How to improve dating quality and reduce noise in tree-ring based debris-flow reconstructions
Résumé
Tree rings have been used to reconstruct the occurrence of debris flows and other mass movements. Identification of past activity was typically based on the presence of growth anomalies in trees, with a focus on scars, stem tilting, trunk burial or apex decapitation. Clear guidelines have been missing so far and the dating of events has only rarely been based on thresholds so as to distinguish signal from noise. In a similar way, the spatial distribution of affected trees has not normally been considered in mass movement reconstructions, and was at best used as a subjective exclusion factor. This study therefore aims at improving dating quality of and reducing noise in debris-flow time series. Based on a dataset of 803 increment cores (385 trees) affected by debris flows, we reconstruct event histories using (i) a classical experts' approach, (ii) a weighted index (Wit) of responding trees as well as (iii) Moran's I and Getis–Ord Local Gi indices. We identify similarities and differences in results and then investigate subsets of the tree-ring sample to define ideal sampling positions on debris-flow cones and guidelines for sample depth.