Moderator: Telma Castro, Universidad Nacional Autónoma de México
Speakers: Mario Molina, MIT; Marisa Ruiz, Instituto Mexicano del Petróleo and Graciela Raga, Universidad Nacional Autónoma de México.

Presentation by Professor Mario Molina

Professor Mario Molina raised a number of questions to open the session:

1. Could particulate matter (rather than ozone) be the primary cause of health impacts?
2. Aside from what is already monitored, what else would be interesting or worthwhile to monitor, that might give us insights into the problem?
3. What are the relative effects on health of indoor vs. outdoor pollution? What exposures result from such activities as cooking and driving?
4. How good are current models to assess control programs? Do we have the right data for these models, and what is the quality of the data (emissions, measurements, reaction rates)? How useful are models fine-tuned elsewhere (esp. in the U.S.) for application in Mexico, and how can we test these models in Mexico?
5. What is the specific situation in Mexico City? The ambient atmosphere appears to be hydrocarbon rich, which suggests that NO_x controls would be more effective, but some results seem to contradict this. What is known about hydrocarbon speciation?
6. How much of the PM is secondary? How do particulates influence gas phase chemistry? How do particles age in the atmosphere? Do they become more acidic? More hydrophilic?

Presentation by Dr. Marisa Ruiz

Dr. Marisa Ruiz offered an overview of the work already done on air pollution in Mexico City, and of research underway at IMP.

• Measurements since 1986 show that overall NO_x concentrations have shown little change, but that NO_x to the northeast and center of the city is slightly increasing, and slightly decreasing southeast. The center has the highest readings.
• Measurements from IMADA­AVER in 1997 show that VOC concentrations are highest in the center of the city. At all sites, paraffins make up roughly half the measured organics, although speciation varies in different sectors of the city.
• Measurements of VOC concentrations from 19921997 show a decreasing trend and higher concentrations in the spring than in winter.
• Smog chamber experiments at IMP were used to make ozone isopleths, which can be used to estimate the effects of changes in NO_x and hydrocarbon emissions on ambient ozone concentrations. (The relationship between NO_x, hydrocarbons, and ozone is non-linear. Depending on where you are on the curve, VOC vs. NO_x reductions have different impacts.) They looked at three groups of conditions: low morning, medium, and high. Results indicate that NO_xreductions would be more effective than hydrocarbon reductions at reducing ozone, and they have suggested decreasing both.
• 1997 VOC speciation data indicate that alkanes (e.g. propane and butane) are more abundant than aromatics, and benzene concentrations are low.
• The measured ratio of ambient VOCs to NO_x in Mexico City is often in the range of 15­35, which is much higher than the ratio in Los Angeles (~8). The decrease in this ratio since 1993 is due to the reduction in VOC emissions.
• Emissions inventories from 1996 show that mobile sources account for a large fraction of emissions of hydrocarbons, NO_x, and total particulates.
• Large differences between emissions inventories over the past decade show the effect of changes in emissions over time, and also differences in the methods used for assessment.
• As an example of the emissions inventory, the IMADA project estimates that most of the ammonia emissions are from animals and sewage.
• Geographic Information System (GIS) mapping shows the daily spatial and temporal distribution of emissions in Mexico City, showing the effect of traffic patterns. GIS mapping has also been used, for example, to show the locations of gas stations in the city.

Presentation by Dr. Graciela Raga

Dr. Graciela Raga addressed two main topics:

Characterization of aerosols and trace gases in Mexico City

• Trace pollutants in the atmosphere come from primary emissions (CO, NO_x, SO₂), and secondary reactions of precursor emissions (ozone formed from NO_x and VOCs). Several measurement campaigns have been carried out to address concentrations of trace gases. Principal participants were RAMA (government), IMP, and UNAM (CCA & IGF).
• Total suspended particulates (TSP) have been measured in the basin for more than 10 years, with three years of hourly measurements of PM₁₀. Columnar concentration of optically active aerosol particles (radii from 0.01­10.0 mm) have been estimated for more than 30 years, using direct solar radiation measurements. Much less is known about size distribution (there are more than 10 years of records of surface size distributions measurements for particles of radii only from 0.003 mm to 0.5 mm) and composition of the particles (microphysical characterization), and this knowledge is needed for better modeling. There is also a need to understand optical characteristics, including extinction coefficients and single scattering albedo.¹
• Measurements suggest that aerosols in Mexico City are highly absorbent of solar radiation (suggesting a high elemental carbon content), which may affect local climate.
• Research/modeling issues: (1) Aerosol modeling efforts are underway but are limited by the availability of data. (2) There are good datasets for trace gases, but little research using these data. (3) There are good data for horizontal surface coverage but limited data for vertical distributions. (4) Other compounds must also be monitored. It is not clear that the substances now measured are the most appropriate to serve as indicators of health impacts or air quality.

Linkage between urban pollution and global climate change

• Modeling suggests a transport of gases to the upper troposphere and that the Mexico City basin cleans itself (to some extent) daily. Despite the mountains, the boundary layer is quite high (2 km), allowing pollutants to escape the basin.
• Export of gases and aerosols from Mexico City over a broader region may be important for climate. Are gases such as NO_x, ozone, and PAN transported outside the city?
• Primary gases such as CO, SO₂, and NMHC, while not greenhouse gases, may be useful as tracers of the urban plume.
• CO₂ is important with respect to global warming but no measurements exist for Mexico City.
• Ozone is an important GHG; is it exported, or created downwind from precursors?
• Little is known about the export of aerosols, how they evolve during transport. There are questions concerning their effect on regional radiative balance, aerosol-cloud interactions (and regional meteorology), and their role in regional photochemical forcing. Do primary particles play a role in the condensation of secondary particles as they are transported outside the basin? Some studies indicate that particles are larger at the top of the boundary layer. In particular, aerosols might affect the stratification of the boundary layer, decreasing vertical transport and therefore decreasing transport out of the basin.
• Regarding the interactions between particles and ozone, aerosols may decrease the radiation flux and therefore decrease ozone formation.

Discussion Summary

Some additional specific information was elicited as part of the discussion:

• During the forest fires in the basin, ozone concentrations were up, along with particles.
• The speciation of hydrocarbons is essential for photochemical modeling. UNAM has measured emissions from ten different industries/services, but except for gasoline distribution they don't have speciation data yet.
• The speciation of hydrocarbons emitted from refueling differs from speciation emitted in exhaust emissions.
• There are high concentrations of polycyclic aromatic hydrocarbons (PAHs) in PM_2.5. Studies carried out by Dr. G. Ruiz Suarez and others also show naphthalene, contrary to common wisdom, and higher concentrations of light aromatics in PM_2.5.

The speciation of hydrocarbons as it relates to fuel formulation and evaporative vs. combustion emissions was discussed. Proposed changes in the refining of PEMEX gasoline could alter the speciation of hydrocarbons. Since Mexico City has the strictest fuel specifications in the country, a lot of imported stock with high octane is used­about 10% at 93 octane (premium) and the rest at 87 octane (regular unleaded). Further changes in the city's fuel specifications are not anticipated.

Gasoline vapor recovery programs are now widespread in Mexico City. Following the US model, Mexico has progressed from Stage 0 (at gasoline distribution points) to Stage I (during delivery of fuel to gas stations) in 1994. Stage II (refueling vapor recovery) is underway. About 100 out of 400 gas stations in the area have operating Stage II vapor recovery equipment. While the requirement was that all gas stations would be equipped by April of 1999, it is projected that roughly 90% would install Stage II by 2000. It was noted that equipment manufacturers overlooked Mexico City's lower atmospheric pressure, and were claiming higher recovery rates than were actually achieved, which is about 85%.

Poorly maintained cars have 4­8 times greater emissions, with diesel vehicles being of particular concern; even EPA '94 type designs with sealed injectors are often tampered with within months of purchase. Some CNG and LPG vehicles have also been found to have worse emissions than gasoline vehicles because of poor maintenance. Part of the problem is that there is no culture of keeping vehicles well-maintained, and mechanics often have little training. In particular, few mechanics are capable of performing maintenance on diesel engines.

Diesel vehicles are suspected of being important sources of PM. There has been some direct and indirect measurement of soot in particles. An effort was conducted by IMP and DRI in 1997 (the IMADA-AVER program) to estimate carbon content of PM_2.5. IMADA measurements show a high content of elemental and organic carbon (~15%). The high absorption of light by small particles suggests a lot of soot. They also examined particles in an electron microscope and visualized the carbon structure of the smallest particles. These techniques, however, are not quantitative. A preliminary inventory of PM₁₀ sources was completed last year and vehicles (both gas and diesel) appear to be very important. Using DRI/IMP information to estimate the soil/dust contribution at 50%, the remaining half was estimated to be about 1/3 from gasoline vehicles, 1/3 from diesel vehicles, and 1/3 from industry. At the PM_2.5 level, the relative proportion from diesel increases, and the exposure potential is even higher.

Issues related to the use of LPG fuel were also discussed. Mexico City is unique because of the widespread use of LPG for cooking. Flames are poorly calibrated, so they might be an important source of carbon. While Mexico City gasoline consumption is 20,000 barrels/day, LPG consumption is 70,000 barrels/day. This is a higher proportion of LPG to gasoline than anywhere in the world, and emissions are not well understood.

The role of weather was brought up. 1998 was a hot year, yet the city saw a constant reduction in ozone concentrations. One study of 18 important ozone episodes at Pedregal (an area in the southwest of the city) noted a consistent wind direction of 185. The most important ozone episodes showed the wind coming from the same direction.

Considerable discussion addressed data quality and emission inventories. There is some concern about the quality of Mexican data. In Mexico, little quality assurance and quality control is carried out, possibly due to prohibitive costs. Data quality must be considered carefully, and it is important to quantify and reflect in the models any resulting uncertainty. Inventories of emissions are also problematic. US EPA methods for assessing emissions may not be correct for Mexico City. Inventories are easy to generate since coefficients are multiplied by economic activities, but there is uncertainty in these numbers. No studies have been conducted on these coefficients, and just last month such a study was cancelled due to cost. One way to improve data quality generally is to have more people looking at the data and doing research with the data, performing statistical analysis, etc. Another possibility is to begin to use inverse methods to work backwards from measurements to assess the quality of the inventory data. Today, there is better chemical specificity and methods to propagate uncertainty. There is a need to go beyond the standard reply that emissions are uncertain and begin to put bounds on those numbers.² This is true for GHG inventories as well as inventories of ozone precursors and particulates.

The connection between monitoring efforts and health effects was raised. For example, if particles are changing blood viscosity (as reported in the health effects session), what elements or compounds should be monitored?

Finally, the role of science was discussed. Three-dimensional modeling of particular air pollution episodes may not be the most appropriate method of assessing effects of changes in emissions from control programs. What other, possibly simpler, techniques could we use, and how reliable are these?

Professor Molina presented the following summary of this session:

• The reliability of existing emissions inventory in Mexico is questionable (as is often the case in the US). Ways to generate better data need to be explored.
• Vehicle maintenance has a large impact on emissions.
• Fuel reformation may also lead to significant air quality improvements.
• Diesel vehicles are probably large sources of fine particles; emission controls should be explored.
• LPG is used extensively for indoor cooking and heating. The corresponding emissions should be better characterized (there is information on the contributions to the hydrocarbon budget from LPG leaks from atmospheric measurements).
• The effects of changes in emissions resulting from various control strategies should be explored using appropriate modeling techniques.

Footnotes

1. Since March 1999 and for periods of 6­8 months (for recalibration necessities), IGF has established a collaboration with NASA AERONET Federal Network (AErosol RObot NETwork) through which a sun photometer was installed at the IGF building roof, forming part of a world network. Goals include the assessment of the basic optical characteristics and size distribution of the aerosol particles.
2. Greg McRae pointed out that Gustavo Sosa of IMP will be spending three months at MIT learning these tools and techniques.