3506B24 Final Report
Download 11.49 Mb.
|
- Navigate this page:
- 4.2 The Fall 2001 Columbus PM 2.5 Pollution Case
- Table 2 and Figure 8
- Table A1
|
Figure 5: PM2.5 mass concentration rose from Fig. 4 projected over the OLC measurement site. Fort Benning’s installation covers about 180,000 acres total, of which 94,000 acres undergo land management by prescribed burning in a 3-year rotation, leaving ~32,000 acres per year (green area). The majority of burns are conducted between November and April. 
Figure 6: Average diurnal cycles of [PM2.5] for the summer periods 2000, 2001, 2002 (left) and winter periods 2000/01 and 2001/02 (right) in comparison of the different FAQS continuous monitoring sites at Griffin, Macon, Columbus-OLC, and Augusta in South-Central Georgia (see map in Fig. 4 for locations). Standard deviations of the hourly averages determined for the winter 2001/02 period are plotted for comparing the Columbus-OLC data with the Augusta profile. It should be noted that the summer 2000 data do not include the first two months, May and June, during which the semi-permanent sites were established. Nevertheless, the summer 2000 was characterized by higher ozone levels and more frequent occurrences of stagnant high pressure periods over the entire South-East compared to the following two summers [Chang et al., FAQS report, 2004]. The higher average PM2.5, levels in 2000 could therefore be a result of more effective SOA formation, which as mentioned above is governed by similar conditions and mechanisms as ozone formation. However, while it is plausible to assume that the SOA source strengths varied, it is possible and valid to consider differences in the effectiveness of loss mechanisms for atmospheric fine PM, which primarily is precipitation. In deed, the National Weather Service (NWS) reported more average rainfall for the state of Georgia during 2001 and 2002 than during the year 2000, as can be seen in Figure 7 [NWS, 2004]. The annual totals for the three consecutive years were 41.7, 42.9, and 50.3 inches, respectively, whereas the 30-year normal (1961-1990) is 50.8 inches. Another interesting feature in the average diurnal cycles of the [PM2.5] shown in Fig. 6 is the difference in profiles: while the average summer profiles show only little variation over the course of the day, with minor maxima in the morning and evening, the winter profiles show a regionally uniform minimum at midday. The lower standard deviation of the midday minima relative to the ones of other times of day seems to point to a frequent occurrence and more regular feature on winter days. Since the average diurnal cycles of horizontal wind speeds (not shown) show midday maxima at all sites, these regional daytime minima are likely a result of atmospheric dilution due to increased mixing into a deeper BL. This dynamic dilution certainly takes place also in summer, with even greater intensity due to deep convective mixing, however, at the same time, more intense SOA formation and mixing of SOA containing air masses from residual layers aloft seems to compensate the dilution effect, preventing the midday minimum to appear at all sites. 
Figure 7: Average monthly precipitation in Georgia during the years 2000 thru 2003, relative to each other and the 30-year normal (1961-1990). The most striking deviations in the [PM2.5] profiles are the excursions to large average fine PM mass concentration levels observed at the Columbus OLC site for the summer 2001 (thin blue trace in left panel of Fig. 6) and even more pronounced for the winter 01/02 period (bold blue trace in right panel of Fig. 6) during late night and early morning hours. The standard deviations associated with those averages are very large, pointing to very rare occurrences of extremely high values. The reason for this behavior is investigated further in the following section. 4.2 The Fall 2001 Columbus PM2.5 Pollution Case During a three-week period in the fall of 2001 the 24h NAAQS for PM2.5 of 65 g m-3 was exceeded at the Columbus OLC site on five different days, with the worst day showing a peak concentration of 448 g/m3, see Table 2 and Figure 8. 24h-Means and ½ h maxima in Table 2 clearly indicate high PM pollution levels, which are not seen (in Fig. 8) at the other FAQS sites and also not at the two regulatory monitoring sites operated in Columbus City by GA-EPD. The two sites are at Cusseta Road and the County Health Department, about 5 km and 11 km north of OLC, respectively. Both sites are part of U.S. EPA’s network of National Air Monitoring Stations/ State and Local Air Monitoring Stations (NAMS/SLAMS), monitoring criteria pollutants, incl. 24-h integrated PM2.5 mass following the Federal Reference Method (FRM) as described in the Code of Federal Regulations [CFR, 1997]. The one in three day sampling schedule of the two Columbus FRM monitors happen to have missed the main pollution episodes during the October-November 2001 episode. Table 2: Daily 24 h means, standard deviations based on ½ h means, and ½ h maxima fine PM (PM2.5) mass concentration in μg/m3, measured at Columbus, OLC; and completeness of records for that particular day in percent. Date 24 h mean Std dev ½ h max Completeness (%) 10/29/01 68.6 49 166 77 10/30/01 141 63 231 50 11/08/01 65.4 41 207 100 11/16/01 71.3 67 231 96 11/17/01 448 92 599 23 
Figure 8: Comparison of measured (OLC) and predicted [GFC, 2003] wind barbs (top), BL height and atmospheric stability indicators (center, with smoke dispersion index, SDI, bottom), areas of prescribed burns and wild fires at Fort Benning with 24 h average [PM2.5] from OLC compared with Griffin, Macon, and Augusta (bottom), for 5 weeks in fall 2001. Note the log scale for [PM2.5], and the discontinuation of the OLC data due to filter overload. The blue open circles and triangles are 24 h [PM2.5] from the state’s regulatory FRM monitoring sites in Columbus at Cusseta Road (5 km N of OLC) and the Health Department (11 km N of OLC), respectively. Witnessed by local residents, all episodes could be attributed to nearby burns originating on the military base of Fort Benning. However, not all burns were intentional, as summarized in Table 3; a large fraction of the burns occurring during the October/November 2001 period were wild fires (1806 acres of 8304 total), originating on or near artillery impact areas, i.e. the red DUD areas shown in above map. The base had planned to burn more than 32,000 acres during the 2001/02 season but bad weather (wet conditions) and complaints brought forward by the Columbus public had only allowed the burning of a little under 28,000 acres. The circumstances surrounding the fall 2001 incidence are being investigated here in more detail. The final decision for conducting a prescribed burn is usually made after retrieving the GFC Forestry Weather and Smoke Management Forecast products on the morning of the planned burn. The above Figure includes the main criteria of the “AM” forecasts published early mornings, i.e. the plume transport speed and trajectory (top panel), boundary layer (BL) height (center), and smoke dispersion index (SDI, bottom). These forecasted model outputs are being compared with actual measured values like maximum daytime wind speed, average wind direction, average daytime solar radiation (PAR), daily maximum and minimum temperatures as combined indicators for near surface atmospheric stability and BL turbulence, and 24 h PM2.5 concentrations from OLC. The latter are compared further with PM2.5 measurements from the other three FAQS sites in Griffin, Macon, and Augusta. Information from the GFC forecasts, Fort Bennings burn activities, and ambient measurements at OLC are being compared on a daily basis for the entire October – December 2001 period in Table A1 of the Appendix. Table 3: Size, distance and direction from OLC of prescribed burns (PB, yellow) and wild fires (WF, orange) that occurred at Fort Benning’s burn units between October 17 and November 22, 2001, as illustrated in Figure 8 above. During this 5 week period, 6498 acres were PB with average 271 ±319 acres per burn day, and 1806 acres were WF with average 301 ±302 acres. The average distances to OLC were 17.4 ±6.8 km and 16.9 ±7.5 km for PB and WF, respectively.
Download 11.49 Mb. Share with your friends:
|