Key Finding 1
The total flux of carbon—which includes gaseous emissions, lateral flux, and burial—from inland waters across the conterminous United States (CONUS) and Alaska is 193 teragrams of carbon (Tg C) per year. The dominant pathway for carbon movement out of inland waters is the emission of carbon dioxide gas across water surfaces of streams, rivers, and lakes (110.1 Tg C per year), a flux not identified in the First State of the Carbon Cycle Report (SOCCR1; CCSP 2007). Second to gaseous emissions are the lateral fluxes of carbon through rivers to coastal environments (59.8 Tg C per year). Total carbon burial in lakes and reservoirs represents the smallest flux for CONUS and Alaska (22.5 Tg C per year) (medium confidence).
Description of evidence base
Estimates for the export of carbon to U.S. coasts have been well documented through long-term observations (Stets and Striegl 2012) and syntheses (Butman et al., 2016; Stackpoole et al., 2016; Zhu and McGuire 2016). Carbon burial is derived from recent model results (Clow et al., 2015). Gaseous emissions of CO2 were originally assessed in Butman and Raymond (2011) for streams and rivers and McDonald et al. (2013) for lakes and reservoirs of CONUS only. Previous data do exist to support inland waters as dominated by supersaturated conditions (Striegl et al., 2012; Tranvik et al., 2009).
The finding that the dominant pathway for carbon loss through inland waters is through surface emissions was identified in Richey et al. (2002) and Cole et al. (2007) and quantified for CONUS in (Butman and Raymond 2011). Estimates that support this finding for Alaska are presented in Zhu and McGuire (2016). McDonald et al. (2012) showed that across CONUS, lake carbon burial and lake emissions are similar in magnitude when considered at the national scale, with regional variation based on the input of dissolved inorganic carbon (DIC) to lake systems.
Major uncertainties
Large uncertainties exist for the emission of CO2 from stream and river systems based on empirical estimates of the gas transfer velocity of CO2 presented in Raymond et al. (2012). The modeling of gas transfer is poorly constrained under high-flow conditions in steep topography. High levels of uncertainty also exist regarding the temporal dynamics of both lentic and lotic CO2 emissions (Battin et al., 2008; Striegl et al., 2012; Tranvik et al., 2009), where limited data exist to assess carbon gas concentrations under ice or storm flow conditions.
Uncertainties also exist regarding the use of the empirical model for carbon burial presented in Clow et al. (2015). Limited concentration data exist for lakes in Alaska, and there may be significant bias in the concentrations used to scale lake fluxes across regions (Stackpoole et al., 2017a; Zhu and McGuire 2016). These constraints may result in overestimates of emissions. In addition, limited data on carbon burial exist for northern latitudes, resulting in the use of empirical models derived from samples that do not capture the level of variability that exists across Alaska (Stackpoole et al., 2016).
Assessment of confidence based on evidence and agreement, including short description of nature of evidence and level of agreement
The overall confidence level of medium reflects 1) advancements in inland water spatial representations in a global information system (GIS) format to develop surface areas, 2) completion of datasets enabling the calculation of lateral fluxes, and 3) advancements in databases relevant to sedimentation rates in U.S. lakes and reservoirs. Confidence is reduced because modeling approaches available to estimate gas transfer velocities used for calculating carbon emissions are limited, and there are few chemical measurements in small stream systems.
Summary sentence or paragraph that integrates the above information
For Key Finding 1, individual flux terms (i.e., lateral flux, CO2 emission, and carbon burial) each have a medium to high level of certainty. This reflects the high confidence in the spatial representation of the chemical data for CONUS and Alaska, as well as the length of monitoring for water chemistry within CONUS and Alaska.
Key Finding 2
Based on estimates presented herein, the carbon flux from inland waters is now understood to be four times larger than estimates presented in SOCCR1. The total flux of carbon from inland waters across North America is estimated to be 507 Tg C per year based on a modeling approach that integrates high-resolution U.S. data and continental-scale estimates of water area, discharge, and carbon emissions. This estimate represents a weighted average of 24 grams of carbon per m2 per year of continental area exported and removed through inland waters in North America (low confidence).
Description of evidence base
Initial data presented in SOCCR1 did not acknowledge emission of carbon across the air-water interface. The estimate of 507 Tg C per year is based on well-constrained estimates of water discharge presented in Mayorga et al. (2010), Seitzinger et al. (2005), and compared with Dai et al. (2009, 2012). Estimates for the export of carbon modeled with water discharge are provided through the Regional Carbon Cycle Assessment and Processes (RECCAP) effort of the Global Carbon Project. Gaseous emissions of CO2 are presented in Raymond et al. (2013) based on similar methods presented in Butman and Raymond (2011). Areal rates of carbon flux through inland waters for CONUS and Alaska match those for North America.
Major uncertainties
Estimates and uncertainties to scale the emissions of CO2 from streams, rivers, and lake systems from CONUS to North America have already been provided. However, the application of CONUS lake carbon burial rates derived from Clow et al. (2015) to the total lake areas from Aufdenkampe et al. (2011) is unique. The methods used an average burial rate of about 110 g C per m2 per year, which is lower than those used in recent global estimates for lake and reservoir burial (Battin et al., 2009a). This burial rate is not dynamic and does not fully capture the spatial heterogeneity found across North America (Clow et al., 2015).
Assessment of confidence based on evidence and agreement, including short description of nature of evidence and level of agreement
Overall level of confidence is lower for the region of North America due to the different modeling approach, lack of data that exist in both Canada and Mexico, and the simplified application of U.S. data to a region that covers many different ecosystem types.
Summary sentence or paragraph that integrates the above information
For Key Finding 2, confidence is low for estimates of inland aquatic carbon fluxes for North America because of a general lack of data available from Mexico and Canada, including CO2 emissions or burial estimates. Methods developed for datasets within CONUS were applied to these two regions.
Key Finding 3
Future research can address critical knowledge gaps and uncertainties related to inland water carbon fluxes. This chapter, for example, does not include methane emissions, which cannot be calculated as precisely as other carbon fluxes because of significant data gaps. Key to reducing uncertainties in estimated carbon fluxes is increased temporal resolution of carbon concentration and discharge sampling to provide better representations of storms and other extreme events for estimates of total inland water carbon fluxes. Improved spatial resolution of sampling also could potentially highlight anthropogenic influences on the quantity and quality of carbon fluxes in inland waters and provide information for land-use planning and management of water resources. Finally, uncertainties could likely be reduced if the community of scientists working in inland waters establishes and adopts standard measurement techniques and protocols similar to those maintained through collaborative efforts of the International Ocean Carbon Coordination Project and relevant governmental agencies from participating nations.
Description of evidence base
Methane CH4 emissions can be a significant source of carbon to the atmosphere from Arctic lakes (Wik et al., 2016). Fixed-interval sampling protocols may miss large storm events and may critically bias estimates for total carbon fluxes to the coast (Raymond et al., 2012). Management of water resources in reservoir systems may influence the magnitude of carbon burial and emissions, driving systems to be more or less effective at storing or releasing carbon over time (Deemer et al., 2016).
Major uncertainties
Uncertainties are presented within the evidence base. Major uncertainties include 1) the relative importance of storm events or perturbations in the hydrological cycle to carbon export to coastal systems, 2) the magnitude of CH4 fluxes over time and across seasonal and latitudinal gradients, 3) the role that management of water resources plays in the movement and storage of carbon over time, and 4) the lack of established protocols for comparable sampling and scaling of carbon emissions across inland waters.
Summary sentence or paragraph that integrates the above information
For Key Finding 3, overall spatial and temporal data are not adequate to estimate the magnitude of CH4 fluxes from inland waters or to capture the influence of storm events or management on inland water carbon fluxes.