Forest management history influences eight decades of shallow landsliding in the northwest Cascade Mountains, USA

Abstract

Forest management‐driven increases in landsliding may impact aquatic habitat and water quality, yet it is difficult to disentangle the effects of hydroclimate from evolving land‐use practices when interpreting landslide rate fluctuations over decadal timescales. We compiled shallow landslide inventories starting in the 1940s from nine small watersheds in the northwest Cascade Mountains, USA, and updated the mapping through 2019. We stratified the landscape by management regime (industrial/state and federally managed forests and forests with no history of management) and compared landslide volume generation rates against timeseries representing regional climate, precipitation and river discharge, timber harvest rate, and instability ratings from the topographic model SHALSTAB. Results demonstrate a bell‐shaped temporal trend in composite landsliding that peaked in the 1970s and 1980s and is unlikely to have arisen from random variation alone. The signal from unmanaged forests correlates with regionally warm/wet climatic conditions at decadal scales, but we found mixed quantitative support for storm hydrology in explaining the landslide trends. Landsliding in federal managed forests was greatly reduced following cessation of clearcut logging in the mid‐1990s in most basins. In contrast, topographic and stand age modelling suggest the private industrial and state forest trend is consistent with reduced landsliding resulting from regulations requiring hazard avoidance in logging unit design—not changes to logging rate—demonstrated by a decline in probability of recent clearcuts on highly unstable terrain relative to more stable terrain following the 1980s in most basins. The analysed covariates could not explain all trends in all watersheds, suggesting the presence of methodological or data‐based limitations in the analyses. Late 20th century landslide sediment delivery acutely impacted tributaries but likely became muted downstream due to lag effects and mixing with other sources; our results suggest that proximal sediment supply impacts may be greatly reduced.