Harvard Forest

Data Archive

HF233

Eddy Covariance and Meteorological Data in the Clearcut Site at Harvard Forest 2009-2012

Related Publications

Data

• hf233-01: eddy flux data (preview)

Overview

• Lead: Christopher Williams
• Investigators:
• Contact: Information Manager
• Start date: 2009
• End date: 2012
• Status: complete
• Location: Prospect Hill Tract (Harvard Forest)
• Latitude: +42.546 degrees
• Longitude: -72.174 degrees
• Elevation: 403 meter
• Datum: WGS84
• Taxa:
• Release date: 2023
• Language: English
• EML file: knb-lter-hfr.233.5
• DOI: digital object identifier
• EDI: data package
• DataONE: data package
• Related links:
• Project Website
• Coarse Woody Debris in the Clearcut Site at Harvard Forest since 2010
• Soil Gas Exchange in the Clearcut Site at Harvard Forest 2011-2013
• Vegetation Cover in the Clearcut Site at Harvard Forest since 2010
• Study type: short-term measurement, long-term measurement, modeling
• Research topic: ecological informatics and modelling; forest-atmosphere exchange; physiological ecology, population dynamics and species interactions; watershed ecology
• LTER core area: primary production, disturbance patterns, human-environment interactions
• Keywords: air temperature, carbon, carbon dioxide, eddy covariance, evapotranspiration, heat flux, net ecosystem exchange, relative humidity
• Abstract:

  Clearcutting and other forest disturbances perturb carbon, water, and energy balances in significant ways, with corresponding influences on Earth’s climate system through biogeochemical and biogeophysical effects. Observations are needed to quantify the precise changes in these balances as they vary across diverse disturbances of different types, severities, and in various climate and ecosystem type settings. This dataset reports eddy covariance and micrometeorological measurements of surface-atmosphere exchanges that can be combined with related datasets from vegetation inventories and chamber-based estimates of soil respiration and leaf gas exchange to collectively quantify and understand how carbon, water, and energy fluxes changed during the first four years following forest clearing in a temperate forest environment of the northeastern US. Associated publications show rapid recovery with sustained increases in gross ecosystem productivity (GEP) over the first three growing seasons post-clearing, coincident with large and relatively stable net emission of CO2 because of overwhelmingly large ecosystem respiration. The rise in GEP was attributed to vegetation changes not environmental conditions (e.g. weather), but attribution to the expansion of leaf area versus changes in vegetation composition remains unclear. Soil respiration was estimated to contribute 44% of total ecosystem respiration during summer months and coarse woody debris accounted for another 18%. Evapotranspiration also recovered rapidly and continued to rise across years with a corresponding decrease in sensible heat flux. Gross shortwave and longwave radiative fluxes were stable across years except for strong wintertime dependence on snow covered conditions and corresponding variation in albedo. Overall, these findings underscore the highly dynamic nature of carbon and water exchanges and vegetation composition during the regrowth following a severe forest disturbance, and sheds light on both the magnitude of such changes and the underlying mechanisms with a unique example from a temperate, deciduous broadleaf forest.

• Methods:

  The study site occupies roughly a 200 m x 400 m area (8 ha) near the top of Prospect Hill, within the Harvard Forest Long Term Ecological Research Site. The following methods description is taken from Williams CA, Vanderhoof M, Khomik M, Ghimire B, 2014, Post-clearcut dynamics of carbon, water and energy exchanges in a mid-latitude temperate, deciduous broadleaf forest environment, Global Change Biology, 2(3), 992-1007, DOI: 10.1111/gcb.12388.

  A 5 m tripod was instrumented with eddy covariance instrumentation including a 3D sonic anemometer (Campbell Scientific CSAT3) measuring three-dimensional, orthogonal components of velocity (u, v, w, m s-1) as well as the ‘sonic’ air temperature (Ta, ˚C), and an open-path infrared gas analyzer (IRGA, LI-COR LI-7500, Lincoln, NE) measuring concentrations of water vapor (q in g H2O kg-1 air), and CO2 (μmol CO2 mol-1 air). Starting in mid-June 2009, eddy covariance instruments were installed at 3.0 to 4.5 m and elevated annually during May-June to maintain a measurement height at or above 1.5 times the canopy height. The tower is located at a local high point of a very gradually sloping landscape. Internal chemicals of the IRGA were changed annually in springtime, followed by zero and span calibrations for CO2 and H2O. Calibration gases included Grade-5, 99.999% pure nitrogen for the zero of CO2 and H2O, and a NIST traceable primary standard (± 1% from Airgas, Inc.) for CO2 that is confirmed with a secondary standard calibration gas (± 0.14ppm) provided by Ameriflux, and a LICOR dew point generator (LI-610) for H2O. Post-processing of measured fluxes is described further below.

  Average half-hourly incoming and outgoing longwave and shortwave radiation fluxes (Rlwin, Rlwout, Rswin, Rswout) were measured with a Kipp and Zonen (Delft, The Netherlands) CNR1 radiometer mounted at 3 m above the ground, the balance of which provides net radiation (Rn, W m-2). The domes were not heated due to insufficient power supply. Average half-hourly photosynthetically active radiation was recorded with a quantum sensor leveled at 2.5 m above the ground (LI-190SB, LICOR). Additional measurement of air temperature and humidity were recorded with a shielded, solid-state sensor at 2.5 m height (Vaisala HMP45C probe, Campbell Scientific). Average half-hourly volumetric soil water content was estimated with 15 cm long Campbell Scientific (Logan, UT) frequency domain reflectometry probes (CS615) installed horizontally in two separate profiles at soil depths (z, cm) of 10, 25, 50, and 94 cm, or 10, 20, 40, and 80 cm. The probes have not been locally calibrated, and while manufacturer notes suggest absolute accuracy to within 2%, estimates of soil water content are approximate with respect to the absolute water content but nonetheless provide a robust measure of the relative moisture dynamics. Half-hourly averaged soil heat flux was obtained with 3 self-calibrating soil heat flux plates (HFP01SC, Hukseflux Thermal Sensors, Delft, The Netherlands) installed 5 cm below the ground surface in representative settings (plant, litter and coarse woody debris environments). Spatially-averaging soil thermocouple probes (TCAV, Campbell Scientific) were installed adjacent to heat flux plates, with probes at 2 and 8 cm below the ground surface. Rainfall per half hour was measured with a tipping bucket precipitation gage (Met One Rain Gage Model 385, Campbell Scientific, Logan, UT) located adjacent to the downwind side of the main sensor tripod at 2 m above the ground and in an environment free of elevated obstructions to quantify open field precipitation. However, because the gage was not heated due to insufficient power supply we rely on the Harvard Forest Fisher meteorological station for additional measurements of precipitation (same gage but heated and 1.6 m above the ground in a pasture) recorded as total precipitation delivered over 15 minute intervals and available since 2001.

  Above-canopy vapor pressure deficit was calculated as the difference between the saturation vapor pressure at the air temperature, and the atmospheric water vapor pressure obtained from q and atmospheric pressure. Total soil water storage (S, cm) was calculated as the sum of Si, where Si = Ɵ1[(z2 - z1)/2 + z1] for i = 1, Si = Ɵi(zi+1 - zi-1)/2 for 1 between 1 and N, and Si = ƟN[65 - ((zN - zN-1)/2 + zN-1)] for i = N, where N is the number of measurement depths (z).

  Post-processing of the raw high frequency (10 Hz) data for calculation of above-canopy, half-hourly turbulent fluxes of sensible heat (H, W m-2), water vapor (LE, W m-2), and carbon dioxide (Fc, mg CO2 m-2 s-1) was consistent with methods described in Lee et al. (2004). It first involved spike filtering based on instrument diagnostics, logical limits, and based on a 4 standard deviation threshold within 3-minute data windows. The velocity field was rotated with the planar method of Wilczak et al. (2001) applied to monthly data for 30 degree sectors. We corrected sonic temperature to air temperature according to the work of Schotanus et al. (1983). LE and Fc were corrected with the Webb-Pearman-Leuning correction (Webb et al., 1980) plus an additional term accounting for surface heating of the instrument (Burba et al. 2006). Frequency response correction of some of the energy lost due to instrument separation and gas analyzer response for LE and Fc was performed with empirical cospectral adjustment to match the H cospectrum (Eugster and Senn, 1995, Su et al., 2004). Heat and mass fluxes were calculated with conventional equations (see e.g. Aubinet et al., 2000, Moncrieff et al., 1997). Net fluxes (LE, H, Fc) are reported as positive when from the land to the atmosphere. Lacking measurements of the vertical profile of CO2 concentration and expecting a modest storage flux given the ecosystem’s low stature, we assumed that net ecosystem-atmosphere CO2 exchange (NEE) equals measured Fc. Lateral and drainage flows of carbon toward groundwater, rivers and streams can also be important for the net carbon balance of landscapes (Butman and Raymond, 2011, Lauerwald et al., 2012) and may be particularly important following forest disturbance (Schelker et al., 2012). However such fluxes have not been measured in this work which has a central focus on ecosystem carbon exchange with the atmosphere.

  Turbulent flux data were excluded based on instrument diagnostics, data exceeding logical bounds, inappropriate wind direction, low turbulence, and large flux footprint extent. The instrument filter ensures proper sonic anemometer and IRGA function. The logical filter removes unrealistic data points. The wind direction filter removes samples unrepresentative of the clearcut as well as flows that may be adversely influenced by the tower-related hardware itself, conservatively defined as a mean half-hourly wind direction arriving from 20° to 160°. Outside of these sectors fluxes were seen to vary little with wind direction. The friction velocity filter removes measurements taken when turbulence was not developed well enough to ensure adequate measurement of the surface-atmosphere exchange during the measurement interval. Insufficient turbulence was defined with the method of Papale et al. (2006), which identifies the friction velocity above which the measured nighttime CO2 flux is within 90% of that measured in conditions of even greater turbulence as the median across 6 levels of temperature. The threshold was separately determined for 3 month segments of data identifying a u* threshold and ranged from 0.12 to 0.18 m s-1. The extent of the flux footprint was estimated with the analytical flux footprint model of Hsieh et al. (2000) and indicates that a source area within 150 m upwind of the tower contributed 80% of the measured flux 95% of the time, well within the clearcut area. Measurements were excluded when the 80% flux footprint extended beyond an upwind distance of 200 m and into the land cover types surrounding the clearcut target. Energy balance closure, defined as (H+LE)/(Rn-G), averaged 62% for half-hourly data (intercept of 12.5 W m-2) and 74% for the daily time scale (intercept of 0.85 W m-2). Lack of closure is a well-known chronic issue with the eddy covariance method, but even so this site’s closure is lower than the norm. It is not clear to what this can be attributed though the large volume of coarse woody debris may contribute a sizeable capacity to store and release energy, a term that is not measured.

  Net ecosystem CO2 exchange (NEE) was separated into ecosystem respiration (Reco) and gross ecosystem productivity (GEP) with the approach of Reichstein et al. (2005). Briefly, this approach equates valid nighttime NEE observations with nighttime Reco, derives temporally-varying (15-day) empirical coefficients relating Reco to temperature with the Lloyd and Taylor (1994) model, applies the model to estimate a continuous series of Reco, and calculates GEP from Reco – NEE.

  Data gaps were filled with the Marginal Distribution Sampling approach of Reichstein et al. (2005). This approach replaces missing values with the mean of temporally-local, valid flux measurements under analogous environmental conditions (Rswin, Ta, Da). The size of the data window is progressively expanded until filling is successful up to a maximum window size, here up to two weeks. Gap-filling, though possible beyond the two week window, was deemed to be of low confidence and thus excluded. We only retained values classified as high confidence (class A) by the quality control rating in Reichstein et al. (2005, see Appendix A therein).

• Organization: Harvard Forest. 324 North Main Street, Petersham, MA 01366, USA. Phone (978) 724-3302. Fax (978) 724-3595.

• Project: The Harvard Forest Long-Term Ecological Research (LTER) program examines ecological dynamics in the New England region resulting from natural disturbances, environmental change, and human impacts. (ROR).

• Funding: National Science Foundation LTER grants: DEB-8811764, DEB-9411975, DEB-0080592, DEB-0620443, DEB-1237491, DEB-1832210.

• Use: This dataset is released to the public under Creative Commons CC0 1.0 (No Rights Reserved). Please keep the dataset creators informed of any plans to use the dataset. Consultation with the original investigators is strongly encouraged. Publications and data products that make use of the dataset should include proper acknowledgement.

• License: Creative Commons Zero v1.0 Universal (CC0-1.0)

• Citation: Williams C. 2023. Eddy Covariance and Meteorological Data in the Clearcut Site at Harvard Forest 2009-2012. Harvard Forest Data Archive: HF233 (v.5). Environmental Data Initiative: https://doi.org/10.6073/pasta/6bdf9ebdaa102cbefa32efa57a69b2d0.

Detailed Metadata

hf233-01: eddy flux data

1. date: date
2. year: year
3. month: month
4. day: day of month
5. half.hour: half-hour of day (unit: nominalHalfHour / missing value: NA)
6. doy: day of year (unit: nominalDay / missing value: NA)
7. fc: net CO2 flux (unit: milligramPerMeterSquaredPerSecond / missing value: NA)
8. le: latent heat flux (unit: wattPerMeterSquared / missing value: NA)
9. h: sensible heat flux (unit: wattPerMeterSquared / missing value: NA)
10. g: ground heat flux (unit: wattPerMeterSquared / missing value: NA)
11. rn: net radiation (unit: wattPerMeterSquared / missing value: NA)
12. sw.in: incoming shortwave radiation (unit: wattPerMeterSquared / missing value: NA)
13. sw.out: outgoing shortwave radiation (unit: wattPerMeterSquared / missing value: NA)
14. lw.net: net longwave radiation (unit: wattPerMeterSquared / missing value: NA)
15. par: photosynthetically active radiation (unit: micromolePerMeterSquaredPerSecond / missing value: NA)
16. tair: air temperature at 2m above surface (unit: celsius / missing value: NA)
17. vpd: vapor pressure deficit (unit: kilopascal / missing value: NA)
18. press: atmospheric pressure (unit: millibar / missing value: NA)
19. rh: relative humidity (unit: dimensionless / missing value: NA)
20. co2: atmospheric co2 concentration (micromoles co2 per mole air) (unit: dimensionless / missing value: NA)
21. q: specific humidity (g h2o per kg air) (unit: dimensionless / missing value: NA)
22. ustar: friction velocity (unit: metersPerSecond / missing value: NA)
23. u: horizontal wind speed (unit: metersPerSecond / missing value: NA)
24. wdir: wind direction (unit: degree / missing value: NA)
25. stability: atmospheric stability parameter (z/L) (unit: dimensionless / missing value: NA)
26. rain: rainfall (unit: millimeterPerHalfHour / missing value: NA)
27. vwc1: ten cm deep, SW trench (% by volume) (unit: dimensionless / missing value: NA)
28. vwc2: five cm deep, SE trench (% by volume) (unit: dimensionless / missing value: NA)
29. vwc3: ten cm deep, SE trench (% by volume) (unit: dimensionless / missing value: NA)
30. vwc4: twenty-five cm deep, SE trench (% by volume) (unit: dimensionless / missing value: NA)
31. vwc5: fifty cm deep, SE trench (% by volume) (unit: dimensionless / missing value: NA)
32. vwc6: ninety-four cm deep, SE trench (% by volume) (unit: dimensionless / missing value: NA)
33. vwc7: twenty cm deep, NE trench (% by volume) (unit: dimensionless / missing value: NA)
34. vwc8: eighty cm deep, NE trench (% by volume) (unit: dimensionless / missing value: NA)
35. vwc9: not working, was between s20ne and s80ne (% by volume) (unit: dimensionless / missing value: NA)
36. soil.wat1: total soil water profile 1 (in a roughly 1 m deep soil column) (unit: centimeter / missing value: NA)
37. soil.wat2: total soil water profile 2 (in a roughly 1 m deep soil column) (unit: centimeter / missing value: NA)
38. tsoil1: soil temperature from profile 1 at 5 cm (unit: celsius / missing value: NA)
39. tsoil2: soil temperature from profile 2 at 5 cm (unit: celsius / missing value: NA)
40. precip: precipitation from the Fisher Met Station heated gauge (unit: millimeterPerHalfHour / missing value: NA)
41. gep: gross ecosystem productivity, R05 method (mg CO2 m-2 s-1) (unit: milligramPerMeterSquaredPerSecond / missing value: NA)
42. reco: ecosystem respiration, R05 method (mg CO2 m-2 s-1) (unit: milligramPerMeterSquaredPerSecond / missing value: NA)
43. le.fill: latent heat flux gap filled with marginal distribution sampling (unit: wattPerMeterSquared / missing value: NA)
44. h.fill: sensible heat flux gap filled with marginal distribution sampling (unit: wattPerMeterSquared / missing value: NA)
45. fc.fill: net co2 flux gap filled with marginal distribution sampling (mg CO2 m-2 s-1) (unit: milligramPerMeterSquaredPerSecond / missing value: NA)
46. gep.fill: gep R05, gap filled with marginal distribution sampling (mg CO2 m-2 s-1) (unit: milligramPerMeterSquaredPerSecond / missing value: NA)
47. reco.fill: reco R05, gap filled with marginal distribution sampling (mg CO2 m-2 s-1) (unit: milligramPerMeterSquaredPerSecond / missing value: NA)
48. qc.flag: quality control flag for gap filling
    • 1: high confidence
    • 2: medium confidence
    • 3: low confidence