- Email: [email protected]
Non-automotive sources of carbon monoxide in urban areas
Chemosphere: Global Change Science 1 (1999) 111±114
Non-automotive sources of carbon monoxide in urban areas M.A.K. Khalil
a,*
, R.A. Rasmussen
b
a
b
Department of Physics, Portland State University, P.O. Box 751, Portland, OR 97207-0751, USA Department of Environmental Science and Engineering, Oregon Graduate Institute, P.O. Box 91000, Portland, OR 97291, USA Received 9 February 1999; accepted 10 February 1999
Importance of this Paper: Carbon monoxide is a criteria pollutant in the US necessitating monitoring and control of emissions. To do so requires an accurate knowledge of the local sources. Although automobiles are a major source in most areas, other sources can contribute signi®cant amounts in some cities. Our research shows that in urban areas of the Northwest, emissions from agricultural burning and wood burning for heat can be comparable to automotive emissions during fall and winter. Abstract Measurements of carbon monoxide were taken for two years at several locations in Portland, Oregon. There is a residual CO that is not accounted for by automobile emissions and the background concentrations. The ratio of excess CH4 (above background) to the residual CO is close to direct measurements of the emission factor of CH4 /CO from biomass burning which supports the idea that the residual CO, observed in our study, is from biomass burning. Biomass burning sources consist of agricultural burning in the surrounding areas and wood burning for heat. The diurnal and seasonal timing of the residual CO is consistent with our knowledge of these sources. We found that in Portland, during Fall and Winter, biomass burning contributes 20±40% of the carbon monoxide excess over background levels. Ó 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction In some urban areas, especially in the Paci®c northwest, there can be substantial contributions of CO from sources other than automobiles. These sources, mostly biomass burning, may now be relatively more important to the CO urban budgets than in previous years because per capita automobile emissions have declined during the last decade.
* Corresponding author. Tel.: +1-503-725-8396; fax: +1-503725-8550; e-mail: [email protected]
Measurements of CO and other gases were taken at four schools around the Portland area. The CO measurements were part of a larger experiment to measure greenhouse gases and ozone-depleting compounds in the atmosphere over continental areas in the GLOBE program (Global Learning and Observations to Bene®t the Environment). The measurements of carbon monoxide were aimed at understanding the seasonality of emissions from the various sources and to determine a continental signature for CO. Unlike measurements at the clean air background sites, these experiments show a complex pattern driven not only
1465-9972/99/$ ± see front matter Ó 1999 Elsevier Science Ltd. All rights reserved. PII: S 1 4 6 5 - 9 9 7 2 ( 9 9 ) 0 0 0 1 4 - 8
112
M.A.K. Khalil, R.A. Rasmussen / Chemosphere: Global Change Science 1 (1999) 111±114
by the well established cycles of trac but also by emissions from sources other than automobiles. At sites in Portland, and in many other parts of the Paci®c northwest, there are substantial contributions of CO from agricultural burning during fall and less so during winters. During winters wood burning for heating homes can contribute large amounts of CO in the Portland area and similar cities in Oregon and Washington. The results of measurements during the last two years are shown in Fig. 1. The data are shown as excess over the values at Cape Meares, which we consider to be a clean air site for this latitude. In previous work we used the concentrations of methyl chloride as an indicator wood burning pollution in Oregon and Washington. This technique was designed primarily for estimating the contribution of wood burning to ®ne particles in the environment but was also extended to evaluate the contribution of wood burning to CO. Our studies for Olympia, Washington, and Medford, Oregon, showed that during winters the contribution of wood burning to ambient CO levels was comparable in magnitude to emissions from automobiles (Khalil and Rasmussen, 1988, 1989). In the present work methyl chloride was not measured so we have taken a dierent approach by
Fig. 1. The monthly average concentrations of carbon monoxide at middle latitude urban and background sites (12/95-12/ 97). The remote continental site is at Minqin in the Gansu Province of China, 1/97-12/97; the urban sites are at GLOBE schools in Portland, Oregon, and the marine background site is at Cape Meares, Oregon. The excesses at the urban sites are likely to be caused by automobiles and biomass burning including agricultural burning and wood burning.
focusing on automobile emissions using total nonmethane hydrocarbons (NMHC, TO-12 method) as an index of automobile pollution. 2. Data analysis and results Air samples used here were obtained once a week from four schools in the Portland area. These schools are Kelly Elementary, Madison High School, Jackson Middle School, and Rose City School. Of these, Kelly school is the closest to a major highway with prevailing winds from the direction of the highway towards the school. Madison is at a slight elevation with prevailing winds from areas of low population outside the city. Data from Rose City School were not used as samples were collected for only a few months and stopped for logistical reasons. Samples were collected in internally electropolished 0.8L stainless steel containers using a pump developed for clean air sampling and used extensively in our research (Rasmussen and Khalil, 1980). To analyze the data (Fig. 1) we use the following model: COamb t COb t COauto t CObio-burn t;
1
COauto BfNMHCamb ÿ NMHCb g;
2
where [CO](t) is the concentration of carbon monoxide due to various sources and adding up to the observed value at time t. B the emission factor for automobile exhaust, expressed as the ratio of CO emissions to total NMHC emissions (TO-12). The subscript ``amb'' is for ``ambient'' air representing the actual measured concentration; b is for ``background''; ``auto'' is for ``automobile'' contributions, and ``bio-burn'' is the residual that we take to be from biomass burning as will be discussed in more detail later. We assume that during summers (June±August), the observed concentrations of CO in the Portland area schools are due entirely to automobiles and the background concentrations. During this time there is no intentional biomass burning, and the background concentrations are
M.A.K. Khalil, R.A. Rasmussen / Chemosphere: Global Change Science 1 (1999) 111±114
relatively constant as measured at Cape Meares, on the Oregon coast. When Eq. (2) is substituted into Eq. (1), it becomes a simple linear expression relating measured CO concentrations to concentrations of NMHC from which we can estimate A, the intercept and B, the slope, which is also the emission factor (as in Eq. (2)). The background concentration of NMHC is then estimated from A as NMHCb COb ÿ A=B:
3
The result of this calculation, for the composite data from all three schools mentioned above, is that the emission factor (excess CO/excess NMHC) is 3.5 0.5 ppbv ± m3 /lg, N 77, correlation coecient of 0.7 and [NMHC]b 6.5 lg/m3 (the value is the standard error). Based on this result we estimate the contribution of automobiles to the observed ambient concentrations using Eq. (2). The results are shown in Table 1 below. 3. Discussion During fall and winter seasons, the residual CO concentration tends to be the highest. We attribute this to biomass burning, which takes two forms in the Portland area. During fall the excess is likely to be contributed by agricultural burning, which is
113
widespread at this time. It includes burning of grass ®eld stubble and orchard prunings. Later in the season and in winters, wood burning for heating homes is the likely source of the excess. There is also forest slash burning throughout this period, but it is generally much farther away from the schools. The biomass burning contribution may be somewhat underestimated by our methods as some NMHCs are also emitted from wood burning, but the emission factors are some 10 times smaller (Andreae, 1991). The results show that there are times of the year when sources other than automobiles contribute signi®cantly to the atmospheric burden of CO in the Portland area. The results are comparable to the previous studies in Olympia (WA) and Medford (OR), although both are much smaller cities than Portland. The residual CO discussed above was compared to the excess methane during the fall and winter seasons. Excess methane was estimated as the dierence between the concentrations measured at the Portland schools and at Cape Meares during the same time. We expect that much of the excess methane during winter and late fall would come from the biomass burning sources. The regression calculation of residual CO vs. excess CH4 gives average slope of 0.12 (dimensionless ppbv/ppbv) for fall and winter, which is close to the middle range of earlier measurements (0.11) reported by
Table 1 Sources and concentrations of carbon monoxide (ppbv): Measurements around Portland, Oregona
a
Season and location
Total
Automobiles
Biomass burning
Bkgd
Non-auto%
Morning Fall Winter Spring Summer All Year Avg
680 510 370 320 430
330 240 210 190 230
250 140 30 50 90
110 130 120 90 110
37 27 8 16 21
Afternoon Fall Winter Spring Summer All Year Avg
400 350 250 230 310
210 180 150 130 170
80 40 0 10 30
110 130 120 90 110
20 11 0 4 10
Notes: (1) Concentrations have been rounded to two signi®cant digits. (2) The period of observations is from December 1995 to November 1997. (3) ``Biomass burning'' is the residual after subtracting the background concentrations and estimated automobile contributions. The last column is the % of CO not attributed to automobiles and includes the background concentrations.
114
M.A.K. Khalil, R.A. Rasmussen / Chemosphere: Global Change Science 1 (1999) 111±114
Andreae (1991). This agreement lends support to the idea that the CO not explained by automobile emissions and background concentrations during the late fall and winter is from biomass burning. Table 1 shows that the contributions of both automobiles and wood burning are greater during the morning hours (before noon) compared to concentrations after noon. This is expected from the diurnal variability of the emissions. For wood burning particularly, there are much larger emissions during the evening and night, which can result in large concentrations of CO and other pollutants during the early morning hours because of slow mixing during the night. At the same time, there are much reduced emissions from automobiles during the night. The CO concentrations, particularly during the early morning hours, can be dominated by wood burning. Commuter trac, after dawn, can add more CO into the atmosphere, reducing the relative impact of wood burning. Our afternoon samples are con®ned to the period between noon and 1600 h. During this time wood burning is greatly reduced and automobile emissions are high. The results in Table 1 are consistent with these observations. Although we have reported numerical results in Table 1, there are substantial uncertainties that cannot be easily quanti®ed. The variabilities of the observed concentrations and the source apportionment are quite large but not random and therefore cannot be used to obtain a quantitative estimate of the errors. For instance, on some particularly cold days there may be a large contribution from wood burning; on other warmer days the wood burning contribution may be small. This dierence creates a large variability but does not tell us what the error of our average estimate is. The uncertainties of the source apportionment arise mostly from the variability of the emission factors and the existence of sources of CO other than the ones we have assumed to exist.
Acknowledgements This work was funded in part by the National Science Foundation (GEO 9696080) and the Department of Energy (DOE DE-FG06-83ER60613). Additional support was provided by the Biospherics Research Corporation and the Andarz Co. We thank the students at the schools for collecting samples and their teachers, particularly Will Costello (VT), Shirley Grin (MA), and Will Mittman and Chris Watson both of Portland. We thank Don Stearns for laboratory work and Martha Shearer for data management.
References Andreae, M.O., 1991. Biomass burning: Its history, use and distribution and its impact on environmental quality and global change. In: Levine, J.S. (Ed.), Global Biomass Burning. MIT Press, Cambridge. Khalil, M.A.K., Rasmussen, R.A., 1988. Carbon monoxide in an urban environment: Application of a receptor model for source apportionment. JAPCA 38, 901±906. Khalil, M.A.K., Rasmussen, R.A., 1989. Urban carbon monoxide: Contributions of automobiles and wood burning. Chemosphere 19, 1383±1386. Rasmussen, R.A., Khalil, M.A.K., 1980. Atmospheric halocarbons: Measurements and analyses of selected trace gases. In: Aikin A.C. (Ed.), Proceedings of the NATO Advanced Study Institute on Atmospheric Ozone: Its Variation and Human In¯uences. Department of Transportation, Washington, DC, pp. 209±231.
Dr. Aslam Khalil is Professor of Physics at Portland State University. His research deals with the global budgets of trace gases, ozone depleting compounds, the greenhouse eect and global warming. Dr. Reinhold Rasmussen is Professor of Environmental Science and Engineering at Oregon Graduate Institute. Some 50 atmospheric trace gases are measured in his laboratory. These gases are involved in global warming, ozone depletion and air pollution. Drs. Khalil and Rasmussen have published more than 130 scienti®c papers together over 20 yr of collaboration.
Non-automotive sources of carbon monoxide in urban areas M.A.K. Khalil
a,*
, R.A. Rasmussen
b
a
b
Department of Physics, Portland State University, P.O. Box 751, Portland, OR 97207-0751, USA Department of Environmental Science and Engineering, Oregon Graduate Institute, P.O. Box 91000, Portland, OR 97291, USA Received 9 February 1999; accepted 10 February 1999
Importance of this Paper: Carbon monoxide is a criteria pollutant in the US necessitating monitoring and control of emissions. To do so requires an accurate knowledge of the local sources. Although automobiles are a major source in most areas, other sources can contribute signi®cant amounts in some cities. Our research shows that in urban areas of the Northwest, emissions from agricultural burning and wood burning for heat can be comparable to automotive emissions during fall and winter. Abstract Measurements of carbon monoxide were taken for two years at several locations in Portland, Oregon. There is a residual CO that is not accounted for by automobile emissions and the background concentrations. The ratio of excess CH4 (above background) to the residual CO is close to direct measurements of the emission factor of CH4 /CO from biomass burning which supports the idea that the residual CO, observed in our study, is from biomass burning. Biomass burning sources consist of agricultural burning in the surrounding areas and wood burning for heat. The diurnal and seasonal timing of the residual CO is consistent with our knowledge of these sources. We found that in Portland, during Fall and Winter, biomass burning contributes 20±40% of the carbon monoxide excess over background levels. Ó 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction In some urban areas, especially in the Paci®c northwest, there can be substantial contributions of CO from sources other than automobiles. These sources, mostly biomass burning, may now be relatively more important to the CO urban budgets than in previous years because per capita automobile emissions have declined during the last decade.
* Corresponding author. Tel.: +1-503-725-8396; fax: +1-503725-8550; e-mail: [email protected]
Measurements of CO and other gases were taken at four schools around the Portland area. The CO measurements were part of a larger experiment to measure greenhouse gases and ozone-depleting compounds in the atmosphere over continental areas in the GLOBE program (Global Learning and Observations to Bene®t the Environment). The measurements of carbon monoxide were aimed at understanding the seasonality of emissions from the various sources and to determine a continental signature for CO. Unlike measurements at the clean air background sites, these experiments show a complex pattern driven not only
1465-9972/99/$ ± see front matter Ó 1999 Elsevier Science Ltd. All rights reserved. PII: S 1 4 6 5 - 9 9 7 2 ( 9 9 ) 0 0 0 1 4 - 8
112
M.A.K. Khalil, R.A. Rasmussen / Chemosphere: Global Change Science 1 (1999) 111±114
by the well established cycles of trac but also by emissions from sources other than automobiles. At sites in Portland, and in many other parts of the Paci®c northwest, there are substantial contributions of CO from agricultural burning during fall and less so during winters. During winters wood burning for heating homes can contribute large amounts of CO in the Portland area and similar cities in Oregon and Washington. The results of measurements during the last two years are shown in Fig. 1. The data are shown as excess over the values at Cape Meares, which we consider to be a clean air site for this latitude. In previous work we used the concentrations of methyl chloride as an indicator wood burning pollution in Oregon and Washington. This technique was designed primarily for estimating the contribution of wood burning to ®ne particles in the environment but was also extended to evaluate the contribution of wood burning to CO. Our studies for Olympia, Washington, and Medford, Oregon, showed that during winters the contribution of wood burning to ambient CO levels was comparable in magnitude to emissions from automobiles (Khalil and Rasmussen, 1988, 1989). In the present work methyl chloride was not measured so we have taken a dierent approach by
Fig. 1. The monthly average concentrations of carbon monoxide at middle latitude urban and background sites (12/95-12/ 97). The remote continental site is at Minqin in the Gansu Province of China, 1/97-12/97; the urban sites are at GLOBE schools in Portland, Oregon, and the marine background site is at Cape Meares, Oregon. The excesses at the urban sites are likely to be caused by automobiles and biomass burning including agricultural burning and wood burning.
focusing on automobile emissions using total nonmethane hydrocarbons (NMHC, TO-12 method) as an index of automobile pollution. 2. Data analysis and results Air samples used here were obtained once a week from four schools in the Portland area. These schools are Kelly Elementary, Madison High School, Jackson Middle School, and Rose City School. Of these, Kelly school is the closest to a major highway with prevailing winds from the direction of the highway towards the school. Madison is at a slight elevation with prevailing winds from areas of low population outside the city. Data from Rose City School were not used as samples were collected for only a few months and stopped for logistical reasons. Samples were collected in internally electropolished 0.8L stainless steel containers using a pump developed for clean air sampling and used extensively in our research (Rasmussen and Khalil, 1980). To analyze the data (Fig. 1) we use the following model: COamb t COb t COauto t CObio-burn t;
1
COauto BfNMHCamb ÿ NMHCb g;
2
where [CO](t) is the concentration of carbon monoxide due to various sources and adding up to the observed value at time t. B the emission factor for automobile exhaust, expressed as the ratio of CO emissions to total NMHC emissions (TO-12). The subscript ``amb'' is for ``ambient'' air representing the actual measured concentration; b is for ``background''; ``auto'' is for ``automobile'' contributions, and ``bio-burn'' is the residual that we take to be from biomass burning as will be discussed in more detail later. We assume that during summers (June±August), the observed concentrations of CO in the Portland area schools are due entirely to automobiles and the background concentrations. During this time there is no intentional biomass burning, and the background concentrations are
M.A.K. Khalil, R.A. Rasmussen / Chemosphere: Global Change Science 1 (1999) 111±114
relatively constant as measured at Cape Meares, on the Oregon coast. When Eq. (2) is substituted into Eq. (1), it becomes a simple linear expression relating measured CO concentrations to concentrations of NMHC from which we can estimate A, the intercept and B, the slope, which is also the emission factor (as in Eq. (2)). The background concentration of NMHC is then estimated from A as NMHCb COb ÿ A=B:
3
The result of this calculation, for the composite data from all three schools mentioned above, is that the emission factor (excess CO/excess NMHC) is 3.5 0.5 ppbv ± m3 /lg, N 77, correlation coecient of 0.7 and [NMHC]b 6.5 lg/m3 (the value is the standard error). Based on this result we estimate the contribution of automobiles to the observed ambient concentrations using Eq. (2). The results are shown in Table 1 below. 3. Discussion During fall and winter seasons, the residual CO concentration tends to be the highest. We attribute this to biomass burning, which takes two forms in the Portland area. During fall the excess is likely to be contributed by agricultural burning, which is
113
widespread at this time. It includes burning of grass ®eld stubble and orchard prunings. Later in the season and in winters, wood burning for heating homes is the likely source of the excess. There is also forest slash burning throughout this period, but it is generally much farther away from the schools. The biomass burning contribution may be somewhat underestimated by our methods as some NMHCs are also emitted from wood burning, but the emission factors are some 10 times smaller (Andreae, 1991). The results show that there are times of the year when sources other than automobiles contribute signi®cantly to the atmospheric burden of CO in the Portland area. The results are comparable to the previous studies in Olympia (WA) and Medford (OR), although both are much smaller cities than Portland. The residual CO discussed above was compared to the excess methane during the fall and winter seasons. Excess methane was estimated as the dierence between the concentrations measured at the Portland schools and at Cape Meares during the same time. We expect that much of the excess methane during winter and late fall would come from the biomass burning sources. The regression calculation of residual CO vs. excess CH4 gives average slope of 0.12 (dimensionless ppbv/ppbv) for fall and winter, which is close to the middle range of earlier measurements (0.11) reported by
Table 1 Sources and concentrations of carbon monoxide (ppbv): Measurements around Portland, Oregona
a
Season and location
Total
Automobiles
Biomass burning
Bkgd
Non-auto%
Morning Fall Winter Spring Summer All Year Avg
680 510 370 320 430
330 240 210 190 230
250 140 30 50 90
110 130 120 90 110
37 27 8 16 21
Afternoon Fall Winter Spring Summer All Year Avg
400 350 250 230 310
210 180 150 130 170
80 40 0 10 30
110 130 120 90 110
20 11 0 4 10
Notes: (1) Concentrations have been rounded to two signi®cant digits. (2) The period of observations is from December 1995 to November 1997. (3) ``Biomass burning'' is the residual after subtracting the background concentrations and estimated automobile contributions. The last column is the % of CO not attributed to automobiles and includes the background concentrations.
114
M.A.K. Khalil, R.A. Rasmussen / Chemosphere: Global Change Science 1 (1999) 111±114
Andreae (1991). This agreement lends support to the idea that the CO not explained by automobile emissions and background concentrations during the late fall and winter is from biomass burning. Table 1 shows that the contributions of both automobiles and wood burning are greater during the morning hours (before noon) compared to concentrations after noon. This is expected from the diurnal variability of the emissions. For wood burning particularly, there are much larger emissions during the evening and night, which can result in large concentrations of CO and other pollutants during the early morning hours because of slow mixing during the night. At the same time, there are much reduced emissions from automobiles during the night. The CO concentrations, particularly during the early morning hours, can be dominated by wood burning. Commuter trac, after dawn, can add more CO into the atmosphere, reducing the relative impact of wood burning. Our afternoon samples are con®ned to the period between noon and 1600 h. During this time wood burning is greatly reduced and automobile emissions are high. The results in Table 1 are consistent with these observations. Although we have reported numerical results in Table 1, there are substantial uncertainties that cannot be easily quanti®ed. The variabilities of the observed concentrations and the source apportionment are quite large but not random and therefore cannot be used to obtain a quantitative estimate of the errors. For instance, on some particularly cold days there may be a large contribution from wood burning; on other warmer days the wood burning contribution may be small. This dierence creates a large variability but does not tell us what the error of our average estimate is. The uncertainties of the source apportionment arise mostly from the variability of the emission factors and the existence of sources of CO other than the ones we have assumed to exist.
Acknowledgements This work was funded in part by the National Science Foundation (GEO 9696080) and the Department of Energy (DOE DE-FG06-83ER60613). Additional support was provided by the Biospherics Research Corporation and the Andarz Co. We thank the students at the schools for collecting samples and their teachers, particularly Will Costello (VT), Shirley Grin (MA), and Will Mittman and Chris Watson both of Portland. We thank Don Stearns for laboratory work and Martha Shearer for data management.
References Andreae, M.O., 1991. Biomass burning: Its history, use and distribution and its impact on environmental quality and global change. In: Levine, J.S. (Ed.), Global Biomass Burning. MIT Press, Cambridge. Khalil, M.A.K., Rasmussen, R.A., 1988. Carbon monoxide in an urban environment: Application of a receptor model for source apportionment. JAPCA 38, 901±906. Khalil, M.A.K., Rasmussen, R.A., 1989. Urban carbon monoxide: Contributions of automobiles and wood burning. Chemosphere 19, 1383±1386. Rasmussen, R.A., Khalil, M.A.K., 1980. Atmospheric halocarbons: Measurements and analyses of selected trace gases. In: Aikin A.C. (Ed.), Proceedings of the NATO Advanced Study Institute on Atmospheric Ozone: Its Variation and Human In¯uences. Department of Transportation, Washington, DC, pp. 209±231.
Dr. Aslam Khalil is Professor of Physics at Portland State University. His research deals with the global budgets of trace gases, ozone depleting compounds, the greenhouse eect and global warming. Dr. Reinhold Rasmussen is Professor of Environmental Science and Engineering at Oregon Graduate Institute. Some 50 atmospheric trace gases are measured in his laboratory. These gases are involved in global warming, ozone depletion and air pollution. Drs. Khalil and Rasmussen have published more than 130 scienti®c papers together over 20 yr of collaboration.