Dam locations are indicated for
two scenarios: definite future, and full-build.
Main stream dams are included
in full-build scenario, but represented separately.
Bedrock channel of the Mekong
River with surficial sand deposits 1-2 m thick, near Xayaburi, Laos. (photo by Kondolf, January 2012)
Abstract
The Mekong River is undergoing
rapid dam construction. Seven main stream
dams are under construction in China and 133 proposed for the Lower Mekong
River Basin (LMRB). We combined geomorphic assessments of the Mekong channel and delta with models of sediment
trapping by reservoirs. We expect the biggest geomorphic changes to occur along
alluvial reaches, though stripping of thin sediment deposits in bedrock reaches
may also have significant consequences for benthic invertebrates, fishes, and
other aquatic organisms dependent on the presence of alluvium in the channel.
If all dams are built as proposed, the resulting 96% reduction in sediment
supply would have profound consequences on productivity of the river and
persistence of the Delta landform itself. Strategies to pass sediment past dams
should be explored to reduce the magnitude of sediment starvation and resulting impacts.
Key Words: Mekong River,
sediment load, reservoir sedimentation, channel change, dam construction
1
Introduction
The Mekong River basin is
undergoing rapid and widespread dam construction (Grumbine and Xu, 2011). On its upper reaches in China (the Lancang),
seven dams have been constructed or are under construction on the mainstem. In
the Lower Mekong River Basin in Laos, Thailand, Cambodia, and Vietnam, another
38 tributary dams are considered certain to be built, with an additional 95
dams at some level of planning for tributaries and the mainstem. How will these
dams alter the sediment load of the Mekong and how will the dams change the
morphology of the downstream channel and delta?
The Delta is currently home to 20
million people whose lives and economy are at risk from subsidence increased flood
risk, and other changes in the delta. Previous authors have estimated the
average annual suspended load of the entire Mekong as 160 million tonnes per
year (Mt y-1), and observed that about half of
this amount was produced by the upper 20% of the basin, the Lancang drainage in
China (Gupta and Liew, 2007; Walling, 2005). These values reflected
conditions prevailing prior to ongoing construction of a cascade of seven dams
on the Lancang, now mostly completed, and which will trap about 83% of the
Lancang basin sediment when complete (Kondolf et al submitted). Thus,
almost half of the natural sediment load of the Mekong will be lost in the
reservoirs of the Lancang, and the sediment load of the Lower Mekong River will
consist largely of sediment derived from sources within the LMRB itself. The
hotspot of the river’s extraordinary fish production is the Tonle Sap system.
This tributary to the Mekong receives monsoon-driven, seasonal backwater
flooding from the Mekong River mainstem, during which time fine-grained
sediments deposit across the Ton Sap basin, making the Tonle Sap tributary
basin net depositional (Tsukawaki 1997, Kummu et al. 2005) and potentially vulnerable
to reduced sediment supply (Baran and Guerin 2012).
Most large rivers in the world are
experiencing decreased sediment loads due to dam- induced sediment starvation.
In the two millennia prior to widespread dam construction, human activities
such as forest clearing and cultivation increased erosion and sediment delivery
to the oceans ((Leopold 1921, Leopold 1923, Wolman and Schick 1967, Milliman
and Syvitski 1992, Syvitski 2008). Worldwide, widespread dam construction has
reversed this historical trend, and substantial reductions in the delivery of
sediment to the oceans are now occurring in many of the world’s rivers
(Milliman and Syvitski, 1992). The Mekong, however, differs from other large
Asian rivers, having produced a relatively consistent sediment yield over the
past three thousand years (Ta et al., 2002), reflecting relatively modest
levels of development that prevailed until very recently.
Reservoirs trap all the bedload
and a percentage of the suspended load carried by a river. The supply of
sediment to the river downstream is reduced by the amount trapped.
Depending on the relative changes
in sediment supply and transport capacity (Schmidt and Wilcock, 2008) erosion
or deposition can occur, but most commonly the reach downstream of the dam is
characterized by sediment-starved, or ‘hungry’ water, which can erode the bed and banks
(Kondolf, 1997). These erosive flows threaten infrastructure and coarsen bed
material, fundamentally altering physical habitat and aquatic food webs (Power
et al., 1996). Sediment starvation
typically affects downstream river channels and deltas by causing incision and
bank erosion, habitat loss, and increased rates of land loss along the coast by
reducing sediment replenishment.
The consequences of delta
subsidence, both natural and accelerated, in combination with discharge
control, sediment-load reduction, and channel stabilization, is to accelerate
shoreline erosion, threaten the health and extent of mangrove swamps and
wetlands, increase salinization of cultivated land, and put human populations
at risk of costly disasters (Syvitski, 2008).
Whereas eustatic sea level rise associated with global warming has
received much focus and interest in recent years, the very land surface that
meets the water has been subsiding more rapidly in recent years, as dam building
reduces sediment supply needed for deposition on the delta plain, distributary
channels are stabilized and dyked so that sediment-laden floodwaters can no
longer disperse over the floodplain, and petroleum and groundwater extraction
induce subsidence. Deltas that develop dense cities and industrial
infrastructure become less resilient to tsunamis and hurricane-induced coastal
surges. Lives and wetlands at risk today in coastal regions will be even more
at risk in the future (Syvitski, 2008). The cumulative impacts of sea-level
rise, sediment starvation from reservoir trapping and instream mining of
construction aggregate, channelization of delta distributary channels, and
groundwater extraction are common to many of the world’s major rivers (Bucx et
al., 2010), and have consequences that are broadly predictable (Table 1).
Prior Work on the Mekong River and Delta
Gupta et al. (2002), Gupta and
Liew (2007), Carling (2005), Gupta (2008), and Carling (2009a) described the
geomorphic framework of the Mekong River (Table 2), noted differences in
geomorphic characteristics of reaches of the Lower Mekong, and to some extent,
explored how reduced sediment supplies might affect different reaches.
Following on Kummu et al.’s (2010)
initial estimates of sediment reduction from planned dams, Kondolf et al. (submitted)
developed geomorphically-based sediment yield estimates for the LMRB,
calculated trap efficiencies for individual reservoirs, and compiling total
storage capacity data for more accurate trap efficiency calculations. They then
applied the 3W model (Minear and Kondolf 2009) to calculate cumulative sediment
deficit from individual reservoirs under different reservoir development
scenarios, accounting for reduced trap efficiencies as reservoirs fill, and
accounting for multiple reservoirs in a given river basin. For the MRC’s
‘definite-future’ scenario of 38 dams already constructed, under construction,
or certain to be built, the sediment load reaching the Delta will be about half
of its pre-1990 level. However, with full build of dams in the Lower Mekong
River basin (Figure 1), including mainstem dams, the cumulative sediment
trapping by dams will be ~96% of its pre-1990 load. Sediment starvation would actually be more severe
owing to the mining of about 27 Mt y-1 of sand and
gravel from the river channel, mostly in
Cambodia (Bravard and Goichot 2012).
Flow alteration from existing and
proposed dams is expected to be more modest than the sediment trapping. Typical
of tropical rivers, Mekong River flow is seasonal, with a monsoon-driven high
flow period from July to October that is responsible for 75% of the annual flow
(Piman et al. 2013). Under a 41-dam “definite future” scenario, and a full-
build scenario of 136 dams in the lower Mekong, Piman et al. (2013), predicted
a dry season flow increase of 22% and 29% for definite future and full build
scenarios respectively at the Kratie station. Wet season flows were predicted
to decrease 4% and 13% for definite future and full build scenarios
respectively. Similar estimates of hydrological changes were also presented by
the Mekong River Commission (2010).
These changes in flow regime may
have significant impacts for the aquatic ecosystem and especially the fishery
of Tonle Sap (Baran and Myschowoda 2009, Lamberts and Koponen 2008), but the
small reduction in wet-season flow is unlikely to change the transport capacity
of the Mekong. As such, the river will still have the capacity to transport a
similar quantity of sediment with future hydropower development. Thus, sediment
trapping by reservoirs is arguably the most important consequence of dams for
the downstream channel.
Anthony et al. (2012) used
sequential satellite images to analyze coastal retreat in the Mekong Delta from
2003 and 2011, finding an average of 4.4 m y-1 of
coastline retreat across the entire delta, with higher rates of 12 m y-1 on the Ca Mau peninsula. Given the stable
sediment supply and the growth of the Delta during the last ~6,000 years,
Anthony et al. (2012) attribute this recent coastal retreat to the reduced
sediment supply caused by massive extraction of sand and gravel from the river
channel for construction aggregate (estimated to exceed 43 Mt y annually by
Bravard and Goichot 2012), and by levees and channel straightening in the
delta, which increase flow velocity and offshore sediment transport. To date, there has been a lack of analysis
(and for that matter, a lack of data. To date, there has been a lack of
analysis (and for that matter, a lack of data upon which to base analysis) to
understand how the Mekong Delta is likely to respond to future sediment
starvation. With better predictions of sediment starvation now available, it is
clear that sediment starvation effects are likely to be severe, and thus there
is an urgent need to draw upon available information for the Mekong Delta and
analogous systems to make initial projections of likely impacts and identify
critical data needs.
2
Methods
Our approach was twofold. We drew
upon prior geomorphic work on the Mekong River basin by Adamson (2001), Gupta
(2004) Gupta and Liew (2007), Gupta (2008), and Carling (2009a) to characterize
channel reaches in terms of their likely response to sediment starvation We
analyzed available data for other deltas as reported in the literature, and
systematically compiled data such as degree of hydrologic alteration,
percentage reduction in sediment supply, documented historical subsidence
rates, and wave energy. Based on these analogous case studies, we made initial
predictions for probable response of the Mekong Delta to the virtual
elimination of its sediment supply.
3
Results
Although the influence of
reservoir-induced sediment starvation on downstream channel change will clearly
be complex and varied, fundamental principles such as Lane’s Balance (Lane,
1955) and the presence or absence of geologic controls can be used as
preliminary predictive tools. The
results of Kondolf et al. (submitted) suggest that sediment trapping will be
substantial while reductions in high flows will be minimal (Mekong River
Commission 2010). Therefore, the Mekong River will continue to have the
capacity to transport sediment in large quantities, but the supply of sediment
for transport will be reduced. Channel adjustment will be limited primarily by
geologic controls.
3.1
Delineation of
Reaches
The Upper Bedrock reach extends
from the Chinese border downstream to about 5 km upstream of Vientiane. In this
reach the Mekong River channel is bedrock controlled, with limited, and
presumably transient, sediment storage (Figure 2). The channel gradient
averages 0.0003, and channel width ranges from 200-2000m. This reach includes
many wide, bedrock-floored reaches where bedrock is discontinuously overlain by
a thin (ca 1-2 m) veneer of sand (Figure 3). The Middle Alluvial Reach extends
downstream from Vientiane to Savannakhet. It is alluvial, with both
single-channel and island-bar sections. Channel gradient averages 0.0001, and
the channel is 800 to 1300m wide.
From Savannakhet downstream to
Kratie, the Middle Bedrock Reach is again bedrock controlled. This reach
includes a wide range of channel forms, as reflected in Gupta and Liew (2007)
having broken this section of river into 4 reaches (their reaches 3, 4, 5, and
part of 6). For our analysis, the key attribute of all these reaches is bedrock
control (and thus we consider it a single reach), though a variety of
sedimentary forms are present including sections with alluvial banks,
anastamosed channels with rock-core islands covered with a relatively thin
veneer of sand and silt (Meshkova and Carling 2012). For example, from Sambor
to Kratie, the bedrock control is largely buried, so the river here displays
many alluvial features. However, the underlying bedrock limits its potential
response to sediment starvation. Channel gradient in the upper portion of Reach
3 is approximately 0.00006 and decreases downstream. Channel width ranges from
750 to 5000 m. The Cambodian Alluvial Reach extends from Kratie downstream to
Phnom Penh. Here, the Mekong is again alluvial, crossing the wide floodplain of
Cambodia to enter the depositional reaches of the delta. Channel gradient is 0.000005 and widths range from 3000 to 4000m. Some
large-scale structural control is provided by bedrock, but channel planform and
position is primarily set by channel migration through alluvium. Downstream of
Phnom Penh is the Mekong Delta, by definition a reach of net deposition. The
delta occupies an area of ~94,000 km2 making it the
third largest delta in the world (Coleman and Wright 1975). The delta begins
~330 km from the sea where the Bassac, the first deltaic distributary,
separates from the mainstem. The two channels flow parallel for 200 km without
additional distributaries or connecting channels. Ultimately, there are four
main channels that reach the sea. As a
result of groundwater extraction and limited sediment starvation, the
Mekong is categorized as a “Delta in Peril” with late 20th Century aggradation at less than
0.5 mm y-1 and relative sea level rise
occurring at approximately 6 mm y -1 (Syvitski et
al., 2009). Sediment trapping under future dam building scenarios will further
limit sediment delivery and distribution in the delta.
3.2
Potential effects on channel Reaches
Definite Future Scenario
Under the “definite future
scenario”, the Upper Bedrock Reach will have an 83% reduction in sediment at
the upstream end of the reach, though as less regulated tributaries enter the
reach the cumulative trapping decreases to 64% at the downstream end of Reach 1
(Kondolf et al. submitted). The relative reduction in sediment supply in
this reach is the greatest of any reach in the definite future scenario, yet
because the reach is bedrock controlled we anticipate only modest channel
adjustment. Loose sediment deposits over bedrock as described by Carling 2009a
(including slack-water deposits on bars, islands, inset floodplains and banks)
(Figure 3) will likely be swept away in the first competent floods post-dam.
Changes in bed-level will likely be confined to accelerated scouring of pools
(Carling, 2009b)
In the Middle Alluvial Reach, the
sediment reduction decreases to 52% at the downstream end of the reach while
sediment reductions in the Middle Bedrock and Cambodian Floodplain Reaches
fluctuate between 51-56%. Under current conditions, bank erosion is not
excessive (Darby et al., 2010) but we anticipate the most substantial post-dam
erosion and channel adjustment in the Middle Alluvial reach and Cambodian
Alluvial reaches, where bank erosion is currently occurring and where coarse
bed sediment is exposed, suggesting
Holocene incision (Carling, 2009a). Large island features are believed to
result from chute cutoffs and suggest a dynamic river system (Carling, 2009a).
Because upstream reservoirs will have a limited influence on flow regime, but
will trap more than half of the total sediment load, we expect channel widening
in alluvial reaches as the river seeks to recover its sediment load by eroding
the channel margin, as commonly observed in sediment-starved rivers (Kondolf,
1997).
Incision is also likely except
where the bed elevation is controlled by bedrock. The Middle Bedrock reach has
single-thread, bedrock-confined reaches, anastomosed reaches of bedrock
islands, and also includes the base-level control of Khoné Falls (a Holocene
lava flow that crosses the river) (Carling, 2009a). Future high flows of
sediment-starved water may erode alluvium on bars, banks, and islands without
replacing. Erosion into bedrock is not expected on the timescale of decades.
The Cambodian Alluvial reach is a floodplain river with active meandering in
anabranching and anastomosed sections.
Individual islands are transient
features, though the island complex is relatively stable (Carling, 2009a).
Without replenishment, new islands will be less likely to develop and loss of
the island features will likely occur. Erosion of the main channel bed and
banks is also expected. The Mekong Delta
will receive about half of its natural sediment load, and can be expected to experience
accelerated subsidence and coastal erosion. Further research is needed on the
size distribution of sediment transported by the river, and the size fractions
most affected by the dams, but we expect the dams to disproportionately affect
bed material load, notably sand, which is most important for building beaches
and nourishing the coast. See further discussion of the Delta response to
reduced sediment supply below.
Full Build Scenario
With all proposed dams constructed
(full build scenario) and cumulative sediment reduction ranging from 83% below
the Chinese boarder to 96% in Vietnam, we expect the most dramatic response in
the alluvial reaches from Vientiane to Savannakhet and from Kratie downstream,
where channel bed and banks will be susceptible to erosion.
While the bedrock-controlled
reaches above Vientiane and from Savannakhet to Kratie will not down cut
(except to remove any layers of erodible alluvium overlying bedrock, and/or to
deepen pools) and will not have dramatic occurrences of erosion or channel instability,
the extensive existing sediment deposits (bars, islands, inset floodplains and
banks) will be stripped away, and bed material size will coarsen, all with
potentially important ecological consequences. If the thin veneer of sediment
in the bedrock reaches is removed, it can substantially alter the substrate,
base flow channel roughness, and water velocity that influence fundamental
elements of habitat availability for the benthic macroinvertebrates, fishes,
and other aquatic biota. Since little is known about many Mekong species, it is
difficult to predict their response to channel change and consequent loss of
habitat.
If the main stream dams are
constructed, they will inundate long reaches of the river and their backwater
effects will extend further upstream. Within these reservoirs and back- water
areas, rather than experiencing erosion from energetic flows, the channel will
become a depositional zone. An important ecological feature of the river are
the deep pools that provide essential habitat for native fishes and river
dolphins (Poulsen and Vlabo-Jorgensen 2001, Baird and Flaherty 2005), and which
are maintained by scour created by local hydraulics. Sediment starvation below
dams is unlikely to negatively affect these pools through increased erosion.
However, within the extensive zones of reservoir inundation and backwater,
local hydraulics will change, likely eliminating the scouring currents that
have maintained these features, and they will begin to fill with sediment and
debris.
3.3
Potential effects on Mekong River delta from
analogous cases
At present, approximately 21,000
km2 of land in the Mekong Delta is less than 2 m
above
sea level and 37,000 km2 is regularly flooded (Syvitski et al., 2009).
The pre-dam sediment accumulation rate across the Mekong Delta was ~0.5 mm yr-1 while relative sea level is rising at ~6 mm yr-1 (Syvitski et al., 2009). While sediment
delivery to the Mekong Delta has remained relatively constant over the 20th century, recent decades have seen accelerated
rates sea level rise and more rapid compaction due to groundwater extraction.
Thus, even under the pre-dam sediment regime, the delta was submerging and flood-prone areas expanding.
Future reductions in sediment discharge from the Mekong River or channelization
in the Delta will exacerbate the rate of land loss.
Similar to “full-build”
predictions for the Mekong River by Kondolf et al. (submitted), the
Colorado, Ebro, Indus, Krishna, Nile, and Yellow River deltas have all
experienced sediment reductions of 90% or more (Table 1). Since those deltas
have comparable or lower rates of relative sea level rise than the Mekong and
similar intensities of wave energy, they provide a reasonable framework for
understanding the likely impacts of unmitigated dam construction. The Indus, Nile, and Yellow River deltas were
all prograding prior to dam construction
and subsequently were net erosional. For example, the Indus coast was
prograding ~100 m y-1 before dam construction and
retreating ~50 m y-1 in recent decades. Pre-dam delta
growth is not reported for the Colorado, Ebro, and Krishna, but all are
actively eroding in the post-dam period (see Table 1 for citations).
Rates range from 1-90 km2 y-1 of area lost per
year and from 10 to 70 m y-1 of coastline
retreat. Detailed modeling of the delta is required to make quantitative
predictions of erosion, but experience from around the world suggests a high
likelihood of widespread erosion unless sediment management practices are
implemented for proposed Mekong dams.
4
Discussion
Deltas evolve through a complex
interplay of river, tides, waves, biological and human factors. For example,
mangrove communities slow wave velocities, thereby efficiently trapping
sediment, and improving water quality and preventing coastal erosion. The extent of mangrove ecosystems in the
Mekong delta has remained stable in recent decades (Shearman et al., 2013).
Elsewhere, mangroves are at risk from rapid sediment deposition (Ellison 1999)
(not likely in the Mekong), as well as sea-level rise and increased storm
intensity, aquaculture and water quality impacts, and sediment reductions from dams (Thampanya et al.,
2006). Such interactions exemplify the
challenges posed to modelers of deltaic systems. These are extremely difficult and important
processes to understand, yet likely impossible
to parameterize accurately given the unpredictable nature of storm events and
other stochastic processes. However, given the magnitude of sediment starvation
likely to occur in the near future, our review of experiences elsewhere,
combined with fundamental geomorphic principles, can provide an initial
prediction of likely effects. In a data-limited, poorly understood system such
as the Mekong, implementation of detailed models may be unrealistic due to the
lack of long- term and/or reliable data for calibration. By relying on the
global dataset we hope to inform decision makers and stakeholders in the Mekong
River basin while advances until accurate modeling and forecasting tools become
available.
