Storage flux is by definition the accumulation or depletion of carbon dioxide [CO2] in the air layer below the eddy covariance flux sensor. It is important as it is added to the measured flux in order to give an accurate picture of the net ecosystem CO2 exchange. Storage is also important in validating carbon sequestration estimates and in assessing the possible role of efforts in offsetting fossil fuel emissions.
A closed path gas analyser measuring CO2 concentrations in profile was setup on the Kings College Strand building in central London. CO2 concentration data was despiked by using the standard deviation from the hourly means to set threshold values for the data. The 30-minute averages from the different heights were then integrated to calculate the storage flux. A mean value of -1.66 X 10-2 mg CO2 m-2 s-1 for the period under study and is therefore negligible for NEE calculations.
Get your grade
or your money back
using our Essay Writing Service!
Future studies would benefit from improved equipment set-up and inclusion of wind speed data to enable more robust analysis.
Keywords: Despiking, eddy covariance, integration, Li840, profile, storage flux
Introduction and Literature Review:
There is a disproportionate amount of carbon dioxide [CO2] emitted from burning fossil fuels, and changes in land use in urban (Matese et al., 2009). Variations of vegetation types in an area, photosynthetic activity, as well as fixed-point [industrial] and mobile emission [traffic] sources are important factors in urban CO2 fluxes (Grimmond et al., 2002). Differences in movement of air to measuring sites also affect diurnal and seasonal variability observed in the CO2 concentrations due to changes in mixing ratios (Bakwin et al., 1998). These factors can be measured by the mass conservation equation.
The mass conservation equation states that the CO2 produced or absorbed by the biological source/ sink is either stored in the air or removed by flux divergence in all directions (Aubinet et al., 2005). When stratification is strong and turbulence levels are low, particularly at night, the storage term assumes greater importance, often being the dominant term in the mass balance calculation (Finnigan, 2006).
According to Burba and Anderson (2010), net ecosystem CO2 exchange [NEE] is the sum of eddy covariance measured flux, Fc as well as the flux associated with accumulation or depletion in CO2 [Fs] in the layer below the level of CO2 flux measurement (Equation 1): -
where Fc is the above-canopy turbulent exchange of CO2 and Fs is change in CO2 storage in canopy air space [mg CO2 m-2 s-1].
CO2 flux is also important in the validation of carbon sequestration estimates and in assessing the possible role of efforts in off-setting fossil fuel emissions (Colls, 2007).
1.2 CO2 Storage Flux
Storage flux is, by definition, a non-turbulent flux that is obtained from the temporal changes in the integrated gas concentrations up to the height of the eddy covariance sensors (Suyker and Verma, 2001). Calculations are especially important during conditions of low winds, stable stratification, in high canopies, or in cases when air mixing is significantly reduced (Burba and Anderson, 2010). Measurement of gas concentrations in profile allows for the determination of such build-ups enabling calculation of the storage flux below the sensor height (Burba and Anderson, 2010; de Araújo et al., 2010).
The storage term is important as it is added to the eddy covariance flux to calculate the final NEE flux number (Burba and Anderson, 2010). If Fs is smaller than Fc and the daily calculated values tend towards zero, it is neglected in the calculation of NEE, but it becomes important when there is low wind speed, particularly around sunrise and sunset (Finnigan, 2006). Taken over a day, the storage term for most scalars of biological interest becomes negligible compared to the eddy flux term but when the aim is to resolve the diurnal cycle over periods like an hour, it can be significant. The storage term is represented by equation 2 (Baldocchi et al., 1996):
where z is the vertical height, c is the CO2 concentration, t is the time interval and zr is the flux measurement sensor height.
This paper demonstrates how to calculate storage flux of NEE. It outlines how to set-up the equipment and the analysis of data collected.
Always on Time
Marked to Standard
Taking CO2 profile measurements is not a simple task because of their large fluctuations (Mölder, et al., 2000). Differential measurements can be obtained by frequent sampling over a long period of time. Alternatively, the data is smoothed by using a mixing system between the tubing and analyser (Mölder, et al., 2000). New technology in the field means that sensitivities to water changes are greatly reduced due to the design of the analyser (Rannik et al., 2004).
2.1 Data Collection Methods
There are several methods of collecting data for storage calculations. For example, de Araújo et al. (2010) used a Gascard II IRGA to obtain simultaneous measurements at different heights and terrain. This method recognised that spatial heterogeneity makes it difficult to calculate storage accurately. Further, horizontal fluxes of gas (advection) may be significant in such terrain and violates neglecting that term in calculating NEE. Their use of the pressure sensors allowed for detection of leaks in the system which could be fixed immediately for reduced error in data collected. However, this set-up can lead to calibration mismatching, resulting from the use of different gas analysers and therefore requires constant IRGA intercalibration (Aubinet, et al., 2005).
Xu, et al. (1999) used a single Licor-6262 IRGA sampling in sequence with reduced internal volume connected to a large purge pump to minimize the time required to switch between air samples taken at different heights. They were also able to program the logger to delay recording as they had tested the response time before set up. However their method required a data logger with a large storage memory. The IRGA had span and drift problems, and had to be calibrated every 2 to 3 days in the laboratory.
The method used in this paper is a single IRGA [Licor-840]. The Licor-840 was used because it is stable over changing pressure and time and does not require frequent offset and span calibration (Licor, 2003). The instrument also has a wide operation range of 0- 3000 ppm and 20- 45 °C. The set-up allowed for the sampling at different heights in sufficiently short time to allow for off-setting the fluctuations in the CO2 measurements (Xu, et al., 1999).
2.2 Data Despiking
High frequency data occasionally has spikes due to both electronic noise and physical reasons. Removing these bad data is an important part of quality control process as the spikes can alter flux calculations and hence the results (Burba and Anderson, 2010). There are two types of spikes for IRGAs; 'hard' resulting from exceeding certain physical limits such as CO2 concentrations outside the range of 300- 3000 ppm and 'soft' spikes (Schmid et al., 2003). The soft spikes are the deviations from the hourly mean of the dataset given a threshold (Schmid et al., 2003). Statistical analysis is employed to despike the data.
2.3 Flux Calculations
Values divergent from the mean should be omitted, because they can influence the results (Perez, 2004). The hourly means are used for setting minimum and maximum thresholds to remove soft spikes observed in the data set. Data can then be integrated using the method described by Hossen et al. (2010), for 30-minute intervals. Data calculated for the storage can then be used to plot a graph.
2.4 Licor-840 Instrument
A closed-path, non-dispersive IRGA, the basic principle of the Licor-840 is that a beam of infrared radiation passes through an optical path containing the air sample of interest. Detectors measure the quantity of infrared radiation interacting with the gas in the path. A sample cell uses an optical filter at the absorption band of CO2, at 4.26 micrometer [µm]. A reference cell with an optical filter at 3.95 µm, with no absorption due to CO2 is used in calibrating the analyser (Licor, 2001). Figure 1 shows a schematic of an IRGA.
Figure 1: 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. It is a fast response and relatively inexpensive instrument that is stable over changes in pressure and time. It also does not require frequent span and zero adjustments (Licor, 2003). It can be setup for profile measurements, enabling calculation of storage flux.
This Essay is
a Student's Work
This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.Examples of our work
2.4.1 Instrument Siting and Calibration
Location and calibration of the instrument are key steps in data acquisition as they affect the quality of data. Located in central London, the roof of Kings College Strand building [KSS] is the site of the instrument setup. Highly urbanised with heavy traffic flowing through it, KSS roof represents a typical landscape of a built-up area. The building itself houses laboratories and cafes and has chillers and exhaust air being expelled via the roof. It has a 3.5 m high fence running around it providing partial shelter from winds. Profile measurements on a tower, however, cater for data collection from maximum fetch for the area (Burba and Anderson, 2010).
Licor-840 calibration is dependent on high-quality standard gases (de Araújo et al., 2010). It is important to enable conversion from units of measurement [millivolts] to unit of interest [mole fraction]. It is also done to verify factory calibrations on purchase of the instrument or to set narrower ranges that may improve resolution of data (Licor, 2001). Calibration is intended to reduce bias in an instrument's readings over a range of continuous values (NIST/ SEMATECH, 2010). Biased measurements can be corrected by using the inverse of the calibration curve obtained.
Instrument calibration was done in the lab before installation of the equipment. For offset calibration, compressed ultra pure nitrogen is used for both CO2 and water. The CO2 span calibration is obtained by using certified standard air, containing 805 parts per million of carbon dioxide (Micromet, 2010). For water span, an airstream of known dew point is fed through the calibration tube of a dew-point generator, Li610 for at least 15-20 minutes (Licor, 2004). The instrument is then calibrated and ready for use.
2.4.2 Air Sampling
Existing profile tubing on the KSS tower was used for the study. Air was sucked from the intakes at 1.78 m, 3.68 m, 6.25 m and 11.36 m, to the instrument through 4 mm plastic tubing. A CR1000 datalogger was programmed to log data from the different height. This was connected to a relay to control the electrical current in the system.
The Li840 was connected to a switcher changing every 2-3minutes between the different intake heights. Particulates from the air were excluded from the Licor by using small inlet filters with a very fine porous structure (Mölder, et al., 1999). Figure 2 shows the setup used [not to scale].
Figure 2: Instrument Set-up for CO2 Profile Measurement
Appendix I shows the detailed wiring diagram for the instrument set-up. In it, Roof 13 communicates with the main computer, Otago. This in turn grabs the profile and concentration data received and merges the two components to obtain the time series of the concentrations at the different heights (Micromet, 2010). Otago also stores the data using MySQL database software.
2.4.3 Data Analysis
High frequency data obtained from Professor Sue Grimmond's Urban Meteorology Group at King's College London for 21 January 2011 was processed using Excel software [Appendix II]. Concentration data are sorted by the different inlet heights. Hourly moving averages were obtained for setting thresholds. The 30-minute averages were then integrated to calculate storage flux.
There were no 'hard' spikes observed in the data. However, hourly average concentration data revealed 'soft' spikes. Figure 3 shows the graph of the hourly concentration data for the different heights.
Figure 3: Hourly Average CO2 Concentrations at KSS [21 January 2011]
The concentrations followed a similar trend through the day with peaks around 9 hours, before 14 hours and around 19 hours regardless of the height. This may be indicative of a well mixed air layer below the flux sensor (Bakwin et al., 1998). The peak concentration times correspond with high traffic volumes and coffee shop and laboratories business hours.
To deal with 'soft spikes', average thresholds were set at 394.39 - 416.88 ppm. Data lying below or above these thresholds were assigned a value of NA and disregarded in further analysis. The 30-minute averages were then integrated to calculate the storage flux (Figure 4).
Figure 4: Calculated Storage Flux at KSS [21 January 2011]
The minimum, maximum and mean storage flux values were; -7.01X 10-2 mg CO2 m-2 s-1, 8.98 x 10-2mg CO2 m-2 s-1 and -1.66 X 10-2 mg CO2 m-2 s-1, respectively, for the period under study. The data for dawn and dusk are important as the wind conditions are more conducive for storage and correspond with the onset of photosynthetic activity (Aubinet et al., 2005). However, the correlation coefficient of 1.01% indicates a very weak relationship between the flux and time of day.
Unlike studies undertaken in stable conditions, vegetation photosynthetic activity does not seem to have as strong an influence on the flux observed during the study period. In spite of initially reducing after 4 am due to photosynthetic activity, the flux accumulates possibly negated by industrial activities in the vicinity. The flux drops to its lowest around 14 hours. It then picks up again, possibly due to heavy traffic at the end of the business day.
The findings could suggest intermittent winds on the day of measurement, consistent with the weather patterns at this time of the year (MetOffice, 2011). This violates conditions for storage measurements (Burba and Anderson, 2010). Being winter, the influence of photosynthetic activity on the results was weak as indicated by the 1.01 % correlation between the storage flux and time of day. The concentration data for the different heights and low mean flux value of -1.66 X10-2 mg CO2 m-2 s-1 suggests a well mixed air layer (Bakwin et al., 1998) or turbulence on the sample date.
There are a number of limitations with the study. First, the setup did not account for response time between heights required to completely purge the Licor. This meant that a mixture of air from different heights could have been present in the detection chamber leading to less than accurate measurements. Tubing of equal length would have served to equalise lag time (Xu, et al., 1999). The response time could then be tested and programmed into the logger before setting up. Alternatively, the tested response time between the heights could have been taken into account and data lying in the time segment would have been omitted. Second, due to time constraints, no wind speed data was included for further analysis.
Despite these limitations, using the same analyser for successive concentration sampling ensured no systematic differences between level readings (Aubinet et al., 2005). The random error introduced by not sampling simultaneously does not lead to bias in storage fluxes (Rannik, et al., 2004). As the mean flux value observed tends towards zero, it would likely be neglected in the calculations of NEE (Finnigan, 2006) for January 21, 2011.
This paper has demonstrated how to set up and calculate storage flux using a single gas analyser measuring concentrations in profile in a built-up environment. The results show that for the period studied the storage flux has the minimum of -7.01X 10-2 mg CO2 m-2 s-1, maximum of 8.98 x 10-2mg CO2 m-2 s-1 and a mean of -1.66 X 10-2 mg CO2 m-2 s-1.
The correlation coefficient of 1.01% indicates a very weak relationship between the flux and time of day. These results suggest well mixed air layer and the calculated storage flux is negligible for the period under study and in calculating the NEE.
The setup would benefit from testing response time to ensure that only measurements from a single height were used in analysis. More robust analysis would also benefit from including wind speed data to determine its impact on the overall calculation of the storage flux.