7.2.1 Atmospheric CO₂ observation

   Atmospheric CO₂ is observed using a non-dispersive infrared analyzer (NDIR) which makes use of the infrared absorption characteristics of CO₂. The system models used at the observation stations are listed in Tables 7.1.1.2, 7.1.1.3, and 7.1.1.4, and the system is depicted in Figure 7.2.1.1. The stability of the observation system does not exceed 0.02 ppm, which is expressed as a 15 minute-average standard deviation; and its repeatability does not exceed ±0.02 ppm, which is the difference of the results of 2 runs of the above stability test.

   Air samples are collected through the intake at a ground height of about 20 m. The air sample is dried to a dew point temperature not exceeding -50°C, and then is introduced to the observation system. Four (Minamitorishima) to five (Ryori and Yonagunijima) working standard gases of CO₂-in-natural-air mixtures in different concentrations are used for CO₂ observation. These working standard gases are analyzed to determine the concentration accurately using a CO₂ calibration system at the JMA headquarters with reference to JMA's standard gases before use at the station. The filling pressures of the working standard gases are typically around 15 MPa (about 12 MPa at Ryori) at the beginning of use. Too low a pressure may change the concentration (cause a drift) due to an interaction with the cylinder wall (due to drift occurring), and therefore use of the gases stops before the gas pressure goes down to below 3.5 MPa. Normally, the use period is from about 3 months (Ryori and Minamitorishima) to 6 months (Yonagunijima). After use, working standard gases are analyzed at the JMA headquarters again with reference to JMA's standard gases. By comparing the concentrations before and after use to determine precisely the drift during the use, the concentrations during the use can be corrected.

   The flow rate of air sample to the analyzer is 500 mL per minute (250 mL per minute at Yonagunijima). As described in Section 7.3.1, JMA's standard gases are calibrated against the WMO reference standard gases maintained by NOAA, which allows the direct comparison of concentration data observed at the stations with those observed at any location in other countries where the WMO reference standard gases are used. Observation is carried out by automatically repeating a 120-minute measurement cycle, which consists of the measurement of the concentrations of working standard gases in 4 to 5 cylinders for 3 to 6 minutes per cylinder, and the measurement of the concentrations of CO₂ in air samples for 90 to 105 minutes. For the measurement of the working standard gases and air samples, the output voltages of the analyzer are measured every second and are averaged over 30 seconds. A quadratic calibration curve, which expresses the relationship between the output voltage and concentration, is determined for every measurement cycle based on the known concentrations of the working standard gases and the corresponding output voltages, and the 30-second averaged output voltages are converted to concentrations using the calibration curve. To minimize errors caused by the temporal variations in the response of the analyzer, two concentrations are calculated using the two calibration curves immediately before and after the measurement of air samples, and these two concentrations are time-interpolated to obtain 30-second averaged concentrations of air samples.

   Large variations in CO₂ concentration are observed even within the order of several hours. This is because local variations in concentration due to anthropogenic or vegetation activities within the surface boundary layer are measured in some weather conditions. In order to obtain atmospheric BG concentrations that represent a horizontally wide space and are thus considered to be the average values within the surface boundary layer, it is necessary to select data without such local variations. An air mass which is affected by local variations is considered to have a larger temporal variation due to the smaller spatial scale. It is desirable to select as many data as possible to obtain highly reliable BG values while removing concentration data like above wherever possible. The BG data selection procedures operated by JMA are as follows:

1. Calculate the hourly mean for all the observations excluding those during inspections or malfunction of the system, and other such situations.

2. If the observed 30-second averaged values forming an hourly mean value are less than half the total number of values that are supposed to be observed, reject that hourly mean value.

3. If the standard deviation of the observed values forming an hourly mean value exceeds a prescribed threshold, then reject that hourly mean value.

4. If an hourly mean value is different from both of the immediately following and preceding values by more than a prescribed threshold of continuity, reject that hourly mean value.

5. Hourly mean values obtained as a result of the above procedure are regarded as BG data.

6. Daily mean values are calculated from hourly values that were selected as BG data.

7. Monthly mean values are calculated from hourly values that were selected as BG data.

8. Annual mean values are calculated from monthly values.

   The thresholds mentioned above are determined by evaluating the past observation data on an observation location basis so as to exclude concentration data that are affected by local factors, while making use of as many as BG data possible. The prescribed thresholds for CO₂ are listed in Table 7.2.1.1.