Talks - Eric Matthes: Using Python to assess landslide risk: A matter of life and death

Learn how Python analyzes stream gauge data to assess landslide risks in Southeast Alaska, providing critical early warnings and helping communities make informed safety decisions.

Key takeaways
  • Project analyzes landslide risk in Southeast Alaska using stream gauge data to monitor watershed conditions during heavy rainfall events

  • Key implementation focuses on two main factors:

    • Total river rise of at least 2.5 feet
    • Sustained rate of rise of 6 inches per hour
  • System provides 30-60 minute advance warning of potential landslide conditions with relatively few false positives

  • Approach complements official forecasts by helping residents understand when current conditions match historical landslide events

  • Stream gauge serves as an effective proxy for soil moisture content and overall watershed conditions

  • Django web framework enabled transition from private analysis tool to public monitoring system

  • Project validated through historical analysis:

    • 9 critical periods identified over 5 years
    • 3 periods associated with actual landslides
    • 6 false positives
    • 35-60 minute warning times
  • Tool helps community members:

    • Know when landslides are unlikely
    • Make informed decisions during heavy rain
    • Reduce anxiety about conditions
  • Limitations include:

    • Only works for small watersheds
    • Requires reliable stream gauge data
    • Not validated by professional geologists
    • Short notification times (15-60 minutes)
  • Project demonstrates how Python tools can create meaningful impact with relatively simple implementation using standard libraries