Potential impacts of climate change on reservoir services and management approaches

Abstract

Reservoirs are important resources for supporting economic growth and provide numerous cultural and ecological services. Currently, loss of storage capacity due to sedimentation, water quality degradation, and toxins from blue-green algal blooms are issues that threaten reservoir services. Climate change is hypothesized to exacerbate these problems by increasing sediment and nutrient export from the surrounding watersheds, changing flow regimes, and increasing summer water temperatures. This report adds to the reservoir management literature by synthesizing the disparate literature on potential impacts of climate change to reservoir services and reviewing both watershed and in-reservoir management strategies to mitigate the impacts of climate change. In addition, this study can serve as a resource for managers who seek to study the impacts of climate change on reservoirs by compiling tools and data sources that have been successfully used to study the impacts of climate change on reservoir systems.

Key words:

• climate change
• reservoir management
• reservoir services

Reservoirs are important economic infrastructures, reflecting billions of dollars invested worldwide. Many large reservoirs in the United States (U.S.) have been built for hydroelectric power, water supply, or flood risk reduction purposes. Yet despite the initial prescribed reservoir purpose, reservoirs are perceived as multi-use infrastructures with the ability to provide several essential services to regional populations (Hargrove et al. 2010). For example, more than 27,000, or 32%, of all reservoirs listed in the National Inventory of Dams (NID) have a primary purpose of recreation (USACE 2013). Many others have purposes such as irrigation, debris control, navigation, fire protection, and fish and wildlife support. Reservoirs can be found in every state in the U.S. Many small- and medium-sized reservoirs (dam < 30.5 m (100 ft)) are clustered in Texas, Oklahoma, Kansas, and Missouri, as well as the Southeastern U.S.; dams from 30.5 to 244 m (100 to 800 ft) are more commonly located in the Western US (USACE 2013). Approximately 50% of all dams listed in the NID were built between 1950 and 1979, with only 10% completed in the last 2 decades (USACE 2013). Although some dams are nearing the end of their prescribed design life, there are economic, social, and environmental incentives to implement reservoir and watershed management practices to ensure continued utility of these existing investments.

Water resource managers are faced with serious issues, such as declining water levels, sedimentation, and algal blooms in reservoirs in many regions of the U.S. and the world that challenge the long-term sustainability of multipurpose reservoirs (Hargrove et al. 2010). In addition, climate change is expected to exacerbate water shortages, erosion (Nearing et al. 2004), and the frequency of algal blooms (Paerl and Huisman 2009), which will create additional complications for reservoir management. The uncertainty of the intensity and duration of future droughts, as well as extreme precipitation events, are both of concern and challenging to planning efforts (Seneviratne et al. 2012).

The impacts of climate change for natural lakes are well known (e.g., Mortsch and Quinn 1996, Blenckner 2005, Pham et al. 2008, Adrian et al. 2009, Schindler 2009); however, reservoirs differ from lakes in several ways. First, reservoirs typically have larger watersheds compared to lakes, which means there is generally higher water, sediment, phosphorus, and nitrogen loads into reservoirs (Kennedy 2005). Second, surface area is also generally greater for reservoirs, which increases evaporative potential. Third, reservoir drawdown zones can be much greater than those of lakes, which can effect erosion and ecological processes in the littoral regions of reservoirs (Furey et al. 2004). Finally, reservoirs typically exhibit longitudinal zonation with 3 ecologically different zones—riverine, transitional, and lacustrine—which need to be examined separately and as a whole to fully understand climate change impacts on a reservoir (Wetzel 2001). The impacts of climate change on reservoirs are often studied from a water supply perspective (Li et al. 2010, Raje and Mujumdar 2010, Georgakakos et al. 2012, Alvarez et al. 2014, Park and Kim 2014), whereas reservoir water quality management issues are infrequently considered in the context of climate change (Zhou and Guo 2013).

The goal of this report is to synthesize the available literature and to review data sources and tools that can be used to understand the possible impacts of climate change specifically related to reservoir services. Specific management decisions for a given system will be case-specific; in other words, it is impossible to make over-arching recommendations for all reservoirs. However, this review can serve as an inventory of possible management solutions, tools, and data sources that may be useful in developing management strategies for reservoir systems affected by climate change. Kansas is used as a case study to discuss particular impacts and management efforts; however, the management strategies and tools are likely applicable to reservoirs within similar landscapes and climate conditions, especially within the Great Plains. Impacts related to changes in ice cover and alterations in snow pack are not considered in this review.

Reservoir-related services

Reservoir services include hydropower, flood risk reduction, recreation, nutrient attenuation, aquatic ecosystem support, and water supply for municipal, industrial, and agricultural uses. The concept of services provided by reservoirs is a useful framework to evaluate the current benefits derived from reservoirs because water quality metrics typically used in reservoir management may not be relevant to the public. For example, metrics such as total phosphorus, total nitrogen, and chlorophyll-a concentrations (Chl-a) are typically used to assess water quality conditions, yet the public is more interested in knowing if they can swim, fish, or boat safely on the water (Keeler et al. 2012).

The framework of reservoir services can capture the link between biological and physical measurements and the economic and social importance of the waterbody. For example, the lack of natural lakes in Kansas amplifies the value of reservoirs within the state. There are few alternative natural waterbodies for outdoor recreation, boating, sailing, fishing, and swimming, other than the limited use of rivers and streams. The annual value of Kansas reservoirs for recreation has been estimated ∼$15 million for Perry Lake, $17 million for Milford Lake, and $12 million for Tuttle Creek Lake (in 2009 dollars; CDM Federal Programs Corporation 2011). Additionally, 93 reservoirs within the state serve as water supply, and ∼60% of the state's population receives drinking water from these reservoirs (deNoyelles and Jakubauskas 2008). For Perry, Milford, and Tuttle Creek reservoirs, the value of water supply is estimated at ∼$294 million when including the avoided costs of constructing new reservoirs and estimated mitigation costs for maintaining water supply (CDM Federal Programs Corporation 2011). Valuing reservoir services has proved useful in other case studies, such as reregulating flow releases from the Glen Canyon Dam on the Colorado River and determining the recreational benefits of maintaining higher water levels on dams operated by the Tennessee Valley Authority (Loomis 2000).

Although all reservoir services are important, some are irreplaceable. Surface water storage is an essential service. In many parts of the U.S., groundwater sources are declining, and reservoirs are relied on to provide freshwater to support growing populations and agricultural production. Reservoir-derived services such as water supply, water quality, recreation, nutrient attenuation, habitat, and navigation will all likely be affected by climate change (Kundzewicz et al. 2007; Table 1).

Table 1 Reservoir services and possible impacts due to climate change.

Reservoir-Derived Ecosystem Services	Possible Effects due to Climate Change
Flood Control	Overwhelm flood control capacity
Water Supply: Municipal	Sedimentation diminishes water supply capacity; uncertainty in drought adds water supply stress; increased nutrient loading will increase eutrophication and algal blooms creating taste and odor events
Water Supply: Industrial	Sedimentation diminishes water supply capacity; uncertainty in drought adds water supply stress; decrease in water quality may require additional treatment before water use
Water Supply: Agriculture	Sedimentation diminishes water supply capacity; uncertainty in drought adds water supply stress; potential for increased salinity
Power Generation	Decreased inflow may bring water levels below turbines and decrease power generating potential
Recreation	Decreased water levels may prevent many aquatic recreation activities; poor water quality and algal blooms limit recreation availability
Habitat for Aquatic Organisms	Poor water quality and turbid waters may stress some aquatic species and allow invasive species to take over, or may cause loss of threatened or endangered species
Nutrient Attenuation	Stressed waterbodies will not be able to effectively attenuate nutrients and may export additional nutrients downstream
Navigation (releases to rivers)	Water supply stress may limit the amount of water available for navigation releases

Possible impacts of climate change on reservoir services

Recent studies support that the earth is warming due to increased concentrations of greenhouse gases in the atmosphere, which could increase the occurrence and magnification of drought, alter geographic and temporal precipitation patterns, and intensify precipitation events (Pachauri and Reisinger 2007, Seneviratne et al. 2012). Although reliably predicting impacts on a local scale is challenging, evidence suggests that climate change will have serious consequences for water management systems due to increased vulnerability to both drought and flooding (Handmer et al. 2012). Research evaluating the potential global and local impacts of climate change scenarios on water resources and water quality through modeling and empirical studies include Firth and Fisher (1992), Schindler (1997), Whitehead et al. (2009), and Brekke (2010).

Climate impacts on erosion and reservoir sedimentation

Reservoirs in agricultural watersheds typically have high sedimentation rates, often higher than original estimates (Hargrove et al. 2010). Major sediment sources come from cropland and grazing lands within the watershed, but also from eroding streambeds and streambanks (Devlin and Barnes 2008, Juracek and Ziegler 2009, Juracek 2011). Soil erosivity for the U.S. as a whole is expected to increase anywhere from 17 to 58% with climate change, with a great deal of variability among regions (Nearing 2001); the sensitivity of runoff and soil loss to precipitation change is also expected to increase, with an expected 1.7% change in erosion for each 1% change in precipitation (Nearing et al. 2004). Simulations in Midwestern watersheds show that a later planting date for crops such as corn and soybeans, in combination with climate change, can have a significant impact on the severity of erosion because cropland soils will be exposed during April and May storms (O’Neal et al. 2005).

Streambank erosion is also a significant problem. A study of Perry Lake in Kansas demonstrated that stream bank erosion above the reservoir was more important than surface soils in the overall amount of transported sediment (Juracek and Ziegler 2009). Stream segments below reservoirs also experience erosion; when the majority of the sediment load is deposited in the reservoir, outflow has low total suspended solids and has the capacity to erode sediment from the channel and streambed to reach equilibrium sediment load (Juracek 2011). The frequency of heavy precipitation is likely to increase with climate change, which would most likely lead to more high flow events and an associated increase in the amount of streambank and bed erosion (Seneviratne et al. 2012).

Increased watershed transport from extreme events

Studies of precipitation patterns from ∼1950 to 2007 in the contiguous U.S. have determined a 20% decrease in the return period for extreme precipitation events (DeGaetano 2009) as well as an increase in the intensity of events above the 95th percentile (Pryor et al. 2009). From a review of available studies, the Intergovernmental Panel on Climate Change (IPCC) has determined that a likely increase in observed heavy precipitation for many regions in North America has occurred over the last half century (Seneviratne et al. 2012). Heavier precipitation is anticipated to become more common with climate change because an increase in greenhouse gases in the atmosphere will lead to higher temperatures, and thus increased evaporation, water-holding capacity, and water content in the atmosphere, which ultimately can lead to enhanced precipitation rates (Trenberth 1999). Although a great deal of uncertainty persists in precipitation predictions from global climate models, projections confirm that the proportion of rainfall from heavy events will likely increase for most areas of the globe in the 21st century (Seneviratne et al. 2012).

To our knowledge, no modeling studies currently exist that estimate water quality impacts from projected extreme event analysis, yet several past studies have empirically drawn a connection between high intensity precipitation events and increased turbidity, sediment loads, and phosphorus export from the watershed (Murdoch et al. 2000, Stutter et al. 2008b). Additionally, with longer dry periods between storms, pollutants can potentially build up in the watershed and result in high pollutant concentrations in storm runoff (Stutter et al. 2008a). Large pulses of nutrient- and sediment-rich storm water could create significant short-term water quality changes that may exceed biologically relevant thresholds and have long-term effects on the ecosystem balance (Murdoch et al. 2000).

Several studies have also demonstrated that an increase in rainfall intensity has a greater effect on erosion rate than only an increase in total rainfall (Pruski and Nearing 2002, Nearing et al. 2005). Empirical evidence also suggests that infrequent, yet large, rainfall and runoff events are already responsible for a greater proportion of overall erosion. A 28-year study found that the 5 largest events contributed 66% of the total erosion within several Ohio watersheds (Edwards and Owens 1991), and a Kansas study of sediment sources to large federal reservoirs, including Kanopolis Lake and Tuttle Creek Lake, indicated that large storms are responsible for the majority of transported sediment. For example, in 2010 at the Ellsworth stream gauge located upstream of Kanopolis Lake, 7 storms accounted for ∼48% of the total discharge and 88% of the total suspended sediment load (Juracek 2011).

Impacts to water quality

Studies linking observed changes in water quality due to recent climatic changes found relationships between eutrophication and algal blooms at higher temperatures and higher nutrient loads with increased storm runoff (Jimenez Cisneros et al. 2014). With future climate change, there is potential for both increased temperature and altered precipitation regimes to increase the export of pollutants such as nutrients, pathogens, pesticides, and heavy metals into surface waterbodies (Kundzewicz et al. 2007, Jimenez Cisneros et al. 2014).

Based on literature cited in this report, along with principles of limnology, we constructed a diagram that explores relationships between increased precipitation and temperature with eutrophication, but for the sake of clarity, not every relationship is represented (Fig. 1). The figure shows that increased temperature will likely increase evaporation rates, leading to diminished water levels, especially during periods of low precipitation. Decreasing water levels have the potential to increase eutrophication by increasing water residence time, which increases nutrient concentrations and the growth potential of algae (Whitehead et al. 2009). Decreasing water levels can also lead to increased turbidity, which has contrasting effects of potentially releasing phosphorus from the sediments as well as reducing light infiltration for algal growth. As an example, Olds et al. (2011) found that water quality parameters varied significantly between drought and normal conditions for Harlan County Reservoir in Nebraska. Chl-a and turbidity were significantly higher and dissolved oxygen was significantly lower during drought conditions, even though water temperature remained constant (Olds et al. 2011).

Figure 1 Positive and negative feedback loops between climate factors, watershed processes, and eutrophication. Negative feedback is indicated with a dotted line and a negative sign (−). For the sake of clarity some relationships were omitted from this figure.

Increased temperature has the potential to increase nitrification in soils and nitrogen availability as well as mineralization, and release phosphorus and carbon from soil organic matter (Delpla et al. 2009). Increased precipitation also will likely increase runoff and the flushing of shallow soils, which would lead to increased nutrient loads to waterbodies (Kundzewicz et al. 2007). Dissolved oxygen (DO) concentrations are also dependent on temperature. A rise in water temperature can decrease DO solubility (Wetzel 2001). If increased water temperatures coincide with high concentrations of nutrients and algae in the water column, DO could reach low levels in bottom waters, risking aquatic life (Whitehead et al. 2009). Low DO concentrations also lead to a reducing environment in which iron hydroxides dissolve, releasing additional phosphorus that was bound to iron-rich sediments and creating an internal phosphorus load, which further exacerbates eutrophication (Wetzel 2001). Internal phosphorus load can also be amplified by suspension of benthic sediments due to lower water levels and increased turbidity from drought conditions (Dzialowski et al. 2008, Wildman and Hering 2011).

Climate change is hypothesized to be a major factor in the rising occurrence of algal blooms globally through increases in water temperature, water residence time, vertical stratification, and nutrient pulses from extreme events (Paerl and Huisman 2009). In a warmer climate, cyanobacteria may have a competitive advantage for resources because cyanobacteria often have optimal growth rates at high temperatures (i.e., >25° C; Jöhnk et al. 2008, Paerl and Huisman 2009). Some blue-green algae, or cyanobacteria, release toxins harmful to both humans and animals when ingested or by skin contact (Codd 2000, Codd et al. 2005). Some cyanobacteria produce taste and odor compounds that create problems for municipal water suppliers, recreational users, and nearby residents or businesses. Taste and odor events can be treated with powdered activated carbon, but conventional treatment methods do not seem to remove all toxins or taste and odor compounds (Jung et al. 2004). In some cases, it may be necessary to invest in expensive water treatment systems that rely on ozone or advanced oxidation treatment technologies (Srinivasan and Sorial 2011). Treatment is further complicated because predicting the occurrence of algal blooms is challenging; many variables, such as nitrogen and phosphorus concentrations, temperature, stratification, and the presence and/or dominance of other competing species, may all be critical factors (Smith 1983, Dzialowski et al. 2011).

Increased likelihood of drought-related impacts

The importance of reservoir management is brought into focus during times of drought. The Colorado River Basin developed a plan for drought and reservoir management due to a prolonged period of drought that began in 2000. The plan was known as the “Interim Guidelines” and ushered forth a new strategy for water conservation and coordinated operation of Lakes Powell and Mead (Rajagopalan et al. 2009). In addition, drought conditions highlighted the need to prepare for future climate change in which diminishing water levels may be the norm (Barnett and Pierce 2008).

Similarly, in a much smaller system, drought conditions during 2012 and 2013 brought the issue of reservoir sustainability into focus for Kansas when water levels in many reservoirs declined to critically low levels (Kansas Water Office 2013). The concern for water management initiated a 50-year water planning effort within the state (Foley et al. 2014). Drought often highlights reservoir vulnerabilities that have been present all along. For example, according to a study by Brikowski (2008), declining reservoir inflow in Kansas has been persistent over time. Several reservoirs in western Kansas have experienced drastic reduction in inflow since construction periods: Cedar Bluff has decreased 88% since 1950, Keith Sebelius has decreased 56% since 1962, Webster has decreased 77% since 1945, and Kirwin has decreased 50% since 1953 (percent change is annual inflow compared from first to last decades of record). It was also estimated that a high percentage of inflow volume was directed toward evaporative demand, ∼68% at Cedar Bluff Reservoir and 83% at Keith Sebelius Reservoir (Brikowski 2008). These statistics reveal that Cedar Bluff and Keith Sebelius are possibly unsustainable reservoirs that could fail during drought conditions.

Planning for drought conditions is extremely challenging due to the limited historical dataset available to study recent drought periods and use as a proxy for future conditions. Using past records to determine the range of variability uses the assumption of stationarity (Milly 2008), the idea that natural systems operate within an envelope of climate variability that can be inferred from historically measured streamflow and weather variations in a designated spatial area. Climate change predictions debunk the assumption of stationarity, and several prominent hydrologists have boldly stated that “stationarity is dead” (Milly 2008, Wagener et al. 2010). Because many of the current water management tools are based on the assumption of stationarity, Milly et al. and Wagener et al. are calling for the creation of new approaches that incorporate the uncertainty of future climate change and the recognition that future patterns may differ greatly from those of the past.

Management approaches for addressing the impacts of climate change on reservoir services

Watershed management

Solutions to water quality problems are often found at the watershed scale. Because watershed-focused management considers climate and landscape factors, it can be used to develop practices that are well-designed to accommodate future climate changes.

Land management

Land treatment and management strategies can vary greatly. Common vegetative treatments, or best management practices (BMPs), for cropland include cover cropping, crop rotations with grasses or legumes, crop residue application, mulching, and planting woody or grass species in critical areas (Vanoni 2006). In general, these vegetative treatments reduce erosion and sediment yield by covering exposed soil surfaces and thus intercepting rainfall and minimizing raindrop impact. Plant material can help increase infiltration rates and reduce runoff. Root mass will also stabilize soil and provide resistance to erosion. Finally, vegetative treatments will increase surface roughness as compared to a bare soil surface and thus reduce overland flow velocity, erosion, and sediment transport capacity (Vanoni 2006). Vegetative BMPs are commonly used in the agricultural sector because they also decrease high nutrient runoff and sustain healthy cropland, but treatments can be especially effective if applied to marginal lands that are uncropped. Other more mechanical BMPs may include contour farming, no-till, or terraces. In addition, grassed waterways, vegetative buffers, sediment traps, and riparian forest buffers can be built to intercept sediment and nutrient-rich runoff from entering waterways (Vanoni 2006, Devlin and Barnes 2008). On grazing land, increasing plant density and adequate cover while also limiting grazing on erosive lands or critical runoff areas can prevent excess sediment loss.

Land management approaches often require personal investment from land owners and are therefore challenging to implement without an economic incentive. BMPs must be targeted to the land that will provide the greatest reduction in sediment and/or nutrient export (Tuppad et al. 2010, Daggupati et al. 2011). For example, a study in Tuttle Creek Lake watershed in Kansas determined through model simulations that targeted BMP application cost-effectively prevented 260,893 tons of sediment transport into Tuttle Creek Lake per year, but that random BMP application was less cost-effective than reservoir dredging (Smith et al. 2013).

Structural measures can also be applied within the watershed to erosive streambanks and reservoir shorelines. Streambank restoration has demonstrated the potential to reduce sediment load to reservoirs and may be necessary to increase stability for high flow events (USEPA 2008). Reservoir shoreline erosion is also a significant issue, caused mainly by surface waves, groundwater seepage, and runoff (Cooke et al. 2005). Installation of breakwaters in strategic locations may stabilize reservoir shorelines and decrease sediment inflow, but it can be costly, ranging from $200 to $855 per linear meter (Pape 2004, Severson et al. 2009). Other treatments such as biotechnical or bioengineering techniques include the use of log breakwaters, certain sediment-stabilizing plants, brush mats, geotextile mats, or using blocks made of plastic, concrete, or fiber (Cooke et al. 2005).

Watershed structures

Upstream debris dams and sediment basins can help trap coarse-grained sediment before it reaches the reservoir. Basins can be periodically dredged of material at a greater convenience and reduced cost compared to large reservoir dredging (Vanoni 2006). Wetlands are also effective at trapping sediment, retaining water during high flow periods and attenuating nutrient loads (Nairn and Mitsch 1999). Strategically placed constructed or restored wetlands in watershed headwaters or near reservoirs could possibly ameliorate the impacts of large precipitation events (Kadlec and Wallace 2008). However, nutrient retention decreases after prolonged periods of high nutrient loading; therefore, without proper maintenance, wetlands may not provide a long-term solution (Verhoeven et al. 2006).

Comprehensive watershed management: A case study of the Kansas WRAPS Program

Kansas has a highly developed watershed-based management program, the Watershed Restoration and Protection Strategy (WRAPS) Program based in the Kansas Department of Health and Environment (KDHE), which involves collaboration among several state agencies. WRAPS involves a planning and management framework based on local stakeholder involvement. Stakeholders are responsible for developing a watershed assessment, establishing goals and identifying necessary actions and costs, preparing a watershed plan, and securing resources needed to execute that plan (KDHE 2012). WRAPS groups are guided by KDHE staff and scientists working through Kansas State Extension Services.

The WRAPS program works primarily to establish BMPs where they are most needed in the watershed. Targeted areas are determined with watershed modeling and then stakeholder groups work to achieve cooperation from necessary landowners. This program seems to be an effective way to implement watershed management on a local scale. Because stakeholders are intimately involved in planning and goal-setting, this program represents a semi-bottom-up approach that can take into account the views and values of watershed residents. Success has already been demonstrated in several WRAPS watersheds where 303d listings (in relation to the Clean Water Act) have been lifted after conditions improved (USEPA 2009b, 2012). For example, DO conditions improved in Toronto Reservoir after repairs to and installation of agricultural ponds, livestock fencing, and watering facility units (USEPA 2012). Land use changes, such as pasture and hay land planting, and critical area planting to reduce runoff, also were utilized in the Toronto Reservoir watershed.

These plans can potentially aid in management of highly variable flow and water quality conditions due to climate change, and although climate change is not addressed in these plans, it is being considered by some of the researchers collaborating with the WRAPS program (Sheshukov et al. 2011). In addition, because watershed models are typically developed during WRAPS plan development, they can potentially be used for further analysis of climate change impacts. Such modeling studies would help advise planners and community leaders to adjust watershed management strategies to accommodate future trends in nutrient and sediment loads. More generally, WRAPS plans have engaged local populations and a variety of stakeholders in watershed management while also increasing awareness of water quality degradation. Involving the community in watershed planning could lead to more adaptive management techniques, which may decrease the vulnerability of the participating watersheds to climate change (Pahl-Wostl 2007).

In-lake management

In-lake sediment management

In many areas, reservoirs are rapidly filling with sediment, and action is needed to preserve these water resource investments (Morris et al. 2006). Reducing incoming sediment and removing excess stored sediment are both options to preserve water storage capacity and increase the usable lifetime of reservoirs (Smith et al. 2013). Some techniques, such as inflow routing and density current venting, require some degree of thermal or density stratification to work correctly (Baker and deNoyelles 2008). Inflow routing and density current routing attempt to route turbid inflow water through the reservoir as a density current, where it is then released downstream, preventing maximum sedimentation in the reservoir (Morris et al. 2006). Sluicing, or flood flushing, moves the sediment load through the reservoir during a high-flow event and requires a low water levels during the flood season to maintain flow velocity. Sluicing takes advantage of the silt-carrying capacity of the floodwaters to flush these particles closer to the dam and then out of the reservoir. Sediment-loaded water is then flushed out of the reservoir as the hydrograph rises, and gates can be closed to trap relatively low-sediment waters on the falling limb of the hydrograph (Durgunoglu and Singh 1993).

Once sediment has settled in the reservoir, it can be removed by hydraulic flushing in which sediment is carried by water through a low-level outlet. Reservoir levels must be low and reservoir inflow must be high for this method to be successful (Palmieri et al. 2001). Any of these techniques that flush sediment through the reservoir could be detrimental to downstream ecosystems, depending on the amount of sediment released (Stanley and Doyle 2003, Crosa et al. 2010). In some cases, managing water levels for flushing or routing may compromise the ability to retain adequate water storage, especially if the reservoir pool is not able to rise and hold seasonal inflow. As mentioned, these methods require some degree of density or thermal stratification, which may be possible in deep reservoirs but possibly not in reservoirs that are adequately wind-mixed (Wetzel 2001).

A more drastic solution for removing sediment includes hydraulic dredging, which employs an underwater cutter head to loosen consolidated sediments and then pumps the sediment-rich slurry through a pipeline to the storage location (Morris et al. 2006). Dredging can be expensive, requires a large area for sediment disposal or storage, and can decrease water quality (Cooke et al. 2005). It can also be challenging to dispose of dredged material without causing further environmental degradation (CDM Federal Programs Corporation 2011).

In-lake nutrient management

Long-term reservoir nutrient management should require reducing nutrient loads into the waterbody through watershed management efforts; however, reducing nutrient loads may not always be sufficient to reduce in-reservoir nutrient concentrations due to internal cycling or long hydraulic residence times (Cooke et al. 2005). In-lake management techniques include selective withdrawal, aeration, and alum treatment (Baker and deNoyelles 2008).

Multilevel selective withdrawal involves discharge of hypolimnetic water, which requires some degree of stratification and therefore may not be a useful technique for all reservoirs. If stratification occurs at the dam, water can be released from a layer that may contain undesired water quality conditions, such as an anoxic hypolimnion or nutrient-rich metalimnion (Nürnberg 2007). Selective withdrawal forces mixing of the water column and has been shown to reduce hypolimnetic anoxia and algal blooms (Lehman 2014). Releasing poor quality waters could cause problems for downstream ecosystems, such as reduced oxygen levels, emissions of hydrogen sulfide or methane, or increased eutrophication potential (Nürnberg 2007).

Hypolimnetic aeration or oxygenation can also be used in cases where the hypolimnion is anoxic. In particular, aeration will oxygenate the water and prevent further release of phosphorus from sediments; in addition it will help keep organic matter suspended in the water column and prevent it from settling into the hypolimnion where bacterial consumption will deplete oxygen (Beutel and Horne 1999). In cases where in-lake phosphorus levels are too high, aluminum salts (alum) can be used as a coagulation and flocculation agent to reduce bioactive phosphorus concentrations in the whole reservoir or in reservoir inflow (Pilgrim and Brezonik 2005, Pilgrim et al. 2007). Many case studies in the literature highlight the success of these methods. For more detail and explanation on implementing techniques, see Cooke et al. (2005).

Climate adaptation in reservoir management

Reservoir sustainability is a critical issue, but long-term sustainability efforts are challenged by the uncertainty of climate change. Past climate data can be used as a proxy for possible future droughts and floods, but this past data may not be sufficient for future planning (Wagener et al. 2010). In an era of uncertainty, reservoir managers will need to use flexible methods to adapt to a changing climate. Adaptive policies and strategies can be informed and developed through simulation modeling. The most common approach is to combine a series of climate, hydrologic, and reservoir and/or ecological models.

First, specific future scenarios are selected to represent possible future climate pathways (Soloman 2007). Then the results of these scenarios in Global Climate Models (GCMs) are downscaled through the use of regional climate models or stochastic weather generators to generate more location-specific climate parameters. Next, the generated climate parameters are used in a calibrated hydrologic model that can generate streamflow and nutrient inputs into the reservoir system; examples of such models include the Soil and Water Assessment Tool (SWAT; Gassman et al. 2007, Douglas-Mankin et al. 2010), the Hydrologic Simulation Program Fortran (HSPF; Donigian et al. 1995), and the Agricultural Non-Point Source Pollution Model (AGNPS; Young et al. 1989). Many other hydrologic models are available that vary slightly in terms of approach, potential applications, and limitations. The choice of model may be dictated by data availability or the accessibility of a predeveloped model for the region of interest. Streamflow results from the utilized hydrologic model can be used as an input into a reservoir optimization model to examine possible management strategies, or into an ecological or water quality model to predict algal biomass, nutrient concentrations, oxygen demand, and a variety of other parameters of concern. Such an approach may be time and resource intensive, depending on the availability of data and calibrated models, but can be flexibly adapted to many regions and systems using a combination of mathematical and statistical models. For example, such studies have been conducted for the Hirakud reservoir in India (Raje and Mujumdar 2010), the Chungju dam in South Korea (Park and Kim 2014), and the Northern California water and power system (Georgakakos et al. 2012). A variety of mathematical and statistical tools have been used to study the impacts of climate change on reservoirs, many with application for specific systems (Table 2).

Table 2 Mathematical and statistical tools used to study impacts of climate change on reservoirs.

Tool	Description	Developer	Example of Application
Integrated Adaptive Optimization Model (IAOM)	Contains 3 modules: weather generator, hydrological simulator, and multipurpose reservoir optimization to develop optimal operating rule curves under climate change	Y. Zhou and S. Guo	Zhou and Guo 2013
Hydrologic Engineering Center – Reservoir System Simulation (HEC-ResSim)	Uses rule-based approach to mimic decision making process	Army Corp of Engineers	Park and Kim 2014
Dynamic Hydroclimatological Assessment Model (DYHAM)	Utilizes system dynamics theories and feedback causal loops to simulate dynamic processes within watershed and reservoir	SP Simonovic and LH Li	Li et al. 2010
Phytoplankton Responses to Environmental Change: PROTECH	Simulates the daily change in Chl-a concentration for up to 10 algal species in response to environmental variability in lakes and reservoirs	Alex Elliott, Colin Reynolds, Tony Irish	Alex Elliott et al. 2005
NAM Rainfall Run-off Model	Deterministic, nondistributed hydrological model	DHI Inc.	Jeppesen et al. 2009
Soil and Water Assessment Tool (SWAT)	Simulates water quality and quantity of surface water and can test scenarios related to land use, land management practices, and climate change	USDA-ARS and Texas A&M AgriLife Research	White et al. 2010a, White et al. 2010b
Hydrologic Simulation Program Fortran (HSPF)	Simulates hydrologic and water quality processes on land, in streams, and in well-mixed impoundments.	USGS	Göncü and Albek 2010
Agricultural Non-Point Source Pollution Model (AGNPS)	Evaluates effect of management decisions that may impact water, sediment, and chemical loadings within a watershed	USDA	Booty et al. 2005
Generalized Watershed Loading Function (GWLF)	Dynamically simulates variations in stream discharge and combines with sources of P to estimate P export	K.H. Reckhow	Pierson et al. 2010
BASINS-CAT: Better Assessment Science Integrating Point and Non-Point Sources Climate Assessment Tool	Integrates GIS, national water data, and watershed modeling tools (HSPF) with the added flexibility to incorporate climate change scenarios	USEPA	Taner et al. 2011
Historical Analog Extended Streamflow Prediction (Analog ESP)	Develops relationship between streamflow and hydro-climate factors, which can then be extended to future climate	The National Weather Service	Yao and Georgakakos 2001
Statistical P-loss Model	Develops relationship between P loss from watershed and critical hydrologic and watershed variables and then can use projected flow to estimate P loss under climate change	E. Jeppesen et al.	Jeppesen et al. 2009
2D Reservoir Water Quality Mathematical Model	Links mathematical equations describing flow and transport, ecological interactions, and water–sediment interchange	E. Komatsu, T. Fukushima, H. Shiraishi	Komatsu et al. 2007

Table 3 Examples of commonly used data sources available to implement tools and approaches outlined in Table 2 and to study climate impacts on reservoirs.

Data Network	Description	Data Source
Geospatial Data Gateway	Census, average climate, easements, elevation, geology, government units, hydrography, land use and land cover, soils, topography, and transportation data layers available for the U.S.	U.S. Department of Agriculture (USDA) Natural Resources Conservation Service
Hydroclimatic Data Network (HCDN)	Subset of streamflow stations that can be used to study climate fluctuations in the U.S.	U.S. Geological Survey (USGS)
U.S. Cooperative Observer Network (COOP)	More than 8000 climate stations operating from as early as 1886, providing daily min and max temperature and precipitation data	National Climatic Data Center (NCDC)
Global Historical Climate Network	Daily climate observations from around the world, including min and max temperatures, total precipitation, snowfall, and depth of snow	NCDC
U.S. Climate Reference Network (USCRN)	114 climate stations in the conterminous U.S., as well as 29 stations in Alaska and 2 in Hawaii to detect the national signal of climate change	National Oceanic and Atmospheric Administration (NOAA)
National Weather Service River Forecast Centers	High resolution gridded hydrologic state variables and flux datasets; both observed and forecasted river conditions and precipitation are available	NOAA
Next Generation Radar (NEXRAD) Level III	40+ data products derived from NEXRAD Level II data including precipitation estimates and storm relative velocity	NOAA

Within this framework, changes in system ecology can be studied using both process-based and empirical approaches. A study by Jeppensen et al. (2009) examined the impacts of climate change for lakes in Denmark using a 4-step process utilizing both process-based and empirical models. First, a climate change scenario was downscaled with a Danish regional climate model, which then fed into the NAM rainfall-runoff model to predict streamflow. Next, a previously developed statistical model was used to estimate diffuse total phosphorus (TP) losses to streams, and the Vollenweider lake model was then used to estimate the in-lake TP concentration with climate change (Jeppesen et al. 2009). In this way, Jeppesen et al. estimated that although TP loads to lakes may increase, overall lake TP concentrations will decrease due to higher flushing rates. This study shows the potential of utilizing previously developed models to analyze the problem of climate change and the advantage of combining both hydrology and lake models to understand impacts to a system.

Alex Elliott et al. (2005) employed a different approach by using results from a downscaled climate model to estimate impacts to algal concentrations using an ecological model, PROTECH, that estimates concentration changes in specific phytoplankton species. The study found that overall Chl-a concentrations were higher in spring, and the summer algal peak occurred earlier in the season with climate change (Alex Elliott et al. 2005). However, this study did not incorporate the effects of any changes in hydrology that may occur due to climate change, which can be critical, as the Jeppesen et al. (2005) study demonstrated. Nevertheless, the PROTECH model combined with climate change predictions may provide a useful starting point for managers to anticipate changes in the timing of peak algal concentrations and species dynamics with climate change.

Other approaches (Table 2) focus on improving decision making in reservoir management by simulating operation rules that incorporate an ecological requirement (Zhou and Guo 2013), or that account for the demands of different stakeholders under varying conditions (Li et al. 2010). Zhou and Guo (2013) used the Integrated Adaptive Optimization Model (IAOM) in their study, which requires on an optimization algorithm (Zhou and Guo 2013). Li et al. (2010) used system dynamics theories and the Dynamic Hydroclimatological Assessment Model (DYHAM) to study the influence of climate on reservoir reliability. The limitations of these approaches are that they require complex mathematical models and a large amount of data, which may not be available for all systems; however, similar studies may provide necessary evidence for amending rule curves or operation rules to develop more sustainable reservoir systems. Information on amending rule curves for U.S. Army Corps of Engineers reservoirs can be found in a study by Mower and Miranda (2013).

The hydrologic tools (Table 2) are currently used to conduct hydrologic, nutrient, or climate change impact analyses for lakes and reservoir systems, such as the SWAT model (White et al. 2010b, Wu et al. 2011) and HSPF (Göncü and Albek 2010). Both of these models require a large time investment to gather and process input data, model development, and perform calibration and are most often used to simulate watershed hydrology and to generate estimated inputs of water, nutrients, and sediment into waterbodies (Douglas-Mankin et al. 2010). The data sources necessary to develop these models vary (Table 3). The user can select among several types of weather data, but precipitation data should be evaluated carefully before selecting a source because stations vary greatly in continuity and may have significant gaps or be missing critical events (based on user experience).

The U.S. Environmental Protection Agency has developed an additional tool called BASINS-CAT, which incorporates available data with the HSPF model and climate change scenarios. BASINS-CAT may be an ideal tool for managers because it may save time and streamline modeling, reducing the need to integrate several modeling platforms. BASINS-CAT is able to adjust historical climate data with arithmetic operators on several time scales to examine long-term or short-term climate changes. Users can also examine events over or under a certain threshold and can increase or decrease the frequency of precipitation events to examine the outcomes (USEPA 2009a). This tool is ideal for examining a variety of climate simulations to improve management decisions. A lake model would need to be integrated to apply the BASINS-CAT results to a lake or reservoir system (Taner et al. 2011).

Conclusion

Reservoirs provide critical services and represent large fiscal investments from previous generations. With often limited locations to develop new reservoirs, it is imperative that current reservoirs are maintained and managed sustainably. Evidence cited in this review shows that climate change may exacerbate reservoir sedimentation, increase nutrient loads, increase the frequency of algal blooms, and challenge the management of adequate water supplies. This literature review presented watershed and in-reservoir management techniques that have been successful for managing current reservoir systems. Managers may also want to prepare for the future through the use of a combined climate–watershed–reservoir modeling framework. Modeling can provide an estimate of the range and probabilities of impacts to local systems, which can be useful for a risk assessment framework and for reservoir planning and management efforts. This review highlights some of the tools that have been used to study climate impacts on reservoir systems, but is not all-inclusive. Managers should carefully review available tools before selecting one and should consult with other resource managers to determine if a model has already been developed for their area of interest. In some cases the data requirements and necessary technical expertise can be limiting. Collaborations between reservoir managers and climate scientists may be necessary to develop simulation modeling platforms that can explore and virtually test adaptive management strategies in the context of altered climate patterns.

Acknowledgments

We would like to acknowledge the insight gained through many valuable conversations with Val Smith, Frank (Jerry) deNoyelles, Mark Jakubauskas, Jennifer Graham, Edward Martinko, and all the staff at the Kansas Water Office. We thank the several anonymous reviewers for their comments, which helped greatly to improve this manuscript.

Funding

Lindsey MW Yasarer was supported by the NSF IGERT C-CHANGE Fellowship (NSF #0801522).