A comparative study of three neural network forecast combination methods for simulated river flows of different rainfall—runoff models

Abstract

The performances of three artificial neural network (NN) methods for combining simulated river flows, based on three different neural network structures, are compared. These network structures are: the simple neural network (SNN), the radial basis function neural network (RBFNN) and the multi-layer perceptron neural network (MLPNN). Daily data of eight catchments, located in different parts of the world, and having different hydrological and climatic conditions, are used to enable comparisons of the performances of these three methods to be made. In the case of each catchment, each neural network combination method synchronously uses the simulated river flows of four rainfall—runoff models operating in design non-updating mode to produce the combined river flows. Two of these four models are black-box, the other two being conceptual models. The results of the study show that the performances of all three combination methods are, on average, better than that of the best individual rainfall—runoff model utilized in the combination, i.e. that the combination concept works. In terms of the Nash-Sutcliffe model efficiency index, the MLPNN combination method generally performs better than the other two combination methods tested. For most of the catchments, the differences in the efficiency index values of the SNN and the RBFNN combination methods are not significant but, on average, the SNN form performs marginally better than the more complex RBFNN alternative. Based on the results obtained for the three NN combination methods, the use of the multi-layer perceptron neural network (MLPNN) is recommended as the appropriate NN form for use in the context of combining simulated river flows.     

Investigating the impact of the Chi‐Chi earthquake on the occurrence of debris flows using artificial neural networks

Debris flows have caused enormous losses of property and human life in Taiwan during the last two decades. An efficient and reliable method for predicting the occurrence of debris flows is required. The major goal of this study is to explore the impact of the Chi‐Chi earthquake on the occurrence of debris flows by applying the artificial neural network (ANN) that takes both hydrological and geomorphologic influences into account. The Chen‐Yu‐Lan River watershed, which is located in central Taiwan, is chosen for evaluating the critical rainfall triggering debris flows. A total of 1151 data sets were collected for calibrating model parameters with two training strategies. Significant differences before and after the earthquake have been found: (1) The size of landslide area is proportioned to the occurrence of debris flows; (2) the amount of critical rainfall required for triggering debris flows has reduced significantly, about half of the original critical rainfall in the study case; and (3) the frequency of the occurrence of debris flows is largely increased. The overall accuracy of model prediction in testing phase has reached 96·5%; moreover, the accuracy of occurrence prediction is largely increased from 24 to 80% as the network trained with data from before the Chi‐Chi earthquake sets and with data from the lumped before and after the earthquake sets. The results demonstrated that the ANN is capable of learning the complex mechanism of debris flows and producing satisfactory predictions. Copyright © 2009 John Wiley & Sons, Ltd.     

The role of the sediment budget in understanding debris flow susceptibility

This study proposes a sediment‐budget model to predict the temporal variation of debris volume stored in a debris‐flow prone watershed. The sediment‐budget is dominated by shallow landslides and debris outflow. The basin topography and the debris volume stored in the source area of the debris‐flow prone watershed help evaluating its debris‐flow susceptibility. The susceptibility model is applied to the Tungshih area of central western Taiwan. The importance of the debris volume in predicting debris‐flow susceptibility is reflected in the standardized coefficients of the proposed statistical discriminant model. The high prediction rate (0·874) for the occurrence of debris flows justifies the capability of the proposed susceptibility models to predict the occurrence of debris flows. This model is then used to evaluate the temporal evolution of the debris‐flow susceptibility index. The analysis results show that the numbers of watershed which are classified as a debris‐flow group correspond well to storage of sediment at different time periods. These numbers are 10 before the occurrence of Chi‐Chi earthquake, 13 after the occurrence of Chi‐Chi earthquake, 16 after the occurrence of landslides induced by Typhoon Mindulle (Typhoon M), and 14 after the occurrence of debris flows induced by Typhoon M. It indicates that the occurrence of 7·6 Chi‐Chi earthquake had significant impact on the debris flow occurrence during subsequent typhoons. Copyright © 2009 John Wiley & Sons, Ltd.     

Information Detection of Seismic Debris Flow by UAV High-resolution Image Based on Transfer Learning

A large number of debris flow disasters (called Seismic debris flows) would occur after an earthquake, which can cause a great amount of damage. UAV low-altitude remote sensing technology has become a means of quickly obtaining disaster information as it has the advantage of convenience and timeliness, but the spectral information of the image is so scarce, making it difficult to accurately detect the information of earthquake debris flow disasters. Based on the above problems, a seismic debris flow detection method based on transfer learning (TL) mechanism is proposed. On the basis of the constructed seismic debris flow disaster database, the features acquired from the training of the convolutional neural network (CNN) are transferred to the disaster information detection of the seismic debris flow. The automatic detection of earthquake debris flow disaster information is then completed, and the results of object-oriented seismic debris flow disaster information detection are compared and analyzed with the detection results supported by transfer learning.     

Development of a remotely controlled debris flow monitoring system in the Dolomites (Acquabona,Italy)

Direct measurements of the hydrological conditions for the occurrence of debris flows and of flow behaviour are of the outmost importance for developing effective flow prevention techniques. An automated and remotely controlled monitoring system was installed in Acquabona Creek in the Dolomites, Italian Eastern Alps, where debris flows occur every year. Its present configuration consists of three on‐site stations, located in the debris‐flow initiation area, in the lower channel and in the retention basin. The monitoring system is equipped with sensors for measuring rainfall, pore‐water pressure in the mobile channel bottom, ground vibrations, debris flow depth, total normal stress and fluid pore‐pressure at the base of the flow. Three video cameras take motion pictures of the events at the initiation zone, in the lower channel and in the deposition area. Data from the on‐site stations are radio‐transmitted to an off‐site station and stored in a host PC, from where they are telemetrically downloaded and used by the Padova University for the study of debris flows. The efficiency of the sensors and of the whole monitoring system has been verified by the analysis of data collected so far. Examples of these data are presented and briefly discussed. If implemented at the numerous debris‐flow sites in the Dolomitic Region, the technology used, derived from the development of this system, will provide civil defence and warn residents of impending debris flows. Copyright © 2003 John Wiley & Sons, Ltd.     

Post-wildfire sediment cascades: A modeling framework linking debris flow generation and network-scale sediment routing

Wildfires represent one of the largest disturbances in watersheds of the Intermountain West. Yet, we lack models capable of predicting post-wildfire impacts on downstream ecosystems and infrastructure. Here we present a novel modeling framework that links new and existing models to simulate the post-wildfire sediment cascade, including spatially explicit predictions of debris flows, storage of debris flow sediment within valleys, delivery of debris flow sediment to active channels, and the downstream routing of sediment through river networks. We apply the model to sediment dynamics in Clear Creek watershed following the 2010 Twitchell Canyon Fire in the Tushar Mountains of southern Utah. The debris flow generation model performed well, correctly predicting 19 out of 20 debris flows from the largest catchments, with only four false positives and two false negatives at observed rainfall intensities. In total, the model predicts the occurrence of 160 post-wildfire debris flows across the Clear Creek watershed, generating more than 650 000 m³ of sediment. Our new storage and delivery model predicts the vast majority of this sediment is stored within valleys, and only 13% is delivered to the river network. The sediment routing model identifies numerous sediment bottlenecks within the network, which alter transport dynamics and may be hotspots for aggradation and aquatic habitat alteration. The volume of sediment exported from the watershed after seven years of simulation totals 17% of that delivered, or 2% of the total generated debris flow sediment. In the case of the Twitchell Canyon Fire, this highlights that significant post-wildfire sediment volumes can be stored in valleys (87%) and within the stream network (11%). Finally, we discuss useful insights that can be gleaned from the model framework, as well as the limitations and need for more monitoring and theory development in order to better constrain essential inputs, process rates, and morphodynamics. © 2019 John Wiley & Sons, Ltd.     

Application of a neural network model in establishing a stage–discharge relationship for a tidal river

This paper presents the applicability of neural network (NN) modelling for forecasting and filtering problems. The multilayer feedforward (MLFF) network was first constructed to forecast the tidal‐level variations at the mouth of the River Chao Phraya in Thailand. Unlike the well‐known conventional harmonic analysis, the NN model uses a set of previous data for learning and then forecasting directly the time‐series of tidal levels. It was found that lead time of 1 to 24 hourly tidal levels can be predicted successfully using only a short‐time hourly learning data. The MLFF network was further used to establish a stage–discharge relationship for the tidal river. The results show a considerably better performance of the NN model over the conventional models. In addition, the stage–discharge relationship obtained by the NN model can indicate reasonably well the important behaviour of the tidal influences. Copyright © 2003 John Wiley & Sons, Ltd.     

Real‐time measurement and preliminary analysis of debris‐flow impact force at Jiangjia Ravine,China

Impact forces associated with major debris flows (Jiangjia Ravine, China, August 25, 2004) were recorded in real time by a system consisting of three strain sensors installed at different flow depths. This provides the first real‐time and long‐duration record of impact forces associated with debris flows. A comprehensive approach including low‐pass filtering and moving average methods were used to preprocess the recorded signals. The upper limit of impact frequency in the debris flows was estimated at 188?66 Hz under the assumption that only coarse grains cause effective impact loadings. Thus, a low‐pass filter with a 200 Hz cut‐off frequency was needed to denoise the original data in order to extract the impact force. Then the moving average method was applied to separate long‐term and random components of the filtered data. These were interpreted as, respectively, the fluid pressure and grain impact loading. It was found that the peak grain impacts at different depths were non‐synchronous within the debris flows. The impact loadings were far greater than, and not proportional to the fluid pressures. Analysis of the impact force of 38 debris flow surges gives an empirical value for the ratio of the hydrodynamic pressure to the momentum flow density, i.e. the product of debris‐flow density and mean velocity square, which provides a very valuable basis for understanding debris flow dynamics and designing debris flow management systems. Copyright © 2011 John Wiley & Sons, Ltd.     

Distinguishing between debris flows and hyperconcentrated flows: an example from the eastern Swiss Alps

Much research has been done on water‐rich mass flows, but the distinction between hyperconcentrated flows and debris flows, and whether the two are indeed different processes, continue to be debated. Here, we contribute to the ongoing discussion of these phenomena by describing and interpreting the deposit of a large landslide‐induced mass flow in the eastern Swiss Alps. About 9400 years ago, 10‐12 km³ of limestone detached from the wall of the Vorderrhein River valley and rapidly fragmented while sliding towards the valley bottom. The rock mass struck the valley floor with enormous force and liquefied at least 1 km³ of valley‐fill sediments. A slurry of liquefied sediment – the ‘Bonaduz gravel’ – traveled tens of kilometres down the Vorderrhein valley from the impact site, carrying huge fragments of rockslide debris that became stranded on the valley floor, forming hills termed ‘tomas’. Part of the flow was deflected by a cross‐valley barrier and traveled 14 km up a tributary of the Vorderrhein valley. The Bonaduz gravel is >65 m thick and fines upward from massive sandy cobble gravel at its base to silty sand at its top. Sedimentologic and geomorphic evidence indicates that Bonaduz gravel was transported as a hyperconcentated flow, likely above a basal carpet of coarse diamictic sediment that behaved as a debris flow. The large amount of water involved in the flow indicates that at least part of the Flims rockslide entered a lake. The Bonaduz deposit shares many properties with sediments left by hyperconcentrated flows generated in flumes, including normal grading and elutriation pipes produced by the rapid escape of fluids when the flow comes to rest. These properties are characteristic of non‐Newtonian laminar flows with high sediment concentrations. Our study reinforces laboratory and theoretical studies showing that debris flows and hyperconcentrated flows are different processes. © 2017 The Authors. Earth Surface Processes and Landforms published by John Wiley & Sons Ltd.     

Sieve deposition by debris flow on a permeable substrate,Leirdalen, Norway

The characteristics of two recent (AD 1994) debris flows in upper Leirdalen, Jotunheimen, Norway, suggest deposition controlled by fluid loss into the underlying, highly permeable, coarse talus. The evidence comprises: (1) drainage holes (sieveholes) up to 44 cm wide and 125 cm deep in the debris‐flow channel floors, which remained open throughout the debris‐flow event; (2) marked channel narrowing, with reduced cross‐sectional areas and termination of the debris flows in flat‐topped, clast‐dominated lobes within a relatively short distance after crossing the junction between impermeable and permeable substrate; (3) the presence of fines deposited in the sieveholes demonstrating the passage of transported matrix; and (4) the absence of substantial lateral drainage through (or dissection of) the levées or the terminal lobes. The term ‘sieve deposition’ is considered particularly well suited to this process involving drainage through the substrate, which is likely to be most effective where debris flows traverse coarse talus either for the first time or only infrequently. Copyright © 2002 John Wiley & Sons, Ltd.     

Tree‐age control on reconstructed debris‐flow frequencies: examples from a regional dendrogeomorphic reconstruction in the Crimean Mountains

The Crimean Mountains (Ukraine) are renowned for the highest occurrence of debris flows along the northern coast of the Black Sea, but information on their origin, frequency and triggers is widely lacking. This study reconstructs a regional time series of debris flows in eight catchments located on the slopes above Yalta. Dendrogeomorphic analyses were performed on 1122 increment cores selected from 566 black pines (Pinus nigra ssp. pallasiana) with clear signs of external damage induced by past debris‐flow activity. The trees sampled were divided into old and young trees. The sample contains 361 young trees with post‐1930 innermost rings and 205 old trees with pre‐1930 germination dates. The two groups of trees were analyzed separately to identify possible age effects in the reconstructed debris‐flow series and to assess the ability of P. nigra to record geomorphic disturbances over time. We date a total of 215 debris flows back to ad 1701 and observe a mean decadal frequency of 6.9 events, with a peak in activity during the 1940s (20 events). The young trees record an increase in debris‐flow activity over the last 70 years, whereas the frequency of events remained fairly constant in the old trees for the same period. By contrast, the formation of reaction wood became increasingly scarce with increasing tree age whereas the occurrence of abrupt growth suppression increased. Copyright © 2014 John Wiley & Sons, Ltd.     

GIS-based risk analysis of debris flow： an application in Sichuan, southwest China

Debris flow is a serious geologic hazard in China. It is estimated that nationally debris flows cause up to 2 billion RMB （250 million US$） in damages and 300-600 deaths and injuries annually. To mitigate debris flow hazards, it is necessary to map, model, and identify zones of debris flow hazards and vulnerability as to inform the local people about the potential risk with a geographic information system. This research presents a regional scale case study modeling debris flow risk （hazard and vulnerability） in Sichuan Province, Southwestern China. In this area, 3,290 debris flows have been identified and the spatial-temporal distribution and activity characteristics of them have been documented. Based on the available meteorological data, a Digital Elevation Model with the rate of 1：250,000 and a regional geological map, the 24-hr rainfall threshold （y） for debris flow occurrence is closely related （significant at 99% confidence level） to the index （x） defined using a geology factor （rock hardness： a） and a topographical factor （channel gradient： d） where y = 21 ＋ 10200 / x, in which x = 2.7 × e^a ＋ 1000 × d. The discipline is constructive in developing the rainfall threshold for debris flow activity in remote mountainous areas that lack data. For a given watershed, a four-level debris flow hazard map is developed by comparing the rainfall threshold to the design rainfall intensities with 50-, 20-, and 5-year average recurrence intervals, respectively. The degree of debris flow vulnerability is determined by the watershed socio-economic conditions. A four-class debris flow risk map, at the final phase of the research, is generated by combining debris flow hazards and vulnerability. With the debris flow risk assessment, the Sichuan Province is classified into the slight, moderate, severe and very severe regions, which accounts for 36%, 19%, 20% and 25% of total area respectively.     

Hyperconcentrated flows on a forested alluvial fan of eastern Canada: geomorphic characteristics,return period,and triggering scenarios

The assessment of the dominant flow type on alluvial fans usually refers to two categories: debris‐flow fans (i.e. sediment gravity flows) and fluvial fans (i.e. fluid gravity flows). Here we report the results of combined morphometric, stratigraphic and sedimentological approaches which suggest that hyperconcentrated flows, a transitional process rheologically distinct from debris flows and floods and sometimes referred to as debris floods, mud floods, or transitional debris flows, are the dominant fan building process in eastern Canada. These flows produce transitional facies between those of debris flows which consist of a cohesive matrix‐supported diamicton, and those of river flows which display more distinct stratification. The size of the blocks in the channels and the abrasion scars at the base of several trees attest to the high transport capacity of these flows. The fan channels are routed according to various obstacles comprised primarily of woody debris that impede sediment transit. However, these conditions of sediment storage are combined with readily available sediment due to the friable nature of the local lithology. Tree‐ring analysis allowed the reconstruction of eight hydrogeomorphic events which are characterized by a return period of 9.25 years for the period 1934–2008, although most of the analyzed events occurred after 1970. Historical weather data analysis indicates that they were related to rare hydrometeorological events at regional and local scales. This evidence led to the elaboration of weather scenarios likely responsible for triggering flows on the fan. According to these scenarios, two distinct hydrologic regimes emerge: the torrential rainfall regime and the nival regime related to snowmelt processes. Hydrogeomorphic processes occurring in a cold‐temperate climate, and particularly on small forested alluvial fans of north‐eastern North America, should receive more attention from land managers given the hazard they represent, as well as because of their sensitivity to various meteorological parameters. Copyright © 2014 John Wiley & Sons, Ltd.     

Exploiting Maximum Entropy method and ASTER data for assessing debris flow and debris slide susceptibility for the Giampilieri catchment (north‐eastern Sicily,Italy)

This study aims at evaluating the performance of the Maximum Entropy method in assessing landslide susceptibility, exploiting topographic and multispectral remote sensing predictors. We selected the catchment of the Giampilieri stream, which is located in the north‐eastern sector of Sicily (southern Italy), as test site. On 1 October 2009, a storm rainfall triggered in this area hundreds of debris flow/avalanche phenomena causing extensive economical damage and loss of life. Within this area a presence‐only‐based statistical method was applied to obtain susceptibility models capable of distinguishing future activation sites of debris flow and debris slide, which where the main source of failure mechanisms for flow or avalanche type propagation. The set of predictors used in this experiment comprised primary and secondary topographic attributes, derived by processing a high resolution digital elevation model, CORINE land cover data and a set of vegetation and mineral indices obtained by processing multispectral ASTER images. All the selected data sources are dated before the disaster. A spatially random partition technique was adopted for validation, generating 50 replicates for each of the two considered movement typologies in order to assess accuracy, precision and reliability of the models. The debris slide and debris flow susceptibility models produced high performances with the first type being the best fit. The evaluation of the probability estimates around the mean value for each mapped pixel shows an inverted relation, with the most robust models corresponding to the debris flows. With respect to the role of each predictor within the modelling phase, debris flows appeared to be primarily controlled by topographic attributes whilst the debris slides were better explained by remotely sensed derived indices, particularly by the occurrence of previous wildfires across the slope. The overall excellent performances of the two models suggest promising perspectives for the application of presence‐only methods and remote sensing derived predictors. Copyright © 2016 John Wiley & Sons, Ltd.     

Comparing threshold definition techniques for rainfall‐induced landslides: A national assessment using radar rainfall

Translational landslides and debris flows are often initiated during intense or prolonged rainfall. Empirical thresholds aim to classify the rain conditions that are commonly associated with landslide occurrence and therefore improve understating of these hazards and predictive ability. Objective techniques that are used to determine these thresholds are likely to be affected by the length of the rain record used, yet this is not routinely considered. Moreover, remotely sensed spatially continuous rainfall observations are under‐exploited. This study compares and evaluates the effect of rain record length on two objective threshold selection techniques in a national assessment of Scotland using weather radar data. Thresholds selected by ‘threat score’ are sensitive to rain record length whereas, in a first application to landslides, ‘optimal point’ (OP) thresholds prove relatively consistent. OP thresholds increase landslide detection and may therefore be applicable in early‐warning systems. Thresholds combining 1‐ and 12‐day antecedence variables best distinguish landslide initiation conditions and indicate that Scottish landslides may be initiated by lower rain accumulation and intensities than previously thought. © 2017 The Authors. Earth Surface Processes and Landforms published by John Wiley & Sons Ltd.     

Methodology of disaster risk assessment for debris flows in a river basin

“Disaster risk assessment” is important in the planning of risk management strategies that reduce societal losses. However, governmental agencies in Taiwan generally assess risks that emerge from debris flows without adequately considering risk management and taking a systems approach. This work proposes an approach to thoroughly consider the interactive influence mechanism of debris flow disaster risk. Additionally, a systematic method for assessing disaster risks is developed. This proposed method can be used in the current risk assessment and as a basis for management strategy planning. Based on systems thinking, the components and attributes of a conceptual system of disaster risk management associated with debris flows in a river basin are identified. Subsequently, a conceptual mitigation–hazard–exposure–resistance framework and an indicator system for assessing the debris flow disaster risks in a river basin are identified. The disaster risks for each exposed community in each drainage zone can be systematically calculated based on the current status or plans of prevention and evacuation measures using the proposed indicator system. A case study of implementing the proposed methodology that involves the Chishan River Basin is presented, in which disaster risk according to the current status of prevention and evacuation measures is assessed. Drainage zones and communities with a significant debris flow disaster risk are located; this risk is associated with a lack of adequate prevention and evacuation measures that have been planned of government agencies. Analytical results indicate that the proposed methodology can systematically and effectively assess the disaster risks of a river basin. The proposed methodology provides a valuable reference for governmental agencies that must manage disaster risk associated with debris flows.     

Bed scour by debris flows: experimental investigation of effects of debris‐flow composition

Debris flows can grow greatly in size by entrainment of bed material, enhancing their runout and hazardous impact. Here, we experimentally investigate the effects of debris‐flow composition on the amount and spatial patterns of bed scour and erosion downstream of a fixed to erodible bed transition. The experimental debris flows were observed to entrain bed particles both grain by grain and en masse, and the majority of entrainment was observed to occur during passage of the flow front. The spatial bed scour patterns are highly variable, but large‐scale patterns are largely similar over 22.5–35° channel slopes for debris flows of similar composition. Scour depth is generally largest slightly downstream of the fixed to erodible bed transition, except for clay‐rich debris flows, which cause a relatively uniform scour pattern. The spatial variability in the scour depth decreases with increasing water, gravel (= grain size) and clay fraction. Basal scour depth increases with channel slope, flow velocity, flow depth, discharge and shear stress in our experiments, whereas there is no correlation with grain collisional stress. The strongest correlation is between basal scour and shear stress and discharge. There are substantial differences in the scour caused by different types of debris flows. In general, mean and maximum scour depths become larger with increasing water fraction and grain size, and decrease with increasing clay content. However, the erodibility of coarse‐grained experimental debris flows (gravel fraction = 0.64) is similar on a wide range of channel slopes, flow depths, flow velocities, discharges and shear stresses. This probably relates to the relatively large influence of grain‐collisional stress to the total bed stress in these flows (30–50%). The relative effect of grain‐collisional stress is low in the other experimental debris flows (<5%), causing erosion to be largely controlled by basal shear stress. Copyright © 2016 John Wiley & Sons, Ltd.     

Field observations,rheological testing and numerical modelling of a debris‐flow event

Debris flows generated from landslides are common processes and represent a severe hazard in mountain regions due to their high mobility and impact energy. We investigate the dynamics and the rheological properties of a 90 000 m³ debris‐flow event triggered by a rapid regressive landslide with high water content. Field evidence revealed a maximum flow depth of 7–8 m, with an estimated peak discharge of 350–400 m³ s^?1. Depositional evidence and grain‐size distribution of the debris pose the debris flow in an intermediate condition between the fluid‐mud and grain‐flow behaviour. The debris‐flow material has silt–clay content up to 15 per cent. The rheological behaviour of the finer matrix was directly assessed with the ball measuring system. The measurements, performed on two samples at 45–63 per cent in sediment concentration by volume, gave values of 3·5–577 Pa for the yield strength, and 0·6–27·9 Pa s for the viscosity. Based on field evidence, we have empirically estimated the yield strength and viscosity ranging between 4000 ± 200 Pa, and 108–134 Pa s, respectively. We used the Flo‐2D code to replicate the debris‐flow event. We applied the model with rheological properties estimated by means of direct measurements and back‐analyses. The results of these models show that the rheological behaviour of a debris‐flow mass containing coarse clasts can not be assessed solely on the contribution of the finer matrix and thus neglecting the effects of direct grain contacts. For debris flows composed by a significant number of coarse clasts a back‐analysis estimation of the rheological parameters is necessary to replicate satisfactorily the depositional extent of debris flows. In these cases, the back‐estimated coefficients do not adequately describe the rheological properties of the debris flow. Copyright © 2006 John Wiley & Sons, Ltd.     

Geomorphological characteristics of small debris flows on Junior's Kop,Marion Island,maritime sub ‐ Antarctic

The geomorphological characteristics of small debris flows in a maritime sub‐Antarctic environment are described. The morphological and sedimentological characteristics of the debris flows are comparable to debris flows documented for other parts of the world; their initiation appears closely linked to the unusual environment in which they are found. Sediment supply is generated by diurnal frost heave of loamy sediment associated with Azorella selago. The debris flows are triggered by sediment mobilization upon saturation of the frost‐heaved surface gravel and overland flow over the low‐permeability and frost‐susceptible slope materials. Morphological effects of the flows are short‐lived due to obliteration by subsequent frost heave activity. Copyright © 2000 John Wiley & Sons, Ltd.     

Dendrogeomorphic reconstruction of past debris‐flow activity using injured broad‐leaved trees

Tree‐ring records from conifers have been regularly used over the last few decades to date debris‐flow events. The reconstruction of past debris‐flow activity was, in contrast, only very rarely based on growth anomalies in broad‐leaved trees. Consequently, this study aimed at dating the occurrence of former debris flows from growth series of broad‐leaved trees and at determining their suitability for dendrogeomorphic research. Results were obtained from gray alder (Alnus incana (L.) Moench), silver birch and pubescent birch (Betula pendula Roth and Betula pubescens Ehrh.), aspen (Populus tremula L.), white poplar, black poplar and gray poplar (Populus alba L., Populus nigra L. and Populus x canescens (Ait.) Sm.), goat willow (Salix caprea L.) and black elder (Sambucus nigra L.) injured by debris‐flow activity at Illgraben (Valais, Swiss Alps). Tree‐ring analysis of 104 increment cores, 118 wedges and 93 cross‐sections from 154 injured broad‐leaved trees allowed the reconstruction of 14 debris‐flow events between AD 1965 and 2007. These events were compared with archival records on debris‐flow activity at Illgraben. It appears that debris flows are very common at Illgraben, but only very rarely left the channel over the period AD 1965–2007. Furthermore, analysis of the spatial distribution of disturbed trees contributed to the identification of six patterns of debris‐flow routing and led to the determination of preferential breakout locations of events. The results of this study demonstrate the high potential of broad‐leaved trees for dendrogeomorphic research and for the assessment of the travel distance and lateral spread of debris‐flow surges. Copyright © 2010 John Wiley & Sons, Ltd.