The sensitivity of dendrogeomorphic approaches to assessing landslide movements

Abstract

An understanding of past landslide behaviour is crucial for the estimation of future landslide activity. In this respect, methods of absolute dating are frequently used for landslide history assessment. Dendrogeomorphic (tree ring-based) approaches are standard methods for the spatio-temporal reconstruction of landslide activity with annual to even sub-annual resolution and timespans of several centuries (depending on the tree age). Nevertheless, the verification of the effectiveness, sensitivity and precision of these dendrogeomorphic approaches has not been addressed in more detail despite their wide spectrum of limitations. This study makes an assessment of tree ring-based chronologies of landslides based on a comparison with the independent data of past landslide activity from instrumental monitoring. The chronology of past landslide events was created from 528 tree ring series from 132 individuals of Norway spruce (Picea abies (L.) Karst.) growing on a monitored landslide in the Outer Western Carpathians (the Ondřejník Mt.) and was expressed as the yearly variation in the standard event-response index. The landslide was monitored at two sites via technical levelling (since 2007) and an automated wire extensometer (since 2014). The obtained results show a high sensitivity of reaction wood formation to movements on the order of several mm (e.g., the abrupt acceleration of movements in February 2017, as recorded by the extensometer). Moreover, a one-year delay in the growth response to movement occurring in the middle part of the growing season of the previous year was detected. This fact fully supports the introduction of the dendrological year to dendrogeomorphic research. The partial lack of correspondence between tree ring records and data from more distant monitoring sites suggests the need for separate dendrogeomorphic analyses of the individual homogenous parts of landslide areas. Finally, the use of four instead of two increment cores per tree did not provide any extra event years but increased the values of the event-response index of the dated events.