Carbon dioxide is the second most important anthropogenic greenhouse gas contributing to climate change. Scientists believe the key to predicting climate change is in understanding the relationship of green house gases and their impact on the environment. Whilst more than half of CO2 emitted is removed from the atmosphere within a century, more than 20 % remains for thousands of years.
In this paper, non-dispersive [NDIR] gas analysers, the eddy covariance technique and remote sensing are discussed as methods of continuous CO2 monitoring. To demonstrate the importance of data collected in studying the environment, R software was used to plot and analyse data obtained from CDIAC website for Jubany Scientific Station [NDIR method]. The paper concludes that the concentrations of CO2 increased at the rate of 0.16 ppm month-1 extrapolated to 1.92 ppm yr-1 for the period under study. The minimum and maximum values were 355.52 and 385.94 ppm, respectively with a mean of 370.73 ppm.
Keywords: CO2, gas analysers, instruments, monitoring, R software, trends
Introduction and Literature Review
Carbon dioxide [CO2] is a major greenhouse gases contributing to climate change (CDIAC, 2010). It is produced naturally from respiration and volcanic action. However, anthropogenic sources include combustion of fossil fuels, changes in land use [e.g. forest clearing] as well as industrial processes such as cement manufacture (CDIAC, 2010).
Over the last three hundred years, the gas has increased by over 35% (CDIAC, 2010) with the global average at 386.3 ppm in 2009 (Tans, 2009). This has been accompanied by an increase in global temperature due to the ability of green house gases to absorb solar and outgoing terrestrial radiation (Baldocchi et al., 1996).
Increases in greenhouse gas concentrations have also been attributed to the imbalance between the rate of emission into, and absorption out of, the system (Baldocchi, et al., 1996). Whilst more than half of CO2 emitted is removed within a century, more than 20 % remains for thousands of years (NASA, 2009). Atmospheric concentrations also change with time and with weather events (SLCO2, 2010). However, the relationship between greenhouse gases and impact on climate change remains uncertain (Brounshtein et al., 1991) but scientists believe the key to predicting climate change is in understanding this relationship (AGU, 2009). Although there are extensive surface networks for monitoring CO2, there is a dearth of data in the southern hemisphere because sensors are placed far from emitting sources like urban areas (Abshire et al., 2003).
1.1 Instrument Characteristics
There are many methods of monitoring the levels of atmospheric CO2. They take advantage of the absorption of infrared by CO2 molecules (SLCO2, 2010). The current study discusses methods of monitoring CO2, the characteristics of the instruments used with a discussion on the considerations in using them.
Non- Dispersive Infrared Gas Analysers [Closed-path]
LI -840 [for example] is a closed path, non-dispersive infrared [NDIR] gas analyser (Licor, 2001) that measure CO2 concentration of a locality in profile. The basic principle of the analyser is that a beam of infrared radiation is passed in an optical path where the air sample of interest is. An infrared radiation detector measures the quantity of infrared radiation reaching it after interaction with the gas in the path. The sample cell uses an optical filter at the absorption band of CO2, at 4.26 micrometers [µm]. The reference cell has an optical filter at 3.95 µm, with no absorption due to CO2 (Licor, 2001). The greater the quantity of carbon dioxide within the path, the more infrared absorption and the less light detected (NOAA, 2008). The reference cell is used to calibrate the instrument. Figure 1 shows a schematic of a non-dispersive infrared gas analyzer.
Figure 1: Schematic of Non-Dispersive Infrared Gas Analyzer (Licor, 2001)
The instrument uses digital single processing techniques to determine temperature and pressure corrected CO2 concentrations using a radiometric computation. The detector signal is measured in millivolts [mV] and this is then passes through a 5th order polynomial that linearizes the signal to mole fraction [µmol mol-1] measurement (NOAA, 2008). Equation 1 summarises the conversion (Licor, 2001):
(1)
where, is the 5th order polynomial, is the pressure correction absorptance of CO2, is the band broadening correction for water vapour and is the temperature [°C].
Closed-path analysers require higher energy consumption than would be required for an open-path analyser (Falk, 2004). The extra power is used to power a pump that lets the gas into the sample chamber. The pressure transducer minimises fluctuations in flow with barometric pressure changes (Licor, 2001). Though user cleanable; non-abrasive cleaners should be used to maintain the gold plated optical path and reflectors key in increasing energy transmission (Licor, 2001).
Eddy Covariance [EC] Technique
The EC technique measures CO2 density fluxes in milligrams per meter squared per second [mg m-2 s-1] (Burba and Anderson, 2008). The method works on the principle that the atmosphere contains turbulents of upward and downward moving air containing trace gases such as CO2 (Falk, 2004). The EC technique uses a gas analyser in conjunction with a sonic anemometer and relates the vertical flux of CO2 as being proportional to the mean covariance of vertical wind velocity and the scalar fluctuations (Baldocchi, et al., 1996). This is represented by Equation 2:
(2)
where is CO2 flux, is wind speed and is the density of CO2.
In the open-path configuration, the gas analyser and sonic anemometer measurements are taken simultaneously. This makes certain that there is no time lag between wind velocity and scalar fluctuations (Baldocchi, et al., 1996). This method measures the CO2 density instead of the mixing ratio, and therefore requires simultaneous measurement of humidity and temperature changes (Baldocchi, et al., 1996).
This configuration is best for remote stations with unreliable power supply. An important consideration is that the sensor head is sensitive to rainfall or dew (Licor, 2004). The gas analyser is therefore installed at a slight angle [10-15°] from vertical to allow for water run-off; and to ensure least obstruction to wind over a range of azimuthal directions (Falk, 2004). It is also mounted downwind of the sonic anemometer (Licor, 2004).
Reference gases [of known concentration, stored under pressure] are, periodically flowed through the analyzer for calibration. The reference gases register a voltage that can then be related to the CO2 mole fraction (NOAA, 2008).
In the closed-path configuration as the air sample is pumped through the chamber [at 1 l min-1], there is a time-lag of the gas against time series due to gas travelling through the path and internal electronics (Baldocchi et al., 1996). Time-lag, temperature and pressure changes in the analyser's chamber have to be corrected (Falk, 2004). The sonic anemometer overheating may also lead to error in the reading (Burba and Anderson, 2008). The closed-path configuration is deployed, preferably in areas where there is undisrupted supply of power (Baldocchi et al., 1996).
An important disadvantage is that EC works best in horizontal, uniform terrain with extended upwind vegetation (Baldocchi, 2003). It is only now being developed for complex terrain, like hills and urban areas (Burba and Anderson, 2008). These types of terrain introduce different eddy measurements that limit the spatial extent of the flux measurement (Burba and Anderson, 2008). In spite of the challenges, the EC technique has emerged as an important tool for studying eco-system physiology (Baldocchi, 2003).
Remote Sensing [Lidar]
Since the 1960s, instruments put on satellites to monitor atmosphere and surface from orbit have been passive imagers and sounders operating in the visible, infrared, and microwave spectral regions (Winkler et al., 1996). The principle of the active system is that a single wavelength Light Detecting and Ranging [Lidar] system emits short laser pulses into the atmosphere and measures the power of the light backscattered into its receiver as a function of distance (Pei-Tao et al., 2006). This allows measurements in profile [observing a spectral line at multiple frequencies of varying optical depth] regardless of time of day or season (Riris et al., 2008).
The instrument measures CO2 near 1.57 µm [free from water vapour interference] and oxygen [O2] near 770 nm (Abshire et al., 2003). The O2 [and sometimes nitrogen] measurement provides the reference signal due to its stable mixing ratio in the atmosphere (Pei-Tao et al., 2006). Pressure broadening provides enhanced sensitivity to lower altitudes (Abshire et al., 2003). A third measurement at 1.064 µm is taken to detect and correct for aerosol and cloud scattering (Abshire et al., 2003). Absorption units [%] are used in measurement. Equation 3 is the simple algorithm used to retrieve real-time CO2 concentrations data (Abshire et al., 2003):
(3)
where K is a calibration constant (determined by making a one-time independent CO2 measurement), SON/SOFF is the measured transmission ratio (ON at 1572.335 nm, OFF at 1572.260 nm), and N is the number of measurements over the spectral line.
In using the laser based method, the diurnal variation of CO2 concentrations, the atmospheric and spectroscopy error [e.g. temperature and aerosol/cloud scattering] as well as systematic and random instrument error (Riris et al., 2008) are taken into consideration. The instrument is insensitive to surface pressure resulting from topographic changes and humidity (Abshire et al., 2003), providing data at high spatial and temporal resolution (Pei-Tao et al., 2006).
The Lidar system can be calibrated in the laboratory or by airborne missions. In the laboratory, it requires the simulation of various earth surfaces and detecting backscatter measurements (Jarzembski and Srivastava, 1998). The ASCENDS satellite is in the proposal phase of its life with hopes of launch of a mission by 2013 (NASA, 2010a). However, the laser based technique has been tested on aircraft and has worked well giving data close to ones from in-situ measurements (Riris et al., 2008).
1.2 Instrument Key Features
Table I shows the key features of the different instruments discussed. The table compares the methods on coverage, power consumption, calibration requirements, accuracy, precision, and cost.
Table I: Instrument Key Features
The accuracy and precision of collected data depends on the instrument used. The more expensive the instrument is, the more accurate the data. These factors are therefore major considerations when procuring equipment for deployment.
Methodology
The mean monthly data [1994-2009] for Jubany Scientific Station [Antarctica] was obtained from the CDIAC website (2010). The station uses the non-dispersive infrared gas analyser method to collect continuous CO2 concentration data (CDAIC, 2010). The analyser is equipped with a serial interface running on computer software and is automated to recalibrate every 3 hours, achieving an accuracy of ± 0.1 ppm (CDAIC, 2010).
In analysing, a column for monthly time-steps was added to allow for plotting the multivariate data obtained in R (R Development Core Team, 2010). Missing data [given a value of -99.99 originally] were re-assigned a value of NA and omitted from the analysis. This is because outliers tend to influence the results of analysis (Perez, 2004).
Some descriptive statistical analyses were performed to determine the extremes [maximum, minimum] from the date they occurred, using the which command in R. From these, the mean value could then be calculated for comparison with the global trends. A regression line was fitted to the data in order to establish the rate of change of CO2 concentration per month (Perez, 2004). The script of the R programme is attached as an appendix to this paper.
3. Results and Discussion
The raw data obtained was used to plot the time series of carbon dioxide concentrations over the study period. Figure 2 shows the graph obtained.
Figure 2: Monthly Average CO2 Concentrations for Jubany Station [1994- 2009]
The missing values are represented by the gaps in the graph. The minimum value occurred in March 1994 [355.52 ppm] and the maximum value in October 2009 [385.94 ppm], giving a mean of 370.73 ppm, which is lower than the global average of 386.3 ppm (Tans, 2009).
The regression line [red] has slope of 0.16 which corresponds with the rate of change of CO2 per month. This is extrapolated to an average increase of 1.92 ppm per year over the study period. In practice however, the actual rate of change from year to year is variable and may be caused by anomalies [e.g. the 1997-98 El Niño and la Niña events] or atmospheric circulation (CDIAC, 2010). Differences in CO2 concentrations could also be attributed to different interactions with plants cover in the different years (Padin, et al., 2007). According to NOAA (2010b), the rate of growth of CO2 concentrations has averaged about 1.91 ppm per year after 1995.
Conclusions
The paper has discusses three instruments for monitoring CO2 concentration levels by highlighting that the accuracy of data has cost implications.
From the study, the maximum and minimum values of were 385.94 and 355.52 ppm, respectively; the mean value was 370.73. The rate change of CO2 concentration was 0.16 ppm month-1. This suggests an average increase of 1.92 ppm year-1 over the study period; a result in line with the figure quoted by NOAA.
