Comment on "Hydrometeorological Triggers of Periglacial Debris Flows in the Zermatt Valley (Switzerland) Since 1864" by Michelle Schneuwly‐Bollschweiler and Markus Stoffel. Issue 3 (10th March 2022)

• Record Type: Journal Article
• Authors: Heiser, M.; Schlögl, M.; Scheidl, C.; Fuchs, S.
• Is Part Of: Journal of geophysical research. Volume 127:Issue 3(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-03-10
• Subjects: trends -- frequency -- seasonality -- completeness analysis -- time series -- data analytic question types; Geomorphology -- Periodicals; 551.3
• DOI: 10.1029/2021JF006562
• Languages: English
• ISSNs: 2169-9003

Abstract

Abstract: Generalizing properties of samples derived via data analysis to advance understanding of underlying processes constitutes the essence of statistical inference. This is especially true in geophysical research, where general findings often emerge from case studies. Consequently, statements about trends in case studies should naturally be grounded on a comprehensive and robust analysis of available data. Here, we focus on trends reported for a debris flow time series of a well‐known case study in the Zermatt Valley, Switzerland (Schneuwly‐Bollschweiler & Stoffel, 2012; https://doi.org/10.1029/2011jf002262 ). We elaborate that we were not able to confirm the findings of the original study with respect to changes in debris flow seasonality. On the contrary, we show that confusion between data analytic question types (exploratory vs. inferential) can lead to the confirmation of trends that do not necessarily hold beyond the data set, and we demonstrate how the reported effect could be an artifact of incomplete data. In doing so we employ a range of methods that can be used for adequately addressing the original question. Plain Language Summary: Mountain hazards such as debris flows repeatedly cause high loss in the European Alps. Time series of past (historical) events are often the most reliable information used for hazard mitigation. These time series, however, are subject to uncertainties in data collection and are likely to be incomplete. Taking a 150‐year data set derived in the Zermatt region, Switzerland, we show the pitfalls of confusing data analytic question types (exploratory vs. inferential) and of analyzing trends in time series without taking the completeness of the time series into account. Failure to avoid these pitfalls can lead to potentially erroneous conclusions about the drivers of changes in debris flow occurrence, such as climate change, land‐use change, or mitigation measures. Key Points: We comment on the implications of mistaking exploratory for inferential data analysis by means of an example We illustrate why the completeness of event time series is a crucial requirement for analyzing trends We outline approaches that foster drawing robust conclusions on temporal trends