Refine
Has Fulltext
- no (73) (remove)
Year of publication
Document Type
- Article (68)
- Other (2)
- Monograph/Edited Volume (1)
- Preprint (1)
- Review (1)
Is part of the Bibliography
- yes (73)
Keywords
- landslide (6)
- Landslide (5)
- Earthquake (4)
- Erosion (4)
- Himalayas (4)
- Chile (3)
- Himalaya (3)
- Patagonia (3)
- climate change (3)
- hydropower (3)
Institute
Probabilistic models to inform landslide early warning systems often rely on rainfall totals observed during past events with landslides. However, these models are generally developed for broad regions using large catalogs, with dozens, hundreds, or even thousands of landslide occurrences. This study evaluates strategies for training landslide forecasting models with a scanty record of landslide-triggering events, which is a typical limitation in remote, sparsely populated regions. We evaluate 136 statistical models trained on a precipitation dataset with five landslide-triggering precipitation events recorded near Sitka, Alaska, USA, as well as 6000 d of non-triggering rainfall (2002–2020). We also conduct extensive statistical evaluation for three primary purposes: (1) to select the best-fitting models, (2) to evaluate performance of the preferred models, and (3) to select and evaluate warning thresholds. We use Akaike, Bayesian, and leave-one-out information criteria to compare the 136 models, which are trained on different cumulative precipitation variables at time intervals ranging from 1 h to 2 weeks, using both frequentist and Bayesian methods to estimate the daily probability and intensity of potential landslide occurrence (logistic regression and Poisson regression). We evaluate the best-fit models using leave-one-out validation as well as by testing a subset of the data. Despite this sparse landslide inventory, we find that probabilistic models can effectively distinguish days with landslides from days without slide activity. Our statistical analyses show that 3 h precipitation totals are the best predictor of elevated landslide hazard, and adding antecedent precipitation (days to weeks) did not improve model performance. This relatively short timescale of precipitation combined with the limited role of antecedent conditions likely reflects the rapid draining of porous colluvial soils on the very steep hillslopes around Sitka. Although frequentist and Bayesian inferences produce similar estimates of landslide hazard, they do have different implications for use and interpretation: frequentist models are familiar and easy to implement, but Bayesian models capture the rare-events problem more explicitly and allow for better understanding of parameter uncertainty given the available data. We use the resulting estimates of daily landslide probability to establish two decision boundaries that define three levels of warning. With these decision boundaries, the frequentist logistic regression model incorporates National Weather Service quantitative precipitation forecasts into a real-time landslide early warning “dashboard” system (https://sitkalandslide.org/, last access: 9 October 2023). This dashboard provides accessible and data-driven situational awareness for community members and emergency managers.
Several thousands of moraine-dammed and supraglacial lakes spread over the Hindu Kush Himalayan (HKH) region, and some have grown rapidly in past decades due to glacier retreat. The sudden emptying of these lakes releases large volumes of water and sediment in destructive glacial lake outburst floods (GLOFs), one of the most publicised natural hazards to the rapidly growing Himalayan population. Despite the growing number and size of glacial lakes, the frequency of documented GLOFs is remarkably constant. We explore this possible reporting bias and offer a new processing chain for establishing a more complete Himalayan GLOF inventory. We make use of the full seasonal archive of Landsat images between 1988 and 2016, and track automatically where GLOFs left shrinking water bodies, and tails of sediment at high elevations. We trained a Random Forest classifier to generate fuzzy land cover maps for 2491 images, achieving overall accuracies of 91%. We developed a likelihood-based change point technique to estimate the timing of GLOFs at the pixel scale. Our method objectively detected ten out of eleven documented GLOFs, and another ten lakes that gave rise to previously unreported GLOFs. We thus nearly doubled the existing GLOF record for a study area covering similar to 10% of the HKH region. Remaining challenges for automatically detecting GLOFs include image insufficiently accurate co-registration, misclassifications in the land cover maps and image noise from clouds, shadows or ice. Yet our processing chain is robust and has the potential for being applied on the greater HKH and mountain ranges elsewhere, opening the door for objectively expanding the knowledge base on GLOF activity over the past three decades.
Shrinking glaciers in the Hindu Kush-Karakoram-Himalaya-Nyainqentanglha (HKKHN) region have formed several thousand moraine-dammed glacial lakes(1-3), some of these having grown rapidly in past decades(3,4). This growth may promote more frequent and potentially destructive glacial lake outburst floods (GLOFs)(5-7). Testing this hypothesis, however, is confounded by incomplete databases of the few reliable, though selective, case studies. Here we present a consistent Himalayan GLOF inventory derived automatically from all available Landsat imagery since the late 1980s. We more than double the known GLOF count and identify the southern Himalayas as a hotspot region, compared to the more rarely affected Hindu Kush-Karakoram ranges. Nevertheless, the average annual frequency of 1.3 GLOFs has no credible posterior trend despite reported increases in glacial lake areas in most of the HKKHN3,8, so that GLOF activity per unit lake area has decreased since the late 1980s. We conclude that learning more about the frequency and magnitude of outburst triggers, rather than focusing solely on rapidly growing glacial lakes, might improve the appraisal of GLOF hazards.
Deforestation is a prominent anthropogenic cause of erosive overland flow and slope instability, boosting rates of soil erosion and concomitant sediment flux. Conventional methods of gauging or estimating post-logging sediment flux often focus on annual timescales but overlook potentially important process response on shorter intervals immediately following timber harvest. We resolve such dynamics with non-parametric quantile regression forests (QRF) based on high-frequency (3 min) discharge measurements and sediment concentration data sampled every 30-60 min in similar-sized (similar to 0.1 km(2)) forested Chilean catchments that were logged during either the rainy or the dry season. The method of QRF builds on the random forest algorithm, and combines quantile regression with repeated random sub-sampling of both cases and predictors. The algorithm belongs to the family of decision-tree classifiers, which allow quantifying relevant predictors in high-dimensional parameter space. We find that, where no logging occurred, similar to 80% of the total sediment load was transported during extremely variable runoff events during only 5% of the monitoring period. In particular, dry-season logging dampened the relative role of these rare, extreme sediment-transport events by increasing load efficiency during more efficient moderate events. We show that QRFs outperform traditional sediment rating curves (SRCs) in terms of accurately simulating short-term dynamics of sediment flux, and conclude that QRF may reliably support forest management recommendations by providing robust simulations of post-logging response of water and sediment fluxes at high temporal resolution.
Bayesian geomorphology
(2020)
The rapidly growing amount and diversity of data are confronting us more than ever with the need to make informed predictions under uncertainty. The adverse impacts of climate change and natural hazards also motivate our search for reliable predictions. The range of statistical techniques that geomorphologists use to tackle this challenge has been growing, but rarely involves Bayesian methods. Instead, many geomorphic models rely on estimated averages that largely miss out on the variability of form and process. Yet seemingly fixed estimates of channel heads, sediment rating curves or glacier equilibrium lines, for example, are all prone to uncertainties. Neighbouring scientific disciplines such as physics, hydrology or ecology have readily embraced Bayesian methods to fully capture and better explain such uncertainties, as the necessary computational tools have advanced greatly. The aim of this article is to introduce the Bayesian toolkit to scientists concerned with Earth surface processes and landforms, and to show how geomorphic models might benefit from probabilistic concepts. I briefly review the use of Bayesian reasoning in geomorphology, and outline the corresponding variants of regression and classification in several worked examples.
Reconciling the paths of extreme rainfall with those of typhoons remains difficult despite advanced forecasting techniques. We use complex networks defined by a nonlinear synchronization measure termed event synchronization to track extreme rainfall over the Japanese islands. Directed networks objectively record patterns of heavy rain brought by frontal storms and typhoons but mask out contributions of local convective storms. We propose a radial rank method to show that paths of extreme rainfall in the typhoon season (August-November, ASON) follow the overall southwest-northeast motion of typhoons and mean rainfall gradient of Japan. The associated eye-of-the-typhoon tracks deviate notably and may thus distort estimates of heavy typhoon rainfall. We mainly found that the lower spread of rainfall tracks in ASON may enable better hindcasting than for westerly-fed frontal storms in June and July.
Reconstructing how rivers respond to changes in runoff or sediment supply by incising or aggrading has been pivotal in gauging the role of the Indian Summer Monsoon (ISM) as a geomorphic driver in the Himalayas. Here we present new chronological data for fluvial aggradation and incision from the Donga alluvial fan and the upper Alaknanda River, as well as a compilation of previous work. In addition to conventional OSL-SAR (Single-Aliquot Regenerative-Dose) dating method, we have tested and applied pulsed OSL (POSL) dating for quartz samples that include K-rich feldspar inclusions, which is expected to improve the applicability and validity of OSL ages in the Lesser Himalaya. For previously dated deposits, our OSL ages are shown to be systematically older than previously reported ages. These results suggest periods of aggradation in the Alaknanda and Dehradun Valleys mainly between similar to 25 and 35 ka. This most likely reflects decreased stream power during periods of weakened monsoon. In addition, in-situ cosmogenic beryllium-10 was used to infer bedrock surface exposure ages, which are interpreted as episodes of active fluvial erosion. Resulting exposure ages span from 3 to 6 ka, suggesting that strath terraces were exposed relatively recently, and incision was dominant through most of the Holocene. In combination, our results support precipitation-driven fluvial dynamics, which regulates the balance between stream power and sediment supply. On a larger spatial scale, however, fluvial dynamics are probably not spatially homogeneous as aggradation could have been taking place in adjacent catchments while incision dominated in the study area. (C) 2018 Elsevier Ltd. All rights reserved.
In this study, we investigate how immersive 3D geovisualization can be used in higher education. Based on MacEachren and Kraak's geovisualization cube, we examine the usage of immersive 3D geovisualization and its usefulness in a research-based learning module on flood risk, called GEOSimulator. Results of a survey among participating students reveal benefits, such as better orientation in the study area, higher interactivity with the data, improved discourse among students and enhanced motivation through immersive 3D geovisualization. This suggests that immersive 3D visualization can effectively be used in higher education and that 3D CAVE settings enhance interactive learning between students.
Roads at risk
(2015)
Globalisation and interregional exchange of people, goods, and services has boosted the importance of and reliance on all kinds of transport networks. The linear structure of road networks is especially sensitive to natural hazards. In southern Norway, steep topography and extreme weather events promote frequent traffic disruption caused by debris flows. Topographic susceptibility and trigger frequency maps serve as input into a hazard appraisal at the scale of first-order catchments to quantify the impact of debris flows on the road network in terms of a failure likelihood of each link connecting two network vertices, e.g. road junctions. We compute total additional traffic loads as a function of traffic volume and excess distance, i.e. the extra length of an alternative path connecting two previously disrupted network vertices using a shortest-path algorithm. Our risk metric of link failure is the total additional annual traffic load, expressed as vehicle kilometres, because of debris-flow-related road closures. We present two scenarios demonstrating the impact of debris flows on the road network and quantify the associated path-failure likelihood between major cities in southern Norway. The scenarios indicate that major routes crossing the central and north-western part of the study area are associated with high link-failure risk. Yet options for detours on major routes are manifold and incur only little additional costs provided that drivers are sufficiently well informed about road closures. Our risk estimates may be of importance to road network managers and transport companies relying on speedy delivery of services and goods.
Supra-glacial deposition and flux of catastrophic rock-slope failure debris, south-central Alaska
(2013)
The ongoing debate over the effects of global environmental change on Earth's cryosphere calls for detailed knowledge about process rates and their variability in cold environments. In this context, appraisals of the coupling between glacier dynamics and para-glacial erosion rates in tectonically active mountains remain rare. We contribute to filling this knowledge gap and present an unprecedented regional-scale inventory of supra-glacial sediment flux and hillslope erosion rates inferred from an analysis of 123 large (> 0 center dot 1km2) catastrophic bedrock landslides that fell onto glaciers in the Chugach Mountains, Alaska, as documented by satellite images obtained between 1972 to 2008. Assuming these supra-glacial landslide deposits to be passive strain markers we infer minimum decadal-scale sediment yields of 190 to 7400tkm-2yr-1 for a given glacier-surface cross-section impacted by episodic rock-slope failure. These rates compare to reported fluvial sediment yields in many mountain rivers, but are an order of magnitude below the extreme sediment yields measured at the snouts of Alaskan glaciers, indicating that the bulk of debris discharged derives from en-glacial, sub-glacial or ice-proximal sources. We estimate an average minimum para-glacial erosion rate by large, episodic rock-slope failures at 0 center dot 5-0 center dot 7mmyr-1 in the Chugach Mountains over a 50-yr period, with earthquakes likely being responsible for up to 73% of this rate. Though ranking amongst the highest decadal landslide erosion rates for this size of study area worldwide, our inferred rates of hillslope erosion in the Chugach Mountains remain an order of magnitude below the pace of extremely rapid glacial sediment export and glacio-isostatic surface uplift previously reported from the region.