Managing Rockfall Raveling Using Ground Based InSAR Velocity and Inverse Velocity Data

TThe raveling of rocks, like landslides, requires acceleration to fail. However, the onset of acceleration to the time of failure may occur in seconds or possibly a few minutes, and highlights one of the difficulties of anticipating and predicting rockfall raveling events. The other difficulty is the small size of the rocks and boulders, making them difficult to resolve with current ground-based radar technologies. There are, however, instances where ground-based radar can provide a level risk for possible small rockfall and raveling events based on hourly observations of velocity data.

The following slope failure event describes the ongoing and daily onslaught of rockfall that construction workers faced, Their concern and request for a solution initiated the development of a rockfall level of risk matrix based on velocity and inverse velocity data.

During the winter of 2017, northern and central California experienced excessive precipitation, which resulted in flooding and numerous landslides across the state. Several landslides of various sizes occurred on the steep slopes of Coast Ranges in central California. By the late spring, many of the unstable slopes were seeping significant amounts of water. On Highway 1 in the Big Sur area of California at a location called Mud Creek, an unstable water seeping slope was deforming and producing abundant rockfall and eventually closing Highway 1 several weeks before the slope collapsed, as shown in the image below.

After the collapse of 70 million cubic meters of rock, an engineering plan started immediately to begin the reopening of Highway 1. The aftermath of the collapse exposed about 600 hundred feet of a near-vertical head scarp surface that was still relaxing and producing daily abundant raveling. Slope monitoring radar was requested to reduce risk from additional slides. On June 19, 2017, about a month after slope collapse, and after the development of a site-specific trigger action response plan (TARP), radar monitoring began.

The image below is a ground-based InSAR velocity map for a 24-hour interval on July 12 and July 13. Similar to the radar's displacement, map velocity rates are color-coded to the values shown on the color bar located on the right side of the image. Similar to pixels in a digital image, the radar scans an entire slope composed of DTM cells, which are overlain on the DTM. Displacement is then calculated for each cell and color-coded according to user preferences. The image below contains tens of thousands of DTM cells.

Nearly 2-months after the failure, the head scarp was still showing significant movement. Many of the velocity hot spots shown below would increase in rate quickly then disappear only to have additional hot spots appear at different locations on the head scarp. The velocity hot spots were typically 2 to 3 DTM cells, which are about 11 square feet each. Most of the hot spots approaching or achieving high rates appeared to be the culprits of raveling and small rockfall. To help verify this, we had the construction site spotters document the approximate time and location of raveling and rockfall. We then correlated their locations to the velocity map. The exercise is quick and easy since the software allows adjusting the velocity map to any interval of time while active scanning and map updates are occurring at the same time.

Once we identified the general location of raveling and small rockfall events, we then zoomed in at a scale that allows for the observation of individual DTM cell. Doing this allows for the creation of individual 2 to 3 DTM cell time series areas to confirm cumulative displacement trends, and inverse velocity's approach to linearity. In addition to the time-series data, the maximum rate of velocity and often, the time of failure for these minimal areas could be determined through this back-analysis method.

The accumulated hot spot velocity data were analyzed and used to create a matrix describing levels of rockfall risk below. Every morning before the start of construction activities, hot spot locations were reviewed with the site engineer through shared screens via the internet; this allowed the site engineer and monitoring professional to discuss risk levels concerning current work locations. Also, the radar data was reviewed often during the day, and any new hot spot acceleration was reported to the site engineer.

The data values listed in the matrix above are unique to the Mud Creek site and should not be used at sites with different rock types, rock strengths, or other unique site-specific characteristics. Also, the individual values listed here should not be used for anticipating or predicting slope failures since there are other, more reliable methods discussed in my other slope monitoring posts.