Abstract
The county of Sogn og Fjordane in Western Norway has a climate and topography which makes it vulnerable for debris avalanches, debris flows, shallow slides and slush flows. The recently established landslide forecasting and warning service at Norwegian Water Resources and Energy Directorate (NVE), at www.varsom.no, is able to predict their regional, spatial and temporal occurrence. The landslide warning levels, which goes from 1 to 4, suggest an expected outcome pursuant to the upcoming hydro-meteorological event. These levels are defined based on the expected number of landslides that will occur in a warning area with an extension of 10.000-15.000 km2. Expected magnitude of the landslides is also included within the definition of warning levels in a qualitative and very general way. For example, a level 3 warning will indicate that “Large landslides that disturb infrastructure and roads may occur”. However, NVE has throughout the first 5 years of the operation observed that, depending on the region, not always only “large” landslides occur during a level 3 warning. It may happen that many small landslides occur that can create severe damages and serious disruptions to the society in that region. The Norwegian landslide database is quite rich with a great number of events registered through many years. However, the events do not contain landslide magnitude information. In this study, the following questions are addressed: How large are the rainfall-induced landslides in this region, and which landslide magnitude is the most frequent? And how can landslide magnitude become incorporated within the definition of the regional landslide warning levels? This thesis aims to find the typical and frequent landslide magnitudes in Sogn og Fjordane by performing a magnitude-frequency analysis. It will also propose a way of incorporating the results into the landslide warning levels and investigate how it can be communicated to public and authorities. A cumulative distribution was applied to investigate the relationship between magnitude and frequency. The results revealed landslides to have an extension of 147-123228 m2 in the region, with the most frequent magnitude being 10000 m2. An incorporation of magnitudes was accomplished by investigating magnitudes from specific events as well as considering their associated warning level. A “small” landslide is proposed to have a range from 0-10.000 5 m2, a “medium” landslide as 10.000-50.000 m2 and a “large” landslide to be greater than 50.000 m2. My findings show that an event with level 4 warning should expect an unusual high number of initiated events at all scales as well as several large magnitude landslides. A warning level 3 should expect multiple landslides with medium magnitude, also with likelihood of dealing with a large landslide. At last, a level 2 event should consider a single event with medium magnitude. This work is part of an ongoing project that aims to map more systematically the magnitude of recent landslide events at national level, because a better understanding of these processes, their spatial distribution, dimensions, mechanisms and frequency are needed to improve the performance of the landslide forecasting and warning service, especially at local scale.