A REVIEW OF WATER QUALITY RESPONSES TO AIR TEMPERATURE AND PRECIPITATION CHANGES 1: FLOW, WATER TEMPERATURE, SALTWATER INTRUSION

Abstract

Anticipated future increases in air temperature and regionally variable changes in precipitation will have direct and cascading effects on U.S. water quality. In this paper, and a companion paper by Coffey et al. (2019), we review technical literature addressing the responses of different water quality attributes to historical and potential future changes in air temperature and precipitation. The goal is to document how different attributes of water quality are sensitive to these drivers, to characterize future risk to inform management responses and to identify research needs to fill gaps in our understanding. Here we focus on potential changes in streamflow, water temperature, and salt water intrusion (SWI). Projected changes in the volume and timing of streamflow vary regionally, with general increases in northern and eastern regions of the U.S., and decreases in the southern Plains, interior Southwest and parts of the Southeast. Water temperatures have increased throughout the U.S. and are expected to continue to increase in the future, with the greatest changes in locations where high summer air temperatures occur together with low streamflow volumes. In coastal areas, especially the mid-Atlantic and Gulf coasts, SWI to rivers and aquifers could be exacerbated by sea level rise, storm surges, and altered freshwater runoff. Management responses for reducing risks to water quality should consider strategies and practices robust to a range of potential future conditions.

INTRODUCTION

Anticipated warming air temperatures, changing precipitation patterns and rising sea levels are expected to alter watershed hydrologic and biogeochemical processes, with direct and cascading effects on water quality (Fant et al., 2017; Murdoch et al., 2000; Worrall et al., 2003; Senhorst and Zwolsman, 2005; Kundzewicz et al., 2007). U.S. air temperatures increased, on average, 0.7 to 1°C from 1986 to 2016. By the end of this century, air temperatures are projected to increase an additional 1.6°C to 6.6°C (USGCRP 2017). Increases are projected to be greatest for higher latitudes and inland, and smaller increases are projected along coastal areas (USGCRP 2017). Warming air temperatures have been linked to intensification of the hydrological cycle (e.g., atmospheric water content and changing precipitation patterns) and altered biogeochemical processes – key drivers of water quality (Watts et al., 2015; USGCRP 2017). Increased air temperatures are also linked to sea level rise (USGCRP 2017), which together with storm surge (movement of seawater landwards due to low pressure and wind associated with storms) can exacerbate salt water intrusion (SWI, movement of saline water into freshwater aquifers) and salinization of coastal rivers, estuaries, and wetlands.

Changes in the amount and timing of precipitation drive watershed hydrologic processes that mobilize and transport pollutants to water bodies. Annual precipitation in the continental U.S. has increased, on average, 4% since 1901. Increases in annual precipitation have been more prevalent in the northern and eastern U.S., while decreases have been observed in parts of the West, Southwest and parts of the Southeast (USGCRP 2017). Heavy precipitation events have increased in intensity and frequency in most locations. In the western U.S, extreme snowfall years, spring snow cover extent, maximum snow depth, and snow water equivalent have decreased; however, extreme snowfall years in parts of the northern U.S. have increased (USGCRP 2017). These trends are expected to continue this century (USGCRP 2017).

In this paper, and a companion paper in this volume (Coffey et al., 2019), we review the scientific literature addressing water quality responses to air temperature and precipitation in the U.S. The objectives are to document how different attributes of water quality are sensitive to climate change, to characterize future risk to inform management responses and to identify research needs to fill gaps in our understanding. This information is useful to identify key vulnerabilities in different regional and watershed settings, and to guide the development of effective management responses to reduce the risk of harmful impacts. This paper focuses on the potential future responses of streamflow, water temperature, and SWI to coastal rivers and aquifers. The companion paper by Coffey et al. (2019) discusses nutrients, algal blooms, sediment, and microbial pathogens.

Water quality changes are complex, with high spatial and temporal variability within and among waterbodies, and influenced by multiple, interacting climatic, watershed (e.g., physiographic setting, land use) and human (e.g., water management infrastructure) factors (Poole and Berman 2001; Caissie 2006; Webb et al. 2008). The focus of this review is sensitivity to changes in air temperature and precipitation. Discussion of changes in land use, water management and other factors affecting water quality is outside the scope of this review.

APPROACH

Relevant studies were identified through a search of recently published peer-reviewed and gray literature (post 2000) examining water quality responses to historic or projected future changes in air temperature and precipitation. Literature searches were conducted in common scientific databases (e.g., PubMed) using appropriate search terms. Our search identified studies that evaluated either observed streamflow, water temperature and SWI/salinity responses to climate drivers, or simulations of potential future responses using hydrological models driven by future climate change scenarios. Studies about responses to observed weather and climatic variability document sensitivity to air temperature and precipitation. In this context, climatic variability refers to the inherent heterogeneity or fluctuation in air temperature, precipitation, etc. over the short-term, days to years. Modeling studies suggest potential changes in response to alternative futures conditions. To the extent possible, we synthesize information from each type of study to make inferences about the risk presented by future changes. Where possible, regional differences are noted. Note that modeling studies typically provide an “ensemble” range of outcomes in response to different future scenarios and time periods etc. In such cases, we make the simplifying assumption that the ensemble mean or median suggests a “more likely” direction of change for the purposes of risk management.

RESULTS

Streamflow

Observed Sensitivity.

Streamflow is a principal driver of changes in water quality (Murdoch et al. 2000; Howarth et al. 2006; Kaushal et al. 2008; Kundzewicz et al. 2008; Ficklin et al. 2010; Wilson and Weng 2011; Joyner and Rohli 2013; Jiang et al. 2014). Changes in streamflow are highly correlated with changes in precipitation (USGS 2005; Zhu and Day 2005; Krakauer and Fung 2008; Dai et al. 2009; WICCI 2009, Patterson et al. 2012; Bassiouni and Oki 2013; Ge et al. 2013; Simpson et al. 2013; Ryberg et al. 2014; Berton et al. 2016). Changes in air temperature also affect streamflow via drivers like evapotranspiration (ET), soil moisture, snow accumulation and snowmelt (Zhu and Day 2005; Fu et al. 2007; Krakauer and Fung 2008).

Streamflow in the U.S. has broadly increased over the past century, but the increase has not been uniform and can be characterized by both short- and long-term variations in precipitation (USGS 2005). Increases in streamflow, particularly for low flow (i.e., baseflow) and average annual flow conditions, have been most prevalent in the Northeast, Southeast, Midwest and Alaska (e.g., Huntington et al. 2009, Marion et al. 2014). Other regions show no consistent trend in average annual streamflow (USGS 2005), however changes in seasonal flows have occurred including, for example, increases in spring and decreases in summer and fall flows in the Central Rockies (e.g., Clark 2010; Leppi et al. 2012) and mountainous Southwest, and Pacific Northwest (USEPA 2017). Observed changes are also likely to have been influenced by human activities (e.g., urban and agricultural land use, dams and other water management infrastructure), which interact with precipitation (Paul and Meyer 2001; Vogel et al. 2011; Jones et al. 2012; Creed et al. 2014; Ford et al. 2011). This can make attribution of observed streamflow trends to specific drivers difficult to distinguish (Ficklin et al. 2018; Fu et al. 2007; Vogel et al. 2011; Jones et al. 2012; Hatcher and Jones 2013).

Decreases in low-flow (baseflow) conditions, which typically occur in late summer or early fall, have been associated with increases in ET associated with warming (Regonda et al. 2005; Stewart et al. 2005; Brabets and Walvoord 2009; Clark 2010; Coats 2010; Hatcher and Jones 2013). Over the past century, changes in low-flow volumes have generally correlated with regional trends in precipitation (USGS 2005; Bassiouni and Oki 2013). Trends in peak streamflow (floods) from 1940 to 2014 show no consistent pattern on a national scale (Villarini et al. 2009, USEPA 2016a), but regional patterns are apparent. In the southern U.S., especially the Southwest, decreases in peak streamflow have been observed, potentially attributable to warmer and drier conditions (Hirsch and Ryberg 2012). In parts of the Northeast and Midwest, the frequency and magnitude of floods have increased in response to increases in precipitation, especially heavy precipitation (USGS 2005; Kalra et al. 2008; Huntington et al. 2009; Kim et al. 2010a; Melillo et al. 2014; Ryberg et al. 2014, USEPA 2016a; Usinowicz et al. 2017).

In snow-dominated watersheds, air temperature affects the annual timing (seasonal distribution) of streamflow through changes in the proportion of precipitation that falls as rain versus snow, snowpack accumulation, the rate and seasonal timing of snow melt (Knowles and Cayan 2002; Coats 2010; Sahoo et al. 2011; Hatcher and Jones 2013). Western mountain regions have experienced shifts in the seasonal timing and distribution of streamflow to earlier in the year, resulting in increased spring discharge volumes and decreased discharge in summer-early fall (Lins and Slack 1999; Maurer 2007; Neiman et al. 2008; Clark 2010; Kim and Jain 2010; Mayer and Naman 2011; Hunsaker et al. 2012; Leppi et al. 2012; Dittmer 2013; Ficklin et al. 2013a; Hatcher and Jones 2013; Null et al. 2013; Melillo et al. 2014; USEPA 2017; CalEPA 2018). Similar trends have also been observed in the Northeast (Huntington et al. 2009; Hamburg et al. 2013). In contrast to snow dominated watersheds, many glaciated basins have experienced increased summer flows due to increased glacial melting as a direct effect of increased temperatures (e.g., Alaska) (Brabets and Walvoord 2009; Hodgkins 2009; Ge et al. 2013).

Decadal oscillations [e.g., El Niño Southern Oscillation (ENSO) and Atlantic Multidecadal Oscillation (AMO)] can also influence streamflow. The AMO, which shifted abruptly in the 1970s, was linked to changes in precipitation and temperature over the Midwest and northeastern U.S., causing changes in streamflow (McCabe and Wolock 2002; Krakauer and Fung 2008; Villarini et al. 2009; Zhang et al. 2010). In the western U.S. and Alaska, the North Atlantic Oscillation and Pacific Decadal Oscillation (PDO) have also been linked to sudden changes in weather which altered streamflow (Woo and Thorne 2008; Coats 2010; Coleman and Budikova 2013). In the southeastern US, complex patterns linking different decadal oscillations to seasonal differences in flow have been described, emphasizing the important link between climate and streamflow (Sheldon and Burd 2014).

Projected future changes.

Compared to water quality, a relatively extensive literature exists addressing potential future streamflow changes in the U.S. Studies include local watershed scale monitoring and modeling, to gridded, continental scale water budget simulations using global climate and land surface models (e.g., USGCRP 2017, Marion et al. 2014, Sun et al. 2015). In this review, we limit our scope to local, watershed scale studies that illustrate the breadth of expected streamflow changes at spatial and temporal scales relevant to water quality. The aim is not a comprehensive review of all literature addressing streamflow response, but rather to provide appropriate hydrologic context for discussions about water quality responses in this and the companion paper by Coffey et al. (2019). See Georgakakos et al. (2014) or USGCRP (2017) for more extensive, national scale review of projected streamflow changes.

Decreases in average annual streamflow are suggested for many watersheds in the interior Southwest, central Rockies, and parts of the Southeast mainly in response to potential decreases in summer-fall precipitation and increases in ET (Figure 1). Potential increases in average annual streamflow are more prevalent for watersheds in the Northeast, Midwest, Pacific Northwest, Northern Plains, and Alaska. There is, however, variability amongst watersheds in all regions. Future streamflow responses will reflect the balance between drying associated with warming air temperature and ET, and regionally variable changes in the amount and direction of future changes in precipitation. Local changes will also be influenced by other factors such as watershed setting (e.g., geology, topography, soil type, and vegetation) and human activities (e.g., land use and water management infrastructure).

Figure 1.

Location of watershed modeling studies in this review assessing streamflow responses to future climate change scenarios. Symbols indicate the suggested direction of change (based on ensemble median, annual loads). Studies not reporting an ensemble median (e.g., a range, or sensitivity) are shown as “Direction indeterminate”. Detailed information about each study are provided in the supplemental materials.

Increases in air temperature are expected to continue to alter the timing and magnitude of seasonal high flows in colder/mountainous areas influenced by snow (e.g., the northern, northeastern, mountain west and pacific northwest areas). Rain-on-snow events are expected to become more frequent, increasing the risk of flooding (Musselman et al. 2018). Some locations could shift from currently snow-dominated systems to either transient systems, which receive a mixture of snow and rain, or rain-dominated systems (Christensen and Lettenmaier 2007; Elsner et al. 2010; Chang et al. 2010). In transient systems, which are near the rain-snow threshold, slight temperature changes can have large effects on the form of precipitation and snow accumulation (Mote and Salathe 2010; Chang et al. 2010). Future changes in these system types will be important for regions and watersheds that depend on snowfall and melt to sustain late summer-fall streamflow as a water supply. Generally, northern and eastern parts of the U.S. are anticipated to experience lower summer flow volumes due to increased winter rainfall and earlier snowmelt (Birsan et al. 2005; Jefferson et al. 2008; Ficklin et al. 2013c, USEPA 2016a).

The risk of water quality degradation can be greater during extreme high and low flow events (Kundzewicz et al., 2008; Watts et al., 2015; Pathak et al., 2016; Fant et al., 2017). Climate change is expected to generally increase flow variability, including a greater proportion of annual precipitation occurring in heavy events, and longer dry periods between events (USGCRP 2017; Naz et al. 2018; Salathé and Mauger 2018). Uncertainty remains, however, concerning specific future changes in distinct locations. In many locations, longer summer dry periods together with increases in air temperature and ET could exacerbate low flow conditions (e.g., Wei et al. 2012, Caldwell et al. 2016), concentrating pollutant inputs. At the same time, warmer temperatures are expected to drive more frequent heavy precipitation events that increase the risk of high-flow and flooding throughout the U.S. (Prein et al. 2016; USGCRP 2017; Wing et al. 2018). Excessive in-stream pollutant loading is commonly associated with such events.

Ecosystem Level Impacts.

Changes in streamflow can have direct and indirect effects on water quality and aquatic ecosystems (Poff et al. 1997). Aquatic organisms are adapted to and depend on natural variability in streamflow; as a result, changes in the natural streamflow regime can alter physical habitat, water quality, food and energy inputs, and aquatic community interactions such as predator-prey dynamics, reproduction, and dispersal (Poff et al. 1997; Lytle and Poff 2004). Figure 2 summarizes how changes in streamflow can affect aquatic communities. In estuaries and coastal systems, freshwater discharge is similarly a major factor determining habitat availability and quality, with shifts in freshwater inflows having a major impact on estuarine species (Sheldon and Burd 2013; Wieşki and Pennings 2013; Beighley et al. 2008).

Figure 2.

A summary of flow alteration effects on aquatic communities. Source: Novak et al. 2016.

Pollutant fluxes within waterbodies typically correlate with changes in streamflow (Murdoch et al. 2000; Howarth et al. 2006; Kaushal et al. 2008; Kundzewicz et al. 2008; Ficklin et al. 2010; Wilson and Weng 2011; Joyner and Rohli 2013; Jiang et al. 2014). In-stream pollutant loads generally increase with streamflow and associated precipitation events which drive non-point source loading. During periods of low flow volume, reduced dilution can in some instances result in higher pollutant concentrations (e.g., downstream of point source discharges). Increased flow residence times together with reduced mixing can also contribute to the formation of algal blooms. These relationships are developed in greater detail in the companion paper by Coffey et al. (2019). Finally, changes in flow volume also affect water body thermal capacity and are a contributing factor to increasing or decreasing water temperatures (Mantua et al. 2010; Wu et al. 2012; Null et al. 2013; Ficklin et al. 2013b; Luo et al. 2013, Butcher et al. 2016).

Water Temperature

Observed Sensitivity.

Solar radiation is the primary driver affecting air and water temperature (Sinokrot and Stefan 1993, Webb and Zhang 1997). Because solar radiation affects both air and water temperature, and sensible heat is transferred between air and water, the two parameters are typically closely correlated. Accordingly, many statistical models use changes in air temperature to predict water temperatures (Caissie 2006). Increases in annual and seasonal water temperatures have been broadly observed throughout the U.S. over the past century, coincident with increases in air temperature over the same time (Figure 3) (Johnson and Stefan 2006; Kaushal et al. 2010; Seekell and Pace 2011, Isaak et al. 2012; USGCRP 2017; Winslow et al. 2017). While many U.S. waterbodies have exhibited warming trends, the magnitude of changes often vary depending on site characteristics, human activities, the thermal metric being evaluated and the time period being analyzed (Arismendi et al. 2012).

Figure 3.

Location of watersheds identified in this review assessing observed historic trends in annual average water temperature. Symbols indicate direction of trends. Direction indeterminate sites did not exhibit a significant trend over time (p>0.05).

In a long-term national scale assessment, increases in annual average water temperature were reported at 33 out of 40 sites scattered across the U.S. (Kaushal et al. 2010). Statistically significant increases were evident at 20 of these sites, with water temperatures rising by +0.009 to +0.077 °C per year. The largest rate of increase was observed at a Delaware River site near Chester, Pennsylvania (Kaushal et al. 2010). In the Northeast, Midwest and Great Lakes, increases in water temperature, decreases in ice cover and earlier ice melt have been observed, corresponding with increases in regional air temperature over the past 30 years (Assel 2005; Austin and Colman 2007; Dobiesz and Lester 2009; Huntington et al. 2009; Magee and Wu 2017; Winslow et al. 2017). Reduced winter ice cover is also causing an earlier onset of lake stratification in the Great Lakes (Austin and Colman 2007) and stronger stratification in Wisconsin (Winslow et al. 2017). Summer water temperatures have increased at a faster rate than regional air temperatures (Dobiesz and Lester 2009). Similar observations have been noted for lakes in other parts of the U.S. (e.g., Lake Tahoe, California – see Coats 2010).

Precipitation driven changes in hydrology also have major effects on water temperature. The effects of precipitation include the direct transfer of thermal energy in runoff (related to air temperature) and changes in hydrology that influence the thermal capacity and energy balance of waterbodies (e.g., streamflow volume, groundwater contribution) (Webb and Zhang, 1997; Mohseni and Stefan 1999). During dry seasons, low streamflow volumes are associated with reduced thermal capacity and longer residence time, which contribute to diurnal increases in water temperature (van Vliet et al. 2011, Butcher et al. 2016). In the northern and mountainous western U.S., reduced snowpack and earlier snow melt associated with warming have shifted the timing of runoff to earlier in the year, resulting in extended summer-fall low flows. These conditions have been linked to high summer water temperatures in many streams (Stewart et al. 2005; Fritze et al. 2011, Isaak et al. 2012).

Conversely, increases in precipitation and runoff can have a buffering effect on water temperature by diluting thermal loads, increasing thermal capacity, and shortening residence time (Webb et al. 2003). In some snow-dominated systems, decreases in spring water temperatures have occurred because of increased cool water inputs and flow volume associated with earlier snowmelt (Isaak et al. 2012). Relatively cool groundwater inputs can also make stream temperatures less sensitive to changes in air temperatures and streamflow, particularly in smaller, shaded streams (Tague et al. 2007; Nichols et al. 2014; Brown 1969; Smith and Lavis 1975; Constantz and Essaid 2007; Sridhar et al. 2004).

Decadal oscillations have been shown to affect water temperature as well. For example, in the western U.S., Bartholow (2005) suggested that warming of the lower main-stem Klamath River might be related to the cyclic PDO, while Coats et al. (2006) found that warming trends in monthly and annual water temperatures in Lake Tahoe were correlated to the PDO, and to a lesser extent, ENSO.

Projected future changes.

Anticipated future changes in water temperature are better studied than other aspects of water quality. Most studies reviewed suggest future increases in annual and seasonal water temperature in response to warmer air temperatures (Figure 4) (van Vliet et al. 2013; Hill et al. 2014; Mantua et al. 2010; Wu et al. 2012; Ficklin et al. 2013b; Caldwell et al. 2015; Fant et al., 2017). Increases are largely attributed to warming air temperatures, and, in some locations, lower streamflow volumes which reduce thermal capacity. Some studies suggest up to 26 percent of the increases in high water temperature (95th percentile) can be indirectly attributed to low flow changes (van Vliet et al. 2013).

Figure 4.

Location of watersheds identified in this review assessing water temperature responses (relative rate of increase in ºC per annum) to future climate change scenarios. Symbols indicate the suggested direction of change (based on ensemble median, annual change). Symbol size reflects relative rates in 3 categories; lower (<0.024 ºC per annum), middle (0.024 to 0.038 ºC per annum), and upper (>0.038 ºC per annum). Detailed information about each study are provided in the supplemental materials.

A national assessment of 569 reference-condition sites by the U.S. Geological Survey suggested average warming of 2.2 °C (ranges were from 0 to +6.2 °C) for summer stream temperature by 2090 (Hill et al. 2014). More than half of the sites (52 percent) are projected to warm by greater than or equal to +2 °C (assuming a high greenhouse gas emission scenario, SRES A2). Water temperatures in mountainous regions especially in the Northwest (Cascades) and Northeast (Appalachian Mountains) are projected to be most responsive, while sites in the Southeast may be least responsive (Hill et al. 2014).

Warmer lake temperatures are expected to initiate an earlier onset and extended duration of stratification. In the Great Lakes, for example, increases in maximum summer lake temperatures and the duration of summer stratification have been projected (Trumpickas et al. 2009). Simulated changes in late century summer plateau temperatures (i.e., the 20^th highest daily water temperatures observed in a year) range from +2.4 ^oC to +3.3 ^oC; in Lake Ontario range from +3.2 ^oC to +4.8 ^oC; in Lake Huron range from +2.6 ^oC to +3.9 ^oC; and in Lake Superior, from +4.6 ^oC to +6.7 ^oC (Trumpickas et al. 2009).

Ecosystem Level Impacts.

Future changes in water temperature could have wide ranging effects on water chemistry, aquatic life and suitability for human use (Hynes 1970; Vannote and Sweeney 1980; Brown et al. 2004). At the ecosystem-level, water temperature influences processes such as primary production, metabolism, and decomposition (Brown et al. 2004; Bott et al. 2006). In some cases, warmer temperatures may favor nuisance taxa, such as toxin-producing cyanobacteria. Cyanobacteria gain a competitive advantage over other phytoplankton groups in warmer temperatures through a variety of adaptations (Jöhnk et al. 2008; Paerl and Huisman 2008; Paerl and Paul 2012). Warmer surface waters strengthen vertical water column stratification which, in combination with nutrient enrichment and other factors, can promote the development cyanobacterial blooms (King et al. 2007; Funari et al. 2012; Paerl and Paul 2012). Increasing water temperatures also decrease dissolved oxygen (DO) concentrations by reducing oxygen solubility and increasing respiration (Ficke et al. 2007). Waterborne pathogen survival rates (Coffey et al. 2014) and the toxicity of many environmental contaminants, such as ammonia, are some of the other water quality attributes affected by water temperature (Rehwoldt et al. 1972; Langford 1990; Emerson et al. 1975). Interactions with nutrients, cyanobacterial blooms and waterborne pathogens are described in greater detail in the companion paper by Coffey et al. (2019).

At the organism-level, temperature affects growth, metabolism, reproduction and behavior (Hynes 1970; Magnuson et al. 1979; Vannote and Sweeney 1980). Most species have a specific range of temperatures they can tolerate, and changes in temperature can result in the loss of suitable habitat for those species (Beechie et al. 2012). Increased water temperatures are expected to reduce suitable habitat for cold- and cool-water fish taxa and increase it for warm-water species (Missaghi et al. 2017; Mohseni et al. 2003; Fang et al. 2004a, 2004b, 2004c).

Salt Water Intrusion and Salinity

Observed Sensitivity.

Sea level rise (SLR) can contribute to SWI and salinization of coastal water bodies. Many U.S. coastal zones have experienced increased SWI (e.g., Long Island, New York; Cape May County, New Jersey; Brunswick, Georgia; Northwest Florida; Fernandina Beach, Florida; Miami-Dade County, Florida; areas along the Louisiana and Texas coasts; Los Angeles, California; and the Salinas and Pajaro Valleys, California), or are considered vulnerable to saltwater intrusion from SLR (Vitousek and Howarth 1991; Childers et al. 2011; Maloney and Preston 2014). Absolute sea level refers to changes in the volume of the ocean due to melting ice or warming seas, while relative SLR (RSLR) (sometimes referred to as local SLR) incorporates changes in local or regional sea level due to vertical land movement [e.g., due to glacial isostatic adjustment (GIA), land subsidence, or tectonic activity] and other local factors (Figure 5). At a global scale, absolute SLR is averaging +3.3 mm yr⁻¹, and observed SLR in the U.S. is generally consistent with this rate.

Figure 5.

Processes influencing sea level rise. Ocean properties refer to ocean temperature, salinity, and density. Source: Church et al. 2013.

Vertical land movement can moderate or amplify SLR and in turn, saltwater intrusion, at the local level. Much of the mid-Atlantic and central Gulf coasts of the U.S., for example, have already experienced rates of RSLR greater than SLR due to a combination of GIA (mid-Atlantic) and regional subsidence, while parts of the Pacific coast and Alaska have observed negative RSLR due, in part, to tectonic uplift (Penland and Ramsey 1990; Morton et al. 2006; Engelhart et al. 2009; Kolker et al. 2011; Ezer and Corlett 2012; Parris et al. 2012; Sallenger et al. 2012). High tides and increasing storm surge also increase the movement, albeit episodic, of saline water landward, exacerbating SLR related salinization of surface and ground waters. Tidal amplitudes in Key West, for example, have increased 57% over pre-1993 levels (Wahl et al. 2014), and storm surges have increased (Melillo et al. 2014).

Local groundwater extraction strongly affects rates of SWI and interacts with the effects of SLR (Ferguson and Gleeson 2012, Uddameri et al. 2014, Sawyer et al. 2016). Some have argued that coastal aquifers are made more susceptible to SWI from groundwater extraction and that SLR is more likely to lead to saltwater inundation (surface movement of saline water) than intrusion (subsurface movement of saline water) (Ferguson and Gleeson 2012).

Projected future changes.

Increasing sea levels will amplify the risk of saltwater inundation of coastal areas and intrusion to rivers and coastal aquifers. Projected future SLR estimates (from global mean SLR projections) generally follow historic trends and increases are expected along most U.S. coastlines (Figure 6). The U.S. Sea Level Rise and Coastal Flood Hazard Scenarios and Tools Interagency Task Force recently projected late century global mean SLR of +0.3m to +2.5m. Regional factors (e.g., vertical land movement, changes in ocean circulation, gravitational changes, etc.) will add to these estimates in some coastal locations. For example, a late century global mean SLR of +1.5 m is projected to result in RSLR of +1.9 to +2.2 m along the East Coast, +1.7 to +2.5 m along the Gulf Coast, and +1.7 to +1.8 m along the West Coast (Sweet et al. 2017). The effects of groundwater pumping on subsidence and hydraulic gradients could exacerbate these estimates. Moreover, groundwater demand is likely to increase in coastal areas where recharge declines, further increasing intrusion (Ferguson and Gleeson 2012, Uddameri et al. 2014, Sawyer et al. 2016).

Figure 6.

U.S. coastal locations and associated projected future sea level rise (relative rate of increase in mm per annum). Symbols indicate the suggested direction of change (based on ensemble median, annual change). Symbol size reflects relative rates in 3 categories; lower (<5.1 mm per annum), middle (5.1 to 8.3 mm per annum), and upper (>8.3 mm per annum) categories. Studies not reporting an ensemble median (e.g., a range, or sensitivity) are shown as “Direction indeterminate”. Detailed information about each study are provided in the supplemental materials.

The effects of SWI are expected to be most pronounced in low elevation, low gradient coastal areas and in areas with high projected RSLR, such as the Mid-Atlantic, central Gulf, Southern California, northern Alaska, and Hawai’i (Burkett and Davidson 2012, Sweet et al. 2017). For example, hydrodynamic simulations driven by different sea level scenarios (from +0.3 to +1.0 m) were used to assess changes in the position of the saline wedge for the James and Chickahominy Rivers in Virginia for a dry and typical flow year (Rice et al. 2012). In the James River, simulations using a +1.0 m SLR scenario moved the 10 parts per thousand (ppt) salt wedge 18 km upstream during dry periods (due to reduced riverine freshwater pushing downstream) and 9 km upstream during periods of typical riverine flow. At Walker’s Dam in the Chickahominy, simulations of the same SLR scenario suggested increases in the number of days with salinity > 0.1 parts per thousand (ppt) from 2 days to over 100 days in a typical year. The 0.1 ppt threshold considered in the study is indicative of a threat to the drinking water standard (Rice et al. 2012). Other modeling work for the Savannah River, Georgia and the Grand Strand region, South Carolina based on 14-year simulations of historic flow data with imposed SLR suggests that the number of days with salinities above 0.5 ppt were likely to increase under various SLR and riverine flow scenarios (Roehl et al. 2013). Projected SLR in the Chesapeake Bay is suggested to increase salinity by 0.5 to 2.0 ppt by late century and to extend saltwater 11 km farther up the Bay during low flow periods (July-October) and 7 km during high flow periods (Hong and Shen 2012). The tidal range is also projected to increase 20 percent by late century and the residence time to increase by 20 days. Other studies in the San Francisco Bay Delta similarly project increases in salinity due to SLR by late century (Cloern et al. 2011; Kibel 2015).

Ecosystem Level Impacts.

Saltwater inundation and intrusion to coastal rivers, wetlands and aquifers, threatens infrastructure and coastal ecosystems (Melillo et al. 2014; Herbert et al. 2015; Wieşki et al. 2010). Shifts in the position of the freshwater-saltwater margin in coastal rivers and aquifers due to SLR, as well as episodic storm surges, are expected to present a risk to sources of drinking water (Vitousek and Howarth 1991; Maloney and Preston 2014). Water utilities relying on near-shore potable freshwater or groundwater sources could experience increases in salinization associated with inundation and intrusion as well as from increased tidal amplitudes and storm surge (Melillo et al. 2014; Blanco et al. 2013). In Florida, for example, a decline in the availability of drinking-quality groundwater has been reported in some locations due the effects of SLR driven saltwater intrusion into aquifers (Blanco et al. 2013). Wastewater, stormwater and drinking water infrastructure could also be vulnerable to storm surges and coastal flooding and increased pollutant loading from coastal storm runoff.

Changes in the salinity, hydrodynamics and mixing of estuaries are anticipated to alter habitat conditions (e.g., dissolved oxygen, salinity, food availability) away from the optima of existing species, including commercially important taxa, resulting in local decreases or extirpation. In the Chesapeake Bay, the pycnocline depth, which separates the oxygenated, less saline, upper water stratum from the lower, more saline stratum where hypoxia occurs, could become shallower, increasing the volume of oxygen stressed habitat (Hong and Shen 2012).

Habitat degradation and decreases in coastal wetlands extent are expected in some locations due to erosion (wave action associated with higher sea levels) and saltwater inundation and intrusion (Michener et al. 1997, Scavia et al. 2002, FitzGerald et al. 2006, USEPA 2011a, 2011b; Thorne et al. 2018; Wieşki and Pennings 2014; Craft et al. 2016; Jun et al. 2013). Increased salinity is likely to increase organic matter mineralization and decrease productivity, which can alter the balance between marsh erosion and accretion, as well as the quality of the habitat. Debates exist as to the extent to which wetland migration and accretion might be able to keep pace with RSLR, with most existing evidence suggesting likely future loss (Michener et al. 1997, Scavia et al. 2002, FitzGerald et al. 2006, USEPA 2011a, 2011b; Thorne et al. 2018). Others, however, suggest some resilience due to accretion or inland migration (Kirwan and Megonigal 2013; Hopkinson et al. 2018). In either case, impacts could be exacerbated by coastal development, which constrains wetland migration (Torio and Chmura 2013; Thorne et al. 2018).

FUTURE RESEARCH

Addressing the hydrologic and water quality challenges outlined in this review requires further assessment of vulnerabilities in different regional and watershed settings, and the development of management strategies to reduce risks. Several research needs are outlined here that can help to improve our knowledge about potential responses to future climate conditions and inform adaptation strategies:

1. Understanding future water quality responses is particularly challenging due to uncertainty about local scale, long-term changes in precipitation, and interactions with local land use, water management infrastructure and other human activities in different watershed settings. Water temperature projections, for example, are currently sparse in north central and northeastern regions of the U.S. Additional studies aimed at improving our spatial and temporal knowledge of potential changes can help decision makers in different regional and watershed settings prepare for future risks and develop targeted management strategies.

2. Observational (monitoring) data are essential to understanding the current water quality trends. This data can inform studies that assess the relationship between water quality trends and climate change. Observational data are also a key component in calibrating hydrologic and water quality models, particularly process-based models, which can subsequently extrapolate responses beyond current conditions when forced by future climate change scenarios. Long-term, continuous observational monitoring should, therefore, be preserved and expanded, potentially contributing to new insights (Webb et al. 2008; Kaushal et al. 2010; Isaak et al. 2012; Arismendi et al. 2012, 2013). Adopting new monitoring advancements, such as sensor technology, can provide more efficient means of collecting year-round, continuous data (e.g., at 15-minute intervals) and help fill spatial/temporal data gaps.

3. Sampling of biological data along with water quality at long-term monitoring sites could improve understanding about potential future impacts on aquatic ecosystems, including ecologically meaningful thresholds or tipping points. A small number of studies have collected long-term contemporaneous biological, temperature, and hydrologic data (USEPA 2016b; USEPA 2014), but long-term freshwater biological data sets useful for assessing the effects of climate variability (e.g., longer than 10 years) are not common, particularly for sites minimally disturbed by humans (Jackson and Füreder 2006).

4. Improving our ability to simulate watershed hydrologic and water quality responses is also important in advancing our understanding of effects. Model-based assessments of water quality are subject to a cascade of uncertainty associated with future climate change scenarios (particularly changes in precipitation), and hydrologic/water quality simulation models. For example, global climate models are currently not capable of accurately predicting long term, local scale changes in the frequency and amount of precipitation falling in intense events or represent processes that lead to the persistence of extended dry periods (Randall et al. 2007; Watts et al. 2015; Salathé and Mauger 2018). Additionally, hydrologic modeling of associated extreme low and high-flow events is particularly difficult (Benham et al. 2006; Beckers et al. 2009; Kim et al. 2010b). Abilities to simulate groundwater also vary, and often include very basic representations of subsurface hydrologic processes. Groundwater contributions to streamflow are known to moderate water temperatures and affect pollutant transport (Tague et al. 2007; Snyder et al. 2015).

5. The relationship between human activities and hydrology or water quality (e.g., water withdrawals and water management infrastructure including dams and reservoirs) has been documented in numerous studies (Poole and Berman 2001, Kaushal et al. 2010; Daraio and Bales 2014); however, water quality responses to the combined effects of future climate, land use, and water management infrastructure are yet to be fully integrated in many modeling studies. Additional modeling studies that integrate future land use, management and weather extremes would broaden our knowledge about potential hydrologic and water quality responses. This information could be used by decision makers to quantify contributions from specific drivers (e.g., climate or land use), assess relative vulnerabilities, and identify robust solutions.

6. This review suggests an increased risk of water quality and ecosystem degradation in the future for many U.S. locations. Ultimately, however, incorporating information about the ability to manage anticipated impacts would provide a more complete assessment of where and how watersheds are most vulnerable. Relatively little is known about the sensitivity of water quality best management practice (BMP) performance to future changes in climate and environmental conditions. In some locations it may be relatively easy to compensate for the anticipated effects of climate change with existing BMPs. In other locations, projected changes will be more difficult to address. Studies addressing BMP resilience to climate change, and the type and scope of management practices necessary to offset the impacts of climate change in different regional, watershed and site-scale settings are therefore necessary to inform adaptation strategies.

SUMMARY AND CONCLUSIONS

Anticipated warming air temperatures, changing precipitation patterns and rising sea levels are expected to have direct and cascading effects on U.S. water quality. In this paper, we review the technical literature addressing the responses of streamflow, water temperature and SWI/salinity (in coastal rivers and aquifers) to potential future changes in air temperature and precipitation. A companion paper by Coffey et al. (2019) reviews the responses of nutrients, algal blooms, sediment and pathogens.

Responses to climate change vary regionally and in different watershed settings in response to different future scenarios and interactions with other watershed (e.g., physiographic setting, land use) and human (e.g., water management infrastructure) factors. Northern and eastern regions are generally projected to experience higher winter precipitation and earlier snowmelt (in areas of accumulating snowfall) that lead to increased winter-spring and decreased summer-fall streamflow. In the southern Plains, interior Southwest and parts of the Southeast, potential decreases in summer-fall precipitation and increases in ET, are expected to contribute to lower annual streamflow volumes. An increased frequency of heavy precipitation events over much of the U.S. also presents an increased risk of high flow and flooding. Changes in streamflow have a major influence on pollutant transport, thermal regimes, ecosystem function and many other water quality factors.

Increases in water temperature are already observed, and are expected to continue throughout this century, driven by warming air temperatures. Impacts are likely to be greatest during summer, and in locations where warming air temperatures occur together with lower streamflow volumes (e.g., Southwest and parts of the mountain west). The projected water temperature increases could have major effects on water chemistry, aquatic life and suitability for human use.

In coastal areas, sea level rise, storm surges, and changes in the volume and timing of freshwater runoff are anticipated to increase the risk of SWI to estuaries and aquifers, altering spatial and temporal patterns of salinity and presenting a risk to water quality and aquatic life. Impacts from SLR and SWI are expected to be greatest in the mid-Atlantic, Gulf, and southern Pacific coasts.

To date, there has been very little effort to synthesize what is known about future water quality implications of shifts in climate drivers on the national scale. The results of this review suggest many US locations could experience substantial changes in flow, water temperature, and SWI, the impacts of which will vary depending on the magnitude of climate change, together with the effectiveness of management responses. Reviewing such impacts on a national scale is important because they have implications for water quality management including 1) where practical and readily implementable solutions are going to be necessary, 2) for which variables are solutions going to be necessary, and 3) for what potential magnitude of water quality response will management be necessary. Unless we continue to improve our ability to understand and answer such needs, we risk under or overpreparing communities, both of which have substantial implications. Understanding these on a national scale is important for guiding decision makers in selecting effective, readily implementable solutions (e.g., water quality BMPs and other approaches) to reduce the risk to water quality management goals and for managing resources and bringing appropriate research and management efforts to bear where needs are likely going to be greatest. For these reasons, reviews like this one are going to be an ongoing and integral part of the adaptation and solution effort.

Research Impact Statement:

Climate change effects on water quality will vary in different regional and watershed settings and could present a risk to human health and the environment.