new paper out
On 24. May 2015, a landslide near the village Baisari, Myagdi District Nepal, dammed the Kali Gandaki. 15 hours later, the dam breached and released a flood wave. Fortunately, people downstream were warned of the imminent flood and no casualties were reported.
In a recent paper by Jeremy Bricker et al. (2017), we hindcasted the event using numerous 1D and 2D hydrodynamic models. Moreover, we used raw and smoothed topographies of the valley floor to investigate the effect of DEM uncertainties on flood wave propagation modelling. We show that using unsmoothed valley thalwegs result in delayed modelled flood wave arrival times that may be critical for the effectiveness early warning systems. A 2D model produced results most in line with field observations.
One of the most striking aspects of our study is the use of crowd-sourced video material available on youtube. We used video material recorded at two bridges crossing the Kali Gandaki to estimate flow depth and speed. If hydrological gauges are unavailable or destroyed, these videos provide an important source of information to assess the magnitude of these extreme events.
Bricker, J.D., Schwanghart, W., Adhikari, B.R., Moriguchi, S., Roeber, V., Giri, S. (2017): Performance of models for flash flood warning and hazard assessment: the 2015 Kali Gandaki landslide breach in Nepal. Mountain Research and Development, 37, 5-15. [DOI: 10.1659/MRD-JOURNAL-D-16-00043.1]
Issyk Kul’s ups and downs: water-filled troughs and peak flows around one of the world’s largest mountain lakes
Text and photos by Angela Landgraf and Swenja Rosenwinkel
Lake Issyk Kul in the northern Kyrgyz Tien Shan is the second-largest mountain lake worldwide and a proposed site for an ICDP drilling project to examine the climatic and erosional history of Central Asia over the past few million years. The lake is endorheic today, but different proxies indicate changing open- and closed-basin conditions through time. Shorelines and lacustrine deposits furnish ample evidence for major lake-level fluctuations during the Quaternary, and suggest the lake was connected to neighboring basins during lake high-stands. Better understanding the dynamics of spill and fill are paramount for correctly assessing changing environmental conditions in the lake basin. A special geomorphic puzzle is the appearance of massive lakebeds west of the present-day sill in the neighboring Kok Moinok basin. The western extent of these lakebeds coincides with a major knickpoint and an abrupt transition from a wide alluviated valley to a narrow bedrock gorge (the Boam Gorge).
In our recently accepted paper in Earth surface Processes and Landforms, we use a variety of field-, laboratory-, and modeling-based methods to assess potential linkages between natural damming to lake highstands and related outburst flood scenarios. These methods include geochronometry of lacustrine and related delta and fluvial deposits around the lake and at the gorge outlet, measurements of entrained mega-clasts for hydraulic paleoflood analysis, and paleoflood modeling using highstand-scenarios based on the elevation of mapped paleoshorelines. We refer to the full paper for a detailed discussion of these methods, and their specific links and assumptions.
We find that one or several catastrophic floods occurred through the Boam gorge in the late Pleistocene. A temporal succession of Issyk Kul’s lake-level drop and boulder deposition at the outlet supports a link between both. Paleoflood modeling, however, shows that catastrophic lake outbursts unconnected to Issyk Kul, i.e., from a separated Kok Moinok lake, could have sustained the necessary peak discharges for moving the boulders at the gorge exit. Although the overall geomorphic and sedimentary evidence around Issyk Kul records some of the largest catastrophic outburst floods in the Tien Shan mountains, if not Central Asia, direct links to documented lake-level changes of Issyk Kul remain elusive.
Rosenwinkel, S., Landgraf, A., Korup, O., Schwanghart, W., Volkmer, F., Dzhumabaeva, A., Merchel, S., Rugel, G., Preusser, F. (2017). Late Pleistocene outburst floods from Issyk Kul, Kyrgyzstan? Earth Surface Processes and Landforms, accepted. [DOI:]
As geomorphologists we have to live with the fact that we can rarely observe the processes that shape the Earth’s surface. Instrumental records cover only a minor time span over which many geomorphic processes act, thus challenging our abilities to disentangle and understand the complex interactions of landscape evolution. Instead, much of our work has to rely on the analysis of landforms, their geometry, their assemblage, and their constituting material, as well as a blend of geochemical and numerical dating techniques of ever-increasing sophistication.
Numerical landscape evolution models (LEM) provide a useful approach to this challenge. LEMs are simulation tools that attempt to model erosion, sediment transport and deposition as well as feedbacks with vegetation and land use. They amalgamate our state of knowledge in a set of physically-based mathematical formulations and allow us to test whether these equations and their parameter values are able to generate output that is plausible and consistent with field evidence.
TTLEM, the TopoToolbox Landscape Evolution Model, is the latest addition to TopoToolbox. Spearheaded by Benjamin Campforts from KU Leuven, we have developed a LEM that allows us to simulate how mountains grow, rivers incise and hillslopes respond to tectonic and climatic forcing. We placed particular emphasis on implementing numerical models that minimize numerical diffusion. To achieve this, Benjamin has adopted a higher order flux limiting total volume method that is total variation diminishing (TVD-TVM) (Campforts and Govers 2015) to solve the partial differential equations of river incision and tectonic displacement.
In our recent manuscript under discussion in ESurf, we show that using these methods is more than just an exercise in numerical modelling. First-order approximations often smooth knickpoints in uncontrollable ways that impact on derived catchment wide erosion rates. Numerical diffusion strongly affects lateral tectonic displacement, thus restricting its simulation to models that use irregular grids and lagrangian approaches. The TVD-TVM approach solves for these issues by reducing numerical diffusion to a minimum, and thus offers a regular-grid based model with wide applications in tectonic geomorphology.
Needless to say that you can directly analyze and visualize the output using TopoToolbox. The implementation comes with a set of examples that you can directly run from the command line. Give it a try and let us know what you think!
Campforts, B., Govers, G., 2015. Keeping the edge: A numerical method that avoids knickpoint smearing when solving the stream power law. Journal of Geophysical Research: Earth Surface 120, 1189–1205. doi:10.1002/2014JF003376
Campforts, B., Schwanghart, W., Govers, G. (2016): Accurate simulation of transient landscape evolution by eliminating numerical diffusion: the TTLEM 1.0 model. Earth Surface Dynamics Discussion, in review. [DOI: 10.5194/esurf-2016-39]
The Himalayan nations’ thirst for energy is increasing. Hydropower is among the most important energy sectors and has experienced massive growth rates during the last years. This growth is expected to further increase in the coming years. More than 400 Himalayan hydropower projects (HPP) are currently registered or at validation for the Clean Development Mechanism (CDM) alone, a global environmental investment and credit scheme to enable countries with emission-reduction commitments to implement emission-reduction projects in developing countries.
In our now published paper in Environmental Research Letters, we show that – as opportune sites along major rivers are already occupied – numerous planned or currently constructed HPP expand into river reaches higher upstream and closer to glacial lakes. These lakes, dammed behind moraines, may sporadically burst upon dam failure and create so-called glacial lake outburst floods (GLOFs). We mapped more than 2000 lakes dammed by terminal or lateral moraines along the Himalayan Arc and calculated outburst scenarios for each using a physically based dam breach model. Since many of the variables of this model are impossible to constrain from remote sensing data alone, we adopted a probabilistic approach and used a Monte-Carlo simulation to derive entire distributions of outburst scenarios. Using TopoToolbox, we determined the flow paths of potential GLOFs from each lake and used these paths as input to a 1D flood propagation model that we fed with the results of the Monte-Carlo simulation. This allowed us to track how uncertainties about lake bathymetry and dam characteristics propagate downstream, and showed that these uncertainties remain significant, for most lakes, until a distance of 80 km downstream. This is the distance within which more and more HPPs are planned or currently constructed.
Clearly, our approach is based on a series of assumptions and limitations, and I encourage you to read the full paper (open access) for a detailed discussion of these. We conclude that the current massive expansion of HPP closer to glaciated areas entails that our lack of knowledge about the stability and geometry of glacial lakes will make it increasingly difficult to estimate potential flood magnitudes and thus compromise reliable risk assessment and safety planning of new HPP projects.
Schwanghart, W., Worni, R., Huggel, C., Stoffel, M., Korup, O. (2016): Uncertainty in the Himalayan energy-water nexus: estimating regional exposure to glacial lake outburst floods. Environmental Research Letters, 11, 074005. [DOI: 10.1088/1748-9326/11/7/074005]
The Pokhara Valley is a very peculiar place in the Lesser Himalayas. Other than the steeply dissected valleys, this place has an unusually broad valley bottom made of tens-of-meter thick sheets of gravel, boulders and lake-like sediments. Nepal’s second largest city and popular tourist destination is built on these deposits.
Inspired and motivated by the 1980s’ works of Monique Fort (Paris Diderot University, Paris) in this area, we had several field campaigns during the last years that aimed at taking a closer look at the Pokhara gravels. Specifically, we were interested in testing the hypothesis of one or several debris-flow and flood events, constraining their chronology by new 14C and 10Be exposure dates, and pinpointing the sediment sources with geochemical sediment fingerprinting.
The results of our study are now published in Science. In brief, we found that Pokhara with its 300.000 citizens is built on 4-5 km3 debris mobilized by three medieval M>8 earthquakes in ~1100, 1255 and 1344 CE. These earthquakes caused catastrophic mass movements from a point source in the Annapurna Massif that traversed >60 km, and invaded tributary river valleys for up to 7 km upstream. This implies that valley fills can offer a unique archive for independently validating both the timing and landscape-changing consequences of large Himalayan earthquakes. Today, rivers still recover from these seismic disturbances by rapidly incising, undermining their banks, and shifting their courses several centuries after.
By the way, the paper is accompanied by another paper by Jeffrey Kargel et al. on the geomorphic consequences of the Gorkha 2015 earthquake. It is definitely worth reading. In addition, a Nature News article by Jane Qiu covers both articles.
Schwanghart, W., Bernhardt, A., Stolle, A., Hoelzmann, P., Adhikari, B.R., Andermann, C., Tofelde, S., Merchel, S., Rugel, G., Fort, M., Korup, O. (2016): Repeated catastrophic valley infill following medieval earthquakes in the Nepal Himalaya. Science, 351, 147-150 [DOI: 10.1126/science.aac9865].
**** edit February 17, 2016 to include final reference for the paper.****
In a previous post, I have highlighted the course Geosimulator at the University of Potsdam led by Ariane Walz. This course was a research-based graduate course focusing on flood risk assessment in the different basins of the Ore Mountains. My task in this course was to teach students how to prepare DEMs for hydrodynamic simulations and how to set up and run the hydrodynamic model LISFLOOD FP.
Central to this course was the use of the university’s 3D lab. This lab offers optical immersion in a 3D CAVE (computer-animated virtual environment) that enables digital 3D objects to hover in mid-air and allows users to move physically around these objects. The CAVE thus provides opportunities for interactive exploration of higher-dimensional data and supports novel ways of gaining insights into complex data structures.
In a now published paper in Journal of Geography in Higher Education we evaluate the possibilities and usefulness of immersive 3D geovisualization in science communication and higher education. We conducted a survey among the Geosimulator students that revealed several benefits of using the 3D CAVE such as better orientation in the study area and higher interactivity with data. In particular, students found that the CAVE enhanced and stimulated discussions in the course and increased motivation, suggesting that working in a 3D lab can effectively enhance the interactive learning among students.
Philips, A., Walz, A., Bergner, A., Graeff, T., Heistermann, M., Kienzler, S., Korup, O., Lipp, T., Schwanghart, W., Zeilinger, G. (2015): Immersive 3D Geovisualisation in Higher Education. Journal of Geography in Higher Education, 39, 437-449. [DOI: 10.1080/03098265.2015.1066314]
In a recent paper, Radoslaw Malinowski et al. investigated ways to detect flooded areas from WorldView-2 image data. While remote sensing of water surfaces is common to delineate the extent of large-scale floods, our study focused on localised inundation patterns as frequently found in lowlands and floodplains during seasonally elevated water levels. These floods are characterized by small inundated patches generated by groundwater seepage, riverine flooding and rain water accumulating in shallow topographic hollows. Frequent and long-lasting inundation alters soil properties and vegetation in these locations and may locally affect infrastructure and agriculture. We tested different flood detection algorithms including methods using object-based image analysis (OBIA) and topographic data in the Nørreå river valley, Denmark. Including topographic data into OBIA turned out to be the approach with the highest overall accuracy. Inundation detection under dense plant canopies, however, proved difficult with all methods.
Malinowski, R., Groom, G., Schwanghart, W., Heckrath, G. (2015): Detection and delineation of localised flooding from WorldView-2 multispectral data. Remote Sensing, 7(11), 14853-14875. [DOI: 10.3390/rs71114853]