Australia: The Land Where Time Began

A biography of the Australian continent 

Deglaciation and seismic activity

According to the author¹ there is strong evidence linking deglaciation and seismic activity, with the rebound of the lithosphere and increased earthquake activity, though it is still not possible to pin down the exact role deglaciation played in either the 1356 destruction of Basle of the early 19^th century or the seismic sequence that occurred in New Madrid in 1811/1812. During the last ice age ice sheets had spread across Europe and North America, stabilising the active faults beneath them and suppressing earthquake activity. This effect can be seen in Greenland and Antarctica where there are presently low levels of seismic activity, though it has yet to be shown if this lack of activity is the result of earthquake faults in the underlying crust, or they are present but locked by the massive weight of the ice.

According to the author¹ there is a large consensus that a combination of spectacular uplift and increased earthquake activity, during which accrued tectonic strain on faults beneath and adjacent to ice sheets is released, results from the melting of large ice sheets. The author¹ suggests Lapland, a transnational wilderness region stretching across Arctic Scandinavia and Finland is probably the best evidence for this relationship. The thickest part of the Fennoscandian Ice Sheet was located here, and it is this place where the relationship between faulting and surface features, such as meltwater channels that are indicative of ice wastage, indicate that major faults moved within a few years to a few decades of being uncovered by the melting of the overlying ice, a response that is virtually geologically instantaneous. It is suggested by the size of the faults, up to 150 km long, and in some cases with movement of 15 km in a single event, that hey would have been the source of some really enormous earthquakes higher than magnitude 8. Similar faults, though on a lesser scale, have been reported from Canada. This suggests that the rebound stresses following the melting of the Fennoscandian Ice Sheet appear to be higher than those associated with the melting of the ice covering much of North America.

Difficulties arise when trying to fix relationships between the earthquake faults and the surface features associated with the melting of the ice sheets when searching for evidence beyond the limits of ice sheets for an indisputable increase of earthquake activity associated with deglaciation. Faults are searched for that cut across sediment of the Late Pleistocene or Holocene, or landforms characteristic of the so-called periglacial environment of the land areas around the ice sheets, though accurate dating of the faults and determining whether the cause was tectonic or glacial rebound is extremely difficult. Increased earthquake activity resulting from deglaciation can also be looked for in the sediments, marine and freshwater, the sequence being looked at to determine if it has been subjected to disturbance by severe shaking. Even moderate earthquakes can stir up unconsolidated sediment, down to about magnitude 5, thereby providing one of the best indicators of post-glacial volcanic activity. It has been found that lake sediments in Canada, Sweden and Scotland deposited in the latest stages of the last ice age, or shortly after, all contain evidence of a strong seismic response to wholesale ice sheet melting.

When evaluating the current risk of earthquake, especially in the stable interior of Europe and North America, it is essential to know the extent, if any, the current pattern of earthquake activity and fault movement is a reflection of post-glacial-bounce-back. Are the interiors of Europe and North America more subject to fault movement and earthquake prone as a result of post-glacial-bounce-back?

It has been suggested by Robert Muir-Wood that the patterns of current earthquake activity around the margins of the former ice sheets, those of the UK being included, can be explained by the combination of stresses related to tectonics and crustal rebound. According to Muir’s suggestion the glacial rebound forces reinforce the background stresses that result from the movement of tectonic plates in some places around the former ice sheets. The 2 forces work against each in other places, ‘shadow zones’, resulting in the suppression of earthquakes. Resulting from this interplay characteristic earthquake patterns have been found in the forebulge surrounding the edge of the former ice sheets that is comprised of 2 zones, called quadrants that are seismically active, alternating with 2 quiescent quadrants in which there is little seismic activity as well as in the lithosphere beneath the ice sheets. In the lithosphere surrounding the British ice sheets the more seismically actively zones cover parts of central and western Scotland, though there is a more notable swaths across almost the entire area of England and into the northern part of France. There is significantly more earthquake activity in the area of this swath than to the west in Ireland or to the east, the North Sea, magnitude 5 earthquake occurring about every decade, the latest of which was in Market Rasen in Lincolnshire in 2008.

In the 14^th and 16^th centuries earthquakes occurred in the Dover Straits, in offshoots of the north-western Rhine Graben that extends into the quadrant that is seismically active, with the result that this provides the opportunity for the stresses arising from glacial rebound to interact with those associated with stretching across the Graben, that possibly explains these events and providing the conditions required for future earthquakes. According to the author the low level of earthquake activity in northern Germany is explained by use of the Muir-Wood model by its location within one of the shadow zones that are seismically quite, associated with the Fennoscandian Ice Sheet. During the Early Holocene there was a bout of earthquake activity further to the south, as well as the large historical earthquakes that are historical, such as the Basel quake, the explanation of which has been suggested to be their location in one of the seismically active zones associated with the reduction of the ice mass covering the Alpine Region, one of the smaller ice sheets. The earthquakes that occurred in New Madrid and Charleston, South Carolina earthquake of 1886 are all located in a Muir-Wood seismically active quadrant southeast of the Laurentide Ice Sheet.

Weight of water

In 1962 construction began on a dam across the Koyna River in the Indian state of Maharashtra at a point where it flows through a deep valley 200 km to the south of Mumbai, with the intension of providing a secure water supply and generating electricity. The water impounded by the dam formed Lake Shivajisagar that occupied a 50 km stretch of the valley. The area around the dam was considered to be geologically stable, the area not being known as one that was prone to earthquakes. Seismic activity began to be noticed once the dam was filled, the size and frequency of the tremors increased from 1963, the magnitude of the tremors increasing until they were exceeding magnitude 3 in November and December 1967, comparatively small compared to earthquakes in seismically active areas, they were occurring in an area that was not previously seismically active. On 11 December a magnitude 6.3 earthquake struck near the site of the dam, though the dam held it had undergone a severe shaking. Though the water wasn’t released catastrophically from the dam there were more than 200 deaths and up to 2,000 injured as a result of the earthquake. The relatively low number of deaths was more the result of the low population density of the area. There was a large amount of damage in nearby villages such as Koyna Nagar. Fissures, rubble and landslides cut bridges and roads, though the power station within the dam survived relatively well, but power supplies to Poona and Mumbai were interrupted.

Problems with storing water

The Hoover Dam on the Colorado River was constructed in 1931to form Lake Mead, the largest reservoir in The USA with more than 35 km³ of water and a surface area of almost 650 km². Before the dam was built the area had no record of earthquake activity, as was the case with the Koyna Dam, but as the dam was filling there were more than 600 earthquakes in the area of the dam by the time the water level reached 150 m in 1939, the biggest of which being magnitude 5.

According to the author¹ it is now apparent that such a response in the underlying crust is not restricted to a few localities with  more than 100 dams being built in previously seismically quite areas where seismic activity increased following the construction of the dam. In the 1950s damaging earthquakes occurred as the Kariba Dam was being built on the Zambezi River between Zimbabwe and Zambia. At the time of its construction the Kariba Dam reservoir behind it was the largest volume of water in a dam in the world, with a volume of 180 billion metric tonnes of water. Strong seismic activity accompanied the filling of the dam, with more than 20 of them being higher than magnitude 5. The Hsingfengkiang Dam in China triggered a magnitude 6 earthquake that caused serious cracks to appear in the dam in 1962. In 1966 when the Kremasta reservoir in Greece was accompanied by a sequence of hundreds of earthquakes that culminated in an earthquake of almost magnitude 6.

Pressure addition

The author¹ suggests it is now apparent that the accumulation of large volumes of water behind dam walls is very effective at triggering earthquakes, apparently in the opposite sense to that of glacial activity. The increase of seismic activity along rejuvenated faults as deglaciation occurred at the end of the last ice age. In the case of large dams it is the addition of water that is responsible for triggering the seismic activity. At first sight it appears that the weight of  the addition of large volumes of water is causing the earthquake activity associated with the construction of dams, but research has shown that the commencement of seismic activity is not the result of the weight of the water, no seismic response being detected at the deepest part of the dam, even though it is the part of the dam that has the greatest weight of water. Most earthquakes have been found to be occurring around the margins of the reservoir. The author¹ suggests this indicates that the faults directly beneath the water are being stabilised by the weight of the water as occurred during the glacial phases when ice sheets stabilised them, though faults around the margins are encouraged to become active. Support has been found for this proposal in the Pakistani section of the Himalayas where the Tarbela reservoir has dammed the Indus River. In this reservoir it has been found that earthquake activity on a small scale on faults beneath the reservoir occur more often during the dry season when the water level is low, suggesting that the weight of the water stabilises the underlying faults when the reservoir is full. The reason for the siting of the earthquakes around the margins has been found to result from water being forced into fractures of all sizes by the pressure of the water when the reservoir begins to fill. The rock is weakened by the new water being forced into it increasing the pressure applied to the pore water that is already present. The increased fluid pressure on any faults that are present and at a suitable critical state can trigger movement, causing an earthquake by the reduction of the stresses that have until that point is reached kept the fault stable. According to the author¹ when water infiltrates the rock it raises the pore pressure and this lubricates the faults allowing them to move more easily.

According to the author¹ in some cases it is possible to observe a pattern of the results of changing pore pressure and changing levels of seismic activity as the level of seismic activity changes in line with the pore pressure, the seismic activity increasing as the reservoir fills and decreasing as the water level falls, as in a dry season. More than 40 yeas after the construction of the Koyna Dam the levels of seismic activity are still raised above those prior to the construction of the dam. The dam at Koyna has been studied by Pradeep Talwani, of the University of South Carolina, a founder of the study of ‘reservoir induce seismicity’ RIS, finding that the levels of seismicity display a recognisable pattern, seismic activity increasing in the wet season, June-August, and decreasing in the dry season covering the remainder of the year. In the rainy season when the lake water levels are high the largest earthquakes occur 6-8 weeks after the water level begins to rise. The proposed link between the water level in a large water body and the level of seismicity in the local area is strongly supported by the observed timing of water level rise and the rise of seismicity; the lag following the rising of the water and the onset on increased seismicity is explained, according to the author¹, by the time required by the increasing pore pressure in the rocks on the margins of the reservoir to achieve sufficiently high values to promote the movement of faults. He says causal relationships are commonly recognised between changing water levels, 2 examples of which have been observed at Lake Meade and the Lake Monticello reservoir in South Carolina.

In 2008 a 7.9 earthquake struck in Wenchuan County in China's Sichuan Province causing 70,000 deaths and making 4 million homeless. In China large, destructive earthquakes are common, including some of the most lethal and destructive to be recorded. Seismic strain accumulates slowly in the area of this large quake, with no major earthquakes being known of in the Holocene, though Wenchuan had experienced more than 6 earthquakes prior to the ‘big one‘of 2008. The Zipingpu Dam had been opened 4 years before the 7.9 quake, the epicentre of which was 21 km from the dam. The reservoir was 120 m deep and was suggested by many as the cause of the magnitude 7.9 earthquake that was one of the most devastating in the first decade of the 21^st century. According to the author¹ research was carried out by Christian Klose of Columbia University, the findings of the study suggesting the earthquake could have been triggered by the 300 million tonnes of water in the reservoir. According to Klose’s proposal the altered crustal stress may have brought closer on a fault that was already in a critical state. This proposal is strongly disputed, with other researchers arguing that the dam may have caused small, more local quakes but could not have been even partly responsible for initiating the 7.9 earthquake that occurred at depth of 20 km.

According to the author there has been much speculation that the Three Gorges Dam in Hubei Province, China, with its 40 km³ of water and a length of 600 km, could possibly promote a seismic response that was serious enough to cause the dam wall to give way. As the water level in the dam rose, reaching 175 m in 2009, the predicted seismic level of activity has increased, Though it is not yet known if the seismic activity will reach a level at which the dam wall will be damaged, or in the worst case, give way.

It is now well established that when pore pressures in rocks are forced up, but in spite of this water is being pumped into the crust at high pressure to tap geothermal energy. At Basle in Switzerland a borehole was drilled to access geothermal energy on the 650^th anniversary of the earthquake that flattened Basle in 1356. Following a magnitude 3.4 earthquake, that caused $US 9 million in damage, which was followed by other earthquakes, the project was eventually shut down to prevent further earthquakes. The aim of the project had been to use high pressure water to open fractures to make the rock more permeable so that as more water was pumped into the ground it would be heated and extracted at the surface where the steam would be used to drive turbines.

A wetter world  

As the global temperature rose rapidly at the close of the last ice age the vast ice sheets melted and the meltwater caused sea levels to rise and the climate became more humid and warmer. As the continental ice sheets melted and drained into the oceans the crust readjusted to the unloading with an increase in earthquake activity. Places where large bodies of meltwater accumulated were also characterised by a seismic response, the place which the author¹ suggests is probably the best known location for this association is the Lochaber area in the western Scottish Highlands, are the valleys of Glen Roy and Glen Spean near the town of Fort William and close to Ben Nevis. Characteristic of these valleys are the parallel horizontal lines following the contours of the topography that are preserved shorelines, that were incised into the sides of the valley by glacial meltwater lakes at various levels, that were fed by glaciers covering much of Scotland in the last ice age, and that drained catastrophically 12,000-11,000 years ago during the Younger Dryas when the Scottish glaciers that had been retreating once again advanced. These ‘parallel roads of Glen Roy’ have been the source of a number of legends in the vacuity. According to one of these the Celtic giant Fingal carved them to facilitate his hunting trips. Another legend says the ancient kings of Scotland constructed them for unknown reasons.

Over the last 150 years there have been many studies of these shorelines, as well as the deposits associated with them, that have resulted in a detailed picture of the lakes and their evolution being constructed.  Ice was the source of these lakes and also formed the barriers that dammed them. During the Younger Dryas about 11,500-12,000 years ago during which glaciers that had been retreating in Scotland began advancing again. The warming of the Holocene caused the ice to retreat again after a bit more than 500 years, the warming eventually melting the ice component of the barriers impounding the water allowing the lakes to empty. Analysis of ancient lake sediments and shorelines by Adrian Palmer et al. of Royal Holloway College, University of London, and the results indicate how the water levels in the lakes rose progressively as the glaciers retreated. The water levels fell as lakes were drained in 3 catastrophic drainage events similar to the jökulhlaups of Iceland.

These lake sediments show evidence that as the water levels were dropping they were strongly churned, a feature of the sediments first noted during a study in the 1980s by Philip Ringrose, that he attributed to liquefaction of the sediment, that was soft and waterlogged, as the result of an earthquake. The author¹ suggests it is possible that as a result of unloading as the ice dams melted faults released by the lessening pressure could have moved to cause earthquakes of up to magnitude 6. An alternative explanation that was also thought possible by Ringrose and others is that the seismic activity indicted by the sediments was actually triggered by the sudden reduction of pressure above the faults as the water broke through the ice dams. As has been demonstrated by the Tarbela reservoir in Pakistan, high lake levels would have stabilised the underlying faults, the loss of the lake water reducing the pressure on the faults that could then move more easily.

In North America the effects on earthquake activity of sudden lake discharge has also been recognised, though on a much larger scale. The Basin and Range Province, a vast region of narrow mountain ranges that are separated by broad valleys aligned in a north-south direction, are situated just to the east of the Rocky Mountains. Some of the faults that bound the mountain chains and valleys display evidence that is believed to indicate they were particularly active at the end of the last glacial phase. This area was not buried beneath a large ice sheet during the ice age so the large-scale melting of ice cannot be an explanation for the increased seismic activity. Such an ice cap covered the Yellowstone Plateau that is nearby, and much of the region was covered by 2 very large lakes, the largest being Lake Bonneville, ancestral to the Great Salt Lake of the present. The lake that was about the same size as Lake Michigan, the second largest of the Great Lakes covered an area of more than 52,000 km² at its greatest extent, at which time it covered most of Utah and some of Nevada and Idaho, with a depth of 350 m, being present for almost 20,000 years it was long-lived, though it eventually succumbed to higher temperatures and reduced precipitation. The water level reached a peak about 17,000 years ago at which point it breached the dam, draining through Red Rock Pass in Idaho as an immense flood. It has been estimated to have formed a wall of water more the 120 m high moving at more than 110 km/hour across the Snake River Plain basalt gouging the Snake River Canyon to a depth of 180m on its way to the Columbia River, and eventually into the Pacific Ocean. A difference between Lake Bonneville, and the smaller Lake Lahontan to its west, and the much smaller lakes in Glen Roy and adjacent valleys was that whereas the Scottish lakes were glacier fed, the American lakes were not.  The water of the American lakes came from a change in the climate to conditions where there was less evaporation and increased rainfall leading to the accumulation of water in areas that would previously have been dry. According to the author¹ evidence has been found that there have been dozens of predecessors of Lake Bonneville over the previous 3 million years apparently indicating that there were previously many times when there were conditions conducive to the formation of lakes.

There is a clear coincidence between the shrinkage of Lakes Bonneville and Lahontan and an outbreak of earthquake activity on local faults, as in the case of meltwater lakes in the Scottish Highlands. The relationship between the large lakes of the Basin and Range Province from the Late Pleistocene, and the activity of the local faults, such as the Wasatch Fault along the eastern margin of Lake Bonneville, has been studied by Andrea Hampel, Ruhr University, Bochum, Germany, with others. The Wasatch Fault is a very active major tectonic structure almost 400 km long on which there have been powerful earthquakes about every 350 years, and according to the author¹ many have concerns that the next earthquake could severely damage Salt Lake City, as well as other population centres along its length, especially as the buildings of the area are constructed on the soft sediments of the floor of the former Lake Bonneville. According to the author¹ a strong seismic shaking is likely to cause liquefaction of these sediments and at least 1 estimate puts the possible death toll to as much as 6,000 and damages up $40 billion. When Lake Bonneville was full during the Late Pleistocene the Wasatch Fault was quiescent, then 17,000 years ago the lake volume was reduced catastrophically and a burst of earthquake activity followed soon after.  According to Hampel et al. this resulted from slippage of parts of this fault at double the normal rate that occurred at this time; a direct consequence of the load reduction as the weight of the water dropped. Once unloading occurred the crust rebounded by up to 70 m, and it was this that freed the fault and initiated a burst of earthquake activity. At Lake Lahontan, that held a lower volume of water, the rebound that resulted from catastrophic lowering of the volume led to a rebound of not more than 20 m and a seismic response of faults in the vicinity of this lake that were similar though less marked than on the Wasatch Fault.

Alaska

The author¹ suggests there is plenty of evidence of the association between earthquakes and water, either liquid or solid, coming from a period of rapid climate change marking the latest part of the Pleistocene and the Holocene. Evidence is also seen on the building of dams and artificial impounding of water in reservoirs. According to the author¹ increases of earthquake activity was a natural response to natural changes that resulted from the wasting of ice sheets that were kilometres thick and the spontaneous draining of vast lakes. He suggests that to discover if the climate changes of the present are capable of triggering a seismic response evidence should be looked for in the parts of the world that are experiencing ice loss at the most rapid rate or the accumulation of glacial meltwater, such as in Alaska.

In Alaska there are 43 volcanoes that have been active historically, such as Pavlof, and there are also snow-capped peaks, permafrost and glaciers, as is the case in Iceland. A wholesale melting of permafrost is occurring in Alaska as a result of a 3º C rise of temperature over the last 50 years that has led to an increasing problem in cities such as Fairbanks of structures including buildings, bridges and roads becoming misaligned or deformed as the ground beneath them become increasingly soggy. So far the Bagley Ice Field (aka the Bagley Ice Valley) at 200 km long, 10 km wide and up to 1 km thick in places is still the largest non-polar ice field in North America, though it is rapidly shrinking. Many valley glaciers are fed by this icefield, branching out from the parent mass, such as the Bering Glacier, among the biggest in the USA. The Bering Glacier and all the other valley glaciers are retreating, and even the parent ice field is beginning to be affected by rising temperatures and precipitation changes, thinning by several hundred metres since 1900. According to the author¹ this retreat is reversed every couple of decades, but the retreat is soon resumed. It has been estimated that it is now adding 30 km³ to the Pacific Ocean annually.

At the surface dynamic change can be seen in the glacier retreat and permafrost thawing, both of which will become more apparent as the Earth warms. What can’t be seen are the natural tectonic forces at play beneath Alaska, as well as around the margins of the state. The powerful geological forces beneath Alaska are hinted at by the many active volcanoes and the high level earthquake activity, the largest of which was Good Friday Earthquake of 1964, the second largest seismic event ever recorded. The largest of all the tectonic plates is the Pacific Plate and it is being subducted to the northwest beneath the North American Plate at 5 cm/year beneath southern Alaska. This rate of subduction is high for plate movements and the massive collision between the 2 plates has pushed up peaks to more than 5,000 m a bit less than 15 km of from the Gulf of Alaska. There are many active faults and many earthquakes. Since the end of the 19^th century there have been 7 earthquakes larger than magnitude 8 as well as the 1964 event, and serious damage was caused by 4 of them. There are annually 4,000 large, shallow earthquakes, making up 6 % of all the earthquakes of this type around the Pacific Rim.

Jeanne Sauber, Space Flight Center, NASA, Maryland and Bruce Molnia, USGS, have been studying how the continued wastage of glaciers and ice fields in southern Alaska are influencing the stability of faults situated at the subduction plate boundary immediately underlying the region, particularly if the seismic record reflects the glacier shrinkage. The ice across much of the region has been thinned by hundreds of metres, possibly 1 km in areas of greatest ice loss has resulted from progressive melting that is driven by a warming climate since the beginning of the 20^th century. As the ice melted over this period, deep beneath the surface the segment of plate boundary fault lying deep beneath the surface was accumulating strain and in 1979 a magnitude 7.2 earthquakes struck. Sauber and Molnia were trying to determine if the wholesale loss of overlying ice mass that occurred over about the previous century directly caused the earthquake, or possibly have a role in its triggering. They concluded that the amount of ice mass loss over this period, a pressure fall that is comparable to 20 times the pressure of the Earth’s atmosphere, would have certainly been sufficient to influence the type of thrust faulting that was characteristic of the 1979 quake. According to the author¹ this would indicates that the ice loss would have facilitated the slippage of the fault, possibly bringing forward the timing of the event though not actually causing it. It would have been the deep tectonic forces that actually drove the slippage.

The author¹ also suggests that in southern Alaska it also becoming apparent that short term fluctuations of ice mass is influencing earthquake activity. In the 1990s the Bering Glacier surged and 14 km³ from the Bagley Ice Field source was moved rapidly to the snout of the glacier. It was suggested by Sauber and Molnia et al. that a noticeable rise in the number of earthquakes of less than magnitude 4 in the underlying crust resulted from the unloading. Since then, according to the author¹, in the early years of the 21^st century a significant rise in the rate ice loss coincided with an increase in the number of seismic shocks beneath Ice Bay on the coast immediately south of the Bagley Ice Field.

The author¹ says the findings in Alaska are of great significance for seismic hazard evaluation throughout the world in places where glaciers are melting as it indicates that even 10,000 years after the close of the last ice age the loss of ice mass is still facilitating, if not directly causing, earthquakes. At locations, such as Alaska, at high latitudes where active faults capable of producing large, destructive earthquakes, are covered by ice, as well as in comparable situations away from the poles where the terrain is covered by glaciers, such as the Andes in South America and the Alps in New Zealand, the author¹ suggests it is advisable that the seismic hazard should be re-evaluated in light of the findings in Alaska. When evaluating the seismic hazard of region it is usual to base it on the seismic record of the place over time, and not take into account changes of environmental factors such as ice cover at the time of the evaluation, and past behaviour of both the ice and the crust. As the continued survival of the glaciers of the world increasingly seems unlikely the evaluation of the likely consequences of their loss on the active faults beneath them has taken on a greater importance.

Crustal rebound

According to the author¹ the city of Basle in Switzerland is situated in a placein which the geology is particularly complicated where there are stresses that stretch the crust along a zone of weakness, the Rhine Graben, that comes into contact with collisional forces caused by the uplift of the Alps, the continued rise of which is in turn caused by the collision of Italy with southern Europe and continues to move north. It was a result of Basle being situated on this highly complicated geology that in 1356 there were a series of massive earthquakes and many aftershocks, some of which were large. The author¹ says most discussion has focused on whether the earthquakes of 1356 were a result of the rupture of a compressional (or thrust) fault south of the city, that was associated with Alpine deformation, or a tensional (or extensional) fault in the Rhine Graben to the north, an ancient rift that is slowly pulling Europe apart, that was a source of the earthquakes. There is now some degree of consensus that the earthquakes were associated with rupturing of a fault within the graben. It now appears the earthquake of 1356 was not unprecedented in comparatively modern history, as there is evidence of at least 2 other comparable events that occurred in the past 10,000 years that were identified during excavations across the fault. There are also other places in western Europe that are now known to be at risk of earthquakes that are large and potentially destructive. There have been many other earthquakes in the northern end of the Rhine Graben such as a 5.8 event that occurred close to the town of Roermond, the Netherlands, in which there was a lot of damage. It has been revealed by fault excavation studies that in about the past 20,000 years there have been 3 larger quakes in the region, and it is suggested by research at other places in and around the northern segment of the graben that faults capable of generating substantial earthquakes crisscross the area. The author¹ suggests the threat of seismic activity may even reach as far as the UK, some faults extending to the west beneath the Dover Straits that have been associated with damaging quakes that occurred in 1382 and 1580. In the 1580 event 2 people were killed in London by a 5.5 magnitude earthquake that caused damage across southeast England. Over about the past 500 years there has been little seismic activity in the area affected by the 1580 earthquake but there remains the potential for earthquakes of comparable size to occur in the future. The population of London is now 50 times greater than it was at the time of the 1580 earthquake, as pointed out by Roger Musson of the British Geological Survey, an earthquake of comparable size would now cause a much greater loss of life and the damage would be in the billions of pounds.

It is now known that the crust of Europe is much less stable than is generally believed, and there is evidence that Europe is now more seismically active that in was in the past. A burst of seismic activity is believed to have occurred in central Switzerland at the time of the Late Pleistocene-Holocene transition when the ice sheets of northern Europe were retreating and the Alpine glaciers were also retreating. It has been suggested that the destructive earthquake of 1356 may be the result of the close of the ice age with the unloading of Europe following the loss of the ice sheets, the accumulated stresses on faults that were previously imprisoned were released leading to European earthquakes in a similar manner to the unloading led to volcanic activity in France and Germany when the ice retreated. Isostatic rebound is the key mechanism involved in the rebound of the lithosphere following unloading as the ice sheets melted, and has been used to explain the earthquakes in Canada, Europe and the UK that occurred in post-glacial times. Once released from the ice there is a vigorous response as the faults can again move resulting in earthquakes.

According to the author¹ isostasy is the principle of buoyancy that has been applied to the outer layers of the Earth. The brittle lithospheric plates are of relatively low density and float on the asthenosphere that is denser, and being plastic deforms more easily than the lithosphere. At any particular point the lithosphere floats at a particular level in the asthenosphere depending on density and thickness of the lithosphere. The lithosphere under high mountains pushes deeper into the asthenosphere than the surrounding lowlands. A state of gravitational or 'isostatic' equilibrium exists between the lithosphere that floats on the asthenosphere and the asthenosphere beneath it under ideal circumstances, though circumstances are not often ideal on the dynamic Earth. As a result of erosion unloading may cause the mountains to rise higher in the asthenosphere. Where the eroded material is accumulated in deposits it increases the weight of the lithosphere which then sinks deeper into the asthenosphere.

The lithosphere is said to be in a state of disequilibrium where is is loaded by ice sheets that are kilometres thick, being forced downwards by the weight of the ice to a degree where its level in the asthenosphere is out of balance with its mass. Isostatic rebound occurs when the ice melts and the lithosphere is unloaded, though at a slow rate. Post-glacial-rebound is the term used for the isostatic rebound that resulted from the unloading as the ice melted, to distinguish it from the broader lithospheric uplift that can occur for a number of other reasons, glacial isostatic readjustment is a more accurate term, as in the case of subsidence of continental margins that are loaded as a result of sea level rise following the end of a glacial phase.

The thickness of the ice sheets across North America and Europe reached up to 3 km at the height of the last glaciation. The uppermost part of the mantle, that is hot and  plastic, was pushed aside as the enormous mass of the overlying ice sheets pushed the lithosphere deeper into the asthenosphere. As the lithosphere was unloaded at the end of the glaciation as the ice sheets were melting the lithosphere flexed back up and mantle material flowing back to its previous position, caused continued long-term uplift, a process that was very slow as a result of the low viscosity of the mantle, the uplift following the ice loss continuing at the present in the land area in much of the Northern Hemisphere, such as Canada and Scandinavia. The initial rate of uplift was up to several centimetres per year, though at the present it is not more than 1 cm per year. The author¹ suggests it could take another 10,000 years before for the lithosphere beneath the deepest parts of the former ice sheets finishes rising.

Examples of physical evidence for rebounding of the lithosphere following the close of the last glacial phase, of which there is a great deal, include marine shore lines and elevated lakes in areas that had previously been glaciated, as well as otherwise unexplained uplift patterns. In about the 12^th century an arm of the Baltic Sea was separated from the marine environment by progressive uplift in Sweden close to the centre of the vast Fennoscandian Ice Sheet that was centred on Scandinavia, in the process transforming the separated arm of the Baltic into a freshwater lake, Lake Mälaren, on the shores of which is Stockholm, the capital of Sweden. The northern arm, the Gulf of Bothnia, will eventually be transformed into a freshwater lake. At Kvarken, where the water has a maximum depth of 25 m, about 80 km separates Sweden from Finland. It has been estimated that it should take more than 25,000 years at the current uplift rate of a bit under 1 cm/year for the northern part to be cut off to form a freshwater lake. According to the author¹ the time should be even longer because of a slowing of the uplift rate over time and rising sea level resulting from anthropogenic climate change.

The author¹ suggests Lake Ontario in North America appears to have been formed in a similar manner. The sea flowed into the St Lawrence Valley after the Laurentide Ice Sheet melted, and Lake Ontario formed a bay of this flooded river valley. Soon after the sea flooded into the St Lawrence Valley uplift separated the bay and it was transformed into the freshwater lake of the present. Uplift has been continuing rapidly in the region, rapidly enough to tilt the bed of the lake to the south, river valleys being converted into bays and causing erosion rates to increase, and this is causing concern among property owners along the shoreline. In the UK post-glacial-rebound continues to cause the UK to pivot across its middle, the south subsiding as Scotland and Northern England are uplifted following the melting of the ice sheets that covered them during the last glaciation. The author¹ suggests that as a consequence of this movement, by 2100 parts of Scotland could have risen by about a further 10 cm while the southeast of England could have subsided by about an extra 5 cm, which would in effect add to the sea level rise that will have occurred by then. It has been established that post-glacial-rebound on a widespread scale has actually occurred, but is there a connection between earthquake occurrence and this lithospheric uplift? The author¹ suggests that a good place to look for evidence of this link is the biggest ice sheet, the Laurentide Ice Sheet in North America.

New Madrid

In the putatively stable central parts of North America, a long way from the known earthquake areas such as California and Alaska, and similar to the case of Western Europe, earthquakes took many by surprise. Shortly before Christmas in 1811 there were severe jolts that shook the upper Mississippi Valley in an area of land, where there were previously unknown faults, between St Louise, Missouri to the north and Memphis, Tennessee to the south. This was followed on the 16^th of December by a major earthquake of magnitude 7-8 centred in north-eastern Arkansas. The population of the area was sparse at that time so property damage was small. At the settlement of New Madrid by the Mississippi River, a bit north of the epicentre, the quake caused the collapse of the river banks and toppled apple trees. The shaking has been estimated to have reached as much as magnitude 9 where the present city of Memphis is located, that is comparable to the magnitude of the earthquake that levelled Basle 450 years earlier, and would have caused massive damage if it occurred at the present. Eliza Bryan, an eyewitness, later wrote of the terror of the population of the settlement.

This earthquake proved to be the beginning of one of the most extraordinary seismic events in the history of the USA. This was followed by major shocks every few minutes and eventually, at around breakfast time, a second major quake about the same size as the first, followed by more aftershocks, and other major quakes on 23^rd  of January and February 7^th, the last possibly being the largest of all. A great deal of property damage was caused in St Louis, with ruptures opening in the ground instantaneously forming waterfalls on the Mississippi, though they were only temporary. Church bells in Richmond, Virginia, more than 1,000 km away, were caused to ring by the earthquakes as the cold, brittle crust of central USA transmitted the seismic waves very efficiently and Windows were rattled in Washington, DC, 1,400 km away, and they were felt in New York and Boston. Aftershocks that were strong enough to be felt continued for more than 5 years.

The source of the New Madrid earthquakes is not known as the area is a very great distance from any plate margins, leading seismologists to look for other possible underlying causes. Research has been focused on the Laurentide ice sheet that covered much of the northern part of North America during the last glacial maximum, in particular on the possibility of post-glacial-rebound being responsible for the earthquake activity many years after the retreat of the glaciers and subsequent unloading as the ice melted.

According to the author¹ a problem with a possible role played by deglaciation in the triggering of the New Madrid earthquakes is that the New Madrid Seismic Zone (NMSZ) was never covered by ice during the glaciation. At the height of the glaciation the glaciers didn’t come closer to the NMSZ than a few hundred kilometres to the North. The application of stress to rocks surrounding the edges of the ice sheets stretches a long way from the glacier margins. When the ice sheets push the lithosphere down into asthenosphere the mantle is pushed to the side to make room for the continental crust. The lithosphere flexes upwards to compensate around the edge of the depressed area, and this uplift is helped by the outward flow of the mantle resulting in the hollow formed beneath the ice being surrounded by a contrasting ‘forebulge’ extending well beyond the ice sheet front. As deglaciation occurs the lithosphere and the forebulge subsides gradually in a wave moving outwards from the area of the former ice sheet. It has been suggested by Robert Muir-Wood that the perfect conditions for the release of tectonic strain that had accumulated during glaciation might be provided by just such as collapse of a forebulge with the result that a corresponding cascade wave of seismic activity migrated progressively away from the ice sheet’s former margins. Muir-Wood also suggested that as applied to North America this model could also explain an earthquake that struck Charleston, South Carolina, in 1886, that was large and caused a lot of damage, as well as others off the Canadian coast. Similar ideas have also been developed by Patrick Wu of University of Calgary, Canada and Paul Johnston, ANU, Canberra, involving rolling waves of seismic activity travelling away from the margin of the Laurentide Ice Sheet over time. They have developed a model that tries to predict the arrival in specific locations in North America of this earthquake activity pulse. According to the author¹ their model predicts accurately the nature and timing of earthquakes that occurred in Quebec and Indiana, though it has problems with those at New Madrid, predicting the timing of the New Madrid sequence but failing to predict the rate at which the rebound stresses decay over distance from the margins of the former ice sheet, suggesting they would be not be large enough to trigger magnitude 8 earthquakes in this area. The magnitudes of the New Madrid earthquakes were later scaled down to about 7 by Susan Hough of the USGS in 2010, a level that can be predicted for the New Madrid quakes by post-glacial-rebound.

According to the author¹ there is apparently good evidence that seismic activity was triggered by the melting of the ice sheets that covered North America; this doesn’t necessarily apply to the earthquakes that struck New Madrid in 1811 and 1812, though it remains to be proved that they actually resulted from post-glacial-rebound. He also suggests that the most scientifically reasonable explanation for the New Madrid earthquakes is that deglaciation and tectonic forces were involved in the triggering of the New Madrid earthquakes by the lithospheric bounce-back being just sufficient to nudge faults that had been weakened by the gradual accumulation of stresses that resulted from the continuous movement of tectonic plates. The result of GPS studies to estimate the rapidity of the accumulation of strain on the NMSZ faults suggests that such strains are accumulating extremely slowly, with fault movements being of a few tenths of a mm per year. The is more than 100 times slower than strain accumulates on the San Andreas Fault in California. According to the author¹ this has been used to play-down that likelihood of another quake on the scale of those of 1811 and 1812 occurring. Some have suggested that earthquakes that have occurred in the NMSZ more recently are very late aftershocks resulting from the resettling of the crust after the 2 big quakes of the early 19^th century. Alternatively it is has been suggested that the NMSZ may be closing down.

 A cautious approach has been taken by the USGS, with the possibility of another large earthquake in the area continuing to be a possibility. They consider that about a decade of strain measurements should not override a record of persistent seismic activity in the NMSZ that covers 4,500 years. There have been more than 4,000 earthquakes detected on the instruments of USGS since the mid-1970s, at least 1/year that was big enough to be felt. The USGS have not found any sign that the activity is diminishing over time, which indicates that the continued tremors are not simply adjustments of the crust. The USGS have not downgraded the perceived risk of another major earthquake in the NMSZ because the possible consequences of any large earthquake that did occur, as the FEMA has warned that a large quake in the area could possibly cost the US more than Hurricane Katrina in New Orleans in 2005. Estimates of the probability of another earthquake comparable to the magnitude of those in the early 19^th century have been put at 10 % by the USGS. It has been suggested by Muir-Wood that the wave of post-glacial strain relief might continue moving south eventually increasing the risk of large earthquakes in Georgia, Arkansas and the northern part of Louisiana.

Overview¹

As in the case of volcanic activity, it is not possible to isolate seismic activity from what occurs on the surface. Water, either solid or liquid, has been shown to be capable of having an influence on potential effects of hazardous geological activity and eliciting responses from the Earth’s interior. Studies of the post-glacial period, as well as of the seismic effects of dam construction, have shown that the addition or removal of sufficiently large amounts of water can significantly affect the activity of the underlying faults and adjacent faults. When an ice sheet is formed or a glacier is thickened the additional pressure will stabilise the underlying faults, their activity being suppressed and the earthquake incidence reduced along that fault. The opposite effect results from unloading, the removal of the ice, the faults being freed up and the facilitating of fault movement resulting in a seismic response. The loading and unloading effects of water are comparable, though with the added complication of pore pressure is increased in the rocks adjacent to new or growing dams or lakes leading to destabilisation of nearby faults.

Assuming present trends in climate change continue until temperature and sea level rises are comparable to those of the post-glacial period it is likely, according to the author¹ that at least some of the many faults around the Earth will respond to the changed redistribution of global water that will no doubt occur as the world continues to get warmer. Among the phenomena that have the potential to trigger a seismic reaction in the underlying crust are melting high latitude ice sheets, the wasting of glaciers in mountainous regions, sea levels that are rising rapidly, dams built to store water to prepare for predicted water shortages. The author suggests it is not certain that a rise of global earthquake activity would be detectable, and any increase in the numbers or frequency of earthquakes on a planet-wide scale will be hidden in the statistical ‘noise’ associated with the 2 million earthquakes that occur, on average, every year. The author¹ suggests that on a regional or local scale it may be more obvious when there is a response from underlying faults, especially if and when the Greenland and West Antarctica ice sheets begin to disintegrate. It is possible that the crust underlying these vast ice sheets may be relatively stable with few active faults capable of triggering an earthquake, though it is also possibly that the ice may be the only stabilising influence on the faults, in which case they may respond with a burst of seismic activity if the ice sheets disintegrate. Even large earthquakes in Greenland or West Antarctica may not be seen as much of a problem for populations on other continents but their effects could be felt much further away. If large earthquakes occur in a watery environment they can trigger tsunamis if the crust is unstable enough to fail and generate a landslide, and such tsunamis can travel large distances across the ocean and cause great damage even thousands of kilometres from the site of the landslide. As Scandinavia continued shedding its ice sheets in the early part of the Holocene isostatic rebound promoted earthquakes that are believed to have triggered massive submarine glacial sediment slides off the Norwegian coast that generated tsunamis that travelled across the northeast Atlantic to the Shetland Islands and mainland Scotland. If such a burst of earthquake activity occurs in Greenland in the future it could prove to be disastrous for coastal communities in Iceland and Newfoundland.

Sources & Further reading

1. McGuire, Prof. Bill, 2012, Waking the Giant: How a changing climate triggers earthquakes, tsunamis, and volcanoes, Oxford University Press.