Tula industrial complex ( Mexico ) emissions of SO 2 and NO 2 during the MCMA 2006 field campaign using a mobile mini-DOAS system

The Mexico City Metropolitan Area (MCMA) has presented severe pollution problems for many years. There are several point and mobile emission sources inside and outside the MCMA which are known to affect air quality in the area. In particular, speculation has risen as to whether the Tula industrial complex, located 60 km northwest of the MCMA has any influence on high SO 2 levels occurring on the northern part of the city, in the winter season mainly. As part of the MILAGRO Field Campaign, from 24 March to 17 April 2006, the differential vertical columns of sulfur dioxide (SO2) and nitrogen dioxide (NO2) were measured during plume transects in the neighborhood of the Tula industrial complex using mobile mini-DOAS instruments. Vertical profiles of wind speed and direction obtained from pilot balloons and radiosondes were used to calculate SO 2 and NO2 emissions. According to our measurements, calculated average emissions of SO2 and NO2 during the field campaign were 384±103 and 24±7 tons day−1, respectively. The standard deviation of these estimations is due to actual variations in the observed emissions from the refinery and power plant, as well as to the uncertainty in the wind fields at the exact time of the measurements. Reported values in recent inventories were found to be in good agreement with calculated emissions during the field campaign. Our measurements were also found to be in good agreement with simulated plumes. Correspondence to: C. Rivera (claudia.rivera@chalmers.se)


Introduction
The Tula industrial complex is located northwest of the Mexico City Metropolitan Area (MCMA), in the State of Hidalgo, Mexico.It is close to a number of other industries in the Tula-Vito-Apasco industrial corridor (Fig. 1).According to the latest information from the environmental authority 323 ktons per year (ktpy) of SO 2 and 44 ktpy of NO x are released in this region.The main emitters are the Miguel Hidalgo Refinery (MHR) and the Francisco P érez Ríos Power Plant (FPRPP) (SEMARNAT, 2002).Other industries such as cement plants, open-sky mines and agricultural activities are also responsible of important particle matter emissions into the atmosphere and for soil degrada-Introduction

Conclusions References
Tables Figures

Back Close
Full tion of the area.The MHR processes 296 thousand barrels per day (TBD) of crude oil, representing 20% of the total refining capacity in the country.Final products are mainly gasoline, diesel, turbosine, kerosene, and other subproducts used to improve fuel specifications.To satisfy internal energy demand, the refinery consumes gas and liquid residuals of the refining processes, often of poor quality (3.8% weight of sulfur content).The FPRPP has an installed capacity of 2000 megawatt (MW), distributed in 9 units combining vapor (five) and combined cycle (four) technologies.These two industries contribute almost 90% of SO 2 and 80% of NO x of the total emission in Hidalgo State (IMP, 2006).NO 2 is of special interest due to its potential for undergoing photochemical reactions and producing, together with volatile organic compounds; ozone, peroxyacetyl nitrate, nitric acid, formaldehyde and formic acid, among others (Finlayson-Pitts and Pitts, 2000).Long term exposure of humans to NO 2 has negative health effects such as lung function decrease and higher risk of respiratory symptoms.SO 2 was shown to lead to reductions in FEV1 (Forced Expiratory Volume in 1 s) and other indices of ventilatory capacity, as well as to increased mortality and hospital emergency admissions for total respiratory cases and chronic obstructive pulmonary disease at lower levels of exposure (WHO, 2000).In addition to their negative effects on human health, SO 2 and NO 2 tend to form sulfuric and nitric acids respectively, which through dry and wet deposition contribute to damage of plants and buildings.
Because of its large emissions, the Tula industrial complex is thought to affect air quality in the MCMA.Since the early 2000's, the atmospheric monitoring network of the MCMA, has been reporting unusual high SO 2 concentrations during night time at the northern part of the MCMA.According to a 2003 report of the "Program to Improve Air Quality in Mexico City Metropolitan Area 2002-2010" (CAM, 2003) in some occasions SO 2 concentrations in the north part of the city have exceeded the Mexican Air Quality Standard (0.13 ppm 24 h-average).It has not been possible to attribute this to irregular operations of industries located in the surrounding area.For this reason, it has been questioned whether the Tula industrial zone is responsible for worsening air quality These were found to be to the northwest of the MCMA, in the direction of the Tula industrial zone.Forward modeling using measured emissions from the complex showed that these could account for the large SO 2 peaks observed in the MCMA, but that they only contribute 20% of the long term average concentration.Local sources are the dominant cause of baseline SO 2 levels.The Popocat épetl volcano is another large point source that affects the MCMA and leads to increased sulfate aerosol production in the city (Raga et al., 1999) even though impacts during MCMA-2003 were shown to be possible but could not be differentiated from the local levels (de Foy et al., 2007).
To address the issue of emissions from the refinery and the power plant, the total fluxes of SO 2 and NO 2 were determined by measurements of their respective integrated vertical columns in the neighborhood of the Tula industrial zone.The importance of this study relies on the possibility to verify published emission inventories and provide detailed emission information for modeling studies.

Mini-DOAS
SO 2 and NO 2 emissions have been determined using Differential Optical Absorption Spectroscopy (DOAS).DOAS is a spectroscopic technique based on the absorption of electromagnetic radiation by matter, allowing the remote detection of trace gases.DOAS instruments have been developed in a wide variety of designs and measurements can be performed using several experimental setups.In this case, passive DOAS using scattered sunlight in Zenith Scattered Light mode has been applied.The Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion key components of passive DOAS instruments are: spectrometer, detector, receiving optics (telescope), additional electronics and computer.Additional software to control the system and perform evaluations is needed, as well as databases of the absorption cross sections of the species to measure (Platt and Stutz, 2008).SO 2 and NO 2 emissions were determined by integrating the total number of molecules in a vertical cross-section of the gas plume, and multiplying them by the wind speed at plume height.Traverses of the plume using a mini-DOAS instrument installed on a mobile platform were performed.The instrument is referred to as mini-DOAS due to its small size and low weight.The mini-DOAS collects scattered ultraviolet sunlight using a telescope equipped with a quartz lens; the light enters a spectrometer through an optical fiber (Galle et al., 2002b).The spectrometers used in this study have a spectral resolution of about 0.6 nanometers (nm) and spectral range of 280-420 nm for SO 2 and 336-480 nm for NO 2 .Similar equipments have been successfully used to quantify SO 2 emissions from volcanoes (Bobrowski et al., 2003;Edmonds et al., 2003;Galle et al., 2002aGalle et al., , 2003aGalle et al., , b, 2006;;McGonigle et al., 2002McGonigle et al., , 2003;;Mori et al., 2006;O'Dwyer et al., 2003), industries (McGonigle et al., 2004;Rivera et al., 2009) and cities (Johansson et al., 2008).
A general spectral evaluation procedure was applied to every spectrum collected during a traverse, starting with dark current correction of every recorded spectrum, division with a "clean-air" reference spectrum, the application of a high pass filter to separate broad and narrow band spectral structures and finally a logarithm of the spectrum.Afterwards a non-linear fitting of a reference spectrum to the measured spectra is made, in this way obtaining the total column of the gases of interest.The fitting intervals used correspond to 307-317 nm and 415-455 nm for SO 2 (Bogumil et al., 2003) and NO 2 (Vandaele et al., 1998) cross sections respectively.The O 3 (Voigt et al., 2001) cross section was included in the fitting procedure as well.
The mini-DOAS instrument is complemented with a Global Positioning System (GPS), both of them connected to a laptop computer and controlled by custom-built software (MobileDOAS, Chalmers University) which collects and evaluates acquired Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion spectra in real time.Measurements were performed traversing the plume, at different distances from the source, perpendicular to the plume direction.For each collected spectrum, GPS data was recorded providing time and position before and after the spectrum was collected.Fluxes from traverses were calculated multiplying total columns by the distance traversed perpendicular to the wind direction, and the wind speed at plume height.The total column is defined as the integral of the species of interest along one line (the average concentration of the species times the length of the path).The corresponding methods for obtaining meteorological information are discussed in the following.

Meteorological measurements
For accurately quantifying emissions, the mini-DOAS technique requires wind speed and wind direction information at the height of the plume.Therefore, these parameters were measured at surface and aloft, using different methods.The meteorological equipment was deployed at an IMP site (Longitude 99 • 16 24.4W and Latitude 20 • 2 48.6 N), located inside the refinery facilities.A surface meteorological station (MAUS-210 from Vaisala) registered continuously those variables 10 m above the ground, while vertical measurements were performed using pilot balloons and radiosondes.
Wind data from pilot balloons was used to calculate SO 2 and NO 2 emissions for measurements performed between 24 and 26 March 2006 because they were more frequently launched than radiosondes.For measurements performed after 27 March 2006, results from radiosondes were used instead, being our only source of wind data available at plume height.amount of commercial helium, resulting in a known uplift force.The ascent rate was estimated by intercomparisons with radiosonde data (see Fig. 2).Once the balloons were launched, they were tracked by theodolites (Tamaya model TD3), and both azimuth and elevation were registered in 10 s intervals.The position of the balloon in space was calculated from these angles and the ascent rate, and wind speed and direction were derived.

Radiosondes
A Digicora II radiosonde system from Vaisala (Mod.SPS-220) was used for upper air sounding measurements of pressure, temperature, relative humidity and the horizontal wind vector.The radiosonde system consists of a ground-base station that receives and stores the incoming signal from the radiosonde transmitter; a radiosonde that supports all meteorological sensors, and a meteorological balloon that raises the sonde from the ground to the upper atmosphere.The radiosondes were launched four times a day from 16 March to 22 April 2006 at 8, 12, 15 and 18 h (local time) by using 300 gr latex-helium-filled balloons.

Modeling
Forward plume simulations were carried out using Lagrangian particle trajectories.3 Results and discussion

Meteorological conditions
Meteorological conditions in the Tula industrial complex were variable during the monitoring period.From 23 March 2006 a Cold Surge over the Gulf of Mexico produced strong northerly winds persisting for two days (Fast et al., 2007).Cold Surges are a characteristic feature in Central Mexico, bringing cold humid air south into the MCMA.This leads to reduced vertical mixing and increased rain and cloudiness.After the Cold Surge, conditions in the basin remained humid with weak southward transport at night and afternoon convection events until the end of March (de Foy et al., 2008).During April, most of the days were clear or partly cloudy without rain due to a persistent high pressure system.Wind speeds were higher during the first days of the field campaign and less intense at the end.Minimum wind speeds were recorded during night time and first morning hours while maximum values were recorded between 19:00-20:00 local time.Maximum temperatures occurred between 16:00-17:00 and minimum temperatures between 07:00-08:00 local time.Relative humidity followed an expected behavior with a maximum during night time and a minimum coinciding with high temperatures.Wind direction before 09:00 was from south-southeast, a transition period was systematically observed between 09:00-11:00 where northerly winds appear.After midday, wind direction turned from north-northwest -north -north-northeast and this condition was maintained until 22:00 where winds turned again from south-southeast until 09:00 h of the next day.This cycle was continually observed over the entire measurement period.

Observed emission factors
A typical measurement of SO 2 and NO 2 columns in the Tula industrial complex is depicted in Fig. 3  ries of NO 2 and SO 2 emission fluxes calculated from the conducted measurements during the field campaign.In addition Table 1 shows a summary of all the measurements performed during the field campaign.NO 2 emissions were measured between 24-27 March, and SO 2 emissions were quantified between 24 March-17 April 2006.A total number of 96 transects were performed in order to determine the flux emission of SO 2 and NO 2 in the region.Actual emission variations due to changes in the processes, as well as atmospheric perturbations along the measurements in a particular transects are main responsible factors of the flux emission variation.From the statistical point of view, the larger the number of transects measurements, the better the flux determination.
During MCMA 2006, SO 2 emissions from Tula's industrial complex yielded 17 786±13 779 kg/h.NO 2 emissions accounted for 1044±970 kg/h.Both SO 2 and NO 2 quantified emissions present large standard deviation, and SO 2 emissions are an order of magnitude larger than the quantified NO 2 fluxes.The high variability on the flux determinations is associated to the actual emission variation on the SO 2 sources in the region, as well as to the uncertainty associated to the wind field at the specific time of the measurements.

Comparison with emission inventories
Table 2 shows a comparison of our results with published emission inventories.In order to develop the first National Inventory of Emissions in Mexico (base year 1999), a multidisciplinary effort between national and international institutions was made.As part of this inventory, PEMEX provided most of the information regarding combustion and process emissions.As for power generation plants emissions, data was provided by Mexico's Energy Secretariat (SENER).Almost 70% of emissions from combustion were calculated using emission factors from US EPA, 1995, Sect. 1 (AP-42), and the rest were based on measurements reported by the power stations themselves (SEMARNAT-INE, 2006).
During MCMA 2003 field experiment, SO 2 emissions from the Tula-Vito-Apasco industrial complex were quantified using zenith sky UV spectroscopy, the same technique applied in this study.
The Mexican Environment and Natural Resources Secretariat (SEMARNAT) provided emission data for 2005, based on DATGEN (General Data).DATGEN is a database containing emissions inventories information (principally from combustion processes) from fixed sources of federal and state jurisdiction, located in areas where air quality management plans have been developed.
Surprisingly, regardless the difference in emission determination approaches applied, both SO 2 and NO 2 emissions quantified by the DOAS technique are comparable with emission inventories reported, particularly those after the year 2002.In the Sustainable Development Annual Report (2006), the Mexican Petroleum Oil Company (PE-MEX) informs an annual average reduction of 6.3% of emissions into the atmosphere during the period 2001-2006 in the company (PEMEX, 2006).Assuming that this annual average rate applies also for the MHR, current emissions reported in 2006 are consistent with those reported in 1999 (approximately 40% in reduction in six years).
In spite of the industries in Mexico reporting SO 2 emissions applying the AP-42 emissions factors, uncertainty on these values is low, because almost 100% of the sulfur content on fuels is emitted as SO x .This may be the reason why the accordance of SO 2 reported emissions and measurements.
It is important to note that emission inventories are given in NO x while our instruments quantify NO 2 .The knowledge of the NO 2 /NO x ratio is then important in order to derive NO x emissions from our measurements.Because of the well known reactions in which NO is oxidized to NO 2 in plumes exiting the stacks of industrial facilities Introduction

Conclusions References
Tables Figures

Back Close
Full Screen / Esc Printer-friendly Version Interactive Discussion (Finlayson-Pitts and Pitts, 2000), we therefore may assume that the quantified NO 2 represents only part of the total released NO x and the discrepancies between NO x inventories and quantified NO 2 may be partially explained by the amount of NO that has not yet been oxidized.

Comparison of measured and modeled plumes
Comparisons between measured and simulated emissions from the Tula industrial complex were made for 26 March 2006 (Fig. 5) and 4 April 2006 (Fig. 6).On 26 March, according to the DOAS measurements, the plume from the industrial complex was continuously shifting: early in the morning the plume dispersed towards the east, moving towards the southsoutheast by noon and turning back towards the southeast by late afternoon.This is correctly simulated by the model, with the plumes initially moving to east, and then turning to the south and back towards the east again.There are slight differences in the timing and strength of the shifts.The model is not able to represent the split plume at 11:55 with one part going east and the other south, although this is to be expected as these features are smaller than resolution of the wind simulations (3 km grid cells, output every 1 h).At 14:18, the simulated plume is much narrower than the measurements, suggesting that there is insufficient dispersion in the model.
As the spatial and temporal scales of this are below the resolution of the model, this suggests that turbulent mixing should be increased in the trajectory simulations.On 4 April, a wide plume was observed at noon, gradually narrowing towards the afternoon.
In general, the plume was more direct moving straight to the southwest throughout the day.This was correctly represented in the model, including the narrowing of the plume as the winds became stronger and the transport faster.In summary, there is good agreement between the measured and simulated plumes suggesting that the model is capable of representing the plume transport, and that the measurements correctly captured the entire plume.
2.2.1 Pilot balloonsPilot balloons were launched during 24-26 March 2006 with a frequency of 1-2 h during daytime, in order to obtain information on the vertical distribution of wind speed and direction at plume height.The pilot balloons were filled with a pre-determined Introduction The mesoscale meteorology was simulated with the Weather Research and Forecast (WRF) model version 3.0.1(Skamarock et al., 2005).Three nested grids were used with resolutions of 27, 9 and 3 km, and winds fields were saved every hour.The simulation options and model evaluations are described in de Foy et al.(2009).Stochastic particle trajectories were calculated with WRF-FLEXPART(Doran et al., 2008;Stohl et al., 2005).1800 particles per hour were released from a single stack representing FPRPP and 720 particles per hour from a stack representing MHR from a height of 70 to 80 m and 30 to 50 m respectively, to account for stack height variations as well as variable plume rise.
for 26 March 2006.This measurement yielded an emission of

a
The standard deviation refers to the variability of single measurements conducted during the same day and not to the uncertainty of the measurements per se.
Mexican Petroleum (PEMEX) and Federal Commission of Electricity (CFE) companies located at the Tula industrial complex however claim to comply with emission regulations, and therefore not to affect air quality.De Foy et al. (2007) used Concentration Field Analysis using backward trajectories during the MCMA-2003 field campaign to identify possible source regions of SO 2 .
estimated emissions from power Introduction on fuel consumption and energy generation data provided by SENER as well as emission factors for specific power generation plants.The methodology used followed the recommendations of the Emissions Inventory Improvement Program of the US EPA; the emission factors used were obtained from the EPA's AP-

Table 1 .
Calculated emissions from Tula's industrial complex during the MCMA 2006 field campaign yielded an average of 155.8±120 ktpy for SO 2 and 9±8.5 ktpy for NO 2 .These emissions are comparable to the more recent emission inventories, but lower than the inventories from 1999 and earlier.SO 2 emissions show average reduction over the years (approximately 40% in six years) and compare well with 2006 emissions reported by PEMEX for the MHR(PEMEX, 2006).On the other hand NO 2 calculated emissions show lower values than previous NO x emission inventories.This discrepancy may be explained by an incomplete oxidation of NO to NO 2 .The plume was simulated with forward particle trajectories using the measured emission rates.The good agreement between the simulated plume transport and the column measurements suggests that the model is capable of reproducing dispersion from the Tula industrial zone and brings supporting evidence that the column measurements correctly captured the plume.A remaining question is the large standard deviation of the measurements performed during the field campaign.It is thought that the reasons for them are associated with changes in real emissions from the refinery and power plant, as well as unavailability of wind fields at the exact time of every measurement.Variability in fluxes could be caused by plume meandering and diagonal transects as well.Quantified emissions during April show less standard deviation, coinciding with more defined plumes.Both simulations and observations during March attest for plumes shifting over short periods of time.Detailed information about production and performance of the Miguel Hidalgo refinery and the Francisco P érez Ríos power plant during the field campaign would yield an improved comparison between our measurements and reported values, however it was not available.Introduction Summary of measurements at Tula industrial complex.