Category: Geomorphology

Several times in Earth Pages News the topic of how erosion contributes to uplift has cropped up. That is more than just the iceberg-like bobbing up of the crust as the load on the underlying asthenosphere is eased by surface rock removal. One oddity is that as large valleys are carved the ridges and peaks that they separate can rise higher than the original lands surface from which they developed (see Erosion and plate tectonics in May 2005 issue of EPN). Now it is becoming clear that sideways movement of the crust beneath mountain ranges can also be a response to erosion; thrusts and nappes can respond to erosion as well as to plate tectonic forces. The most likely place where this might be happening is in the Himalaya, which produce a huge contrast in climate and erosion rate between their southern and northern sides by creating the world’s largest rain shadow. The evidence for this possibility is nicely reviewed by Kip Hodges of Arizona Sate University (Hodges, K. 2006. Climate and the evolution of mountains. Scientific American, v. 295 (August 2006 issue), p. 54-61).

The highest erosion rates take place where rainfall during the Indian monsoon is greatest, on the SSW face of the Himalaya, especially in the foothills between about 1000 and 3500 m. The Tibetan Plateau lies in the rain shadow of the Himalaya, and erosion is far less intense. Yet the Tibetan plateau is buoyed up by crust that is double the normal thickness, to an average elevation of around 5 km. In a crude way Tibet can be regarded as having a pressure head ‘dammed’ to the north of the Himalaya. Intense erosion at the foot of the mountain ‘dam’ is likewise akin to one cause of landslides: erosion of the toe of a slope. The gravitational potential of Tibet, combined with continual undermining of the Himalayan front must create a lateral force. Where the crust is able to behave in a plastic fashion, i.e. at depth, and if there are surfaces on which movement is possible — the north-dipping frontal thrusts of the Himalaya — then deep crust should be extruded sideways. In fact there are faults systems just to the north of the Himalaya that have the same dip as the thrusts, but an opposite sense of movement, directed northwards to create extensional detachments. The crustal zone in-between is the most likely to undergo extrusion. GPS measurements there and cosmogenic dating of the surface reveal that indeed this zone is experiencing  anomalously high rates of uplift. It is producing extremely high gradients on both hillsides and valley floors.

When Cenozoic mountain belts and high plateaux began to rise and be eroded has become a disputed topic about most of them, such as the Himalaya and Tibetan Plateau, the North American Western Cordillera, the Andes and the Ethiopian Plateau. A host of techniques have been used, including plant-leaf stomatal indices, the bubbles in lavas, cosmogenic isotopes and various stratigraphic approaches. For the Tibetan Plateau (see When did Tibet rise? and Tibetan uplift: looking a gift horse in the mouth in the March and May 2006 issues of EPN) opinion is polarised: as soon as India collided with Asia, around 40-50 Ma ago, due to crustal thickening; as a result of a slab of lithosphere delaminating from the region as recently as the Late Miocene (8-10 Ma), when the crust rose gravitationally. There is support for both hypotheses. Much the same divergence of opinion applies to the Sierra Nevada of the Western USA, which may have been a major feature throughout the Cenozoic, or subject to delamination in the Pliocene (3 to 5 Ma ago). In its case, a new way of estimating topographic elevation in the past may resolve the disputation (Mulch, A. et al. 2006. Hydrogen isotopes in Eocene river gravels and paleoelevation of the Sierra Nevada. Science, v. 313, p. 87-89).

The Stanford University group led by Andreas Mulch have focused on the way in which the proportion of deuterium (²H) in rainwater changes as clouds rise over high areas. This is known quite precisely from modern hydrological studies. Of course, you cannot find ancient rainwater ponded in the place where it once fell, but that rain does find its way into clay minerals during weathering. The western flank of the Sierra Nevada is mantled by extensive Eocene graves deposited by rivers that drained the area now occupied by the mountains. Handily, those gravels were the main target for the 1849 California Gold Rush and subsequent alluvial gold mining, so they have been well exposed by the prospectors throughout the Sierra Nevada. At high elevations they are preserved in river terraces, so it is possible to trace roughly the ancient drainage courses and sample a range of modern elevations. The hydrogen isotope data from kaolinite in cobbles of weathered granitic rocks are extremely interesting. They show that the Eocene rainwater that entered kaolinite molecules at different modern heights had fallen at a very similar range some 50 Ma ago. There seems little doubt that the Sierra Nevada had risen more than 2 km before the Eocene

The land’s present topography is not just the frontier between the lithosphere and the atmosphere and hydrosphere, but where plants of many different kinds grow. Whether in the form of cyanobacteria, lichens or luxuriant tropical rain forest, vegetation affects weathering, erosion and the deposition of sediments. Animals – leaving out humans – also have some influence, whether they be subterranean rabbits, moles and worms, or heavy-footed beasts that force soils to move downslope. Inevitably life-land interactions affect landforms, although rock-type and active geological processes tend to dominate. Nonetheless, a planet with life ought to show different styles of surface shapes from one that is organically dead. The central issues for geomorphologists is whether or not it is possible to define absolutely the differences, and then to use them as a means of detecting the likely former influence of life on other worlds.

Central to such a venture (Dietrich, W.E. & Perron, J.t. 2006. The search for a topographic signature of life. Nature, v. 439, p. 411-418) is the ability to map in detail the variation of topographic elevation. Digital topographic elevation data is now available for most of the Earth’s land surface at a resolution of between 90 and 30 m, the second only publicly available for the USA, from the groundbreaking Shuttle Radar Topography Mission of 2000. Aerial photography and high-resolution stereoscopic images from satellite such as Quickbird and Ikonos, allow resolution as sharp as a few metres.  Laser scanning from aircraft potentially can even improve that to the scale of a few tens of centimetres, but such high-resolution data are far from global. The planet Mars is now better endowed with elevation data than is our own planet, thanks to photogrammetric instruments carried by ESA’s Mars Express mission, and the shyness of various intelligence agencies to share publicly what they have gleaned from high-altitude aircraft and spy satellites. Nonetheless, it is now possible to analyse elevation data from the entire range of terrestrial biomes to see what signal vegetation has imposed on surface shape. An easy way to visualise that is simple – just use Google Earth (see The Digital Earth revolution above).

Dietrich and Perron review the mathematical approaches to modelling life’s topographic influences, beginning with an equation that relates elevation and time to rates of uplift, erosion and entry of sediment into storage, thereby expressing conservation of mass.  All the variables are themselves governed by a variety of processes, theoretically amenable to quantification, summarised in Dietrich and Perron’s review. In each there will be some potential biotic influence. On Earth there are sufficient landscapes devoid of all but a minute veneer of organisms to assess both end-members clearly. Mars and Venus ought to be good tests.  But, should such a rigorous quantification of lifeless and lively surfaces at a spectrum of scales be achieved, where would we deploy it?

The world’s most awesome natural spectacle is probably the Brahmaputra River in full spate.  Unlike most large rivers, it is constrained for most of its course within a deep, narrow gorge that has to take the snow melt from a huge catchment on the northern flank of the High Himalaya, brought partly by the Tibetan Tsangpo River.  Each spate hurtles onto the plains of  Bangladesh, loaded with debris, at a rate of around 70 thousand cubic metres per second.  Although that is but a third of the flood discharge of the Amazon, for much of the  Brahmaputra’s course it must pass through a gorge only a few hundred metres wide in places.  This gives not inconsiderable erosive power, indeed probably the highest anywhere.  Not surprisingly, little is known about the Tsangpo-Brahmaputra valley, because of its inhospitable character.  With the recent release of ~90m resolution elevation data from the Shuttle Radar Topography Mission, it is now possible to analyse the whole catchment’s morphology in detail, without needing to follow the individual rivers.  Parts of the lower Tsangpo have remarkably high gradients, including a 100 km stretch with a fall of more than 2 km, through a gorge with almost 7 km of relief on either flank that cuts N-S across the axis of the Eastern Syntaxis of the High Himalaya. The gorge lies downstream of a west to east stretch with lower gradients, falling around 1 km in 300 km, which suggests some dramatic incision begins at the junction of the two sections.  US and Chinese geomorphologists visited the area and discovered that high on the flanks of the upper Tsangpo are terraces of lacustrine sediments, at about 3100 and 3500 m (200 and 600 m higher than the river) (Montgomery, D.R. et al. 2004.  Evidence for Holocene megafloods down the Tsangpo River gorge, southeastern Tibet.  Quaternary Research, 9 September 2004 issue).  Charcoal in the sediments gives radiocarbon ages between 1200 to 1600 BP and 8800 to 9800 BP for the lower and higher terrace levels, so the lakes formed during the Holocene.  The terraces stop at a zone of thick glacial moraine, cut by the Tsangpo, which suggests that both formed in lakes behind two ice dams.  Using SRTM data allows the volume of water ponded in both ice-dammed lakes to be estimated.  The older and higher level indicates about 830 km³, and the lower some 80 km³.  Breaching of the dams would have caused the largest recorded erosive events in recent Earth history, and explains the gorge below.  Each flood discharge would have been between 1 and 5 million cubic metres per second, equivalent 3 to 15 times the maximum flood discharge of the Amazon.

The greater the rainfall, the more effective streams become as agents of erosion.  So, “common sense” suggests that very wet mountain areas should be eroded more quickly and develop a more profound relief than those that are drier.  With the advent of detailed digital elevation models that cover the world, it is very easy to calculate slope angles and relief over huge areas, and match them with rainfall records.  Geomorphologists from the Universities of Montana and California have done this for the wettest and most rugged area in the world, the Annapurna area of the Himalaya (Gabet, E.J. et al. 2004.  Climatic controls on hillslope angle and relief in the Himalayas.  Geology, v. 32, p. 629-632).  The main agent of erosion there as streams cut downwards is by landslides.  The region also shows a profound gradient in annual rainfall from about 1000 mm in the High Himalaya to 4500 at the front of the range, where the monsoon rains hit hardest.  “Common sense” is wrong, for the slopes decrease from an average of 35º to 25º as rainfall increases.  The authors believe this is due to the influence of deeper weathering in more humid parts that reduces the strength of slope materials so that they must stabilise at lower angles than those in dry areas.  Their other finding is that relief (elevation difference in small segments of an area) and slope angle have a strong positive correlation, so that relief itself is inversely related to rainfall.  They are able to comment interestingly on various ideas about mountain evolution.  Their main conclusion is that in any particular area, a transition from dry to wet conditions lowers mountain ridges faster than valley incision can shift the debris, whereas during drying, ridges are barely lowered, while streams cut unhindered into bedrock, thereby sharpening up the landscape.

Formation of gorges in tectonically quiet areas

The flanks of the North Atlantic probably became tectonically inactive in Mesozoic times, yet rivers large and small have cut large gorges, often through highly resistant bedrock.  But they also have developed broad valleys over millions of years, and it is into them that the gorges are incised.  Slow upward flexing caused by sediment loading on the continental shelves, a general lowering of sea level since Antarctica first formed a permanent ice cap, and isostatic response to gradual denudation help explain the full extent and shape of the rivers drainage basins.  The gorges are young, and must have developed rapidly.  Old ideas focussed on W.M. Davis’ theories of landscape evolution, particularly rejuvenation associated with changing base levels of erosion, but with no quantitative backing.  The development of means of dating eroded surfaces using the decay of short-lived radioactive isotopes that cosmic-ray bombardment creates now offers an opportunity to test hypotheses rigorously and come up with others.  Quite a few published works on cosmogenic dating applied to landform development seem to add little to geomorphological knowledge, so it is a relief to find one that does (Reusser, L.J. et al. 2004.  Rapid late Pleistocene incision of Atlantic passive-margin river gorges.  Science, v. 305, p. 499-502).  The authors, from the Universities of Vermont and Maryland, the USGS and the Lawrence Livermore National Laboratory, focus on impressive gorges in the lower reaches of the Susquehanna and Potomac Rivers as they drain the eastern US into the Atlantic, and a series of higher surfaces which they cut into to leave as rocky straths.  The oldest ages occur on the highest of these straths, as expected, and age decreases on successively lower ones to the rocky flood plain of the modern rivers just above their current channels.  The highest levels are between 85 and 97 ka, the most prominent strath formed between 30 and 33 ka, succeeded by  one at 19 ka and the lowest level seems to have formed between 13 and 14 ka.  Interpreting the periods of intense erosion that cut each level must involve late Pleistocene climate change, sea-level shifts, and the bulging effect due to the North American ice sheet which reached its maximum extent in the northernmost part of the Susquehanna basin.  It seems that during the early part of the last glacial episode, incision was slow, although probably faster than during the Holocene.  But around 30 to 33 ka ago it accelerated rapidly to half a metre every thousand years, some 1 to 2 orders of magnitude greater than at present.  This was at a time when ice loading was only half that at the glacial maximum around 20 ka, so it seems likely to have been initiated more by increased storminess and torrents, and indeed correlates with an abrupt increase in sea-salt content in the Greenland ice cap brought in by winds at that time.  Lasting through the glacial maximum, increased frequency of flooding combined with more rapid sea-level fall, also beginning at around 32 ka, were probably the main driving forces for gorge incision.  This still leaves a puzzle.  Both drainage basins had been in existence since well before the cycles of glacial and interglacial periods began on the flanks of the North Atlantic around 2.5 Ma ago.  Similar periods of accelerated incision must have been repeated, at least during the last 6 or 7 glaciations which were the most extensive.  Did earlier topographic features exert any control over later ones, and do any relics of them remain?

If ever there was “received wisdom” in the geosciences the most pervasive is the notion that the weak acid formed when carbon dioxide dissolves in rainwater is the cause of carbonate solution.  Anyone hearing it in the spiel from a cave guide, while admiring caverns as big as cathedrals, is not surprisingly awe-struck by such an innocuous sculpting agent.  Many speleologists have long wondered if there might be other mechanisms, and the discovery of bacterial films that generate strong sulphuric acid provides a good candidate.  They can take the form of floppy, stalactite-like masses, that have become fondly known as “snoticles”.  However, their role in cave formation had not been substantiated until April 2004.  Microbial geochemists at the University of Texas carefully studied the geochemical balances in a cave system in Wyoming where such bacteria are abundant (Summers Engel, A. et al. 2004.  Microbial contributions to Cave formation: New insights into sulfuric acid speleogenesis.  Geology, v. 32, p. 369-372).  The bacteria are members of two groups that live in aerated conditions and use the oxidation of sulphide ions (from hydrogen sulphide) as a source of metabolic energy.  Oxidation results in sulphuric acid, which rapidly dissociates in water to generate abundant hydrogen ions (the source of acidity and low pH) and sulphate ions.  So, to thrive the bacteria need a continuous source of hydrogen sulphide, of which more later.  The study by Annette Summers Engels and two colleagues shows that hydrogen sulphide is efficiently consumed by the bacteria, so that little if any enters the cave’s atmosphere.  Interestingly, water flowing through the cave isn’t particularly acid either, yet the bacteria generate a great deal of sulphuric acid.  It is rapidly neutralised by reaction with calcium carbonate near the colonial mats, to increase the flux of calcium and sulphate ions into solution.  The effect extends to limestone pebbles on the beds of the cave streams, so the bacteria encourage solution beneath water as well as near snoticles hanging from the roof.  That suggests that they can live below the water table, where many caves are thought to have formed in the past, being left as open caverns as the water table fell as bulk permeability increased with solution.  The studied cave does experience a constant flux of hydrogen sulphide, but where does that come from?  There are other groups of bacteria that generate sulphide from dissolved sulphate ions, but under highly reducing conditions.  They are the source of the “sour gas” that is a constant danger in oil production in some petroleum fields, consumed gleefully in dissolved form at a great many spas and generated in our own guts.  These sulphate-sulphide reducing bacteria get their energy from dead organic matter, that many sediments deposited under reducing conditions contain in substantial volumes.  Interestingly, connectivity between oxygen-rich and oxygen-starved groundwater might create a recycling of sulphur that involves both bacterial groups.  Many limestones contain strata that are rich in organic remains and metal sulphides, in which conditions become reducing.  Equally, interbedded, black shales might play a role.

In many areas of active deformation, landforms that suggest that uplift and river down-cutting keep pace are very common.  Stream courses cross zones of uplift, rather than being diverted or ponded up to form lakes.  Traditionally, geomorphologists have described such drainages as “antecedent”, i.e. rivers that were present before uplift began.  They can be seen on all scales up to examples such as the Indus and Brahmaputra rivers that carve their way across the actively rising Himalaya.  The most common are anticlines through which streams flow in canyons perpendicular to the fold axes.  A curious and common feature is that the canyons are not haphazard, but often cut the fold where its amplitude is greatest and its axis plunges away from the site of incision.  The stupendous rates at which crustal rocks are eroded and transported away in the courses of the Indus and Brahmaputra, and in lesser drainages on the flanks of major extensional orogens, such as the Red Sea, clearly removes load from the crust.  Consequently there is an isostatic component to the uplift involved in the two cases at a grand scale.  Peter Molnar and Phillip England suggested an erosional role in large-scale uplift over a decade ago.  Intervening ridges rise higher than they would if erosion was slower or non-existent.  In major rift systems, the highest peaks are often within the escarpments rather than at the lip of uplift, sometimes more than 500 m higher.   Bearing this well-known process in mind, Guy Simpson of ETH Zurich, has sought evidence that it functions on much smaller scales (Simpson, G. 2004.  Role of river incision in enhancing deformation.  Geology, v. 32, p. 341-344).  That comes from the surprising symmetry of doubly plunging anticlines that are cut by rivers at their highest point.  His modelling suggests that the phenomenon can occur when the crust deforms plastically, allowing isostatic response to erosion on even minor scales during compression.  When deformation is by brittle means, any uplift of rigid crust is flexural and has long wavelengths, so that rivers bear no relation to local structures

One of the most widely quoted bits of geological information that appear in non-specialist literature is that the oldest land surface on Earth is that of interior Australia.  Vast tracts are Precambrian capped by horizontal Permian glaciogenic rocks in places, but for the most part by relics of lateritic palaeosols that give it is famous red appearance.  The oldest outlying platform sediments are 1100 Ma old, so the actual surface does date back at least as far, but has it been exposed at the surface for that long?  Dating the present surface has not been easy.  New methods involving the creation of unstable isotopes by cosmic-ray bombardment offer a solution (see Measuring erosion rates, February 2002 issue of EPN), combined with apatite fission-track dating (Belton, D.X. et al. 2004. Quantitative resolution of the debate over antiquity of the central Australian landscape: implications for the tectonic and geomorphic stability of cratonic interiors.  Earth and Planetary Science Letters, v. 219, p. 21-34).  The results suggest that Australian landscape antiquity is a myth.  Erosion rates since the Cambrian varied over most of the Red Centre from 0.4 to 4.0 metres per million years, and reached as high as 17 m per Ma on occasion.  They suggest a common or garden history, comparable with those of most continental interiors.  Again and again it has been buried by sediments, albeit on a flat surface, and equally it has been exhumed several times by erosion.  Only at the outset of the Cenozoic did much of it sit unchanged for long, which enabled its red surface to develop.  The present surface is covered with what is termed regolith by Australians, but much of that is reworked material from the Palaeocene laterites that sits in a network of shallow drainage systems, including huge ephemeral lakes.  It might seem that recourse to Hutton’s “the present is the key to the past” should long ago have staved off the myth of the gnarled old place of which Australians have become inordinately proud.

It is rare for one issue of a “journal of record”, such as Nature to contain three papers on closely related topics, especially when they are geoscientific, but its 11 December issue of 2003 did.  All were about the way in which mountains erode, and attempted to measure the rates involved in three different settings.  Insofar as it is possible in Earth science, they try a reductionist approach in terms of the climatic and tectonic forces that are involved in denudation.  Getting useful timings is not as easy as it might seem with measuring fission tracks and the amount of radiogenic helium generated by decay of uranium and thorium isotopes in grains of apatite.  The principle lies in estimating when unroofed rocks rose and cooled below the temperatures at which apatite loses noble gases and the tracks in it formed by alpha particle emission heal up. In an exposed section subjected to erosion and isostatic uplift the higher rocks should record older ages than those lower down, the difference representing the pace of erosion and uplift.  There is, as yet, no way that periods less that 500 thousand years can be resolved by either method, and in terms of recent climate that can cover several glacial-interglacial cycles.

The simplest of the case studies was in the Cascade mountains of the NW USA, where there has been minimal tectonic activity, but a great deal of rain over the last few million years.  The crust has risen as material was stripped off the mountains. The average rates of erosion on time scales of millions to tens of million years closely follow the modern variation in precipitation over the area (Reiners, P.W. 2003.  Coupled spatial variations in precipitation and long-term erosion rates across the Washington Cascades.  Nature, v. 426, p. 645-647).  As a result, western parts of the range where rainfall is far higher than in the eastern rain shadow could be expected to be rising as much as three times faster, if a balance between erosion and isostatic uplift has been achieved. Since erosional power is expressed by rainfall and surface gradient, the fact that average erosion rates do not correlate well with topographic relief suggests that precipitation has outweighed the effects of slope steepness.  The opposite seems to hold in the Himalaya of central Nepal, which show the most gross variations in precipitation, due to monsoonal conditions (Burbank, D.W. and 7 others 2003.  Decoupling of erosion and precipitation in the Himalayas.  Nature, v. 426, p. 652-655), yet long-term erosion rates do not vary very much, except between the topographically distinct Lesser and Greater Himalaya ranges.  The Himalaya are altogether more geologically and tectonically complex than the NW USA, so finding such little variation is as interesting as it seems currently inexplicable.  The lack of correlation in the Greater Himalaya between precipitation (a five-fold decrease from south to north across the range) and erosion rates (more or less constant and high) suggests that tectonic uplift is the main driving force.  Much the same findings from the area immediately to the east in the Nepalese Himalaya, though using a mica Ar-Ar thermochronology method that spans a longer period, have been interpreted very differently (Wobus, C.W. et al. 2003.  Has focused denudation sustained active thrusting at the Himalayan topographic front?  Geology, v. 31, p. 861-864).  Wobus and his colleagues from MIT suggest that rapid rise of the Greater Himalaya (~10 km in the last 10 Ma) was induced by isostatic uplift driven by erosion, even maintaining movement on the huge bounding thrusts to the orogenic belt.  Altogether more complicated is the erosion of Taiwan, which is seismically active, has a complex tectonic history that affected rocks of very different strengths in different areas and is subject to a highly variable maritime climate (Dadson, S.J. and 11 others 2003.  Links between erosion, runoff variability and seismicity in the Taiwan orogen.  Nature, v. 426, p. 648-651).  They detect changing patterns of erosion as deformation has migrated.  Attempts at correlation between modern erosion rates and various factors came up with only two of significance, with recent seismicity and typhoons.  Each triggers landslips that instantaneously add debris to flowing rivers.  Precipitation rates, river discharge, slopes and stream power showed little link with erosion rates.  Of the four papers, only one (Wobus et al.) is able to relate differences in the erosive power of streams to the contrasting erosion rates of the Greater and Lesser Himalaya.

Such a hodge-podge of seemingly conflicting findings, based on studies that use supposedly revolutionising techniques, must worry agencies who have been induced to part with large funds to support fission-track and (U-Th/He) dating facilities supposedly to advance geomorphological studies.  Peter Molnar, who with Phillip England first reviewed the complex interplay between erosion, tectonics and uplift, and their counter-intuitive outcomes, made the following pithy comment, “The differences among these papers call attention to the inadequacy of current theory, without which one gropes for a way to plot data”.  Plainly, there has been over-excitement about techniques in the hope of empirically deriving theories, which has resulted in half-cocked research, and some gullibility among funding bodies.

See also:  Molnar, P. 2003.  Nature, nurture and landscape.  Nature, v. 426, p. 612-614.  New Scientist (31 January 2004) includes a 12 page special report on the technological issues involved in the Bush vision.

Working out the rate at which landscapes evolve depends on some means of dating surfaces formed at different stages in the cutting down of topography.   Modern studies rely to a large extent on the build up of isotopes, such as ¹⁰Be, that form in minerals no more than a metre or so beneath the Earth’s surface when they are exposed to cosmic ray bombardment.  If such transmuted nuclides stay in place, for instance on a relic surface or a series of alluvial terraces, cosmogenic isotope analysis dates the formation of that surface.  No matter how precise such surface dating can be, and currently there is a slop of around 20% either side of an age, there is a limit to the number of suitable surfaces.  So, the continual degradation of landscape can only be sampled at a few isolated times.  At best, an average denudation rate over long periods is all that geomorphologists can hope for.  The same goes for analysis of the range of times during which grains in a sediment were exposed to cosmic rays, before they were eroded, transported and finally protected from bombardment when they were buried in alluvium, that is another approach to timing erosion.  Average rates of erosion are useful in assessing some aspects of landscape development, but they are not much good for judging how it took place.  Standing in some awesome scenery easily gives the impression that it must have evolved by some continuous, steady process, and there is a long tradition dating back to James Hutton that views surface processes in that way.  Changes in rates have been seen as responses to “rejuvenation”, either by falls in the base-level of erosion or tectonic uplift to add gravitational potential to a region that makes flowing water more energetic.

Another approach is to look at the actual transfer of mass, either carried by rivers during different seasons or in the volumes of sediments that were deposited by recognisable individual events, such as a flood.  In large river basins that have a low average gradient it is well-accepted that occasional floods don’t have much effect on sediment movement in the long term, but most of the sediment moved in such basins is alluvium already supplied by earlier processes.  In mountainous areas rivers carry material directly from bedrock and the regolith that lies on it.  Anyone who has witnessed flash floods in a normally crystal clear mountain river knows their awesome power.  They become mud torrents studded with boulders that even fly through the air; they are debris flows rather than streams in the normally accepted sense.  Such flows are episodic, but frequently annual, and exert a major influence over denuding the landscape.  Yet over millennia, they too should maintain a consistent down wearing.  In the Appalachian mountains of the eastern USA denudation rates seem to average out between 2.5 to 5 centimetres per thousand years.  However, four catastrophic Appalachian storms in the late 20^th century, related to hurricanes, had an astonishing effect on erosion there (Eaton, L.S. et al. 2003.  Role of debris flows in long-term landscape denudation in the central Appalachians of Virginia.  Geology, v. 31, p. 339-342).  Carbon-14 dating of ancient mass-flow deposits formed in Virginia by comparable storms indicates recurrences in particular drainage basins around every 2500 to 3500 years.  During that time the average rate of erosion would have denuded the surface by a measurable amount (5 to 10 cm), yet the recent storms removed between 47 to 63% of that expected during periods measured in millennia.  The Appalachians are well vegetated, and therefore well protected from the effects of extreme floods compared with the surfaces of really big mountains such as the Himalaya and rugged areas in arid regions.  The obvious question is, “Are average denudation rates, no matter how precise, very relevant to the way landscapes actually develop?”  It is an important one, because weathering of debris from mountains is regarded by many geochemists as a means of taking carbon dioxide from the atmosphere – silicate weathering that involves CO₂ dissolved in rainwater locks atmospheric carbon in bicarbonate ions that carbonate-secreting creatures in the sea can sequester to deep storage when they die.  If about half the erosion of mountains is in widely separated catastrophes, which shift and then dump debris in a matter of days, then it is possible that the sums based on equating rates of weathering with those of erosion are not entirely valid.  Weathering of continental silicates is one means of forcing global cooling by reducing the greenhouse effect, and understandably mountain rivers teem with researchers sampling the water and sediment load, especially in the Himalaya.  If the bulk of debris shed to plains, such as those of the Indus, Ganges and Brahmaputra, never had time to be weathered at high altitude because it moved in catastrophic pulses, then maybe the sampling should be done somewhere else.  Processes in the vast alluvial tracts far below high mountains are slower and more constant wit time, so maybe looking at groundwater that moves through them might add to the current research..