Divergent patterns of landslide activity and triggering factors at a local scale of a single mountain massif (Island Beskid Mts., Western Carpathians, Poland)

Landslide protection and mitigation are critical issues in many regions worldwide, where lives and livelihoods depend on correctly determining landslide hazards. However, local differences in the response of landslide slopes to meteorological or seismic triggering factors can decrease the performance of landslide models and forecasts. Thus, recognising and understanding spatial patterns of landslide activity at a local scale should be one of the priorities of landslide research. In this study, we analyse the activity and triggering factors of two landslides in one mountain massif, <2 km apart, in the Western Carpathians, Poland. We analysed 61-year-long (1959–2019) records of the activity of study landslides, i.e. dendrochronological reconstruction based on ring eccentricity and compression wood dated in 65 specimens of European silver fir (Abies alba). We compared the dendrochronological proxy of landslide activity at two study slopes with precipitation and earthquakes in the study area. Based on statistical correlations with 560 precipitation indicators, we found that the similarities between the two study landslides include only the direct triggers in June–July and March. The Wiśnia landslide, unlike Hajdowska, can also be triggered by precipitation in October. However, the main difference between the study landslides lies in the preparatory precipitation. Hajdowska shows significant dependence from precipitation of the previous December and current February, while at the Wiśnia study site, the preparatory precipitation is less important and covers the previous October–November and current January. Our results show that although study landslides could be reactivated at one time by similar direct triggers of spring or summer, such simultaneous reactivation would require preconditions of antecedent precipitation of the previous winter half-year (October–March) different for each landslide. Suppose, the requirements of antecedent precipitation are fulfilled only for one of the landslides during the winter half-year. In that case, the other will remain inactive despite the later occurrence of a shared direct trigger in spring or summer. The example of Hajdowska and Wiśnia shows that significant differences in landslide activity, hazard and triggering factors can exist even at the local scale. We argue that omitting even such subtle, unevident differences in the conditions of landslide activity leads to errors in landslide modelling and forecasting. For example, linking Hajdowska and Wiśnia landslides with one precipitation threshold or model of slope instability based on their adjacent location would not be effective in predicting reactivations. The precision of landslide modelling and forecasting depends on the accuracy of preceding analyses of landslide activity and triggering factors. We argue that in the case of the two adjacent landslides studied the divergent patterns of their activity and triggers would not be revealed without analysing long-term (> 60 years long) data sets.