Stream Length-gradient Hotspot and Cluster Analysis (SL-HCA) to fine-tune the detection and interpretation of knickzones on longitudinal profiles

Abstract This work presents a new approach for fine-tuning the analysis of stream longitudinal profiles. We show that applying Hotspot and Cluster Analysis (HCA), based on the Getis-Ord Gi* statistic, to the stream length-gradient (SL) index improves the visualization of anomalous values, assisting in the identification of tectonic structures and large landslides. High positive Gi* values indicate the clustering of SL anomalies (hotspots), and mirror the occurrence of knickzones on the stream long-profiles. We applied this methodology to a mountainous sector of the eastern Emilia-Romagna region, in northern Italy. Remote sensing and field surveys conducted on hotspot sites indicate that large landslides are the main process associated to over-steepened long-profile segments along streams connected to the valley slopes. Along-stream changes in bedrock resistance accounts for the main anomalies within sectors where hillslopes and valley floors are disconnected. We demonstrate that specific relationships between geometry and intensity of SL hotspots are indicative of the process responsible for the knickzone formation and, in particular that tectonic structures generally provide the longest and highest anomalies. The results of this work suggest that SL-HCA maps are more advantageous for detecting and interpreting knickzones compared with traditional SL maps, since: i) they need less input data to be computed, thus making them useful to investigate regions poorly covered by detailed geological data and/or where field surveys are difficult to be carried out and ii) the hotspot geometry can help discriminate the knickzones attributable to gravitational mass movements from litho-structural ones.