EMERGING PERSPECTIVES ON CONTINENTAL-SCALE RIVERINE CARBON FLUXES David Butman, Yale University Edward G. Stets, U.S. Geological Survey Peter Raymond, Yale University Robert G. Striegl, U.S. Geological Survey
Jan 12, 2016
EMERGING PERSPECTIVES ON
CONTINENTAL-SCALE
RIVERINE CARBON FLUXES
David Butman, Yale UniversityEdward G. Stets, U.S. Geological SurveyPeter Raymond, Yale UniversityRobert G. Striegl, U.S. Geological Survey
Acknowledgements -
NASA Earth and Space Science Fellowship
NSF-Ecosystems CAREER
Yale Center for Earth Observation
Christopher Zappa, Tom Bott, Jody Potter, Patrick Mulholland and William McDowell (Gas Transfer Meta-Analysis), Charlie Crawford, Kathleen Johnson, Cory McDonald.
AQUATIC CARBON CYCLINGAQUATIC CARBON CYCLING
CO2 IC OC
Stream flow
CO2 IC OC
UpstreamTerrestrialHyporheic
DownstreamorCoastal
IC – Inorganic carbonOC – Organic carbon
AQUATIC CARBON CYCLINGAQUATIC CARBON CYCLING
CO2 CO2
CO2 IC OC
Stream flow
CO2 IC OC
IC OCCO2CO2
UpstreamTerrestrialHyporheic
DownstreamorCoastal
IC – Inorganic carbonOC – Organic carbon
Sedimentation
AQUATIC CARBON CYCLINGAQUATIC CARBON CYCLING
CO2 CO2
CO2 IC OC
Stream flow
CO2 IC OC
IC OCCO2CO2
UpstreamTerrestrialHyporheic
DownstreamorCoastal
IC – Inorganic carbonOC – Organic carbon
Sedimentation
AQUATIC CARBON CYCLINGAQUATIC CARBON CYCLING
CO2 IC OC
Stream flow
CO2 IC OC
UpstreamTerrestrialHyporheic
DownstreamorCoastal
IC – Inorganic carbonOC – Organic carbon
CO2
“CO2 Flux”“Lateral Flux”
Sedimentation
Water quality and streamflow data–• USGS National Water Information System (NWIS).
• National Stream Quality Accounting Network (NASQAN).
DATA SOURCES & TERMINOLOGYDATA SOURCES & TERMINOLOGY
Carbon fractions –
• Inorganic carbon (IC) and CO2
• Calculated from Alkalinity, temperature, pH
• Assume alkalinity arises from ΣCO2
• Assume particulate fraction of alkalinity is negligible
• Organic carbon (OC).
• Total organic carbon (TOC) – unfiltered
• Dissolved organic carbon (DOC) - filtered
Inorganic carb(g C m-2 yr-1)
>
4
<
No LATERAL CARBON FLUXESLATERAL CARBON FLUXES
COCO2 2 FLUX FROM STREAMS FLUX FROM STREAMS
otal lux: .7 Tg C yr-1
Tota (Tg
ICTOCCO2
RIVERINE C FLUX IN PERSPECTIVERIVERINE C FLUX IN PERSPECTIVE
USGS Load Estimator (LOADEST)-
1. Calibration data.
2. Multiple regression.
3. Calculate daily loads.
4. Aggregate into annual fluxes.
DateStream Flow (cfs)
TOC (mg L-1)
04/24/2008 17,100 5.04
05/20/2008 12,500 4.45
06/11/2008 5,390 4.69
07/10/2008 6,520 6.23
08/05/2008 6,520 4.69
08/20/2008 8,290 6.87
09/17/2008 10,600 7.31
10/14/2008 8,270 6.2
12/23/2008 37,300 7.7
01/15/2009 25,600 5.44
02/05/2009 13,300 3.92
02/17/2009 13,500 16.99
FLUX CALCULATIONSFLUX CALCULATIONS
USGS Load Estimator (LOADEST)-
1. Calibration data.
2. Multiple regression.
)cos()sin(lnlnln 432
210 timeatimeaQaQaaLOAD
3. Calculate daily loads.
4. Aggregate into annual fluxes.
Relationship between load and stream flow
Seasonal variability
FLUX CALCULATIONSFLUX CALCULATIONS
USGS Load Estimator (LOADEST)-
1. Calibration data.
2. Multiple regression.
3. Calculate daily loads.
4. Aggregate into annual fluxes.
Daily TOC loads (x106 g d-1)
Jan-08 Jul-08 Jan-09 Jul-09 0
2000
4000
6000
FLUX CALCULATIONSFLUX CALCULATIONS
USGS Load Estimator (LOADEST)-
1. Calibration data.
2. Multiple regression.
3. Calculate daily loads.
4. Aggregate into annual fluxes.
2008 20090.0
0.1
0.2
Annual TOC loads (Tg C yr-1)
FLUX CALCULATIONSFLUX CALCULATIONS
1950 1970 1990 20100
200
400
600
800
1000 TOC (n = 5,126; 151 sites)
IC (n = 4,552; 161 sites)
DOC (n = 3,121; 144 sites)
DATABASE FEATURESDATABASE FEATURES
Observationsper year
DATABASE FEATURESDATABASE FEATURES
TOC
IC
DOC
Weighted bydrainage area
DATABASE FEATURESDATABASE FEATURES
Drainage Area Discharge0
20
40
60
80
100
IC TOC DOC
Percent includedin lateral flux database
0
10
20
30
40
50
IC TOC
SUMMARY OF FINDINGSSUMMARY OF FINDINGS
Observed Corrected fordrainage area
Corrected fordischarge
Lateral fluxTg C yr-1
28 32 34
11
910
Total =37
Total =42
Total =45
COMPARISON WITH OTHER STUDIESCOMPARISON WITH OTHER STUDIES
0 5 10 15 20 25 30
This study
Mulholland 1982
Boyer et al. (In prep.)
Meybeck 1981
Schlesinger and Melack 1982
TOC Flux from Conterminous U.S. (Tg C yr-1)
† - Disaggregated by watershed area.
†
§ - Disaggregated by biome type.
§
§
Inorganic carbon yield(g C m-2 yr-1)
Dissolved organic carbon yield(g C m-2 yr-1)
> 7
4 – 72 - 4
< 2
No data
> 4
2 – 40.8 – 2
< 0.8
No data
REGIONAL PATTERNSREGIONAL PATTERNS
Carbon yieldg C m-2 yr-1
St Law
renc
e
Miss
issipp
i
South
west
North
east
North
west
Easte
rn G
ulf
Wes
tern
Gulf
South
east
Red
Colora
do0
2
4
6
8
Entire
U.S
.
IC
TOC
Calculating COCalculating CO22 flux flux
• Stream surface area• CO2 concentrations• Transfer velocity
Flux = stream surface area * ([CO2]water - [CO2]air)* k
High Resolution NHDPlus data•Inventory of streams•Stream order•Modeled discharge•Modeled slope•Modeled velocity
Stream areaStream area
Use modeled discharge to calculate stream width.
Sum all stream segments to obtain total stream area.
Stream areaStream area
ln discharge
n = 1,026
Verification data from USGS streamgaging stations
Raymond et al. In Prep.
Use modeled discharge to calculate stream width.
Sum all stream segments to obtain total stream area.
Stream areaStream area
ln discharge
n = 1,026
Verification data from USGS streamgaging stations
Raymond et al. In Prep.
Surface Area of Streams: 40,560 km2
~1/2 of Lake Superior (US)~Costa Rica
•Calculated from alkalinity, temperature, and pH
•4,200 sites
•440,000 observations
COCO22 concentrations concentrations
COCO22 concentrations concentrations
Northern
K600 (m d-1) = 2841(SV) + 2.025
S= Slope ; V = Velocity
Gas transfer velocityGas transfer velocity
Raymond et al. in prep
Gas transfer velocityGas transfer velocity
Very high k in low-order Western streams.
Total Flux: 96.7 ± 32 Tg C yr-1
Total COTotal CO22 flux from streams flux from streams
Flux = stream surface area * ([CO2]water - [CO2]air)* k
Regional CORegional CO22 yield yield
CO2 yieldg C m-2 yr-1
Centra
lGulf
North
ern
Midw
est
Wes
t
South
west
0
10
20
Entire
US
LATERAL VERSUS COLATERAL VERSUS CO22 FLUXES FLUXES
Lateral flux
CO2
CO2
CO2
CO2
CO2
CO2
CO2 flux• Diffuse• Dominated by low-order streams.
Lateral flux• Focused• Top 10 streams carry
• 75% of IC flux.• 60% of TOC flux.
LATERAL VERSUS COLATERAL VERSUS CO22 FLUXES FLUXES
CO2 and lateral fluxes• Correlated with precipitation and
runoff.• Water discharge.
• Carbon delivery from terrestrial environment.
• In-stream carbon transformations.
CO2
r2 = 0.86
TOCr = 0.72
g C
m-2 y
r-1
Runoff (mm yr-1)
Total carbon flux in riversTotal carbon flux in rivers
Total flux (Tg C yr-1)
IC 34TOC 11CO2 97
Total 142
142 Tg C yr-1
Total carbon flux in riversTotal carbon flux in rivers
Yield (g C m-2 yr-1)
IC 4.0TOC 1.9CO2 14.9
Total 20.8
20.8 g C m-2 yr-1
Carbon fluxes in contextCarbon fluxes in context
>142 Tg C yr-1
Stream flow
45 Tg C yr-1
97 Tg C yr-1
“CO2 Flux”“Lateral Flux”
Sedimentation
?
Terrestrial and aquatic carbon fluxTerrestrial and aquatic carbon flux
NEP (g C m-2 yr-1)
Grasslands 24 ± 14 Zhang et al. 2011
Total flux Rivers & Streams
20.8 This study
Predictor of Basin Flux Slope r2
Stream Surface Area %1 0.72 0.78
Average Annual Precipitation2
23.1 0.86
% Forested Land3 0.56 0.64
• Precipitation controls the regional differences in CO2 flux
• Short-term: flushing of soil CO2
• Long-term: geomorphology of the density of stream networks (stream surface area).
Regional CORegional CO22 yield yield
LATERAL VERSUS COLATERAL VERSUS CO22 FLUXES FLUXES
Lateral flux
CO2
CO2
CO2
CO2
CO2
CO2
CO2 flux• Dominated by fluxes in small-order
streams.• Strong correlation with precipitation.
Carbonate Weathering
Soil Respiration
Acidity:Precipitation /
Mining
Allochthonous DOM Respiration
Wetlands / Riparian
Vegetation
• The dominant source of pCO2 will depend on scale
• Headwater systems will show terrestrial respiration• Large river systems will incorporate internal and external
sources.
Contribution of Soil CO2 to River Efflux
• Soil Water pCO2 20,000 – 30,000 ppm• Derived land surface run-off from USGS – Waterwatch –
Wolcock et al. – in prep.• Total US discharge 1722 km3
– Amazon discharge ~ 5000 km3
Region Basin Area thousand km2
Predicted q (m)
Predicted Max Total Flux Tg C
Predicted CO2
Evasion Flux Tg C
Max % of CO2 efflux
exported laterally (Soil)
Central 1020 0.36 6.6 23.3 28.1%
Northern 909 0.39 6.3 15.5 40.8%
Midwest 1965 0.06 2.2 15.3 14.4%
Gulf 1425 0.32 1.72 26.8 30.5%
West 1788 0.22 8.1 12.6 56.2%
Southwest 722 0.02 7.1 3.3 7.0%
Total US 7828 0.22 31.0 96.7 32.1%
Contribution of Soil CO2 to River Efflux
Total carbon flux in riversTotal carbon flux in rivers
Yield (g C m-2 yr-1)IC 3.9TOC 2.8CO2 25.5
Total 32.2
Stream w = e(0.42(lnQ)+2.55) Raymond et al. in prep
Stream Surface Area:
F(g C yr-1) = ([CO2]water - [CO2]air)* k * Surface Area
1026 sites
REGIONAL PATTERNSREGIONAL PATTERNS
North
east
South
east
Easte
rn G
ulf
Wes
tern
Gulf
Miss
issipp
i
St Law
renc
eRed
Colora
do
South
west
North
west
0
2
4
6
8
All
Carbon yieldg C m-2 yr-1
REGIONAL PATTERNSREGIONAL PATTERNS
0
10
20
30
40
50
IC POC DOC
Observed Corrected fordrainage area
Corrected fordischarge
Lateral fluxTg C yr-1
28 32
7
34
38
36 3
Gulf
Wes
t
Centra
l
North
Midw
est
South
west
0
10
20
Entire
US
Regional CORegional CO22 yield yield
CO2 yieldg C m-2 yr-1
DATABASE FEATURESDATABASE FEATURES
Observationsper year
Willamette River(at Salem, OR)
Illinois River(at Peoria, IL)(at Valley City, IL)
Schuylkill River(at Philadelphia, PA)
Acknowledgements –
Whitney Broussard (U of Louisiana, Lafayette)
Thor Smith (USGS Vermont / New Hampshire)
Valerie Kelly (USGS Oregon)
Evan Hornig (USGS Austin, TX)
Kate Halm (USGS Colorado)
Sources of data and methods of analysis –
SourcesWater quality -• Clarke (1924) compilation of Dole & Stabler survey (USGS).
• U.S. Geological Survey (National Water Information System).
• Illinois State Water Survey.• Illinois Environmental Protection Agency.• Philadelphia Water Department.
Stream discharge –• U.S. Geological Survey.
Methodological considerations –
Nitrate analysis
Pre-1970 Colorimetric phenoldisulphonic acid method.After 1970 Colorimetric cadmium reduction.
Organic nitrogen
• “Albuminoid ammonia” method used on most samples before 1940.Total N = Alb. NH3 + NO3 + NO2 + NH3
• Kjeldahl nitrogen used on all samples since 1975.Total N = TKN + NO3 + NO2
Sample handling• “Composite samples” – time-averaged.
• Discrete samples – single time point.
• Holding time in older samples.
Willamette River at Salem, ORDrainage area 19,000 km2
Human impacts in the watershed• Slaughterhouse and animal processing
facilities.• Urban wastewater and runoff.• Pulp and paper mill waste.• Logging / forestry.• Some agriculture in lower basin.
Noted historical water quality problems• Low oxygen concentration around
Salem (largely solved by 1972 through discharge permit system).
• Untreated urban wastewater (successive rounds of treatment plant updates 1930s through 1960s).
• Depleted salmon population (ongoing, improvements after 1970).
Illinois RiverDrainage area 41,000 km2 (upper)
69,000 km2 (lower)
Human impacts in the watershed• Diversion of Chicago River (1890).• Urban runoff and wastewater.• Intense row crop agriculture.• Bottom land farms maintained through
levee & drainage districts.• Industrial effluent.
Noted historical water quality problems• Water-borne disease outbreaks in late
1800s.• Fisheries virtually destroyed by 1920s.• Low oxygen associated with
slaughterhouse wastes in Pekin/Peoria area.
Schuylkill RiverDrainage area 4,902 km2
Human impacts in the watershed• First municipal water works in the nation
(1802).• Mine waste in upper watershed (1860s).• Heavy industry (1880s).• Moderate agriculture in lower watershed
outside of Philadelphia.• Intense urban development in lower
watershed.
Noted historical water quality problems• Numerous government reports between
1866 and 1946 recommended abandoning the Schuylkill as a drinking water source.
• Coal mine silt pollution in upper basin.• Dramatic water quality improvements
resulted in the Schuylkill River being designated as a Federal Scenic River by 1978.