With increasing populations and climate change reservoir storage is a more valuable resource than ever. However, that resource is dwindling.  Despite construction of new reservoirs, storage space per capita has been decreasing since 2000 [1]. This is due not only to growth in population and demand, but also because storage space in existing projects is continuously lost due to sedimentation. Therefore, when evaluating the importance of reservoir storage loss, reservoir storage should be considered a non-renewable, finite resource [2]: as most of the promising reservoir locations are already developed, marginal costs of adding storage increase, while the number of suitable dam sites still available decreases.

Sedimentation also poses major maintenance and operation challenges. When sediment accumulates behind the dam, dam safety may be compromised. Sediment that reaches turbines or other equipment (both in storage and diversion projects) can damage hydroelectric equipment and require costly additional maintenance. Sedimentation at the upstream end of a reservoir can form delta-like deposits, which can reduce the active storage and reduce the operational capacity of a reservoir to buffer floods or to provide peak energy.

Sediment trapping in dams impacts rivers downstream. Riverbeds, banks, bars, and deltas are formed through natural sediment transport through a river system. Interrupting sediment transport by means of dam construction and releasing sediment starved, “hungry” water [3] will hence directly impact river forms and processes (morpho-dynamics) and the ecosystems, infrastructures, and livelihoods rivers support. Limiting these downstream impacts is key in situations where downstream degradation of river forms damages infrastructures [4], but also for relicensing under stricter environmental regulations.

Thus, sediment accumulation in reservoirs can have severe impacts on operation and maintenance [5] of dams and on their downstream impacts [6]. Understanding and minimising dam sediment trapping and downstream impacts for both existing and new projects is of interest for planners, operators, owners and financers of dams and reservoirs alike.

Approaches for sediment management

How much sediment is trapped in a reservoir is controlled (1) by its technical design, operation, and reservoir sediment management, and (2) the sediment input from the reservoir’s catchment. The technical design (e.g., volume and residence time of the impoundment, availability of bottom gates), together with operational and management strategies (e.g., sediment sluicing, flushing, or dredging) will control how effectively sediment that enters the reservoir can be released downstream.

Sediment inputs into a reservoir are, instead, controlled by a number of factors acting on the scale of the reservoir’s entire catchment. First, natural sediment yield can vary widely over the landscape as a function of, for example, lithologic and climatic gradients. Second, human disturbance such as land use changes or road construction can greatly increase natural sediment yields. Last, upstream reservoirs and dams can trap parts of the original sediment load and greatly reduce the sediment yield for a downstream reservoir (that effect will decrease over time: as upstream reservoirs fill with sediment their trap efficiency will decrease and more and more sediment will be passed downstream). As a result, sedimentation in reservoirs varies widely between different locations. For Californian dams, for example, Minear and Kondolf [7] showed that sedimentation risk for reservoirs varies greatly as a function of heterogeneous natural sediment yields and the presence of other dams in the catchment. 

Operational and technical strategies to reservoir sediment management have been discussed extensively (for an overview see Kondolf et al., 2014 [6]) and are of key importance to reduce sediment trapping in existing and planned dams. However, even the best reservoir sediment management cannot avoid some sediment trapping in most cases. Therefore, we propose to adopt a catchment perspective on sediment management. Carefully selecting dam sites in such a way that the final reservoir cascade cumulatively traps the least possible amount of sediment might be a very effective way to reduce sediment trapping in dams, especially in cases where decision makers can select from multiple potential dam sites.

New approaches

Quantifying sediment trapping in hydropower cascades requires integrated sediment modeling at the scale of the entire river catchment.  While catchment-scale processes are often poorly monitored and few hydro-dynamic numerical models are applicable on such scales, new global datasets from remote sensing and new modeling approaches can help us understand sediment transport and to manage reservoir sedimentation on entire catchment scales. 

The CASCADE (CAtchment Sediment Connectivity And DElivery) model [8,9], for example, was developed for network-scale sediment assessment and management tasks. CASCADE provides a flexible, numerically efficient, and spatially distributed screening tool for modeling sediment connectivity and transport in large river networks. Numerical efficiency is key to modeling entire river networks, apply optimization based methods for portfolio sediment management, and to quantify uncertainty in modelled sediment transport via stochastic modeling approaches. 

CASCADE aims to fill a gap where common hydro-dynamic models are not practical because of their data and computational demand while common empirical approaches for estimating sediment yield and reservoir sedimentation fail to represent the spatio-temporal component of sediment trapping in reservoir cascades. CASCADE separately models sediment transport from many distinct sediment sources deriving information on how each reservoir is connected to specific sediment sources in its upstream catchment. It requires only data that will be available for most river systems –  a digital elevation model and gauged or modelled time-series of river discharge. Additional data, for example on reservoir operating policies, or separation in dead and active storage can be used if available. CASCADE supports a 1-D representation of reservoir geometry that can be extracted from globally available digital elevation models [9].

Network scales

The Se San, Se Kong, Sre Pok Rivers join before flowing into the Mekong, forming one of the river’s largest tributary basins.  Commonly referred to as the “3S” basin, it is only 10%  (around 80,000km2) of the Mekong’s drainage area, but contributes nearly 20% of sediment and water flow in the lower Mekong. Construction of 41 reservoirs (with a total production capacity of around 31,000GWh/yr) in the basin raises concerns with regard to the impact of sediment trapping on the Mekong’s sediment budget as well as to future storage loss in the 3S reservoir cascade.

However, where sediment originates in the basin and how sediment would be trapped in the planned hydropower cascade was previously poorly understood. We applied CASCADE to model connectivity for the sand fraction (Figure 1), which is more likely to be trapped by dams than silt and clay, and which has most relevance for downstream river and delta morphology. Based on a stochastic ensemble forecast, we found that most of the total sand load (around 15 Mt/yr) likely originates from the Sre Pok River [10].

We also applied CASCADE to estimate cumulative sediment trapping and storage loss in the 3S reservoir cascade over the 2010-2050 period. The result is a map of bottom level change in all 41 reservoirs over the 40-year modeling period. While results indicate that only few reservoirs are likely to lose a significant amount of their total storage over the study period (Figure 2 a), CASCADE gives a detailed spatially distributed assessment of reservoir sedimentation (i.e., where in each reservoir a certain grain size fractions will deposit, Figure 2 b and c).

Strategic selection of dam portfolios

We also used the 3S basin as a case study to demonstrate how different dam portfolios (i.e., different sets of dam sites) can result in variable sediment trapping in reservoirs, despite yielding similar economic benefits. We created 17,000 different dam portfolios from the 41 dam sites in the 3S (Figure 3a) and quantified sediment trapping and hydropower production for each portfolio. Sixty of 17000 portfolios result in an optimal trade-off between sediment trapping and hydro-power production, hence a minimal sediment trapping is achieved for a given level of hydropower production (Figure 3b, dots).  Optimal portfolios of upstream dams (Figure 3d) that exploit 70% of the full hydropower potential while trapping around 20 % of the total sediment flux. 

However, the actual development pattern in the 3S to date has been far from optimal.  Dams constructed to date (including the massive Lower Se San 2) will trap 90 % of the transported sediment, while generating only 30% of the total hydropower production potential (Figure 3b, squares)., focusing most hydropower development in the Se San and Sre Pok Rivers (see Figure 3a and c).  However, an optimal portfolio would have focused development in the Se Kong River, with lower sediment loads (Figure 3d)

Fostering future sustainable dam sediment management

The 3S case study demonstrates that strategic network-scale assessments, supported by new modeling approaches, have the potential to reduce economic and ecological impacts of sediment trapping in dams. For the future, we envision a multi-stage, multi-stakeholder approach for sustainable sediment management on network scales. Rather than evaluating and approaching sedimentation for a single dam site only, strategic early selection of dam portfolios that minimise dam sediment trapping, preserve storage space, and reduce downstream impacts is in the interest of national planners, power providers, and financers. Network scale models of connectivity can also be useful for existing dam cascades, e.g., to identify origins of sediment that accumulates in dams or to evaluate the effectiveness of catchment/cascade scale sediment management.

Such a first screening can help to strategically focus more costly numerical models and field-based assessments (e.g., of reservoir hydraulics, sediment yield, geology, and river processes) to make decisions in dam placement and sediment management that maximize reservoir lifetime, minimise O&M costs, and compromise downstream rivers the least possible – hence working towards optimal infrastructure decisions and more sustainable hydropower.

 

The authors are Rafael Schmitt and Matt Kondolf from the University of California Berkeley College of Environmental Design. www.riverlab.berkeley.edu

 

References

  1. Wisser, D., Frolking, S., Hagen, S. & Bierkens, M. F. P. Beyond peak reservoir storage? A global estimate of declining water storage capacity in large reservoirs. Water Resour. Res. 49, 5732–5739 (2013).
  2. Annandale, G. Quenching the thirst: sustainable water supply and climate change. (CreateSpace Independent Publishing Platform, 2013).
  3. Kondolf, G. M. PROFILE: hungry water: effects of dams and gravel mining on river channels. Environ. Manage. 21, 533–551 (1997).
  4. Rinaldi, M., Wyżga, B. & Surian, N. Sediment mining in alluvial channels: physical effects and management perspectives. River Res. Appl. 21, 805–828 (2005).
  5. Schleiss, A. J., Franca, M. J., Juez, C. & De Cesare, G. Reservoir sedimentation. J. Hydraul. Res. 54, 595–614 (2016).
  6. Kondolf, G. M. et al. Sustainable sediment management in reservoirs and regulated rivers: Experiences from five continents. Earths Future 2, 256–280 (2014).
  7. Minear, J. T. & Kondolf, G. M. Estimating reservoir sedimentation rates at large spatial and temporal scales: A case study of California. Water Resour. Res. 45, W12502 (2009).
  8. Schmitt, R. J. P., Bizzi, S. & Castelletti, A. Tracking multiple sediment cascades at the river network scale identifies controls and emerging patterns of sediment connectivity. Water Resour. Res. 3941–3965 (2016). doi:10.1002/2015WR018097
  9. Schmitt, R. J. P. CASCADE – A framework for modeling fluvial sediment connectivity and its application for designing low impact hydropower portfolios. (Politecnico di Milano, 2016).
  10. Schmitt, R. J. P., Bizzi, S., Castelletti, A. & Kondolf, G. M. Stochastic modeling of sediment connectivity for reconstructing sand fluxes and origins in the unmonitored Se Kong, Se San, and Sre Pok tributaries of the Mekong River. J. Geophys. Res. Earth Surf. (in Review).