B11F-0499
Comparing Stream DOC Fluxes from Sensor- and Sample-Based Approaches
Monday, 14 December 2015
Poster Hall (Moscone South)
James B Shanley1, JohnFranco Saraceno2, Brent T Aulenbach3, Alisa Mast4, David W Clow4, Krista Hood5, John F Walker6, Sheila F Murphy7, Angel Torres-Sanchez8, George Aiken9 and William H McDowell10, (1)U.S. Geological Survey, Montpelier, VT, United States, (2)USGS California Water Science Center Sacramento, Sacramento, CA, United States, (3)USGS, Georgia Water Science Center, Norcross, GA, United States, (4)USGS Colorado Water Science Center Denver, Denver, CO, United States, (5)USGS, Middleton, WI, United States, (6)USGS Wisconsin Water Science Center, Middleton, WI, United States, (7)USGS Central Region Offices Denver, Denver, CO, United States, (8)US Geological Survey, Guaynabo, PR, United States, (9)USGS Colorado Water Science Center Boulder, Boulder, CO, United States, (10)University of New Hampshire Main Campus, Durham, NH, United States
Abstract:
DOC transport by streamwater is a significant flux that does not consistently show up in ecosystem carbon budgets. In an effort to quantify stream DOC flux, we analyzed three to four years of high-frequency in situ fluorescing dissolved organic matter (FDOM) concentrations and turbidity measured by optical sensors at the five diverse forested and/or alpine headwater sites of the U.S. Geological Survey (USGS) Water, Energy, and Biogeochemical Budgets (WEBB) program. FDOM serves as a proxy for DOC. We also took discrete samples over a range of hydrologic conditions, using both manual weekly and automated event-based sampling. After compensating FDOM for temperature effects and turbidity interference – which was successful even at the high-turbidity Luquillo, PR site -- we evaluated the DOC-FDOM relation based on discrete sample DOC analyses matched to corrected FDOM at the time of sampling. FDOM was a moderately robust predictor of DOC, with r2 from 0.60 to more than 0.95 among sites. We then formed continuous DOC time series by two independent approaches: (1) DOC predicted from FDOM; and (2) the composite method, based on modeled DOC from regression on stream discharge, season, air temperature, and time, forcing the model to observations and adjusting modeled concentrations between observations by linearly-interpolated model residuals. DOC flux from each approach was then computed directly as concentration times discharge. DOC fluxes based on the sensor approach were consistently greater than the sample-based approach. At Loch Vale, CO (2.5 years) and Panola Mountain GA (1 year), the difference was 5-17%. At Sleepers River, VT (3 years), preliminary differences were greater than 20%. The difference is driven by the highest events, but we are investigating these results further. We will also present comparisons from Luquillo, PR, and Allequash Creek, WI. The higher sensor-based DOC fluxes could result from their accuracy during hysteresis, which is difficult to model. In at least one case the higher sensor-based DOC flux was linked to an unsampled event outside the range of the concentration model. Sensors require upkeep and vigilance with the data, but have the potential to yield more accurate fluxes than sample-based approaches.
