This file describes the following data set: ATMOSPHERIC METHANE AT CAPE MEARES, OREGON, U.S.A.: A HIGH-RESOLUTION DATA BASE FOR THE PERIOD 1979-1992. CDIAC DB-1007 M.A.K. Khalil and R. A. Rasmussen Oregon Graduate Institute of Science and Technology Portland, Oregon 97291 U.S.A. DOI: 10.3334/CDIAC/atg.db1007 CONTENTS I. Introduction II. Methods III. File Descriptions "cmeares.all" (all individual methane measurements at Cape Meares) "cmeares.day" (daily averages of methane measurements at Cape Meares) "cmeares.mon" (monthly averages of methane measurements at Cape Meares) IV. Data checks performed by CDIAC V. References I. INTRODUCTION The atmospheric monitoring facility at Cape Meares, Oregon, U.S.A., is operated by the Oregon Graduate Institute of Science and Technology and was the first station in the world to carry out continuous automated measurements of atmospheric methane. The location (45 degrees 29 minutes N, 123 degrees 58 minutes W) is representative for atmospheric methane concentrations in the northern latitudes (from 30 degrees N to 90 degrees N). The site is isolated from populated areas by the presence of a coastal mountain range, and due to the dominance of westerly winds, is influenced by largely unpolluted Pacific air masses. These data collected at Cape Meares represent nearly 119,000 individual atmospheric methane measurements carried out during 1979-92. Daily and monthly averages of the data are also presented. The data are contained in three files: (1) "cmeares.all" (a 2.14 MB data file containing all individual atmospheric methane measurements, carried out 12 to 72 times daily during 1979-92 at Cape Meares); (2) "cmeares.day" (a 189 kB data file containing daily averages of the data that were presented in the file "cmeares.all"); (3) "cmeares.mon" (a 10 kB data file containing monthly averages of the data that were presented in the files "cmeares.day" and "cmeares.all"). II. METHODS Analysis of ambient air was carried out by means of an automated sampling and measurement system. At regular intervals, air samples were pumped into the apparatus, dried to a dew point of -30 C (<1% relative humidity) by a Nafion dryer, and injected onto the column of a gas chromatograph (either a Carle 211-MS or a Hewlett-Packard unit) equipped with a flame ionization detector. Chromatographic peak areas were determined by means of an electronic integrator. Methane concentrations were then calculated from the integrated peak areas, by comparison with peaks of a standard reference gas run before and after the ambient sample. At the beginning of the experiment (January 1979), 72 measurements of ambient air were taken per day. This frequency was reduced to 24/day in 1984, and to 12/day in 1986. Data were discarded for periods when the instruments were found to be malfunctioning and for a few cases where the values were judged to be outliers. Sufficient data were available to calculate monthly averages for all months over the period 1979-1991. For most months, sampling was carried out on 80-90% of days. During winter months, however, storms occasionally caused power outages and suspended station operation, sometimes for several weeks at a time. The primary calibration standards consisted of precisely calibrated air stored in high-pressure stainless steel tanks. The continuous measurement procedure also employed secondary calibration standards that were used at the monitoring site. During the course of the record, several primary and many secondary calibration tanks were used, but all may be linked to a single absolute calibration standard. The overall precision of the measurements, including instrumental, atmospheric, and sampling variabilities, was estimated from the variability of the observed daily concentrations. The standard deviation of the individual measurements from a given day is, on average, about 10 parts per billion by volume (ppbv), and 90% of the standard deviations are between 4 and 21 ppbv. More detailed descriptions of the analytical system and experimental methodology are given by Khalil et al. (1993) and by Rasmussen and Khalil (1981). III. FILE DESCRIPTIONS "cmeares.all" This is a 2.14 MB data file containing all individual atmospheric methane measurements, carried out 12 to 72 times daily from January 1979 to January 1992 at Cape Meares. Each data record presents two variables, time and atmospheric methane concentration. The file can be read by using the following FORTRAN code: C FORTRAN data retrieval routine to read the file named "cmeares.all". C C Unit 1 is input. C REAL TIME, CH4 OPEN (UNIT=1, FILE='cmeares.all') READ (1, 1) TIME, CH4 1 FORMAT (18(/)F10.7,1X,F6.1) 10 READ (1, 2, END=99) TIME, CH4 2 FORMAT (F10.7,1X,F6.1) GO TO 10 99 STOP END The data can also be read by using the following SAS code: * SAS data retrieval routine to read the file named "cmeares.all"; *; DATA ALL; INFILE 'cmeares.all'; IF _N_=1 THEN INPUT #19 @1 TIME 10.7 @12 CH4 6.1; ELSE INPUT @1 TIME 10.7 @12 CH4 6.1; RUN; Stated in tabular form, the contents include the following: __________________________________________________________ Variable Variable Starting Ending Variable type width column column __________________________________________________________ TIME Numeric 10 1 10 CH4 Numeric 6 12 17 __________________________________________________________ where TIME is the decimal year. The integer portion is the final two digits of the year (since 1900) in which the measurement was taken. The decimal portion is the fractional year, defined as: ((day of year - 1)+(hour of day / 24)+(minute of hour / 1440)) / total number of days in year. CH4 is the atmospheric methane concentration in parts per billion by volume (ppbv). "cmeares.day" This a 189 kB data file containing daily averages of the atmospheric methane data presented in the file "cmeares.all". The data are available for the period February 3, 1979 to January 1, 1992. In addition to the date (i.e., year, month, and day) and the daily average concentration of methane, each data record presents the standard deviation, the number of samples obtained during that day, and the number of days since the beginning of the record. There is also a column that contains interpolations for the days when data were not taken. The interpolation scheme is based on a parametric model of the data as discussed in detail by Khalil et al. (1993). The file can be read by using the following FORTRAN code: C FORTRAN data retrieval routine to read the file named "cmeares.day". C C Unit 1 is input. C INTEGER MONTH, DAY, YEAR, EXPDAY, N REAL CH4, INTERP, SD OPEN (UNIT=1, FILE='cmeares.day') READ (1, 1) MONTH, DAY, YEAR, EXPDAY, CH4, INTERP, SD, N 1 FORMAT (18(/)I2,1X,I2,1X,I2,1X,I4,2X,F6.1,2X,F6.1,2X,F4.1,2X,I2) 10 READ (1, 2, END=99) MONTH, DAY, YEAR, EXPDAY, CH4, INTERP, SD, N 2 FORMAT (I2,1X,I2,1X,I2,1X,I4,2X,F6.1,2X,F6.1,2X,F4.1,2X,I2) GO TO 10 99 STOP END The data can also be read by using the following SAS code: * SAS data retrieval routine to read the file named "cmeares.day"; *; DATA DAY; INFILE 'cmeares.day'; IF _N_=1 THEN INPUT #19 MONTH 1-2 DAY 4-5 YEAR 7-8 EXPDAY 10-13 @16 CH4 6.1 @24 INTERP 6.1 @32 SD 4.1 N 38-39; ELSE INPUT MONTH 1-2 DAY 4-5 YEAR 7-8 EXPDAY 10-13 @16 CH4 6.1 @24 INTERP 6.1 @32 SD 4.1 N 38-39; RUN; Stated in tabular form, the contents include the following: __________________________________________________________ Variable Variable Starting Ending Variable type width column column __________________________________________________________ MONTH Numeric 2 1 2 DAY Numeric 2 4 5 YEAR Numeric 2 7 8 EXPDAY Numeric 4 10 13 CH4 Numeric 6 16 21 INTERP Numeric 6 24 29 SD Numeric 4 32 35 N Numeric 2 38 39 __________________________________________________________ where MONTH is the numeric month of the year (i.e., 1=January, 2=February, etc.) in which the measurements were taken. DAY is the numeric day of the month for which the average of the individual measurements (taken from 0000 to 2359, local time) is presented. YEAR is the final two digits of the year (since 1900) in which the measurements were taken. EXPDAY is the total number of days elapsed since the beginning of methane measurements at Cape Meares. Values range from 32 to 4747. CH4 is the daily mean atmospheric methane concentration in parts per billion by volume (ppbv). Days on which no methane measurements were made are assigned missing values, denoted as -999.9. INTERP is the interpolated daily mean atmospheric methane concentration (given in ppbv), calculated for the days when measurements were not taken. The interpolation scheme is based on a parametric model of the data as discussed in detail by Khalil et al. (1993). For days on which methane measurements were made, values for INTERP and CH4 are identical. SD is the standard deviation among the individual measurements used to calculate the daily mean. Days on which no methane measurements were made are assigned missing values, denoted as -9.9. N is the number of individual measurements used to compute the daily mean and the standard deviation. Days on which no methane measurements were made are assigned missing values, denoted as -9. "cmeares.mon" This a 10 kB data file containing monthly averages of the atmospheric methane data presented in the files "cmeares.all" and "cmeares.day". The data are available for all months from February 1979 through December 1991. Each data record presents the year, month, monthly mean and 90% confidence interval, monthly median and associated 5th and 95th percentiles, and an additional monthly average and 90% confidence interval calculated by a non-parametric statistical method. The file can be read by using the following FORTRAN code: C FORTRAN data retrieval routine to read the file named "cmeares.mon". C C Unit 1 is input. C INTEGER YEAR, MONTH REAL MEAN, CONF90, MEDIAN, PCT05, PCT95, NONPARA, NP90 OPEN (UNIT=1, FILE='cmeares.mon') READ (1, 1) YEAR, MONTH, MEAN, CONF90, MEDIAN, PCT05, PCT95, 2 NONPARA, NP90 1 FORMAT (18(/)I2,1X,I2,3X,F6.1,2X,F3.1,3X,F6.1,2X,F6.1,2X,F6.1,3X, 2 F6.1,2X,F4.1) 10 READ (1, 2, END=99) YEAR, MONTH, MEAN, CONF90, MEDIAN, PCT05, 2 PCT95, NONPARA, NP90 2 FORMAT (I2,1X,I2,3X,F6.1,2X,F3.1,3X,F6.1,2X,F6.1,2X,F6.1,3X, 2 F6.1,2X,F4.1) GO TO 10 99 STOP END The data can also be read by using the following SAS code: * SAS data retrieval routine to read the file named "cmeares.mon"; *; DATA MON; INFILE 'cmeares.mon'; IF _N_=1 THEN INPUT #19 YEAR 1-2 MONTH 4-5 @9 MEAN 6.1 @17 CONF90 3.1 @23 MEDIAN 6.1 @31 PCT05 6.1 @39 PCT95 6.1 @48 NONPARA 6.1 @56 NP90 4.1; ELSE INPUT YEAR 1-2 MONTH 4-5 @9 MEAN 6.1 @17 CONF90 3.1 @23 MEDIAN 6.1 @31 PCT05 6.1 @39 PCT95 6.1 @48 NONPARA 6.1 @56 NP90 4.1; RUN; Stated in tabular form, the contents include the following: __________________________________________________________ Variable Variable Starting Ending Variable type width column column __________________________________________________________ YEAR Numeric 2 1 2 MONTH Numeric 2 4 5 MEAN Numeric 6 9 14 CONF90 Numeric 3 17 19 MEDIAN Numeric 6 23 28 PCT05 Numeric 6 31 36 PCT95 Numeric 6 39 44 NONPARA Numeric 6 48 53 NP90 Numeric 4 56 59 __________________________________________________________ where YEAR is the final two digits of the year (since 1900) in which the measurements were taken. MONTH is the numeric month of the year (i.e., 1=January, 2=February, etc.) for which averages are presented. MEAN is the monthly mean atmospheric methane concentration in parts per billion by volume (ppbv). Values represent arithmetic means calculated from daily averages presented in the file "cmeares.day." Only measured values (i.e., non-missing values for the variable CH4 in the file "cmeares.day") were used in calculating the means: interpolated values were excluded. CONF90 is the quantity (also in parts per billion by volume) which, when added to or subtracted from the value for MEAN, gives the upper and lower limits of the 90% confidence interval of the mean. MEDIAN is the monthly median atmospheric methane concentration in parts per billion by volume (ppbv). Values are determined by ranking all individual CH4 measurements for a given month (from the file "cmeares.all") from lowest to highest and selecting the middle (50th percentile) value. PCT05 is the fifth percentile monthly methane concentration, given in parts per billion by volume (ppbv). Fifth percentile values are determined following a ranking of all individual CH4 measurements for a given month (from the file "cmeares.all") from lowest to highest. PCT95 is the 95th percentile monthly methane concentration, given in parts per billion by volume (ppbv). Ninety-fifth percentile values are determined following a ranking of all individual CH4 measurements for a given month (from the file "cmeares.all") from lowest to highest. NONPARA is a nonparametric estimate of the monthly average methane concentration, given in parts per billion by volume (ppbv). Details concerning the calculation are given by Khalil and Rasmussen (1993). NP90 is the quantity (also in parts per billion by volume) which, when added to or subtracted from the value for NONPARA, gives the upper and lower limits of the 90% confidence interval for the nonparametric estimate. IV. DATA CHECKS PERFORMED BY CDIAC The Carbon Dioxide Information Analysis Center (CDIAC) endeavors to provide quality assurance (QA) of all data before their distribution. To ensure the highest possible quality in the data, CDIAC conducts extensive reviews for reasonableness, accuracy, completeness, and consistency of form. While having common objectives, the specific form of these reviews must be tailored to each data set; this tailoring process may involve considerable programming efforts. The entire QA process is an important part of CDIAC's effort to assure accurate, usable data for researchers. The following summarizes the QA checks performed on the Cape Meares atmospheric methane data by CDIAC. 1. The format of all information was checked to ensure consistency throughout each data file. 2. Numeric data values were examined graphically and mathematically for reasonableness, absence of typographical errors, and absence of outliers. 3. Derived data such as daily means, monthly means, and monthly medians were checked by comparing the values received from the Oregon Graduate Institute of Science and Technology with values generated mathematically from the primary data (i.e., the individual methane measurements). 4. When possible, daily means were compared with measurements derived from flask air samples collected at Cape Meares on the same day by other investigators. A minor error was found. The date assigned to the daily numbers was shifted by two days. The correction of this error led to slightly different monthly averages from those reported in the original paper by Khalil et al. (1993). The differences of monthly averages were in most cases less than 2 ppbv or 0.1% and about the same order as the range of errors introduced by interpolation of missing data. The present monthly data distributed in this database therefore replace the original table in Khalil et al. (1993). The changes, however, do not in any way alter the conclusions presented in the original paper. V. REFERENCES Khalil, M.A.K., R.A. Rasmussen, and F. Moraes. 1993. Atmospheric methane at Cape Meares: Analysis of a high resolution data base and its environmental implications. Journal of Geophysical Research 98:14,753-14,770. Khalil, M.A.K., and R.A. Rasmussen. 1992. Atmospheric methane at Cape Meares, Oregon. Rep. 92-111. Oregon Graduate Institute, Beaverton, Oregon. Khalil, M.A.K., and R.A. Rasmussen. 1990. Atmospheric methane: Recent global trends. Environmental Science and Technology 24:549-553. Khalil, M.A.K., and R.A. Rasmussen. 1983. Sources, sinks, and seasonal cycles of atmospheric methane. Journal of Geophysical Research 88:5131-5144. Rasmussen, R.A., and M.A.K. Khalil. 1981. Atmospheric methane: Trends and seasonal cycles. Journal of Geophysical Research 86:9826-9832.