Application of satellite and airborne technologies for the development of probabilistic rainfall thresholds and susceptibility maps for landslides in Papua New Guinea

Landslides pose a significant risk to life and infrastructure in Papua New Guinea (PNG). The combination of rugged topography and high seismicity makes PNG highly susceptible to large-volume, earthquake-induced landslides, while the climate encourages widespread rainfall-induced landslides. Of the two triggering mechanisms, rainfall offers the best opportunities for the development of early warning/forecasting systems, as meteorological models continue to improve in skill and resolution. To understand the relationships between rainfall and landslides, studies have conventionally looked to develop landslide-triggering rainfall event thresholds. Such thresholds can form the basis for early warning/forecasting systems. In this thesis, satellite-based precipitation estimates from the Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis (TMPA) are used to examine characteristics and relationships between rainfall and landslides. These data, in conjunction with Bayesian statistical approaches, were then used to develop landslide probabilities based on rainfall events of varying duration and magnitude. Understanding the landslide triggers is only one aspect of slope instability. Environmental control factors, such as slope or curvature, can enhance or reduce the likelihood of slopes to fail. Therefore, multispectral imagery and high resolution GeoSAR (synthetic aperture radar) digital elevation models have been exploited to verify and map landslide scars in different regions of PNG. These data also support the development of landslide susceptibility maps, providing detailed information on the terrain and structures important for slope instability. Using fuzzy relation-based assessments these data were used to produce landslide susceptibility maps which differentiate areas of low/no susceptibility from those with high susceptibility. Overall, the satellite and airborne techniques have provided the tools to understand landslide occurrence relative to changes in rainfall and susceptibility, which can form the basis of early warning/forecasting models.