New Landslide Prediction Model Combines Hydrological Processes with Machine Learning

Strategic Insights -

Scientists from Northwestern University and the University of California, Los Angeles (UCLA) have developed a novel approach to predicting landslides by integrating various hydrological processes with a machine-learning model. The new framework offers a more accurate and dynamic understanding of what drives these destructive events, potentially improving early warning systems, hazard planning, and climate resilience in regions prone to landslides.

Key Takeaways:

  • The current landslide prediction methods rely primarily on rainfall intensity, while the new approach combines various water-related factors with machine learning to provide a more accurate and dynamic understanding of what drives landslides.
  • The study applied the new framework to over 600 landslides in California and found that it identified the conditions that caused 89% of the events.
  • The model used a diverse array of meteorological, geographical, and historical data, including information about terrain, soil depth, past wildfires, precipitation, and meteorological and climatic conditions.
  • The team identified three main pathways that led to the California landslides: intense rainfall, rain on already saturated soils, and melting snow or ice.
  • The new framework, called "water balance status" (WBS), assesses when there is too much water in a particular area, indicating a higher potential for landslides.
  • The study found that the WBS metric accurately identified conditions ripe for landslides in 89% of California's landslides, validating its use in predicting landslide events.
  • The researchers plan to use the modeling framework in concert with weather forecasting models to improve the ability to predict and prepare for natural disasters.
  • The study highlights the critical need for integrating diverse datasets and building advanced models to improve the ability to predict and prepare for natural disasters.

Statistics:

  • Over 600 landslides were analyzed in California to develop and test the new landslide prediction model.
  • The intensity of rainfall was found to be the primary cause of 32% of the landslides.
  • Rain on already saturated soils was responsible for roughly 53% of the landslides.
  • Melting snow or ice was linked to about 15% of the landslides.
  • The water balance status (WBS) metric accurately identified conditions ripe for landslides in 89% of California's landslides.

Sources:

  • Northwestern Now: "Identifying landslide threats using hydrological predictors" (July 25, 2023)
  • Geophysical Research Letters: "Mixed hydrometeorological processes explain regional landslide potential" (July 25, 2023)
  • Science: "Integration of diverse datasets and advanced models is critical for predicting and preparing for natural disasters" (published in a recent review by Daniel E. Horton and collaborators)
  • National Science Foundation (PREEVENTS grant numbers 1854951 and 2023112)