C:\TEMP\98f16mem.doc
Memorandum
To: CC: From: Date: Re: George C. Murray, Jr Alan Klimek, Brock Nicholson Wayne L. Cornelius 21 August 1998 Detection in NC of Smoke from Central America
Contents Introduction...................................................................................................................................................2 Methods .........................................................................................................................................................2 Results............................................................................................................................................................2 Discussion ......................................................................................................................................................3 Recommendations.........................................................................................................................................3
List of Tables Table 1. Block Means of PM-10 Concentrations, Using Sites That Have Been Active in Every Covered Year. ................................................................................................................................................................4 Table 2. Number of Sites Sampled That Are Active Every Year. ..................................................................4 Table 3. Block Means of PM-10 concentrations, Using all NC active sites. ..................................................5 Table 4. Number of Sites Sampled. ................................................................................................................6 Table 5. Central America Smoke Episode Samples........................................................................................7 Table 6. Samples From 26 May 1994.............................................................................................................9 Table 7. Samples from 21 May 1996............................................................................................................10
List of Figures Figure 1. Locations of PM-10 Monitors Active in 1998...............................................................................11 Figure 2. PM-10 Averages Shown as a Continuous Time Series.. ...............................................................12 Figure 3. PM-10 Concentrations Surrounding the Central America Smoke Episode...................................13
1
November 29, 2000
Introduction Throughout the Spring of 1998, as happens every Spring, thousands of fires in Central America have been burning. However, the 1998 fires are said to be about twice as intense as the normal year. As a result, smoke has been lingering over southern Mexico, Guatemala and Honduras since late February 1998 throughout the Spring, and smaller incursions of smoke palls have reached the US Gulf Coast, particularly Florida. Unlike earlier years, the research community has followed with keen interest the 1998 Central American fires by a variety of UV, visible and infrared remote sensors from satellites1. Brett Schichtel, Washington University, St. Louis, MO. requested through Rich Poirot continuous PM-10 data to track the spread of smoke subsequent to a stagnation episode in Central America 1012 May. (The smoke was transported North, reaching Western Ontario by 14 May. On 16-17 May, a moving cold front pushed the pall of smoke eastward.) After responding to this request, I had my staff assemble a database of daily, not continuous, PM-10 data to see whether we could recognize the effect of this smoke. In particular, do PM-10 samples from 17 May in NC have substantially higher concentrations than on 11 May not attributable to localized effects, normal seasonal variation, or other influences that can be distinguished from the phenomenon?
Methods We retrieved all the NC daily PM-10 data samples available in AIRS for the 7 years 1992-98, between the dates 2 March through 29 June. We divided this 120-day period into 20 6-day blocks, and analyzed the samples as representing the time block in which they fell. (Because of the 1996 Leap Year, this didn't quite succeed at putting 1 sample from every day of the week in every block.) The sites active in 1998 are shown on a map in Figure 1. We examined two groupings of sites: (i) 15 sites at which monitoring data exist in every one of the 7 years, and (ii) about 50 sites at which monitoring data exist in any one or more of the 7 years. Results In Table 1 (and also Figure 2), average concentrations are shown for each time block and year. Marginal column and row maximums are also shown. The averages are arithmetic means over the 15 sites shown in the header of the table. Table 2 shows the number of sites that were averaged for each cell of Table 1. In Table 3, average concentrations are shown for each time block and year. Marginal column and row maximums are also shown. The averages are arithmetic means over the sites shown in the header of the table. Table 4 shows the number of sites that were averaged for each cell of Table 3. Because the difference between concentrations of 11 May and 17 May 1998 are of special interest, the specific samples for those dates are listed in Table 5 and shown as boxplots in Figure 3. The 11 May 1998 samples range from 4 to 23 µg/m3 (mean 15.0, s.e. 4.6) , and the 17 May 1998 samples range from 30 to 61 µg/m3 (mean 46.9, s.e. 7.9). The 17 May concentrations exceed all other data in Table 1 and Table 3, with the possible exception of 24-29 June in 1992 and 1993, 25-30 May 1994, and 19-24 May 1996; whereas the concentrations on 11 May 1998 are among the smallest concentrations found on any date within the range covered. Table 6 lists the samples of 26 May 1994, and Table 7 lists the samples of 21 May 1996. , accessed 16 June 1998.
1
2
November 29, 2000
Discussion PM-10 concentrations appear to have more than tripled from 11 May to 17 May, and this increase appears to be uniform statewide. On 24 May the concentrations decreased to merely double those of 11 May. This finding does not prove that Central American smoke caused the increased PM-10 concentrations, but it is very strong correlative evidence for a relationship between the smoke episode observed by satellite data and the concentrations observed by PM-10 sampling. Table 1 and Table 3 show that on 2 other occasions a rather high average concentration (over 40 µg/m3) is sandwiched between immediately earlier and later averages that are only about ½ as large, for example 25-30 May 1994 and 19-24 May 1996. Figure 2 shows a very emphatic view of the pattern (although the dates are not explicitly labeled).This pattern, which we have never noticed before, because we never looked for it before, may be cautiously interpreted as evidence that some widespread PM-10 "event" probably occurred during the time that the high concentration was observed. However, closer inspection of the data seems to rule out such an event on these 2 dates.We are not certain that candidate events can be identified for these dates. Moreover, when the individual samples are examined (Table 6 and Table 7), it is clear that the high concentrations are not found uniformly at all sites. The 26 May 1994 samples range from 24 to 81 (mean 44.8, s.e. 10.3) and the 21 may 1996 samples range from 23 to 54 (mean 39.4, s.e. 7.9).
Recommendations Media interest in this topic seems to have passed away already, before we could assemble the data to study it most effectively. This will be useful information for reference in the 1997 Annual Air Quality Report, if nowhere else. Also, although the episode did not produce PM-10 exceedances, we have hereby developed a case for validity-flagging our AIRS PM-10 data for 17 May 1998 to indicate abnormal observations or a statewide "exceptional event", if we wish to do so. (Unfortunately, the only suitable AIRS code available is the one for "misc.".)
3
November 29, 2000
Table 1. Block Means of PM-10 Concentrations, Using Sites That Have Been Active in Every Covered Year2.
Dates 02-07.Mar 08-13.Mar 14-19.Mar 20-25.Mar 26-31.Mar 01-06.Apr 07-12.Apr 13-18.Apr 19-24.Apr 25-30.Apr 01-06.May 07-12.May 13-18.May 19-24.May 25-30.May 31-05.Jun 06-11.Jun 12-17.Jun 18-23.Jun 24-29.Jun MAX 92 14.1 19.1 15.7 25.6 21.0 25.3 16.7 23.0 31.7 32.9 17.0 35.3 30.4 35.3 14.9 22.4 29.1 30.6 39.1 40.1 40.1 93 30.7 18.3 15.8 21.5 22.9 17.7 17.7 29.0 23.0 24.7 31.9 36.0 24.4 24.9 31.0 20.9 21.6 32.4 35.3 48.5 48.5 94 14.7 20.7 26.4 29.4 17.1 26.9 22.0 21.5 35.3 31.1 24.1 17.8 29.0 18.0 45.8 29.8 21.1 35.0 26.2 22.4 45.8 95 18.8 22.6 32.3 21.9 26.1 34.2 22.8 19.6 26.9 29.1 19.1 34.1 27.3 22.2 25.6 19.8 32.9 25.3 22.1 27.4 34.2 96 22.80 17.00 24.40 19.50 9.13 23.70 20.00 24.30 17.60 21.10 29.50 31.00 23.10 41.60 23.60 18.10 18.30 32.30 25.50 30.10 41.60 97 22.1 21.3 21.5 16.2 21.2 33.9 22.1 32.1 23.6 12.4 14.9 15.1 20.3 21.9 17.3 13.5 21.4 28.9 35.3 24.6 35.3 98 27.8 15.8 14.9 20.1 29.9 15.6 16.9 22.4 13.1 22.8 20.4 15.0 46.9 33.5 38.2 28.0 21.0 29.1 32.7 41.2 46.9 MAX 30.7 22.6 32.3 29.4 29.9 34.2 22.8 32.1 35.3 32.9 31.9 36.0 46.9 41.6 45.8 29.8 32.9 35.0 39.1 48.5 48.5
Table 2. Number of Sites Sampled That Are Active Every Year.
Dates4 02-07.Mar 08-13.Mar 14-19.Mar 20-25.Mar 26-31.Mar 01-06.Apr 07-12.Apr 13-18.Apr 19-24.Apr 25-30.Apr 01-06.May
2
[1] "370210003" "370510004" "370670009" "370810009" "370850001" [6] "370870002" "371110002" "371190010" "371191001" "371191005" [11] "371210001" "371330004" "371390001" "371830003" "371910004"
92 13 14 11 14 14 15 13 13 15 15 15 93 13 14 13 13 14 14 15 14 14 15 14 94 15 14 14 15 13 15 15 14 13 14 14 95 13 13 14 15 13 13 15 15 15 15 15 96 15 14 13 15 15 15 15 15 15 15 15 97 15 15 15 14 15 15 14 15 15 13 14 98 15 15 14 15 15 14 11 13 14 13 14
4
November 29, 2000
Dates4 07-12.May 13-18.May 19-24.May 25-30.May 31-05.Jun 06-11.Jun 12-17.Jun 18-23.Jun 24-29.Jun 92 15 14 15 15 14 15 14 15 14 93 12 14 14 15 14 15 15 14 15 94 14 15 15 15 15 15 14 14 13 95 15 15 15 15 14 14 15 15 15 96 14 15 14 13 14 14 14 14 13 97 15 15 15 15 15 15 14 15 14 98 14 14 13 14 14 13 14 14 13
Table 3. Block Means of PM-10 concentrations, Using all NC active sites3.
Dates 02-07.Mar 08-13.Mar 14-19.Mar 20-25.Mar 26-31.Mar 01-06.Apr 07-12.Apr 13-18.Apr 19-24.Apr 25-30.Apr 01-06.May 07-12.May 13-18.May 19-24.May 25-30.May 31-05.Jun 06-11.Jun 12-17.Jun 18-23.Jun 24-29.Jun 92 13.7 18.1 15.6 25.8 21.2 25.9 16.9 24.1 32.3 31.8 17.7 36.0 30.7 35.2 16.2 22.0 29.9 31.7 38.7 41.0 93 31.7 17.4 13.9 22.7 27.0 17.2 17.9 27.7 21.9 24.4 29.9 36.2 26.9 24.4 30.8 23.7 23.6 34.3 36.1 46.6 94 12.7 18.8 23.1 28.6 16.4 26.3 20.2 19.4 33.8 31.1 22.6 17.7 28.6 17.7 44.8 28.7 22.2 34.9 26.2 24.9 95 18.2 19.3 31.5 22.2 22.6 30.2 23.5 18.6 26.6 25.4 19.4 30.1 26.7 23.1 25.7 19.7 31.5 24.2 23.5 26.2 96 19.8 15.6 21.4 15.7 8.3 21.0 17.8 23.0 16.0 19.0 26.2 31.9 23.7 39.4 22.3 17.6 18.5 32.0 25.5 29.8 97 20.3 19.2 19.8 14.3 19.3 33.3 20.7 29.1 23.1 11.8 12.9 13.9 19.1 21.7 16.0 12.6 20.4 28.6 35.1 24.4 98 27.7 15.2 14.7 19.3 28.0 15.3 16.3 21.8 13.3 22.2 19.8 15.0 46.9 34.2 36.9 27.8 21.7 27.6 31.2 41.3
3
[1] "370010002" "370030003" "370130005" "370210003" "370210032" [6] "370210033" "370250004" "370350004" "370370004" "370510004" [11] "370570002" "370571002" "370590099" "370630001" "370650002" [16] "370650099" "370670009" "370670013" "370670020" "370670023" [21] "370670024" "370671001" "370710014" "370770002" "370810009" [26] "370811005" "370830002" "370850001" "370870002" "370891006" [31] "371090003" "371110002" "371190001" "371190003" "371190010" [36] "371191001" "371191005" "371210001" "371290005" "371290007" [41] "371330004" "371390001" "371470003" "371550003" "371591006" [46] "371730002" "371830003" "371890003" "371910004" "371950002"
5
November 29, 2000
Table 4. Number of Sites Sampled4.
Dates 02-07.Mar 08-13.Mar 14-19.Mar 20-25.Mar 26-31.Mar 01-06.Apr 07-12.Apr 13-18.Apr 19-24.Apr 25-30.Apr 01-06.May 07-12.May 13-18.May 19-24.May 25-30.May 31-05.Jun 06-11.Jun 12-17.Jun 18-23.Jun 24-29.Jun 92 19 21 18 21 21 21 19 20 22 22 22 22 21 22 22 21 22 22 23 21 93 33 36 33 35 35 36 43 39 40 42 41 36 44 45 47 47 46 48 43 46 94 39 38 37 36 36 37 38 37 36 38 39 38 39 40 39 39 39 36 38 39 95 49 51 53 53 53 51 53 54 54 54 53 54 50 52 53 49 46 47 48 48 96 40 38 39 43 43 42 41 40 36 40 43 42 42 41 42 41 40 38 39 37 97 43 42 40 42 41 41 41 42 41 40 41 44 41 42 42 41 42 40 40 39 98 28 27 26 26 28 25 24 23 26 24 27 27 27 26 27 27 26 27 27 26
4 Some date blocks in 1993 and 1995 have more samples than the number of sites, because the two
continuous monitors (370210032, 370210033) provided up to 6 samples per block. I have not investigated why these monitors are represented in only these two years.
6
November 29, 2000
Table 5. Central America Smoke Episode Samples. Boldface rows are used in Table 1. All rows are used in Table 3. AIRS Number PM-10 Concentration
370250004 370350004 370571002 370630001 370650002 370710014 370811005 370891006 371190001 371190003 371290007 371470003 371730002 370210003 370510004 370810009 370850001 370870002 371110002 371190010 371191001 371191005
11 May 17 17 14 11 13 17 15 17 19 18 10 11 15 17 11 12 10 23 22 18 16 22 17 May 53 46 55 42 43 46 61 43 58 56 34 35 37 47 50 57 47 51 48 54 51 54 23 May 31 35 32 30 32 38 35 42 45 45 29 27 33 36 28 34 28 40 34 42 38 52
7
November 29, 2000
AIRS Number PM-10 Concentration
11 May 22 10 04 13 10 15.0 4.6 17 May 45 37 30 45 41 46.9 7.9 23 May NA 24 18 34 28 34.2 7.3
371210001 371330004 371390001 371830003 371910004 Mean Std. Dev.
8
November 29, 2000
Table 6. Samples From 26 May 1994.
AIRS year month day PM10
15 370010002 112 370030003 306 370210003 753 370250004 864 370350004 944 370370004 1044 370510004 1155 370570002 1356 370590099 1416 370630001 1499 370650002 1608 370650099 1726 370670009 1844 370670013 2003 370670023 2099 370670024 2196 370671001 2279 370770002 2400 370810009 2510 370811005 2619 370830002 2728 370850001 2856 370870002 2939 370891006 3043 371090003 3159 371110002 3332 371190003 3457 371190010 3571 371191001 3690 371191005 3807 371210001 4050 371330004 4178 371390001 4276 371470003 4437 371591006 4611 371830003 4690 371890003 4807 371910004 4916 371950002 mean std.err. 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 94 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 26 43 46 26 60 41 45 58 48 37 47 53 47 40 42 45 36 41 45 41 43 47 52 30 34 43 40 56 56 47 54 27 37 81 43 44 48 24 50 49 44.77 10.27
9
November 29, 2000
Table 7. Samples from 21 May 1996.
AIRS year month day PM10
51 370010002 150 370030003 227 370130005 343 370210003 791 370250004 900 370350004 1081 370510004 1192 370570002 1285 370571002 1537 370650002 1647 370650099 1764 370670009 1880 370670013 2037 370670023 2115 370670024 2233 370710014 2318 370770002 2438 370810009 2548 370811005 2650 370830002 2764 370850001 2893 370870002 2976 370891006 3081 371090003 3198 371110002 3266 371190001 3371 371190003 3610 371191001 3729 371191005 3845 371210001 3960 371290007 4088 371330004 4211 371390001 4315 371470003 4377 371550003 4476 371591006 4517 371730002 4650 371830003 4727 371890003 4846 371910004 4952 371950002 mean std.err. 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 96 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 21 50 32 33 29 31 38 43 50 47 38 43 44 46 46 41 35 40 49 54 34 39 41 33 36 45 40 45 42 53 42 23 26 38 29 42 42 24 49 27 43 34 39.41 7.85
10
November 29, 2000
Figure 1. Locations of PM-10 Monitors Active in 1998.
11
November 29, 2000
0
C.A. smoke event
unusual event?
unusual event?
05
PM-10 04
mean
03
statewide
021
1992 1993 1994 1995 1996 1997 1998
0
0 20 40 60 80 100 120 140
2.Mar - 29.June Samples
Figure 2. PM-10 Averages Shown as a Continuous Time Series. From Table 1.
12
November 29, 2000
PM-10 concentrations in NC during 1998 episode of Central America Smoke
0
065
0 hr)
04
g/cu.m.-24 (µ
032
01
0
05/11/1998 (before) 05/17/1998 (during) 05/23/1998 (after)
Figure 3. PM-10 Concentrations Surrounding the Central America Smoke Episode. From Table 5.
13
