Subglacial Outburst Floods Research Articles

Tracking the Filling, Outburst Flood and Resulting Subglacial Water Channel From a Large Canadian Arctic Subglacial Lake

AbstractWe use digital elevation models (DEMs) and ICESat‐2 data to study the filling and outflow from a large subglacial lake under Manson Icefield in the Canadian Arctic. When full, the lake is ∼17 × 3 km with an area of 52 km2. Early in 2021 the ice surface over the center of the lake sank by >140 m implying a subglacial outburst flood of ∼4 km3. Rapid outflow occurred over ∼30 days at an average rate of ∼1,500 m3s−1 resulting in the formation of a single ∼15 km subglacial outflow path detectable from post‐outflow surface depression. The shape of the surface depression, 600–800 m wide by 2–4 m deep, reflects the shape of the subglacial channel prior to closure. Downstream ice movement appears unaffected by the outflow. After outflow ends the surface depression persisted over weeks, apparently dependent on the difference between water and overburden pressures.

A rare signature of subglacial outburst floods developed along structural ice weaknesses in the southern sector of the Scandinavian Ice Sheet during the Drenthian Glaciation, S Poland

We present the study of a subglacial drainage system that developed in the base of the Scandinavian Ice Sheet at the border of the Silesian Lowland and Silesian Upland (southern Poland) during the Drenthian Glaciation as a result of large outburst flood events. This drainage system was characterised by a spatially complex structure, i.e. it was composed of an individual an N-channel (tunnel channel) incised in Mesozoic rocks, i.e. Cretaceous and Triassic marls and limestones, and an R-channel, deposits of which form a set of large esker ridges which also occur on bedrock in the SE prolongation of the tunnel channel. Unlike most known cases of the co-occurrence of such forms, the studied tunnel channel and eskers are directly related, having been formed during the same flow events. The studied forms are unique within the entire southern sector of the Scandinavian Ice Sheet, which, in Central Europe, advanced mainly over soft unconsolidated Quaternary and Neogene sediments. The tunnel channel was formed as a result of enormous erosion of the ice sheet bed. The material eroded from the channel was partially deposited in a subsequent part of the drainage system. The location and orientation of the drainage system did not result directly from water pressure gradients, but was strongly determined by the internal structure of the ice sheet, i.e. it presumably developed along the marginal part of the ice stream supplying the Upper Odra ice lobe, which represented distinct weaknesses within the ice sheet. The development of the drainage system was the effect of the drainage of meltwater stored in the ice sheet system, most probably as a supraglacial lake. Gravelly-sandy rhythms recorded within the esker succession indicate that the process of lake drainage took place in several stages, or that the lake developed and drained several times.

Estimation of the water discharge during the development of the glacial and subglacial outburst floods.

Estimation of the water discharge during the development of the glacial and subglacial outburst floods.

Оценка расхода воды в процессе развития прорывного паводка ледниковых и подледниковых водоемов

Оценка расхода воды в процессе развития прорывного паводка ледниковых и подледниковых водоемов

Irish Ice Sheet dynamics during deglaciation of the central Irish Midlands: Evidence of ice streaming and surging from airborne LiDAR

High resolution digital terrain models (DTMs) generated from airborne LiDAR data and supplemented by field evidence are used to map glacial landform assemblages dating from the last glaciation (Midlandian glaciation; OI stages 2–3) in the central Irish Midlands. The DTMs reveal previously unrecognised low-amplitude landforms, including crevasse-squeeze ridges and mega-scale glacial lineations overprinted by conduit fills leading to ice-marginal subaqueous deposits. We interpret this landform assemblage as evidence for surging behaviour during ice recession. The data indicate that two separate phases of accelerated ice flow were followed by ice sheet stagnation during overall deglaciation. The second surge event was followed by a subglacial outburst flood, forming an intricate esker and crevasse-fill network. The data provide the first clear evidence that ice flow direction was eastward along the eastern watershed of the Shannon River basin, at odds with previous models, and raise the possibility that an ice stream existed in this area. Our work demonstrates the potential for airborne LiDAR surveys to produce detailed paleoglaciological reconstructions and to enhance our understanding of complex palaeo-ice sheet dynamics.

Ice-proximal Labrador Sea Heinrich layers: a sedimentological approach

During Late Pleistocene Heinrich events (H-events), distinct, decimetre- to centimetre-thick layers of ice-rafted debris (IRD) were deposited in the North Atlantic as Heinrich layers (H-layers). These layers are characterized by high detrital carbonate content, low foraminifera content, a high percentage of Neogloboquadrina pachyderma (sinistral) among the planktonic foraminifera, high magnetic susceptibility, and high grey colour values. In contrast, H-layers in the Labrador Sea reach metre thickness at core sites proximal to the iceberg source off the Hudson Strait ice stream (HSIS), and show low magnetic susceptibility and relatively low grey levels on the colour scale. To provide the reader with some background information, four hypotheses concerning the origin of H-events are discussed at the outset: (1) the binge–purge (internal forcing) model, (2) the subglacial outburst flood model, (3) the external forcing model, and (4) the catastrophic ice shelf breakup model. The higher thickness of ice-proximal H-layers is due to the supply of large amounts of terrigenous sediments that were eroded from country rocks underlying the northeastern sector of the Laurentide Ice Sheet (LIS). These sediments were transported to the deep Labrador Sea by the efficient processes of bottom-following mass and surface plume movement, where they mixed with ice-rafted sediment. Four distinct depositional facies of H-layers (Types I to IV) have been identified: Type I H-layers occur within 300 km from the presumed HSIS terminus and consist of stacked thin layers of graded muds containing IRD. The graded muds that are spiked with IRD are the result of deposition of fine-grained sediment from lofting sediment columns that collected dropstones and grains under the iceberg route. Type II H-layers occur on the slope and rise at a greater distance south of the Hudson Strait outlet, on the levees of tributary canyons to the Northwest Atlantic Mid-Ocean Channel (NAMOC). These layers consist of alternating thin mud turbidites with intercalated laminae of IRD. Type III H-layers exist on the levees of the main channel of the NAMOC, and consist of layers of IRD alternating with fewer fine-grained spillover turbidites, reflecting the lower spillover frequency from the deep channel compared to the less deep slope canyons. Type IV H-layers are made up of bioturbated hemipelagic muds with IRD, and occur in regions between canyons not reached by spillover turbidity currents, and in distal regions of the open ocean or on seamounts. The anomalously high thickness of individual H-layers on the slope and rise off Hudson Strait is explained by the transport of significant portions of H-layer sediment by suspended sediment columns lofted from sand-carrying freshwater turbidity currents (Type I), and by low density turbidity currents (Types II and III). Isopach maps for H-layers 1–3 give hints of the drift routes of the lofted suspended sediment during its ascent to the surface, and of iceberg drift directions in the Labrador Sea.

The geomorphic signature of massive subglacial floods in Victoria Land, Antarctica.

We describe a landscape in the Dry Valleys area consisting of channels, potholes, plunge pools, stripped bedrock, and scabland created by subglacial outburst floods on a scale similar to that of the massive outbursts of Lake Missoula in Washington, United States. The features are dated to the mid-Miocene and occurred when a thicker East Antarctic ice sheet overrode this part of the Transantarctic Mountains. Ice sheet modeling is used to reconstruct potential source areas for meltwater discharge. Results suggest that meltwater from Lake Vostok, or other interior basins, would have bypassed the Dry Valleys area and instead flowed out beneath the troughs of David and Byrd glaciers, north and south of the Dry Valleys, respectively. A more likely source might have been lakes on the inner flank of the Transantarctic Mountains, with periodic outbursts associated with local breaches in the cold-based rim of the ice sheet. The efficacy of erosion by such outbursts is remarkable. The most concentrated effects were in the labyrinth on the bed of the former Wright outlet glacier. Such outbursts are likely to contribute to the erosion of present-day outlet glacier troughs.

Quantifying the predictability of the timing of jökulhlaups from Merzbacher Lake, Kyrgyzstan

AbstractGlacier-dammed lakes can yield subglacial outburst floods (jökulhlaups) repeatedly. Predicting flood timing is crucial for hazard mitigation, but incomplete understanding of flood-initiation physics makes this challenging. Here we examine the predictability of the timing of jökulhlaups from Merzbacher Lake, Kyrgyzstan, using five flood-date prediction models of varying complexity. The simplest model, which offers a benchmark against which the other models are compared, assumes that floods occur on the same date each year. The other four models predict flood dates using a flood-initiation threshold approach and incorporate weather forcing (approximated by the output of two climate reanalyses) behind the meltwater input to the lake; the most complex of these models accounts for a moving subglacial water divide beneath the glacier that dams the lake. Each model is optimized against recorded flood dates to maximize its prediction ability. In terms of their flood prediction ability, our two best models are those that assume a variable outburst threshold governed by the rate of meltwater input to the lake and the rate of lake-level rise. They excel over the simplest and most complex models and correctly predict flood dates to within ±20 days 57.4% of the time. We also quantify the impact of weather uncertainty on prediction success. Our findings can inform practical flood-forecasting schemes and future investigations of flood-initiation physics.

Hypothesis for mega-outburst flooding from a palaeo-subglacial lake beneath the East Antarctic Ice Sheet

Subglacial outburst floods at the margins of the Antarctic and Laurentide Ice Sheets have been linked to changes in global ocean circulation and climate. The impact of palaeo-hydrological systems beneath the East Antarctic Ice Sheet (EAIS) has, however, remained elusive. By analysing bed morphology in the Wilkes Subglacial Basin, we hypothesise the occurrence of a major palaeo-subglacial lake and associated outburst floods in the interior of East Antarctica. At similar to 70 km wide and over 100 km long, the inferred area of outburst flooding is the most extensive in Antarctica, and the palaeo-subglacial lake from which the floods originated is the second largest after Lake Vostok. The scale of inferred outburst flooding is similar to mega-floods beneath the Laurentide Ice Sheet. We suggest that this major hydrological system developed during expansion of the EAIS in the middle Miocene and probably affected ice sheet stability, ocean circulation and climate evolution.

Investigation of Li/Ca variations in aragonitic shells of the ocean quahog Arctica islandica, northeast Iceland

Interannual and intra‐annual variations in lithium‐to‐calcium ratio were investigated with high temporal resolution in the aragonitic outer shell layer of juvenile Arctica islandica (Mollusca; Bivalvia) collected alive in 2006 off northeast Iceland. Li/Cashell ranged between 7.00 and 11.12 μmol mol−1 and presented well‐marked seasonal cycles with minimum values recorded at the annual growth lines; a general pattern was a progressive increase in Li/Cashell from March to May, followed by a plateau in June and a decrease down to minimum values in July–August. Li/Cashell was correlated with δ18Oshell‐derived temperature, but the strength of this relationship was weak (r2 < 0.25 and p < 0.05). It covaried significantly with microgrowth increment width and with the discharge from one of the closest rivers. Seasonal variations of Li/Cashell in A. islandica may most likely be explained (1) by calcification rate and/or (2) by significant river inputs of Li‐rich silicate particles flowing to the sea as soon as snow melts. In the first case, Li/Cashell may be a useful proxy for addressing seasonal variations of growth rate in bivalves that lack discernable microgrowth patterns. Abrupt decreases of Li/Cashell may, in turn, help identify growth retardations due to harsh environmental conditions. Alternatively, if Li/Cashell variations are linked to particulate Li inputs by rivers, this could be a new proxy for the intensity of mechanical weathering of Icelandic basalts, with interesting perspectives for the reconstruction of frequency and intensity of past jökulhlaups (subglacial outburst floods). Further works, including experimental studies, are needed to test these hypotheses.

Glacier ecosystem response to episodic nitrogen enrichment in Svalbard, European High Arctic

We describe the climatology, hydrology and biogeochemistry of an extreme nitrogen deposition event that occurred in the highly glacierised environment of the European High Arctic during June 1999. Meteorological analysis, three-dimensional air mass trajectories and a 3D transport model show that blocking high pressures over Scandinavia and the rapid advection of western European pollution toward Svalbard were sufficient to cause the most concentrated (1.15 ppm NO3–N and 1.20 ppm NH4–N), high magnitude (total 26 mm and up to 2.4 mm h−1 at 30 m above sea level) nitrogen deposition event on record in this sensitive, high Arctic environment (78.91° N, 11.93° E). Since the event occurred when much of the catchment remained frozen or under snow cover, microbial utilisation of nitrogen within snowpacks and perennially unfrozen subglacial sediments, rather than soils, were mostly responsible for reducing N export. The rainfall event occurred long before the annual subglacial outburst flood and so prolonged (ca. 10 day) water storage at the glacier bed further enhanced the microbial assimilation. When the subglacial outburst eventually occurred, high runoff and concentrations of NO3 − (but not NH4 +) returned in the downstream rivers. Assimilation accounted for between 53 and 72% of the total inorganic nitrogen deposited during the event, but the annual NO3 − and NH4 + runoff yields were still enhanced by up to 5 and 40 times respectively. Episodic atmospheric inputs of reactive nitrogen can therefore directly influence the biogeochemical functioning of High Arctic catchments, even when microbial activity takes place beneath a glacier at a time when terrestrial soil ecosystems remain frozen and unresponsive.

Climatic control on the peak discharge of glacier outburst floods

Lakes impounded by natural ice dams occur in many glacier regions. Their sudden emptying along subglacial paths can unleash ∼1 km3 of floodwater, but predicting the peak discharge of these subglacial outburst floods (‘jökulhlaups’) is notoriously difficult. To study how environmental factors control jökulhlaup magnitude, we use thermo‐mechanical modelling to interpret a 40‐year flood record from Merzbacher Lake in the Tian Shan. We show that the mean air temperature during each flood modulates its peak discharge, by influencing both the rate of meltwater input to the lake as it drains, and the lake‐water temperature. The flood devastation potential thus depends sensitively on weather, and this dependence explains how regional climatic warming drives the rising trend of peak discharges in our dataset. For other subaerial ice‐dammed lakes worldwide, regional warming will also promote higher‐impact jökulhlaups by raising the likelihood of warm weather during their occurrence, unless other factors reduce lake volumes at flood initiation to outweigh this effect.

Stratigraphic and sedimentological evidence for late Wisconsinan sub-glacial outburst floods to Laurentian Fan

Sub-glacial meltwater produces a distinctive stratigraphic and sedimentological response on the continental margin. Seismo-stratigraphy of Laurentian Channel reveals thick till deposits at its seaward end that pass laterally into stratified sediment in deeper basins, that may record periods of water build up beneath the ice. Two scales of meltwater discharge are recognised: large scale that caused catastrophic erosion and transported large volumes of coarse sediment to the abyssal plain and smaller scale, yielding principally muddy sediment. Sub-glacial outburst floods from the Laurentian Channel ice stream delivered distinctive red sediment derived from Permian–Carboniferous strata of the Gulf of St. Lawrence directly to Laurentian Fan between ca. 17 and 14 14C ka, separate from North Atlantic Heinrich events. On levees of Laurentian Fan, three major pulses of meltwater plume muds are separated by intervals dominated by hemipelagic sediments. These meltwater intervals are recognised distally as periods of plume sedimentation on the Scotian Slope and ice-rafting of hematite-stained quartz to the North Atlantic Ocean. In channels of Laurentian Fan, at least one major sediment transport event is recognised that eroded the upper slope and the major fan valleys, depositing a bed of gravel at least 3 m thick in the characteristically wide fan valleys and thick sand on the Sohm Abyssal Plain. The same event was probably responsible for giant flute-like scours. The age of the gravel bed is directly constrained only by the presence of local overlying Holocene sediment. Much of the surface of the gravel bed was re-worked by the 1929 “Grand Banks” turbidity current. An erosional event on the upper slope, likely correlative with the flood-generated gravel bed, has been dated at 16.5 14C ka. Such large scale erosional flood events can be recognised back through several glacial cycles and have played an important role in the architectural evolution of Laurentian Fan.

Hydraulics of subglacial outburst floods: new insights from the Spring–Hutter formulation

AbstractUsing a slightly modified form of the Spring–Hutter equations, glacial outburst floods are simulated from three classic sites, “Hazard Lake”, Yukon, Canada, Summit Lake, British Columbia, Canada, and Grímsvötn, Iceland, in order to calibrate the hydraulic roughness associated with subglacial conduits. Previous work has suggested that the Manning roughness of the conduits is remarkably high, but the new calibration yields substantially lower values that are representative of those for natural streams and rivers. The discrepancy can be traced to a poor assumption about the effectiveness of heat transfer at the conduit walls. The simulations reveal behaviour that cannot be inferred from simplified theories: (1) During flood onset, water pressure over much of the conduit can exceed the confining pressure of surrounding ice. (2) Local values of fluid potential gradient can differ substantially from the value averaged over the length of the conduit, contradicting an assumption of simple theories. (3) As the flood progresses, the location of flow constrictions that effectively control the flood magnitude can jump rapidly over large distances. (4) Predicted water temperature at the conduit outlet exceeds that suggested by measurements of exit water temperature.

Palimpsest tunnel valleys: evidence for relative timing of advances in an interlobate area of the Laurentide ice sheet

AbstractDuring retreat from the lateWisconsinan maximum advance in the Great Lakes region of North America, the Laurentide ice sheet margin became distinctly lobate. The Lake Michigan, Saginaw, and Huron—Erie lobes converged in southern Michigan and Indiana, U.S.A. to form a complex interlobate region. Some time after the glacial maximum, the Lake Michigan lobe advanced over landscapes previously formed by the Saginaw lobe. This can be explained by an asynchronous advance of the Lake Michigan lobe during a Saginaw lobe retreat or by an increase in size of the Lake Michigan lobe relative to the Saginaw lobe during a synchronous readvance.Cross-cutting relationships in southwestern Michigan, including palimpsest tunnel valleys, document the overriding of Saginaw lobe terrain. Deep, generally straight trenches parallel to glacial flow lines with hummocky, irregular sides and bottoms are interpreted as tunnel valleys. Saginaw lobe tunnel valleys trend northeast—southwest and Lake Michigan lobe tunnel valleys generally trend east—west.At some time after a Saginaw lobe retreat in southern Michigan, the drumlinized landscape was overridden by an advance of the Lake Michigan lobe to an ice-marginal position at the Tekonsha moraine. Saginaw lobe tunnel valleys in the overridden area were completely filled with ice and debris from the Saginaw lobe retreat at the time of the Lake Michigan lobe advance. Supraglacial and proglacial sediments were deposited over the buried valleys by the Lake Michigan lobe, sometimes by meltwater streams that flowed at high angles to the trends of the valleys. After entrenchment of the Kalamazoo River valley, probably by a subglacial outburst flood, short tributaries were cut nearly at right angles across and through the debris and ice within several buried Saginaw lobe tunnel valleys. After the retreat of the Lake Michigan lobe, subsequent melting of ice in the palimpsest tunnel valleys exhumed the valleys, creating the cross-cutting relationships with the Lake Michigan lobe deposits.

Evidence for catastrophic subglacial meltwater sheetflood events on the Bruce Peninsula, Ontario

The Bruce Peninsula, a carbonate bedrock escarpment, lies "downflow" from a sculpted bedrock terrain at the French River. The sculpted forms are attributed to a hypothesis of erosion by regional-scale, subglacial meltwater flooding. This paper presents new data from the Bruce Peninsula that tests the meltwater outburst hypothesis in a downflow direction of the predicted flood path. The bedrock surface of the Bruce Peninsula shows extensive development of sculpted features that bear a striking resemblance to s-forms at the mouth of the French River. They are self-similar and hierarchical in scale, ranging in dimensions from a few centimetres to several kilometres. Remarkable concentrations of potholes are located near the brow of the escarpment. The Bruce Peninsula lacks a pervasive cover of unconsolidated sediment. What little sediment exists has been modified into long, narrow drumlins. The Niagara Escarpment on the peninsula has been back wasted into the edge of the Paleozoic Michigan Basin. Along its east-facing slope, the escarpment is marked by more overdeepened reentrant valleys and intervening promontories than is normal for the rest of the escarpment. Clusters of rounded, percussion-marked boulders of exotic origin are concentrated at the heads of the reentrant valleys. Taken together, these features are inferred to support the hypothesis that subglacial outburst floods beneath the Laurentide ice sheet crossed Georgian Bay and strongly sculpted the Bruce Peninsula. The consistent orientation of the reentrant valleys, aligned with the French River sculpting across the basin to the northeast, and the backwasting of its caprock attest to the power and directional stability of the sheetfloods.

Erratum: Laurentide subglacial outburst floods: landform evidence from digital elevation models

Erratum: Laurentide subglacial outburst floods: landform evidence from digital elevation models

Regional-scale meltwater erosion and deposition patterns, northern Quebec, Canada

Glaciofluvial ridges, several hundred kilometres long, are commonly referred to as interlobate moraines because they appear to have formed at the convergence of two distinct ice lobes. Flow convergence is indicated by patterns of striations, streamlined forms and eskers. The so-called interlobate moraines are also thought to have formed asynchronously as the ice margin retreated. By contrast, we argue that the Harricana moraine of northern Quebec, Canada, formed following flow convergence in a regional-scale subglacial outburst flood. Flowlines constructed from streamlined bedforms mapped on the glacial map of Canada, reinterpretation of these streamlined forms as products of meltwater erosion, and field records of erosional marks (S-forms) in bedrock and glaciofluvial deposits to the lee of bedrock highs support this model. The effects of this flow convergence on the ice-sheet topography and drainage controlled the location of the broad conduit in which the Harricana moraine was deposited. Continued flow in this conduit and melting of the conduit walls explain the local patterns of striae, the supply of debris to the conduit, and the morphological and sedimentary characteristics of the moraine itself. From these characteristics, we conclude that the moraine was formed synchronously. This conclusion, if correct, is instructive regarding the deglacial hydrological organization of a large sector of the Laurentide ice sheet.

Regional-scale meltwater erosion and deposition patterns, northern Quebec, Canada

Glaciofluvial ridges, several hundred kilometres long, are commonly referred to as interlobate moraines because they appear to have formed at the convergence of two distinct ice lobes. Flow convergence is indicated by patterns of striations, streamlined forms and eskers. The so-called interlobate moraines are also thought to have formed asynchronously as the ice margin retreated. By contrast, we argue that the Harricana moraine of northern Quebec, Canada, formed following flow convergence in a regional-scale subglacial outburst flood. Flowlines constructed from streamlined bedforms mapped on the glacial map of Canada, reinterpretation of these streamlined forms as products of meltwater erosion, and field records of erosional marks (S-forms) in bedrock and glaciofluvial deposits to the lee of bedrock highs support this model. The effects of this flow convergence on the ice-sheet topography and drainage controlled the location of the broad conduit in which the Harricana moraine was deposited. Continued flow in this conduit and melting of the conduit walls explain the local patterns of striae, the supply of debris to the conduit, and the morphological and sedimentary characteristics of the moraine itself. From these characteristics, we conclude that the moraine was formed synchronously. This conclusion, if correct, is instructive regarding the deglacial hydrological organization of a large sector of the Laurentide ice sheet.