Northwestern University and the University of California, Los Angeles (UCLA) have introduced a new framework to predict landslides with greater accuracy. This approach integrates various water-related processes into a machine-learning model, moving beyond traditional methods that focus solely on rainfall intensity.
The research, published in Geophysical Research Letters, emphasizes the need for comprehensive data to understand the factors driving landslides. Chuxuan Li, first author of the study, explained that existing early warning systems rely heavily on historical precipitation data and do not account for changing climate conditions. "Our model considers a wider range of factors," Li stated.
Daniel E. Horton from Northwestern's Weinberg College noted that different hydrological processes could trigger landslides. The goal is to develop tools useful across broad regions like California.
The team simulated extreme weather conditions experienced in California during the winter of 2022-23 using a computer model that assesses how water moves through various environmental processes. They developed a metric called “water balance status” (WBS) to determine when an area has excess water.
Through machine learning techniques, they identified three primary pathways leading to landslides: intense rainfall, rain on already saturated soils, and melting snow or ice. Their findings revealed that 89% of landslides occurred in areas with positive WBS values.
Horton emphasized the importance of forward-looking models in light of climate change impacts on storm intensity and frequency. In Science journal's recent review, Horton highlighted how atmospheric rivers often lead to chain reactions triggering other natural disasters.
The study received support from the National Science Foundation under grant numbers 1854951 and 2023112.