Predicting underwater landslides before they strike

Below ocean wind farms, oil rigs and other offshore installations are mammoth networks of underwater structures, including pipelines, anchors, risers and cables, that are essential to harnessing the energy source. But much like terrestrial structures, these subsea constructions are also vulnerable to natural events, like submarine landslides, that can hamper the productivity of installations below the sea.

Researchers at Texas A&M may now be able to accurately predict the occurrence of marine landslides using underwater site characterization data.

“One of the main events threatening onshore and offshore facilities is landslides. They can completely wipe out all these installations,” said Zenon Medina-Cetina, associate professor in the Department of Civil & Environmental Engineering. “We show in our paper that information from multiple disciplines in the correct sequence is needed to better understand the probability of landslide development at any place and time.”

The researchers have published their work in the journal Landslides.

Before any offshore project begins, such as oil and gas operations or wind farms, a team gathers information about the seabed, sub-seabed and environmental conditions. This site characterization helps to mitigate potential geohazards and informs the design, construction and installation of offshore structures.

This process involves the collaborative efforts of a number of personnel, including geophysicists, geomatic technologists, geotechnical engineers and geologists. Medina-Cetina’s model calibration methodology uses site characterization information to predict the occurrence of underwater landslides.

Although data from personnel with different expertise is needed to tell the story of the land below the sea, the order in which they perform their tasks towards site characterization is very important. This sequence, if violated due to budgetary or time constraints, could lead to uncertainty in the prediction of landslides.

“It is very important to start with the geophysicist and then bring in the geologist and then have the geomatics group working with the geotechnical engineers,” said Medina-Cetina. “As an analogy, imagine that I need to train a baby to walk while teaching it how to run. This is going to be much harder, right? A systematic sequence on the use of evidence ensures that the landslide models are better calibrated by learning from the data as they are being produced.”

The researchers noted that companies funding offshore projects typically lose money when they are not confident that the designs of the subsea civil infrastructures can withstand geohazards. Thus, the model calibration methodology introduced by Medina-Cetina and his team uses a probabilistic approach called Bayesian statistics to maximize the information produced in site investigation data. This methodology, they demonstrated, increases the accuracy and confidence of the landslide model when it makes predictions.

“My job is to make sure that under any geo-hazardous conditions, these offshore structures are going to be safe and are going to remain where they were designed to be,” said Medina-Cetina. “What we’re trying to say is it matters how you do the sequence of these site investigations, and how you integrate those data to train the landslide models, so that you can be more confident about the occurrence of potential submarine landslides.”

Other contributors to this research are Patricia Varela from Geosyntec Consultants, Inc and Billy Hernawan, a student from the civil and environmental engineering department at Texas A&M.