Dendrogeomorphic dating vs. low-magnitude landsliding

Abstract

Landsliding is a major natural hazard; therefore, understanding its activity is an important objective worldwide. For the investigation of the current landslide events, dendrogeomorphic methods are commonly used as they allow quite a precise dating of individual events. Nevertheless, there is still a question of whether dendrogeomorphic methods can successfully work for landslides with low-magnitude movements. To determine their usability and shortcomings, four commonly used dendrogeomorphologic approaches were put to the test: Approach I is based on the detection of abrupt growth suppression; Approach II is focused on the determination of compression wood; and Approaches III and IV that work with eccentric increments. For the detection of landslide events itself, I_t index thresholds and spatial statistics were used. In total, 329 individuals Picea abies (L.) Karst. Growing on a seemingly dormant landslide in the Outer Western Carpathians were sampled and processed. Overall, obtained landslide chronologies varied considerably although the same trees were used, which allowed pinpointing of the main limitations of each approach. Approach I showed a high sensitivity to water shortages, causing false noise signals. Approach II was not sensitive enough to low-magnitude movements. In contrast, Approaches III and IV recorded many possible landslide events, but most of the events were just noise signals induced by creep movements (and non-geomorphological influences). These conditions made it impossible to filter the real landslide events based on I_t index thresholds; however, as a substitute, spatial statistics combined with a detailed analysis of real landslide morphology were successfully used. Last but not least, a sensitivity of trees to record possible landslide movements in various stages of their lives was analysed for each approach. Except for Approach IV, all approaches showed high variability in changing sensitivity during a tree's life; thus, during a certain period of the tree's growth, landslide events can hardly be detected. All things considered, findings in our study are crucial for the strategy of the dendrogeomorphic sampling conducted on landslides with low-magnitude movements.

Introduction

The assessment of past landslide activity is actually one of the top aspects of landslide research (Pánek et al., 2008; Du et al., 2020; Wood et al., 2020). Detailed chronological information about past landslide events is crucial for the determination of possible landslide triggers (Corominas and Moya, 1999), assessment of climate relationships (Lopez Saez et al., 2013), modelling of future landslide development (Stoffel et al., 2014) and landslide hazard assessment (Klimeš and Blahůt, 2012). Unfortunately, data about past landslide behaviour are usually very rare or totally absent. The reasons for this state are the incompleteness of historical archives (Raška et al., 2015), selective human memory (Mayer et al., 2010), and the remoteness of landslide areas from eyewitnesses. This state calls for an extensive use of various methods of absolute dating (Pánek, 2015). Nevertheless, radiometric approaches (AMS ¹⁴C, CRN, OSL) enable to obtain very long chronologies, but the precision is usually not adequate for the abovementioned purposes.

From the precision of a chronological range ratio point of view, dendrogeomorphic (tree ring based) methods are the optimal solution (Alestalo, 1971). The applicability of these methods covers all regions where landslides are overgrown with trees and climatic conditions enable creation of annual tree rings. By using the tree rings, it is possible to detect from several centuries (Šilhán et al., 2012) up to a millennium (Zhang et al., 2019) of old landslide events. Dendrogeomorphic methods were almost exclusively used to detect past landslide movements of distinctly active landslides (Šilhán, 2020a, 2020b). However, most landslides probably move in very low-magnitude events that give the impression of their recent dormancy. Pawlik et al. (2013) suggested a high sensitivity of trees to slope movements. Subsequently, Šilhán et al. (2019) verified this hypothesis and provided evidence about the ability of trees to record landslide movements in the order of mm via formation of reaction wood. This finding suggests the possibility of dendrogeomorphic method application to even seemingly inactive landslides with a very limited magnitude of movements. Thus, the possibility is now open to test this assumption (that has not yet been verified) and to extend the spectrum of dendrogeomorphic processes datable by a tree-ring based approach.

Various approaches of tree-ring based landslide dating were historically developed (Šilhán, 2020a) including analysis of reaction wood, abrupt growth suppression or tree ring eccentricity. Šilhán and Stoffel (2015) compared results from the analysis of reaction wood with results obtained from one of the methods working with tree ring eccentricity, and Šilhán (2019) compared results obtained from various tree-ring eccentricity based approaches. The results of these studies provided various ability of each approach to detect landslide movements. Thus, testing to determine the effectivity of various approaches for low-magnitude landslide events detection is desired. This research is of particular importance because the analysis of slope movements focused on landslides is often complicated by the presence of another type of movements. Specifically, creep movements can affect landslide slopes and induce the same growth responses in trees as high magnitude landslide movements (e.g., m. year⁻¹) (Pawlik et al., 2013). Distinguishing between landslide- and creep-induced growth disturbances is practically impossible based only on tree ring analysis. Šilhán (2017) tried to extract both sources of slope movement signal based on the position of analysed trees within landslide morphological zones. Although the presented approach expressed some limits, an analysis of spatial patterns of affected trees seems to be a good option for distinguishing creep-induced signals from resulting tree-ring chronology. Landslide events (including the low-magnitude ones) generally express spatially concentrated reactivations (Van Den Eeckhaut et al., 2007; Rosone et al., 2018) whereas creep movements generally express a spatially dispersed or random pattern (Saunders and Young, 1983; Pawlik and Šamonil, 2018). Thus, the application of spatial statistics of affected trees could help distinguish the effects of both processes on the resulting chronology of slope movements.

Finally, as analysed trees are living organisms, it is necessary to take into consideration their development in time connected with physiological changes. These aspects of trees lead to key changes in ability to record a landslide signal within their tree ring series. Some pioneering studies presenting first insight into this phenomenon have already been published (Šilhán and Stoffel, 2015; Šilhán, 2016, 2019). The first results generally confirm the expected age-dependent sensitivity of trees to landslide movements and suggest sensitivity depending on the method used. Nevertheless, some complex test of the effect of the method used to detect tree sensitivity to slope movements has not yet been realized.

Based on the presented possibilities and challenges of dendrogeomorphic dating of seemingly dormant landslides that truly move in the form of very low-magnitude movement events, the main aims of this paper are as follows: i) to compare chronological results obtained by using various tree-ring based approaches and to provide an overview of their advantages and limitations for these specific types of landslides, ii) to test the possibility of spatial pattern analysis of affected trees to distinguish between low-magnitude landslide and creep-induced signals, and iii) to assess the changes of age-dependent sensitivity of trees to low-magnitude movements (landslide and creep) induced by the use of various dendrogeomorphic approaches. Solving these mentioned aims should extend the applicability of dendrogeomorphic methods to many localities with the presence of seemingly inactive landslides.