3506B24 Final Report
Download 11.49 Mb.
|
- Navigate this page:
- Figure 42 and Tables A7
- 6.6 Particulate Organic Compounds (POC) from GC/MS
- Figure 43
Figure 41: PCM derived discrete data of PM2.5 mass and composition, incl. average TEOM mass (top) and reactive gases concentrations (bottom) for the 30 individual samples collected during the January background event and the six prescribed burn events (incl. each total acres burnt) between early February and late May 2003. Also shown are the mixing ratios of gaseous species for each measurement period, as determined from the phosphorous acid and sodium carbonate coated denuders (the former capturing NH3 the latter all other, acidic species). Acetic and formic acids are closely correlated with each other (slope = 0.86 0.10, intercept = 0.02 0.13 ppbv formic acid, at r2 = 0.71), and still significantly correlated with ambient temperature, especially later in the season from March to May, at r2 = 0.70 and 0.53, and slopes of 0.17 0.02 and 0.14 0.03 (95 % confidence level) for formic and acetic acid, respectively. Similar relationships have been reported in previous studies indicating that these specific carboxylic acids can be formed via secondary photochemical pathways as well as emitted from common sources, e.g. microbal soil activity and decay [Talbot et al., 1995], or combustion processes [Khwaja, 1995]. The temperature dependence was less clear earlier in the season during January and February, when similarly high formic and acetic acid levels correlated more closely with CO, NOy and HONO levels during traffic rush hour times, suggesting a potential sensitivity to nearby vehicular sources and lack of mixing, in support of earlier findings from Khwaja [1995]. The possibility of oxygenated VOC (OVOC) such as formic and acidic acids being directly emitted from both automotive combustion and biomass burning, has implications for other less volatile OVOC and oxygenated organics being emitted as primary PM from these sources as well. The following will address this question only on the basis of indirect evidence and hence in a very speculative way, since no oxygenated POC (OPOC) or SOA species were actually identified nor measured. The OOE term is particularly sensitive to the amount of oxygen associated with the particulate organics, as O atom contributes significant mass to a mainly C and H atoms containing molecule, and can therefore indicate a certain level of oxidation that the carbonaceous aerosol might have experienced before sampling. The organic mass (OM) contributing to the total PM2.5 mass is therefore the sum of OC+OOE, and the OM/OC ratio which is considered to correct the measured OC, takes into account all non-carbon elements contributing to an average molecular weight per carbon weight of a typical organic aerosol. The factor 1.4 that has been widely used in the past, originates from very limited theoretical and laboratory studies from more than 20 years ago, suggesting it to be the lowest reasonable estimate for urban aerosols [White and Roberts, 1977; Countess et al., 1980; Japar et al., 1984]. A more recent investigation by Turpin and Lim [2001], however, suggests a factor of 1.6 0.2 to be more accurate in an urban environment. Since this factor’s uncertainty is larger than any one of the identified species (see Table A7c), the mass closure approach was chosen to calculate it here for each sample of this data set. Assuming that the original unidentified mass fraction is entirely associated with OC, and that no other non-detected unknown species contributes to the total PM2.5 mass, the organic mass to OC ratio (OM/OC) is assessed from mass closure for each of the samples. In most atmospheric instances, this is a reasonable assumption, since the trace metals and crustal materials are typically less than 3 % of the total mass. The uncertainty of the OOE estimate determined from error propagation of the uncertainties of all identified species, varied among all individual samples between 1.5 and 5.4 g m-3 or between 16 and ±21 % relative to the total mass concentration. The OOE uncertainties governed the uncertainties of the OM/OC ratio, i.e. the “OC correction factor”, which varied individually between 1.3 and 3.1, and between 1.5 ±0.1 and 2.1 ±0.7 for the averages of the 7 different events; see Figure 42 and Tables A7. The mass balance concentrations and relative mass fractions in Fig. 42 follow the same color-coding for individual species: what is labeled “Others” includes potassium, sodium, and chloride, and “LOA” includes the particle-phase light organic acids acetate, formate and oxalate. The top panel also shows numbers of acres burned, and maxima ½ h O3 for each event scaled relative to the right axis. The OM/OC ratios derived from mass closure provide some measure of the degree of POC oxygenation relative to each sampling period. Despite its relatively high uncertainty, this ratio yielded a trend toward higher values later in the season with all values being clearly above 1.4. Despite less burn activity towards the end of the season, the OM/OC was 2.1 ±0.7 late April and 2.0 ±0.3 in late May, only insignificantly higher than 1.9 ±0.3 in early April, when a maximum 4,000 acres were burned. Only slightly less was burned in late March (3,770 acres), but the level of POC oxygenation seemed significantly lower at 1.6 ±0.2. Combining both observations, more SOA predominantly from non-biomass burn sources in aged air masses contributed to total POM at the end, while more primary OPOC contributed earlier in the season. 
Figure 42: Absolute (top) and fractional (bottom) PM2.5 mass and composition data from the PCM averaged for the 7 different sampling events between mid January and late May 2003. The three samples between March 10 and 11 varied the most, increasing from 1.3 to 3.3, and caused the largest average of 2.2 ±1.0. These values were accompanied by also the lowest and equally variable OC/EC ratios of 7.6 ±3.7, a less uncertain quantity. The lowest OC/EC of 3.4 coincided with the highest OM/OC of 3.3 during 0800 and 1300 of 3/11, which was also the period with the highest absolute CO, NOy and NO values. OLC received low westerly flow during this period while the burns were conducted between 13 and 23 km to the east, unlikely to cause a direct plume impact. With CO/NOy of ~9 and NO ~90 % of NOy, these polluted air masses were clearly of an origin other than prescribed burning, indicating automotive emissions from relatively near-by containing extremely high fractions of OPOC. The fact that OC/EC was low and that it was still early in the season, i.e. photo-chemical activity was low as indicated by max-O3, the high OM/OC is suspected to be due to primary OPOC and not SOA. 6.6 Particulate Organic Compounds (POC) from GC/MS Eighty seven organic species (from Table 14), including n-alkanes, n-alkanoic acids, polycyclic aromatic hydrocarbons (PAH), steranes, hopanes, resin acids, aromatic diacids, and some key tracer compounds serving as specific markers for biomass burning, e.g. levoglucosan, certain resin acids (pimaric acid, sandaraco-pimaric acid, abietic acid), and the PAH retene, were identified and quantified in the samples. The concentrations of some organic compound groups are shown in Figure 43 and in Table A7f using the same format as for the other quantities in previous Tables A7. The individual POC are listed in Tables A8. It is interesting and important to note, that the fraction of the sum total of all identified POC to the total measured OC (ΣPOC/OC in last column of Table A7f) varied from sample to sample between 1 and 17 %, and averaged at 6.2 ±4.9 % for the 15 samples compared. It is quite remarkable that approximately 94 % of the measured OC mass, and even more if related to actual organics mass (remembering that OM is an upper limit estimate), remain unidentified by GC/MS. The implications of this fact are, as mentioned in the introductory background information, that most POC are of more polar character, which are undetectable to current GC/MS techniques. New laboratory extraction and analytical techniques are being investigated to reduce this lack of speciation capability inherent to the GC/MS principle, and other approaches are being developed such as Proton Nuclear Magnetic Resonance (1HNMR), Electro-spray Ionization with Mass Spectrometry (ESI/MS), or High Pressure Liquid Chromatography coupled to a Time-Of-Flight MS (HPLC-TOF/MS). The time axis in Fig. 43 marks the start time of each HVS sample, and is labeled according to combustion stages of the prescribed burns conducted on the military installation. The start times labeled BKG (background) and FL&SM (flaming and smoldering) sampled for 24 h, all other ones sampled between 5 h in February and 9 to 12 h any other time. Hence, FL&SM denotes a 24 h period covering both flaming and smoldering stages of the PB. However, as mentioned and elaborated earlier, this does not automatically imply that the sampling location (OLC) was actually impacted by the PB plume that could have originated as far as 28 km away. From the CO/NOy ratios discussed earlier, the biggest impact occurred typically at night, with the highest value of 26.6 (r2= 0.51) during the evening of March 27. Unfortunately, no HVS sample for POC analysis was taken then but earlier that week on the 24th, when combined 956 acres were burned at TA K22 and M4 (25 and 14 km to the ENE from OLC). The CO/NOy indicated still a strong influence under weak northerly flow later veering over east to south. Both levoglucosan, as one of monosaccharide derivatives from cellulose burning/thermal degradation regarded as a main tracer for biomass burning, and resin acids peaked in that 24 h sample. For shorter samples that tried to separate the flaming from the smoldering phases, both compounds showed a trend of higher concentrations during smoldering, which is the more incomplete combustion process. Not surprisingly, the PAH followed this trend as well. A slightly different trend was observed for the n-alkanes, the n-alkanoic acids, the hopanes and steranes, which were highest during the early flaming stages, significantly lower during the smoldering stages, and dropping back to the background levels. A striking exception occurred on March 10th, when the site was impacted by a completely different air mass and pollutant mix as discussed earlier. The following discussion will focus around three case studies, in which POC data are being evaluated in conjunction with burn activities, meteorological conditions and supporting indicators. 
Download 11.49 Mb. Share with your friends:
|