammetry and terrestrial laser scanning
Abstract
Rockfalls are a dominant control on rates of cliff erosion, but documenting their occurrence can be challenging, especially on large and/or remote cliffs. Repeat application of remote sensing methods such as Structure-from-Motion (SfM) photogrammetry and Terrestrial Laser Scanning (TLS) can detect a wide range of rockfall volumes, but typically these applications span only a few years and may not accurately reflect long-term rates. Inventory databases can provide longer-term rockfall records, but are commonly incomplete and prone to observation and volume assignment biases. We employed SFM and TLS on two adjacent 900+ m-tall cliffs (El Capitan and Middle Brother) in Yosemite Valley (California, USA), integrating frequent data collections from 2010 to 2017 with "historical" SfM models derived from oblique photographs taken in 1976. By comparing the 1976 SfM models against recent data, we detect 40 years of rockfall in high fidelity. Change detection reveals that 235 rockfalls occurred from the two monitored cliffs, more than twice as many events as are recorded in Yosemite's inventory database; most of the unrecorded rockfalls were <1 m3 in volume. Although individual rock fall volumes measured by SfM-TLS vary from those reported in the inventory database, measured cumulative volumes (16,653 m3 and 60,133 m3 for El Capitan and Middle Brother, respectively) are similar to reported volumes, likely because the large-volume events that account for most of the cumulative volume tend to be widely observed and reported. Robust volume-frequency relationships enabled by the SfM-TLS analyses indicate that the cliffs erode predominantly by less frequent, larger-volume rock falls, at rates of 0.9 to 1.7 mm/yr. Our study demonstrates how SfM models derived from historical imagery allow quantification of rock falls spanning several decades, thus improving inventory databases, informing rock fall hazard assessment, and recording longer-term rates of cliff erosion.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFMNH51B0776S
- Keywords:
-
- 1810 Debris flow and landslides;
- HYDROLOGY;
- 1826 Geomorphology: hillslope;
- HYDROLOGY;
- 4302 Geological;
- NATURAL HAZARDS;
- 4315 Monitoring;
- forecasting;
- prediction;
- NATURAL HAZARDS