The Suli Gad Valley and Lake Phoksundo
The Suli Gad Valley forms a deep gorge between Dunai and Lake Phoksundo (Fig. 3.4). The valley exposes a transect through some of the main geological features of the Himalaya (Fig. 3): its southern end at Dunai is in the Lesser Himalaya (LH), and the rest of the valley is within the Tibetan Sedimentary Zone (TSZ). At Lake Phoksundo, the Tethyan Sedimentary Sequence (TSS) or TSZ is exposed, in particular the mainly Ordovician Dhaulagiri limestone (Fuchs, 1977). The Suli Gad Valley (Fig. 3.4) has been the site of detailed geological and tectonic investigations by Carosi et al. (2002, 2006), but no detailed geomorphological studies have been performed. During our trek through the valley, we identified glacial landforms and fluvial-glacial deposits. Due to the presence of numerous landslides, the water level of the rivers, the dense vegetation and the steepness of the cliffs we were not able to collect samples for dating. In the vicinity of Lake Phoksundo, numerous hummocky moraines cover the hillside of the valley. As the valley opens up, it gives way to a glacial deposit of several hundred metres high, suggesting a glacial knob. At the top of this till deposit, Lake Phoksundo lies between the limestone slopes of the TSS. Lake Phoksundo is situated in the upper course of the Suli Gad River, at the border of the lower and higher Dolpo (Fig. 3.4). It is the second-largest lake of Nepal, with very steep sides plunging to a 135-m deep flat lake bottom (Fig. 3.6), giving the lake a fjord-like appearance.
Figure 3.6. Aerial photograph and bathymetric map of Lake Phoksundo. Yellow and green transparent areas represent frontal moraines and rockslide deposits, respectively. White dots are the locations of 36Cl samples collected by Fort et al. (2013) on the rockslide deposit. The white line is the 2016 front of an unnamed glacier. Thick yellow lines are inferred Little Ice Age (LIA) lateral moraines. The dashed yellow lines represent lateral and hummocky moraines that we attribute to the maximum glacial extent.
The lake has been suggested by Yagi (1977) and Weidinger and Ibetsberger (2000) to result from damming of the Suli Gad River by the collapse of a mountain wall of Dhaulagiri limestone (Fuchs, 1977), culminating at 5148 m SE of the lake (Fort et al., 2013). The collapse, which led to a 4.5-km3 rockslide dam, was suggested to have occurred at 30 to 40 ka (Yagi, 1977), although more recent dating by Monique Fort (results published with this study) has revised this chronology (see below). The rockslide morphology presents a series of complex landforms, including mounds and depressions of varying size. Massive interlocked blocks of several tens of meters dominate the deposit, included in fine-grained material.West of and beneath the rockslide deposit, fine sediments appearing as till/moraine material and overlain by orange conglomerates including metre-sized dolomite boulders, in contrast, suggest a glacial origin.
The Bheri Valley
The Bheri River is a major tributary of the Karnali River, draining the western Dhaulagiri Range in western Nepal (Fig. 3.4) from its source in the Dolpo highlands. Between the villages of Khani Gaun and Kukot, numerous fluvial-glacial deposits, lake sediments and moraines have been preserved (Fig. 4). The valley section between Khani Gaun and Kukkot presents a series of rapid and significant differences in elevation that resemble glacial knobs, rising rapidly from ~2700 to ~3300 m. However, a large number of landslides, which vary greatly in magnitude, also occur on both sides of the river. From Khani Gaun to Mukot, the Bheri River flows along the South Tibetan Detachment (STD) and within the Tethyan Sedimentary Sequence (TSS) (Fig. 3.3). In Kukkot, the valley opens up rapidly and gives way to a relatively large alluvial plain with numerous channels. On the edges of this plain, a series of small fluvial terraces are developed on top of a thick deposit of white lacustrine sediments, mainly composed of clay. The characteristic rhythmic parallel laminated sediment deposit (e.g. Ashley, 2002), which are characteristic of lake bottom deposits, are visible from afar and can be followed along the alluvial plain. The limits of the lacustrine deposits could, however, not be established due to the widespread presence of landslides.
Samples and methods
Our analysis and interpretation is based on field observation and interpretation of satellite imagery, as well as existing and new surface-exposure dating using the Terrestrial Cosmogenic Nuclides (TCN) 10Be and 36Cl.
Field observations and mapping
Field observations were collected both in a notebook and using FieldMove software developed by Midland Valley. Photos were described, geolocated and directly plotted on a map. Geomorphological mapping was also performed using FieldMove, drawing the boundaries of each unit and describing them.
10Be Terrestrial Cosmogenic Nuclide (TCN) surface-exposure dating.
Four samples were collected from quarzitic boulders on the crests of the frontal and lateral moraines exposed east of Lake Rara (Fig. 3.7) for 10Be TCN surface-exposure dating following the protocol outlined below. Monique Fort provided the 36Cl TCN ages of the Phoksundo rockslide; a short description of the dating methods for these samples is available in Fort et al. (2013). Samples were crushed and sieved to obtain the 200–500 μm size fraction. The chemical extraction protocol is adapted from Brown et al. (1991) and Merchel and Herpers (1999) and was carried out at Ghent University, Belgium and the ISTerre cosmogenic laboratory in Grenoble, France. Quartz was isolated through repeated leaching in an H2SiF6-HCl (2/3-1/3) mixture. Meteoric Be was removed with three sequential baths in diluted HF (Kohl and Nishiizumi, 1992). The purified quartz samples (weighing between 14 g and 52 g) were spiked with ∼300 μl of a 1 mg.g−1 Be carrier solution (Scharlab ICP Standard) before being totally dissolved in concentrated HF. After evaporation of HF, perchloric and nitric acids were added and evaporated to remove organic compounds and fluorides, respectively. Anion and cation exchange columns allowed the separation of Fe and Ti and the isolation of the Be fraction. Be hydroxide was extracted by alkaline precipitation (Von Blanckenburg et al., 1996). The final BeO targets were oxidized and mixed with Nb powder prior to loading them on cathodes for Accelerator Mass Spectrometer (AMS) measurements, which were carried out at ASTER, the French National AMS facility at CEREGE, Aix-en-Provence. The measured 10Be/9Be ratios were calibrated against a CEREGE in-house standard, using an assigned value of 1.191 ± 0.01 × 10−11 (Braucher et al., 2015) and a 10Be half-life of 1.387 ± 0.012 × 106 yr (Chmeleff et al., 2010; Korschinek et al., 2010). The 10Be concentrations inferred from the measured 10Be/9Be ratios were corrected for the corresponding full process blank ratios (3.901 × 10−15). AMS analytical uncertainties (reported as 1σ) include the uncertainties associated with the AMS counting statistics, the chemical blank corrections, and the ASTER AMS external error (0.5%; Arnold et al., 2010).
Figure 3.7. (a) View of Lake Rara from the north summit bordering the lake. Moraines are outlined in yellow and the 10Be sample locations are indicated. The dashed arrow indicates the past outflow direction, the plain one the modern outflow. (b, c) Quartzitic boulders in Lake Rara moraine; (b) Example of boulder not sampled, present at the shore of the frontal moraine and (c) sampled boulder on the crest of the frontal moraine.
Exposure ages were computed with the online CREp calculator (Martin et al., 2017; http://crep.crpg.cnrs-nancy.fr). Production-rate scaling to the sample locations was made according to the recent, physically based, LSD model (Lifton et al., 2014), which performs similarly to older empirical models (Borchers et al., 2016). Chosen parameters include the ERA40 atmospheric model (Uppala et al., 2005) and the Lifton-VDM2016 geomagnetic database (Lifton, 2016). Topographic shielding was estimated in the field through skyline survey using a clinometer. We retained the production rate derived by Balco et al. (2009), as no regional production rates are available for the Himalaya (Owen et al., 2010). The Balco et al. (2009) production rate has a value of 3.93 ± 0.19 at g−1 yr−1 for LSD scaling and is consistent with other recently derived Northern-Hemisphere production rates (Fenton et al., 2011; Ballantyne and Stone, 2012; Briner et al., 2012; Goehring et al., 2012; Young et al., 2013; Small and Fabel, 2015; Stroeven et al., 2015). We did not apply erosion- or snow corrections, because we have no indication for significant erosion of the sampled blocks and most sampled boulders lie in windswept locations where little snow would build up.
Reconstruction of the paleo-Equilibrium Line Altitude (ELA)
Using our field observations and satellite imagery, we attempted to reconstruct the paleo-Equilibrium Line Altitude (ELA) during the inferred most recent glacial extension as well as depicting roughly the maximum glacial extension of the glaciers in the vicinity of the Suli Gad and Bheri Valleys (Fig. 3.4). To do so, we used the toe-to-headwall altitude ratio (THAR; Louis, 1955) based on the assumption that the ELA lies at a fixed altitudinal ratio between the lowest and highest altitude (i.e., the headwall) of a glacier (Louis, 1955). We used the THAR value of 0.4 suggested by Owen and Benn (2005) for the reconstruction of paleo-ELAs along the Himalaya.
Reconstruction of the maximum glacial extent
Given the relatively good preservation of lateral moraines compared to that of frontal moraines in the study areas, we used the latter to estimate and reconstruct the maximum glacial extent. As lateral moraines are deposited below the Equilibrium Line Altitude (ELA), the maximum altitude of lateral moraine (MALM) method (Benn and Evans, 1998) assumes that the uppermost elevation of a remnant lateral moraine marks the paleo-ELA (Andrews, 1975; Dahl et al., 2003). However, our intention was not to reconstruct the paleo-ELA given the data at our disposal, but rather to estimate the maximum expansion of the glaciers from the minimum elevation of lateral moraines. In addition, we used the occurrence of cirque lakes as a marker of past glaciations. In both cases, we used both our field observations and satellite imagery.
Results and discussion
Rock samples were taken from the frontal moraine rising about ten metres above the lake (Figs. 3.5 and 3.7). The moraine is composed of eroded boulders on which glacial erosion figures (i.e., striae) can be discerned and is covered by a pine forest. The blocks are metre-sized or larger and their lithology is quarzitic. The whole is consolidated by glacial till.
From the five dated samples, four have valid measurements, whereas one (MS14-05) showed to low and instable current during AMS measurement and is therefore not reported. The 10Be exposure ages of these four boulders are listed in Table 2.1.
Two boulders from the frontal moraine of Lake Rara (MS14-01 and 02; Fig. 2.7; Tab. 2.1) yielded consistent exposure ages of 62 ± 6 ka and 60 ± 12 ka, respectively. The other two boulders (MS14-03 and 04), collected from the lateral moraine, have significantly older ages of 355 ± 22 ka and 242 ± 13 ka, respectively. These unexpectedly old ages raise questions on issues associated with the application of surface exposure dating methods to date moraines in the Himalaya, Tibet and elsewhere, which have been discussed at length in several studies (e.g. Benn and Owen, 2002; Putkonen et al., 2008; Owen and Dortch, 2014). Such issues can be related to uncertainty introduced in the calculation of the TCN production rates, geological complexities affecting surfaces such as weathering as well as previous exposure and shielding of the surface by snow and/or sediments (Owen et al., 2010). In previous studies, Balco et al. (2008) and Owen and Dortch (2014) showed that the uncertainty related with different scaling models for low-latitude and high-altitude areas such as the Himalaya can reach 40% between scaling models over the last glacial cycle. In view of these issues, we applied the Lifton et al. (2014) time-independent production-rate model, acknowledging the uncertainty associated to our calculated ages, and use caution while assigning our numerically dated moraines to a specific climatic stage.
Table of contents :
Chapter 1 ─ Introduction
1.3 Thesis outline
Chapter 2 ─ Glacial and landslide controls on the geomorphology of Lakes Rara and Phoksundo, western Nepal
2.2 Tectonic and climatic context
2.3 Study area
2.3.1 Lake Rara
2.3.2 The Suli Gad Valley and Lake Phoksundo
2.3.3 The Bheri Valley
2.4 Samples and methods
2.4.1 Field observations and mapping.
2.4.2 10Be Terrestrial Cosmogenic Nuclide (TCN) surface-exposure dating.
2.4.3 Reconstruction of the paleo-Equilibrium Line Altitude (ELA)
2.4.4 Reconstruction of the maximum glacial extent
2.5 Results and discussion
2.5.1 Lake Rara
2.5.2 Suli Gad Valley and Lake Phoksundo
2.5.3 Bheri Valley
2.5.4 Reconstruction of the Holocene Equilibrium Line Altitude (ELA)
2.5.5 Reconstruction of the maximum glacial extension
Chapter 3 ─ Large post-1505 AD earthquakes in western Nepal revealed by a new lake sediment record
Origin of turbidites.
Earthquake turbidite-triggering threshold.
Possible correlation with other historical earthquakes.
Significance of previously unknown events.
3.3 Implications for the notion of a seismic gap in western Nepal
Sediment core collection and analysis.
Modelling shaking intensity.
Modelling sensitivity to near-field background seismicity.
3.5 Supplementary Information
Identification of turbidite layers.
Turbidite triggering mechanism.
Chapter 4 ─ Seismic hazard minimized by the cycle concept
Time distribution analysis.
4.7 Supplementary Information
Identification of turbidite layers.
Chapter 5 ─ Correlation between Holocene climate changes and global seismicity
5.2 Data and analysis
5.3 Potential regional variations in temporal distributions
5.4 Holocene paleo-seismicity clustering
5.5.1 Correlation with other natural periodic phenomena
5.5.2 Ice sheets, crustal deformation and seismicity
5.6 Conclusions and perspectives
Chapter 6─Conclusions and Perspectives
On the seismic-gap hypothesis, from a regional to a continental scale.
Himalayan earthquake time-distribution models.
Global seismic modulation through climate changes.