How do surface processes shape the landscapes in which we live? Is it the every-day flow of rivers that gently, yet persistently, erodes and transports sediment from highlands to ocean basins, dissecting the land surface into networks of ridges and valleys? Or is it cataclysmic events of incredible magnitude that, despite their infrequency, conspire to shape Earth’s surface? These questions highlight the debate over the relative importance of extreme events in sculpting Earth’s surface, and are as old as the science of geology. Although geologists have gathered data and proposed theories supporting both Hutton’s (1795) and Lyell’s (1830) uniformitarianism and Cuvier’s (1818) catastrophism for over 200 years, the paper by Anderson et al. (2015, p. 391 in this issue of Geology) shows that the debate is still active and that, even with new tools, we have much to learn about the degree to which observations of modern sediment transport processes quantify the full range of formative geomorphic events.

If infrequent, extreme events dominate the long-term (103–106 yr) rate of erosion and sediment transport, then historic records of sediment flux that have not captured an extreme event might grossly underestimate the actual long-term sediment flux from a landscape (Kirchner et al., 2001; Carretier et al., 2013). Misunderstanding such a discrepancy between modern and long-term erosion rates can lead to inaccuracies in: predicting the life span of reservoirs; determining the impact of changing land use; setting attainable water-quality standards; and mitigating sediment-related hazards, such as rapid mass movements like landslides or debris flows, and extreme river channel aggradation. At longer time scales, accurate portrayal of the magnitudes and spatial-temporal patterns of sediment fluxes is critical for understanding how landscapes evolve, how sediment fluxes might change with a changing climate, and what flux of sediment and nutrients is required to maintain healthy ecosystems.

Debris flows and related rapid mass movements are perfect archetypes of extreme events, as too clearly demonstrated by the recent mass movement disaster in Oso Washington, USA (Iverson et al., 2015). Debris flows are gravity-driven mixtures of mud, rock, water, and other incidental debris that flow down slope at speeds that can exceed tens of meters per second (Iverson, 1997). They commonly initiate in response to intense or prolonged precipitation events as either shallow landslides that liquefy upon initiation of motion (Iverson et al., 1997) or from intense surface-water runoff that rapidly mobilizes large quantities of sediment (Cannon et al., 2001; Kean et al., 2013). But quantifying the importance of rare, extreme events including large debris-flows is complicated by the obvious factors of infrequent, sporadic occurrence, and by the hazardous nature of trying to observe them directly.

Critical to resolving the importance of infrequent large events in setting the pace of landscape evolution are accurate measurements of both net sediment transport occurring during an extreme event, and of the long-term erosion rate. Anderson et al. managed to obtain both these crucial pieces of information for an extreme precipitation event that triggered over 1100 landslides and debris flows in the Colorado Front Range (Colorado, USA; Coe et al., 2014). Their strategy called for the use of tools that simply didn’t exist 20 years ago: (1) multi-temporal meter-scale resolution digital elevation models obtained from airborne lidar (Glennie et al., 2013), and (2) millennial-scale denudation rates determined from the concentration of 10Be in sediment (Granger et al., 2013). By differencing the digital elevation models obtained before and after the event, Anderson et al. were able to quantify the net volume of sediment evacuated from the landscape by debris flows, as well as the equivalent basin-average lowering depths. Using recently published long-term erosion rates from the Colorado Front Range, the authors were then able to calculate how many years of hillslope erosion and transport it took to produce the regolith evacuated by the debris flows.

The intriguing result is that, in a single event, debris flows transported hundreds to thousands of years worth of accumulated hillslope material into the main stem rivers. From this result, they reached the conclusion that debris-flow erosion and transport do the majority of the geomorphic work in the studied steep valley networks, despite the low frequency of their occurrence.

The results and conclusion from Anderson et al. compare favorably with other studies that have attempted comparisons between net sediment transport from large debris-flow events and long-term erosion rates. Eaton et al. (2003) compared net sediment transport by debris flows initiated during hurricane strikes in the Appalachian Mountains (eastern United States) to decadal sediment fluxes. They found that, even with thousand-year recurrence intervals, as determined from radiocarbon dating of hillslope material in headwater valleys, debris flows still transported a majority of the sediment out of the steep valley networks. Similarly long recurrence intervals and dominance of sediment transport by debris flows were found for small basins in the Coast Ranges of the western United States (Dietrich and Dunne, 1978; Benda, 1990; Reneau and Dietrich, 1991; Lancaster and Casebeer, 2007).

Taken together, a clear picture is emerging. In steep valley networks, debris flows can erode and transport the majority of weathering products produced by the landscape. What is less clear from these studies of sediment budgets in steep lands is whether the episodic passage of debris flows contributes directly to incision of bedrock-floored valleys by impact wear versus simply accelerating bedrock weathering in valley bottoms by removing thick regolith cover. As noted by Anderson et al., most debris-flow paths in the Colorado flood area were scoured clean to bedrock. Similar qualitative observations of apparent bedrock scour by debris flows are commonly reported in valleys with slopes steeper than ∼5° (Stock and Dietrich, 2003, 2006; Hsu, 2015). In one four-year monitoring study, measurements of bedrock incision made in a steep bedrock-floored valley demonstrated that debris-flow scour caused cm-scale bedrock lowering (McCoy et al., 2013). In many steep lands, valleys with gradients steeper than 5° can make up the dominant portion of the valley network relief and network length (Fig. 1; Stock and Dietrich, 2003). This implies the incision of ridge-valley topography in unglaciated mountain ranges might not be derived purely from rivers, but also by debris flows.

Significant consideration has been given to bedrock incision by rivers because it is thought to set the relief of mountain ranges, as well as the pace of landscape evolution, by transmitting tectonic and climactic changes throughout the landscape (Whipple and Tucker, 1999). But there is no agreed-upon mechanistic framework to describe the controls of bedrock incision by debris flows, unlike rivers, which in turn raises questions about the accuracy of predictions regarding the pace and spatial pattern of steep land evolution from models that do not consider the effects of episodic debris flows.

Over the past decade, rapid progress has been made in understanding event-scale debris-flow mechanics needed to formulate and test mechanistic theories depicting the integrated effect of debris-flow scour over hundreds to millions of years. Field monitoring, laboratory experiments, numerical modeling, and theory have demonstrated important mechanics controlling debris-flow initiation (Baum et al., 2010; Kean et al., 2013), entrainment and transport of sediment (Iverson et al., 2011; McCoy et al., 2012; George and Iverson, 2014), and the expected distribution of impact forces at the base of debris flows (Yohannes et al., 2012; McCoy et al., 2013; Hsu et al., 2014). With the ever-expanding coverage of high-resolution topographic data, our increasing knowledge of debris-flow mechanics, and the continued clever use of new tools as demonstrated by Anderson et al., we should continue to produce testable hypotheses on the role of debris flows and other extreme events in controlling steep land evolution.