It is unlikely that low frequency, episodic events that trigger landslides and debris flows can be found in "typical hydrological years". Our paper on sediment from a small ephemeral gully indicated the mechanisms to erode and deliver sediment correlated years with extreme flows. Roughly 5 named Tropical Storms were the episodes that produced conditions to deliver sediment from a small hydrologic unit in South Carolina. From my years in Oregon coast range, it was both severe rainfall events and sometimes activities like road building or clearcutting that contributed. Severe wildfires were another contributor. Because of this, we had several geologists and a hydrologist to help identify conditions relative to geology contacts and hyper steepness of terrain. There were instances when we avoided roading or clearcutting in sensitive terrain, or used special mitigation measures to end haul fill and buttress sensitive slopes. DF in high gradient streams may aggrade sections downstream to extent that slope undercutting and failure may occur in steep terrain, with unstable slopes.

It has been a few decades since I was involved with DF, so I am not current on present research, but I do not remember any models or hearing about models that would do well predicting episodic driven DF and the sediment delivery from them. The geologists who I worked with were good at identifying sensitive terrain with slope and geologic contacts/types, so in the age of GIS and improved inventory of geology, slope and soil detail, identifying sensitive or high risk conditions at landscape levels for modeling may be possible. Areas with higher rainfall risk are also fairly predictable in mountainous landscapes, yet conditions exposed to hurricanes, monsoons, typhones, and other high intensity or long duration events are not necessarily predictable.

In terrain with episodic DF, I would suggest using LiDAR flights periodically to identify changes, employing Change Detection in GIS when needed to address landform and channel changes periodically. These changes in headwater, unstable landscapes could be approximated and measurable. Both DF and sediment movement tend to be episodic, and some channel sections have low gradient or geologic constrictions that naturally tend to accumulate sediment, even appearing much like a pond landform with accumulated and stored sediment rather than water, or a braided channel system where grade changes abruptly from steep to low gradient.

On a local analysis scale of past DF activity, past aerial photos conducted and georeferenced at regular intervals may help define boundaries of failures, and then site visits to define more specifics as to depths of slope failure, estimated depths of channel scour and fill downstream that help define volumes of sediment. Sometimes layers of sediment can be defined, dated. The further downstream or down velley one goes, typically the greater the stream power to make changes and more area such as floodplains to incorporate sediments in less noticeable ways. Training received from Dr. David Rosgen' work suggested use of channel chains to help define and measure conditions where channels aggrade, degrade, or conditions of channel degrading during severe event, then aggrading after the event in recession. His work or normalizing flow and sediment to bankfull conditions appeared promising relative to studying specific circumstances. Measuring steamflow and sediment under conditions of DF also has many issues for normal stream and channel instrumentation.

Dear William F. Hansen , thanks for such a complete description of field methods. That is very worthy. However, it would be interesting to find integrative approaches relating episodic supply with downstream transport, kind of following work of Benda and Dunne (1997) (attached), because sometimes you want to estimate future regime while lack of aereal imagery from the basin. For example, how can current river morphology in several sites of drainage network allow to estimate rates and timing of episodic supply?

Thanks for reference of interest. Dunne is well known in my early work Interests. As mentioned, there are places in channel morphology, where sediments tend to accumulate and store. Roehl (1962) is example of researcher who used ponds and reservoirs to help measure efficiency of sediment delivery. Possibly, using LiDAR to locate failure areas, and locate areas of likely channel filling as progressing downstream, one should eventually reach circumstances where likely storage areas are not full. If you locate areas aggraded, the layers of sediment might be identified, and various methods exist to date sediment layers, so frequency and approximate dates of storms realized. I will read reference, but may be delayed due to personal reasons.