Inferring precipitation thresholds of landslide activity from long-term dendrochronological and precipitation data: Case study on the unstable slope at Karpenciny, Poland

Abstract

An extended statistical comparison of numerous precipitation parameters (absolute, minimum, and maximum totals; number of days with precipitation; and duration of precipitation in particular months, seasons, and longer periods) and the 48-y long dendrochronological reconstruction of landslide activity was conducted for a generally unmonitored landslide slope at Karpenciny, Western Carpathians, Poland. The Karpenciny landslide is a deep (approximately 40 m) translational landslide consisting of large, poorly fragmented, and almost intact landslide blocks. It is located only 2.4 km from the meteorological gauging station. We aimed to explore the long-term dependence of this landslide on precipitation conditions and reveal the precipitation parameters crucial for its activity, that is, direct triggers and preparatory antecedent precipitation. Therefore, dendrochronological records of tree-ring eccentricity and compression wood in 35 Norway spruce trees were correlated with 520 precipitation parameters (including seasonality, duration, and total), which might influence the activity of the studied landslide. Ten best-correlated precipitation parameters and seasonal sums of summer half-years and preceding winter half-years (separately and in total, i.e. as 12-month sums), were then tested for their potential in establishing precipitation thresholds for landslide activity. The analysis of multiple precipitation parameters enabled us to develop precipitation thresholds based on both direct causes and long-term preparatory precipitation necessary to accelerate the studied landslide. From the tree-ring data, we established landslide-safe conditions and 0.5-probability thresholds for a particular landslide slope. This study also generated comprehensive data on the seasonality of antecedent and triggering precipitation, long-term periods of antecedent precipitation, critical minimum levels of precipitation that sustain slope imbalance, and the importance of generally wet conditions (demonstrated as a high total number of days with precipitation regardless of its totals) as a factor in landslide activity. These patterns of landslide-precipitation dependence would not have been revealed with standard methods and standard precipitation parameters applied in threshold analyses. Thus, long-term reconstructions can be a valuable source of data for precipitation thresholds of landslide activity.