Sie verwenden einen veralteten Browser mit Sicherheitsschwachstellen und können die Funktionen dieser Webseite nicht nutzen.
Timescale: Oct.2018 – Sept.2021
Prof. Dr. Ariane Walz, University of Potsdam,
Dr. Sigrid Rössner, GFZ Potsdam,
Prof. Dr. Oliver Korup, University of Potsdam
Among the many natural hazards in the Himalayan region, glacial lake outburst floods (GLOFs) have attracted some of the most research attention in recent decades. New advances in satellite-based monitoring have revealed in detail the pattern of glacier melt along the mountain belt, and ongoing campaigns focus on finding diagnostics of which glacier lakes are most prone to catastrophic emptying, thus releasing potentially destructive GLOFs. While the debate about useful predictors of GLOFs remains heated, in Phase 1 of NatRiskChange we have aimed at systematically expanding the previously censored knowledge of Himalayan GLOFs. For this purpose we developed an automated remote sensing based method using a random forest classifier coupled to a probabilistic change-point model to analyse the full seasonal LANDSAT archive comprising almost three decades of multi-temporal imagery for traces of GLOFs. We thus derived a new database of past events that nearly doubled the existing GLOF record within the analysed study regions.
In Phase 2 of NatRiskChange we plan on taking this study farther by (1) reconstructing the peak discharges of GLOFs at all known, including the newly detected, locations as a function of dam type and geometry; (2) numerically modelling the runout dynamics of selected GLOFs; and (3) obtaining spatially explicit risk estimates from intersecting these results with various data on population density, buildings, and infrastructure. For estimating the peak discharges we test various topographic predictors that are routinely used in dam-break modelling. Preliminary results from a pixel-based simulation of peak discharge for some 2.5 million locations in the Himalayan drainage network reveal a distinctly heavy-tailed distribution of peak discharge depending on parameters such as glacier-dam type, height, local topographic relief, or contributing catchment area. Our analysis of LANDSAT imagery has revealed hundreds to thousands of ephemeral supraglacial lakes and revealed unprecedented insights into glacier dynamics that we consider for distinguishing various mechanisms, or at least sources, of lake outbursts. We also consider improving this analysis by using higher resolution Sentinel-2 imagery, which has been available since 2015. For the numerical runout modelling we will use HEC-RAS, a tried and tested code for flood routing and reconstructing palaeofloods calibrated by flood-stage indicators. Our main target region will be the Pokhara valley, which sustains Nepal’s second largest city, and which was the site of a destructive GLOF event in 2012. We have field-surveyed dozens of channel cross-sections with a terrestrial LiDAR scanner and use these topographic data as input for the modelling. We will consider various inundation scenarios and also include changes in channel geometry brought about by road cutting, bridge building, and gravel mining. Finally, we will have access to dozens of time slices, partly derived from remote sensing analysis showing how the built environment expanded rapidly in the Pokhara region. We will use this information together with OpenStreetMap and LandScan data to quantitatively estimate the degrees of exposure and vulnerability of people, buildings, and infrastructure. These results will feed into a first appraisal of how GLOF risk changes with natural drivers (magnitude and frequency of GLOFs, valley geometry, etc.) and socio-economic drivers (projected population growth, land-use changes, etc.). Such an appraisal is likely to be the first and most detailed of its kind for a major Himalayan watershed. Together with our improved regional assessment of GLOF locations, possible peak discharges, and impact lengths we will also be able to provide some pioneering susceptibility maps for the Himalayas.
Objectives and Methods
Melanie Fischer is based at the research team “Landscape Management” of the University of Potsdam.