This research aims to improve our knowledge of debris flow occurrence and behaviour in New Zealand. Detailed field data collected in four debris flow prone areas in New Zealand are presented and compared. The travel distance of these events is then modelled with an empiricalstatistical model, UBCDflow, and an analytical, “equivalent fluid” continuum model DAN-W. While field studies are useful, they are often not linked to the underlying mechanics of debris flow motion or compared with the behavior of small scale flows due to the inherent complexity and unknown boundary conditions in field scale flows. Physical modelling simplifies the situation and allows boundary conditions to be controlled. The second part of this research uses physical modelling, including a series of novel debris flow tests in a geotechnical centrifuge, to compare and contrast flow behaviour and mechanics of laboratory and field scale flows. The debris flows events investigated in the field were categorized into hillslope, torrent, or intermediate-type events. Hillslope events were less channelized and progressively deposited on high slope angles. Consequently, high friction coefficients were needed to model their mobility. Torrent flows entrained more material than hillslope flows and deposited on lower angle slopes in response to unconfinement on the debris flow fan. Friction coefficients back-calculated for torrent events were lower than for the hillslope flows, but still larger than most of the friction coefficients given for large, channelized, debris flow events in the literature. Intermediate events were similar to hillslope events in terms of deposition angle and best-fit friction coefficients, but were very confined. Both UBCDflow and DAN-W were found to be useful decision support tools, but the capability of each model was limited. Greater modelling capability was gained by using the volume change behaviour predicted by UBCDflow in DAN-W, as DAN-W simulates flow heights and velocities, but does not predict the depth of erosion. In the second part of the research, a geotechnical centrifuge is used to model debris flow processes in a larger acceleration field than earth’s gravity. While centrifuges have been used to model a variety of processes in other geotechnical problems, debris flows are a relatively new phenomenon to be tested on a centrifuge. The centrifuge was successful in increasing the frictional properties of flow, but viscous forces were still the dominant form of shear stress with the materials used. Markedly different flow behaviour of tests using different pore-fluid rheologies suggested that the dominant mechanism of shear resistance may have changed between confined, downslope movement and unconfined runout. The results also showed that in geotechnical centrifuge testing, the viscosity of the pore fluid scales with the g-level, N. This research is an important step in developing centrifuge testing as an accepted method of modelling debris flow processes. Finally, a brief comparison of friction slopes between small-scale 1-g flume tests and field scale flows suggests that 1-g flume experiments are able to model the mobility of field scale flows if the soil used is well-graded and the pore-fluid is not too viscous. This research shows that the the ability of laboratory scale flows to model large scale processes may not be as limited as previously suggested by some investigators.

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第17866号 / 工博第3775号 / 新制||工||1577(附属図書館) / 30686 / 京都大学大学院工学研究科社会基盤工学専攻 / (主査)教授 中川 一, 教授 藤田 正治, 准教授 川池 健司 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

A 3-D seismic volume, wireline logs and core data were integrated to determine the spatial distribution of porous reservoirs within the Wolfcampian-Leonardian detrital carbonate slope and basin strata in Glasscock County, Texas. A 3-D seismic amplitude volume was used to construct a seismic facies analysis of the detrital carbonate section, and generated attribute volumes helped identify detrital carbonate depositional trends, as well as establish a potential correlation between thick detrital carbonate intervals and associated amplitude response. Eight lithofacies were identified in core and were subsequently classified into three main facies: debris flow, grain flow/turbidite, and basinal shale. A facies type log was then created, which was used to supervise the creation of facies logs within other wells to ultimately use in the creation of a 3-D facies model. Cross sections through the study area show an increase in bathymetric relief beginning in Wolfcampian time and continuing through the Leonardian. Detrital carbonate deposition increases dramatically during the Leonardian, consisting of large gravity flows deposited basinward in a northwest-southeast linear trend, rapidly thinning basinward. Individual flows are discontinuous and bounded by basinal shale facies. Four seismic facies were identified within the interval of interest using a structurally smoothed attribute volume, while an RMS amplitude attribute volume provided a correlation between high RMS amplitude values and detrital carbonate thickness. A high RMS amplitude value corresponding to the debris flow facies was extracted from the RMS attribute volume in the form of a seismic geobody. Two facies models and one porosity model were generated by using upscaled values from the gamma ray, total porosity, and lithofacies logs, which were applied over areas with the densest well control. Although the facies model populated from upscaled GR values was useful in stratigraphic interpretation, it is determined that the models should be applied over areas with denser well spacing in order to provide a more accurate and geologically viable subsurface model.

Reaches dominated by debris flow scour and incision tend to greatly influence landscape form in steepland basins. Debris flow networks, despite their ubiquity, have not been exploited to develop erosion rate proxies. To bridge this gap, I applied a proposed empirical function that describes the variation of valley slope with drainage area in fluvial and debris flow reaches of steepland channel networks in the Oregon Coast Range. I calibrated a relationship between profile concavity and erosion rate to map spatial patterns of long-term uplift rates assuming steady state. I also estimated the magnitude and inland extent of coseismic subsidence in my study area. My estimates agree with field measurements in the same area along the Cascadia margin, indicating that debris flow valley profiles can be used to make interpretations from spatial patterns of rock uplift that may better constrain physical models of crustal deformation. This thesis includes unpublished co-authored material.

Rainfall-induced landslides that change into debris flows and travel large distances are one of the treacherous natural calamities that can occur in mountainous areas, particularly in Nepal’s mountains. Debris flow was the second highest cause of human death in Nepal after epidemics between 1971 and 2016. Because debris flow is common in mountainous regions, its prediction and remedial measures through land use plans are important factors to consider for saving lives and properties. The spatial distribution of the initial landslides that change into debris flow, on a watershed scale, is still an important area of study in this mountainous region to develop essential land use plan. In this research, hydrologic, slope stability and Flow-R models are applied in GIS modeling to locate potential landslide and debris flow areas for a given threshold rainfall in a mountainous watershed-Kulekhani, Nepal. Soil samples from 73 locations within the watershed and a geotechnical investigation on one old landslide area were considered to determine the Soil Water Characteristics Curve (SWCC), friction angle, cohesion, and infiltration characteristics of the subsurface soils in the study area. This information is applied in an unsaturated slope stability model to find unstable locations in the study watershed in a GIS environment. The model is tested on a recorded 24-hour rainfall of 540 mm in the watershed, and potential landslide locations are obtained. The validation results show that there is a good agreement between the predicted and mapped landslides. For debris flow run out, Flow-R model, which has the capability to analyze debris flow inundation with limited input information, and the model software is readily available in the public domain, was chosen for further analysis. Two recent debris flow events and the study watershed are taken as case studies to identify the appropriate algorithms of Flow-R for runout analysis of the study areas. Landslide-triggering threshold rainfall frequency is related to the frequency of landslides and the debris flow hazard in these mountains. The above validated models are applied in a GIS environment to locate potential debris flow areas in expected threshold rainfall. Rainfall records from 1980 to 2013 are computed for one- to seven-day cumulative annual maximum rainfall. The probable rainfalls for 1 in 10 to 1 in 200 years return periods are identified. The anticipated probable rainfalls are modeled in the GIS environment to identify the factor of safety of mountain slopes for landslide susceptibility in the study watershed. The Flow-R model with user-defined landslide-susceptible areas was chosen for debris flow runout analysis. A relation between the frequency of rainfall and landslide-induced debris flow hazard area is derived for return periods of 25, 50, 100, and 200 years. Also, the debris flow hazard results from the analysis are compared with a known event in the watershed and found to agree. This developed method can be applied to anticipated landslide and landslide-induced debris flow from the live rainfall record to warn hazard-prone communities for saving lives and regulating hazardous transportation corridors in these mountains. In addition to this, this methodology will be a useful tool to help policy makers create appropriate land use plans.

Postwildfire debris flows are frequently triggered by runoff following high-intensity rainfall, but the physical mechanisms by which water-dominated flows transition to debris flows are poorly understood relative to debris flow initiation from shallow landslides. In this study, we combined a numerical model with high-resolution hydrologic and geomorphic data sets to test two different hypotheses for debris flow initiation during a rainfall event that produced numerous debris flows within a recently burned drainage basin. Based on simulations, large volumes of sediment eroded from the hillslopes were redeposited within the channel network throughout the storm, leading to the initiation of numerous debris flows as a result of the mass failure of sediment dams that built up within the channel. More generally, results provide a quantitative framework for assessing the potential of runoff-generated debris flows based on sediment supply and hydrologic conditions.

Large-scale characteristics of Ladore Canyon debris fans are dependent upon the bedrock and lithology of the mainstem and tributary canyons. The largest fans occur in the widest section of the mainstem canyon, which typically correlates with the location of large faults. The steepest fans are found at the mouths of tributaries where cliffs are formed by resistant lithologies. Smaller-scale fan characteristics are dependent upon the magnitude and frequency of events from the respective drainage basin, which is controlled primarily by climate. Three distinct deposit ages ( oldest, intermediate, youngest) were distinguished on individual fans and were tentatively correlated throughout the canyon based on observations of boulder weathering, boulder concentration, soil development, vegetation, and topography. During fall 1997 and late spring 1998, four debris flows aggraded fans in Ladore Canyon. The largest of the four events, Wild Mountain, deposited a 3,800-m2 fan in the mainstem canyon, significantly constricting the Green River. Three of the four debris flows occurred in drainages that had been burned by forest fires during summer 1996. The debris flows were initiated during rainfall events with precipitation totaling more than 3 cm. Events of this magnitude have rarely been recorded in the region during the period of record. Measurements from the Wild Mountain debris fan indicate that under current operating conditions of Flaming Gorge Dam, the Green River has a limited capacity to mobilize newly deposited debris-flow material; therefore, particles eroded from the fan face cannot replenish downstream gravel bars. High release discharges equivalent to the 1997 high releases from Flaming Gorge Dam have a greater potential to rework newly deposited debris fans. (155 pages)

El texto completo de este trabajo no está disponible en el Repositorio Académico UPC por restricciones de la casa editorial donde ha sido publicado. / Recurrent economic and human losses occur in populated areas caused by the debris flow, known in Peru as “huayco” and for which there is little information. To determine the effectiveness of debris flow barriers installed in 2016 in three high-risk creeks with slopes ranging from 29 to 35%, it was analyzed satellite images with the Geographical Information System (GIS). For that, it is necessary to obtain the volumes of both soil erosion in the upper basin transported by the debris flow and the retained solids by the barriers. Topographic, geological, geomorphological, and hydrological characteristics were evaluated, as well as the evolution of the population in the dejection cone. It was no possible to obtain results for all the destructive events because there are no cartographies, and the available satellite images do not have enough temporal or spatial resolution or present cloudiness greater than 20%. The field investigations after the ENSO 2017 occurrence made it possible to verify that the debris flow barriers allowed to avoid the loss of human life and material damage. The field measured volumes of the solids retained by the barriers and the previously estimated volumes of erosion were compared, and retention efficiency of 80–90% was obtained. It recommends continuing with this research because it is necessary to know debris flow characteristics for analyzing the convenience and technical requirements for the implementation of retention barriers in other dangerous streams that are difficult to access.

Mountain watersheds recently burned by wildfire often experience greater amounts of runoff and increased rates of sediment transport relative to similar unburned areas. Given the sedimentation and debris flow threats caused by increases in erosion, more work is needed to better understand the physical mechanisms responsible for the observed increase in sediment transport in burned environments and the time scale over which a heightened geomorphic response can be expected. In this study, we quantified the relative importance of different hillslope erosion mechanisms during two postwildfire rainstorms at a drainage basin in Southern California by combining terrestrial laser scanner-derived maps of topographic change, field measurements, and numerical modeling of overland flow and sediment transport. Numerous debris flows were initiated by runoff at our study area during a long-duration storm of relatively modest intensity. Despite the presence of a well-developed rill network, numerical model results suggest that the majority of eroded hillslope sediment during this long-duration rainstorm was transported by raindrop-induced sediment transport processes, highlighting the importance of raindrop-driven processes in supplying channels with potential debris flow material. We also used the numerical model to explore relationships between postwildfire storm characteristics, vegetation cover, soil infiltration capacity, and the total volume of eroded sediment from a synthetic hillslope for different end-member erosion regimes. This study adds to our understanding of sediment transport in steep, postwildfire landscapes and shows how data from field monitoring can be combined with numerical modeling of sediment transport to isolate the processes leading to increased erosion in burned areas.

Debris flows are a known hazard in southern California where growing numbers of people are moving into the urban-wildland interface, threatening lives and property. A common location to see a debris flow head scarp is the upper one-third to one-half of an unburned slope at or near the head of a first-order catchment, particularly in areas of relatively shallow soils overlying bedrock. Unburned, relatively steep slopes with gently rounded shoulders and thin soil over bedrock in southern California were investigated to determine if there is a position on these types of slopes where near-surface water levels and the associated pore pressures are relatively and consistently higher during and after rainfall events than the rest of the slope, resulting in an area of preferential shallow slope failure and debris flow initiation. It was hypothesized that this position, if it exists, would be on the upper one-third to one-half of the slope near a change from a shallower slope to a steeper slope (the slope shoulder). It was further hypothesized that elevated subsurface pore pressures at this location would contribute to it being an area of preferential shallow slope failure. The near-surface water levels at two field sites in southern California were monitored for three field seasons. In the laboratory, a meso-scale simulator was constructed and used to replicate field conditions using an adjustable artificial slope and simulated rainfall. The field research showed that areas of higher water levels can exist on the upper one-third to one-half of hillslopes meeting the designated criteria. The laboratory simulations showed elevated water levels in the same general area as the field data. Laboratory simulations also suggested that this is an area of preferential shallow slope failure. The angle of the slope influenced how long a slope took to fail and how much water was needed to do so, with gentler slopes requiring more time and approximately double the amount of water than steeper slopes.