You are here

Harvard Forest Data Archive


EMS - Automated High-Frequency Methane Data

Related Publications



  • Lead: Patrick Crill
  • Investigators:
  • Contact: Information Manager
  • Start date: 1991
  • End date: 1994
  • Status: completed
  • Location: Prospect Hill Tract (Harvard Forest)
  • Latitude: +42.54
  • Longitude: -72.17
  • Elevation: 340 meter
  • Taxa:
  • Release date: 1999
  • Revisions:
  • EML file: knb-lter-hfr.60.18
  • DOI: digital object identifier
  • EDI: data package
  • DataONE: data package
  • Related links:
  • Study type: short-term measurement
  • Research topic: forest-atmosphere exchange
  • LTER core area: inorganic nutrients
  • Keywords: gas flux, methane
  • Abstract:

    Automated high-frequency methane (CH4) in ambient air measurements were made at the Harvard Forest (HF) research site since 1992. The proximity and the relative location of the site to numerous industrial/urban areas presents the opportunity to sample air flows that have been influenced by known CH4 sources on a regular and repeatable basis and to characterize the atmospheric chemical signature of the sampling location and assess its sensitivity to both local and regional sources at different time scales.

  • Methods:

    The automated CH4 analysis system was designed and maintained by Patrick M. Crill then at the University of New Hampshire, Durham (now at Dept of Geological Sciences, Stockholm University).

    The system is built around a Shimadzu Mini-2 gas chromatograph equipped with a flame ionization detector. Ambien air was continuously drawn through an inlet 20 m above the ground surface (a few meters above the forest canopy) through 1/4" OD plastic-coated aluminum tubing (Dekoron 041943-1 tubing) with an electric diaphragm pump at a rate of 25 L/min. This flow was sampled every 8-11 min by diverting flow with a 1/4” Valco stream select valve and dried across a 1m Perma Pure nation drier to rinse and fill a 1ml sample loop on a second 1/8” Valco 6- port injection valve. After five second pause to allow time for sample loop to relax to atmospheric pressure 1 mL of sample was is injected into the carrier gas stream of the chromatograph. The procedure was repeated with a standard gas after sufficient time (3 min) for the sample to elute through the detector. Every sample was followed by a standard in order to compensate for ambient pressure and temperature changes as well as instrument variation. The sampling procedure yielded an independent measurement of ambient CH4 every 8-11min.

    CH4 was separated from ambient air on a 2 m by 3.2 mm OD stainless steel column packed with HayeSepQ at 40o C. Detector temperature was 125o C. Valve timing, analog to digital conversion of the detector output and signal integration were controlled by a Hewlett- Packard HP 3395A series II integrator with an HP19405A event controller. Raw data and integration reports were transferred and logged on a personal computer until the data were transferred to the University of New Hampshire via modem every3 days.

    During operation a field calibration sample was analyzed with each ambient air sample. The field standards were cylinders of hospital grade breathing air containing near ambient mixing ratios of CH4. The field standards were calibrated by comparing them to Niwot Ridge air standards prepared by NOAA/CMDL in Boulder, Colorado before and after deployment to Harvard Forest. Over the course of a day, the standard response of the gas chromatograph varied about 1% due to diel heating and cooling. The standards were kept in an unheated pump room next to the instrument room in order to mimic the temperature of the outside air sample. Therefore, to evaluate the precision of the CH4 measurements, the responses of an individual analysis of the standards are compared to the 24-min running average of the standard response. The coefficient of variation (CV) (the standard deviation/mean) expressed as the percent variation from the 24-minute running mean ranged annually from 0.18 to 0.29%. If this CV represents the sum of errors in the system measurement (an assumption not entirely true since the standards do not f low through the sampling pump), then the precision of the analysis is considered to be 3.6 ± 0.4 ppbv CH4 . Additionally, all data points were filtered for instrument response by calculating the difference between each standard response and its average with the two points on either side. This difference was then divided by the sample standard deviation for these same five points. The resulting number was then compared with the mixing ratios obtained by the Grubbs test for outliers at a significance level of 95% [Sokal and Rohlf, 1981] and removed as an outlier if it exceeded that value. No data were removed in 1992 and 1993, 1.4% were removed in 1994, and 1.55% were removed in 1995.

  • 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.

  • Citation:

    Crill P. 1999. EMS - Automated High-Frequency Methane Data. Harvard Forest Data Archive: HF060 (v.18). Environmental Data Initiative:

Detailed Metadata

hf060-01: methane data hourly

  1. datetime: date and time
  2. year: year
  3. doy: Julian day (unit: nominalDay )
  4. methane: hourly averaged methane value (ppmv) (unit: dimensionless / missing value: NA)

hf060-02: methane data

  1. datetime: date and time
  2. month: month
  3. year: year
  4. time: time
  5. doy: Julian day (unit: nominalDay )
  6. ch4: methane value (ppm) (unit: dimensionless / missing value: NA)