Data Package Metadata   View Summary

An eddy covariance (EC) system (LiCor Biosciences, Lincoln, Nebraska) was used to collect greenhouse gas fluxes (carbon dioxide, methane) from Falling Creek Reservoir from April 2020 to May 2022. The EC instrumentation was deployed at the deepest site of Falling Creek Reservoir on a permanent metal platform that extends about 45 m from the dam and approximately 2.9 m over the reservoir's surface. The reservoir was maintained at full pond during the monitoring period, resulting in negligible change in distance between the EC system and the water's surface.

The EC instrumentation included an ultrasonic anemometer to measure 3D wind speed and direction (CSAT3, Campbell Scientific Inc., Logan, Utah, USA), an open-path infrared gas analyzer for measuring methane concentration (LI-7700, LiCor Biosciences, Lincoln, Nebraska, USA), and an enclosed path infrared gas analyzer for measuring carbon dioxide and water vapor concentrations (LI-7200, LiCor Biosciences, Lincoln, Nebraska, USA). The data streams (anemometer, methane, carbon dioxide, water vapor) were recorded at 10-Hz using a data logger that included a temperature sensor and pressure transducer (LI-7550, LiCor Biosciences, Lincoln, Nebraska, USA). The open path methane sensor was manually cleaned approximately weekly from April through October and approximately monthly from November to April. The carbon dioxide sensor was cleaned approximately every three months or when the sensor strength dropped below ~95%.

The collected, raw 10-Hz data were processed into 30-minute fluxes using the EddyPro v. 7.0.6 software (LiCor Biosciences, Lincoln, Nebraska, USA; LiCor Biosciences 2019) using the 'Express' settings. Following processing in EddyPro software, we excluded some redundant measurements and calculations in R using EddyPro_CleanUp.R. The EddyPro processed data is the data published as part of this data product (Apr2020toMay2022_EddyPro_Cleaned.csv).

Additional data processing following standard best practices can be applied using the FCR_Process_BD.R script. Additional data processing included: 1) removing wind directions which originated behind the dam (i.e., outside of the reservoir; wind direction between 80-250 degrees removed); 2) removing extreme flux values (carbon dioxide fluxes > abs(100) umol C m-2 s-1; methane fluxes > abs(0.25) umol C m-2 s-1); 3) removing methane fluxes when the signal strength <20%; 4) removing carbon dioxide and methane fluxes when they did not pass the test for stationarity or developed turbulent conditions (quality control, QC level 2, per Foken et al. 2004) in addition to when the latent heat flux (LE) or sensible heat (H) had QC level <2; 4) removing open path methane fluxes during periods of rainfall, which was determined based on the rain gauge deployed at the FCR dam; 5) correction for high-pass and low-pass filtering effects (Moncrieff et al. 2004; using the function defined in despike.R), and 6) removing data that corresponded to flux footprints that extended significantly beyond the reservoir. Flux footprints were modeled every half-hour using a simple, two-dimensional parameterization developed by Kljun et al. (2015). This model builds on the lagrangian stochastic particle dispersion model (Kljun et al., 2002), and provides information on the upwind and crosswind spread of the footprint. All the variables needed for the model were obtained directly from the dataset, or calculated following Kljun et al. (2015). Fluxes were excluded when the along-wind distance providing 90% cumulative contribution to turbulent fluxes was outside the reservoir, based on the footprint analysis. Finally, 7) we filtered out additional periods of low turbulence friction velocity (ustar) using REddyProc as described below (Wutzler et al. 2018).

We include the quality control flags for each calculated flux as assigned by EddyPro software such that: 0 = best quality fluxes; 1 = fluxes suitable for general analysis; and 2 = remove fluxes following Mauder and Foken (2006). These quality control flags were used for further data QA/QC as described above.

Following 30-minute flux conversions in Eddy Pro and additional post-processing as described above, the script can also be used for additional data processing using the R package REddyProc (Wutzler et al. 2018) to conduct gap-filling of missing data. First, we used the meteorological data (Carey et al. 2021) measured at the dam (located ~45 m from the EC sensors) to gap-fill any missing wind speed, direction, temperature, and relative humidity from the EC data. Second, we calculated the vapor pressure deficit from measured air temperature and relative humidity and calculated net radiation balance from upwelling and downwelling shortwave and longwave radiation. Using REddyProc, we then gap-filled any remaining gaps in the air temperature, shortwave radiation, total PAR, net radiation, sensible heat flux, and latent heat flux using the marginal distribution sampling (MDS) following Wutzler et al. (2018). We then used REddyProc to estimate the ustar threshold distribution and removed any fluxes where ustar was too low (Wutzler et al. 2018). Finally, we gap-filled any missing fluxes using the estimated ustar distributions using the MDS method (Wutzler et al. 2018).

On 10 August 2020, the data logger was removed for maintenance and was re-deployed on 2 September 2020. Additionally, a thermocouple on the CO2 sensor (LI-7500) was inoperable starting on 5 April 2021 and was reparied on 26 April 2021. We note that power interruptions or instrument malfunction resulted in 83% and 70% raw data coverage for carbon dioxide and methane, respectively. Ultimately, all data processing as described above, resulted in a total of 23% data coverage for carbon dioxide and 19% for methane fluxes prior to gap-filling.

References

Carey C.C., Breef-Pilz A, Bookout BJ, Lofton ME, McClure RP. 2021. Time series of high-frequency meteorological data at Falling Creek Reservoir, Virginia, USA 2015-2020 ver 5. Environmental Data Initiative. https://doi.org/10.6073/pasta/890e4c11f4348b3ceda802732ffa48b4 (Accessed 2021-10-05).

Foken T., Goockede M., Mauder M., Mahrt L., Amiro B., Munger W. 2004. Post-Field Data Quality Control. In: Lee X., Massman W., Law B. (eds) Handbook of Micrometeorology. Atmospheric and Oceanographic Sciences Library, vol 29. Springer, Dordrecht. https://doi.org/10.1007/1-4020-2265-4_9

Kljun N., Rotach M.W., Schmid H.P. 2002. A 3D Backward Lagrangian Footprint Model for a Wide Range of Boundary Layer Stratifications. Boundary Layer Meteorology, 103, 205-226.

Kljun N., Calanca P., Rotach M.W., Schmid H.P. 2015. A simple two-dimensional parameterisation for Flux Footprint Prediction (FFP), Geoscience Model Development, 8, 3695-3713. https://doi.org/10.5194/gmd-8-3695-2015

LiCor Biosciences. 2019. Eddy Pro v. 7.0.6 [Computer software]. Available: https://www.licor.com/env/support/EddyPro/software.html. Accessed: 22 December 2021.

Mauder M, Foken T. 2006. Impact of post-field data processing on eddy covariance flux estimates and energy balance closure. Meteorologische Zeitschrift, 15: 597-609.

Moncrieff J., Clement R., Finnigan J., Meyers T. 2004. Averaging, Detrending, and Filtering of Eddy Covariance Time Series. In: Lee X., Massman W., Law B. (eds) Handbook of Micrometeorology. Atmospheric and Oceanographic Sciences Library, vol 29. Springer, Dordrecht. https://doi.org/10.1007/1-4020-2265-4_2

Wutzler T, Lucas-Moffat A, Migliavacca M, Knauer J, Sickel K, Sigut L, Menzer O, Reichstien M. 2018. Basic and extensible post-processing of eddy covariance flux data with REddyProc. Biogeosciences, 15, 5015-5030. https://doi.org/10.5194/bg-15-5015-2018