Sources and transport of 1 14 C in CO 2 within the Mexico City Basin and vicinity

. Radiocarbon samples taken over Mexico City and the surrounding region during the MILAGRO ﬁeld campaign in March 2006 exhibited an unexpected distribution: (1) relatively few samples (23%) were below the North American free tropospheric background value (57 ± 2‰) de-spite the fossil fuel emissions from one of the world’s most highly polluted environments; and (2) frequent enrichment well above the background value was observed. Correlate source tracer species and air transport characteristics were examined to elucidate inﬂuences on the radiocarbon distribution. Our analysis suggests that a combination of radiocarbon sources biased the “regional radiocarbon background” above the North American value thereby decreasing the ap-parent fossil fuel signature. Likely sources include the release of 14 C-enhanced carbon from bomb 14 C sequestered in plant carbon pools via the ubiquitous biomass burning in the region as well as the direct release of radiocarbon as CO 2 from other “hot” sources. Plausible perturbations from local point “hot” sources include the burning of hazardous waste in cement kilns; medical waste incineration; and


Introduction
With over 10 million inhabitants, megacities are a major source of air pollution given their high traffic densities, energy consumption rates, industrial processes, and elevated levels of biomass combustion from land clearing, trash burning and both domestic and industrial use of wood fuels . Emissions from megacities located within developing countries can differ from US and European cities as less stringent regulations allow higher levels of more complex gases and aerosols to be emitted (Raga et al., 2001). As their number increases worldwide, there is a growing recognition that airborne emissions from these large urban and industrial centers change the chemical content of the downwind troposphere, influencing both air quality and climate change on multiple scales (Molina et al., 2008).
The Mexico City metropolitan area (MCMA) is the second largest megacity worldwide, the most populous city in North America, and was the initial case study for MILAGRO (Megacity Initiative: Local And Global Research Observations); an international collaborative project to examine the behavior and export of atmospheric pollutants generated in megacities. The interaction of physical and cultural factors in Published by Copernicus Publications on behalf of the European Geosciences Union. 4974 S. A. Vay et al.: Sources and transport of 14 C in CO 2 the Valley of Mexico and surrounding region creates a spectrum of air pollutants extremely diverse in chemical characteristics (deBauer and Krupa, 1990;Collins and Scott, 1993). The NASA DC-8 was one of seven instrumented research aircraft participating in MILAGRO, providing in situ and remote observations of trace gases and aerosols from a regional-scale perspective. Instrumentation for measurement of carbon dioxide and the collection of whole air samples for radiocarbon analyses was integrated on the DC-8 to investigate the various local and regional sources contributing to the measured total atmospheric CO 2 signal.
Anthropogenic perturbations, mainly attributable to landuse changes, fossil fuel burning, and atmospheric nuclear detonations, have markedly affected the natural balance of the 14 C in CO 2 cycle (Suess, 1955;Tans et al., 1979;Stuiver and Quay, 1981;Rozanski et al., 1995;and Caldeira et al., 1998). The isotopic analysis of CO 2 in the lower atmosphere has resulted in a better understanding of the 14 C exchange with the biosphere and the role of anthropogenic activity (Zondervan and Meijer, 1996;Dutta et al., 2002). More than 40 years of direct tropospheric 14 CO 2 observations from maritime and continental stations in both hemispheres are now available since the nuclear test ban treaty in 1962 (Mook, 1980;Rozanski et al., 1995;Levin and Hesshaimer, 2000;Kitagawa et al., 2004). Seasonal and regional variability in atmospheric 14 C is clearly visible in many contemporary observations and has the potential to provide additional information about the source, age, and magnitude of regional fluxes (Randerson et al. 2002).
The direct effect of fossil fuel emissions is arguably the most significant influence of urbanization on the carbon cycle (Pataki et al., 2006). 14 C is a particularly sensitive tracer of fossil fuel emissions (Hsueh et al., 2007) because fossil fuel-derived CO 2 is the only source of atmospheric CO 2 that is devoid of 14 C (Suess, 1955;Levin et al., 1980;Turnbull et al., 2009). When combined with atmospheric CO 2 concentration measurements, observations of 14 C can potentially be used to constrain fossil fuel emissions at local and regional scales. Quantifying fossil fuel CO 2 emissions by isotopic analysis requires knowledge of the clean air isotopic composition (Levin et al., 1980). A measured 14 C/ 12 C ratio below the background value has been used to directly yield the fraction of fossil fuel CO 2 (Levin et al., 1980;Zondervan and Meijer, 1996).
The purpose of this paper is to present insight into the processes regulating the distribution of atmospheric 14 C in the MCMA environment in March 2006, and to discuss the implications of those measurements as they pertain to alterations to the carbon cycle and the quantification of anthropogenic carbon sources.

Field deployments
In 2006, NASA's Tropospheric Chemistry Program (TCP) sponsored the INTEX-B (Intercontinental Chemical Transport Experiment-B) mission which consisted of two separate field deployments. The initial phase investigated the extent and persistence of outflow of pollution from Mexico during MILAGRO (1-30 March 2006) with flights based out of Houston, Texas (4-19 March 2006). The DC-8 then transited to Moffett Field, California, and after a three-week break, began making measurements over the North Pacific to quantify the impact of the long-range transport of accelerating Asian emissions on the changing chemical composition of the troposphere. The DC-8 conducted three local flights out of Honolulu, Hawaii followed by four local sorties out of Anchorage, Alaska. This study focuses on data obtained from the Houston-based flights and invokes the transit flight data over the North Pacific. Data from the local flights out of Hawaii and Alaska are the topic of a forthcoming manuscript.

Airborne measurements
A modified LI-COR model 6252 non-dispersive infrared gas analyzer was used to determine CO 2 mixing ratios. This dual-cell instrument achieves high precision by measuring the differential absorption between sample air and a calibrated reference gas that is traceable to the World Meteorological Organization primary CO 2 standards maintained by NOAA ESRL. The LI-COR-based CO 2 sampling system was operated at constant pressure (250 torr) and had a precision of ±0.1 ppm (1σ ) and accuracy of ±0.25 ppm. Experimental procedures are described in detail by Anderson et al. (1996) and Vay et al. (1999Vay et al. ( , 2003. The MILAGRO/INTEX-B in situ CO 2 data are archived at 1 s resolution and are publically available via http://www-air.larc.nasa.gov. The 14 C content of CO 2 was determined from air remaining in a select subset of whole air canister samples collected onboard the DC-8 for measurements of hydrocarbons and halocarbons (Blake and Rowland, 1995). Our radiocarbon sample collection strategy targeted pollution events in an attempt to quantify fossil fuel CO 2 . Pollution plumes were identified by enhancements in various chemical tracers measured in real-time onboard the DC-8. Post-mission, measurements of the radiocarbon content on CO 2 were made at the W. M. Keck Carbon Isotope Accelerator Mass Spectrometer Facility at the University of CA-Irvine and are described below.

Analytical procedures
Aliquots of air for radiocarbon CO 2 analyses were taken from sample canisters after all other hydrocarbon measurements were completed. In the selected canisters, 70 to 140 kPa above atmospheric pressure was left post hydrocarbon analyses easily allowing for aliquots to be withdrawn for radiocarbon analyses with an optimum sample size for high precision 14 C measurements, i.e., ca. 1000 uL CO 2 gas at STP which provides ca. 0.5 mg C target for AMS analysis. Preparation of the samples for isotopic analysis was done on a combustion vacuum line that separates CH 4 , CO, and CO 2 trace gases from whole air (and from each other) while further converting CH 4 to CO 2 and H 2 O and converting CO to CO 2 as the air stream moves toward the pump. Most of the details regarding the vacuum line design and procedure as well as improvements made over time have been reported previously (Tyler et al., 1999 and references therein). In this study, only the CO 2 fraction from the total air sample was recovered. In brief, for a given sample, a series of cryo traps condensed trace gases such as CO 2 and H 2 O vapor at liquid N 2 temperature (−196 • C), leaving less-readily condensable gases such as CO and CH 4 in the air stream. The CO 2 was then separated from H 2 O using a dry ice/ethanol slush bath (−80 • C) and saved for isotopic analysis.
Measurements of 14 C content in samples of CO 2 from selected air samples were made following previously described methods (Southon et al., 2003). Briefly, CO 2 gas is graphitized to pure carbon in the presence of H 2 and hot zinc metal. From there, graphite targets are formed and then mounted into a sample target wheel where they are subsequently measured in the accelerator mass spectrometer. Overall 14 C measurement precision is ±2‰ or ±0.2 percent Modern Carbon, where 14 C and pMC are defined by Stuiver and Polach (1977).
Repeated tests of radiocarbon dead CO 2 from either old coal or calcite were made along with the air sample analyses. These 14 C-free calibration samples were provided by the Keck AMS facility and have a sample size of ca. 1000 uL CO 2 gas at STP, which closely matched the size of the unknown CO 2 in the air samples measured. Some of these calibration blanks were introduced directly into the vacuum line used to recover CO 2 in the samples, while others were introduced as pseudo-air samples by first injecting them into randomly selected canisters from D. Blake's research group and pressurizing to ca. 105 kPa with Zero Air to ascertain whether his sample vessels could have a high background of 14 C from previous usage on other projects. Our vacuum line blank was 14 C=-995±2‰ (0.5±0.2 pMC), a value consistent with other low blank determinations made in years past in our lab. Note that this value includes the quoted Keck AMS measurement precision of ±2‰ and is the combined blank and uncertainty from sampling, processing, graphitizing, and measuring 14 C content. It should be pointed out that the exact age of either the calcite or graphite derived calibration gases is not known. In the past we have analyzed natural gas from petroleum deposits and determined values as low as 14 C=−999 to −1001‰ (−0.1 to +0.1 pMC), indicating that total sample blank from vacuum line may be even lower than the value listed above.

Clean air isotopic composition
The INTEX-B transit flight data from Moffett Field, California to Honolulu, Hawaii (17 April 2006) and from Honolulu to Anchorage, Alaska (30 April 2006) are invoked to obtain a representative background radiocarbon value of 57±2‰ for this study. This value is in good agreement with a mean March 2006 14 C value of 56.6‰ determined from observations at Niwot Ridge, Colorado (40 • N, 105 • W, 3526 m a.s.l.), a proxy for North American free tropospheric air (Turnbull et al., 2007).
Though this background value is determined from midlatitude observations (32-43 • N), 57±2‰ is our best assessment of the clean air isotopic composition based on available data. The monitoring network for 14 C within the tropical Northern Hemisphere is very sparse; the closest station to Mexico being Izana, Tenerife (8 • N) located ∼300 km west of the African coast (March 2006 data are unavailable, I. Levin, personal communication, 2009). Earlier observations by Rozanski et al. (1995) and Levin and Hesshaimer (2000) revealed <4‰ difference between 14 CO 2 levels within the Northern Hemisphere tropical troposphere compared to mid-latitudes. Hsueh et al. (2007) reported one 14 CO 2 measurement from Mexico in a non-polluted setting near Tapalpa (19 • N, 103 • W) in July 2004 of 64.6‰; the corresponding July 2004 value reported for Niwot Ridge was 65.3‰ i.e. less than 1‰ difference between 19 • N and 40 • N values.

Mixing model
Raw data measurement results for the MILAGRO phase of INTEX-B appear in Table 1. For the 21 samples collected, the background 14 C content in CO 2 over Mexico City and from the surrounding region had a surprisingly high average value of ca. 72±26‰ (2σ )= 14 C (107.2±2.6 pMC). Several had 14 C values much higher than background for 14 C (>1σ outside mean). This result was quite unexpected as these samples also had elevated CO 2 concentration above background level, an observation which is normally associated with fossil fuel sources in urban regions and would necessarily lead to depleted rather than elevated 14 C content. In comparison, our results for 22 whole air samples collected during the North Pacific phase of INTEX-B in April and May 2006 indicated that the background 14 C content in CO 2 was ca. 59±6‰ (2σ )= 14 C (105.9±0.6 percent Modern Representative March surface background concentrations (and atmospheric lifetimes) for C 2 Cl 4 (τ ∼1 day), CH 3 CN (τ >few months), HCN (τ >few months), and C 2 H 6 (τ ∼3 months) are 3 pptv, 100 pptv, 140 pptv, and 750 pptv, respectively. Background<detection limit (DL) for isoprene (τ <1 h) and isopentane (τ ∼1 h). Radar (ALT R) and pressure (ALT P) altitude.
Carbon). This is in excellent agreement with recent Northern Hemispheric studies of background CO 2 in air (Levin and Kromer, 2004;Xu et al., , 2006) that indicate a level of between 105 to 106 pMC over the whole year by direct monitoring at specific sites. Many of the air samples measured over Mexico indicate that we potentially have at least two CO 2 sources contributing to the elevated CO 2 mixing ratios observed in those cases. For example, one can see the need for two sources most readily by a simple calculation made for the three samples with lowest measured 14 C values in Table 1. Assuming background air is affected by only one source, fossil fuel, one would expect the excess CO 2 above background to have only a radiocarbon dead 14 C signature. Equations (1) and (2) can be used to calculate the expected value of 14 C in the sample air when fossil fuel CO 2 is added to the background.
Using background values of 382.8 ppmv and 57‰ for mixing ratio and 14 C of CO 2 to conform to Northern Hemispheric averages measured in March 2006 by others and a 14 C FF value of −1000‰ (radiocarbon dead), one calculates that the three measured values of 14 C at 20, 47, and 49‰ would have to have been much lower. Hence, a mixture of both radiocarbon dead and radiocarbon enhanced sources would reasonably lead mathematically to the 3 values listed above.
Similarly, our measured values of 14 CO 2 much above the background value must necessarily include source contributions from a radiocarbon enhanced source such as biomass burning (which was observed extensively in the region and could easily contain a slightly higher 14 C than modern background air depending on the biomass age) to balance any radiocarbon dead contribution from fossil fuels. Yet another possibility is that the air contains a very small amount of an appreciably "hotter" 14 C source such as medical waste released as 14 CO 2 . In this case the excess CO 2 mixing ratio from the hot source could be very small with most of the excess attributable to fossil fuel CO 2 source and yet the 14 C value would remain elevated above background from the very "hot" 14 C source.
To analyze our results further we constructed a twocomponent CO 2 source mixture equation with sources both 14 C-depleted and 14 C-enhanced contributing to a CO 2 background of well-mixed air which makes up the majority of the sample. Calculations partitioning burning biomass (BIO) and fossil fuel (FF) sources of CO 2 can be made using Eqs. (3) and (4) below.
Assigning exact values to the radiocarbon-enhanced source is problematic because we are uncertain what that source is. Assuming that CO 2 from biomass burning of recent wood is the main source of 14 C-enhanced CO 2 in the region is a reasonable one based on visual observation and supporting data of other HC measurements. In that case, endpoints for our two-point source mixing curve were chosen as 14 C=−1000‰ for fossil fuel and 14 C=499‰ for biomass, where 499‰ represents stored carbon of about 35 years ago with the imprint from bomb-test 14 C suddenly being released as CO 2 .
For observed 14 C of CO 2 values <50‰ listed in Table 1, calculations made with Eqs. (3) and (4) yield very plausible partitioning between fossil fuel and biomass burning sources. In those three cases, the partitioning is close to 50:50 for the two at 47 and 49‰ while it is ca. 75:25 in favor of fossil fuel over biomass burning for the observed 14 C of CO 2 value at 20‰. The above calculations also yield a reasonable partitioning between biomass and fossil fuel source contributions for observed values of 14 C of CO 2 between 55 and 70‰. In those cases, most of the excess CO 2 is calculated as coming from the biomass burning source. (Note that other partitioning ratios are possible if the 14 C of CO 2 from biomass burning is taken to be a value other than 499‰.) However, for samples >70‰ in 14 C, we have some physically impossible splits between the two types of sources in most cases. The problem stems from the fact that many of the observed 14 C of CO 2 values are very high even without much of an increase in the mixing ratio of CO 2 (only a few ppm) above background. In order to attribute enough source CO 2 to a 14 C-enhanced source, the fossil fuel contribution becomes negative in the partitioning which is impossible. This weakens our argument that the source of air with 14 C>+100‰ can all be explained merely by somewhat aged wood from biomass burning.
One possible solution is to push the biomass age further back in time toward even higher assigned 14 C values, but there is a limit to this. The endpoint 14 C value for the biomass burning source in Eq. (4) is not really known, unlike the situation for most two-point mixing curves. We are estimating to obtain a fit however, since the bomb curve is not monotonically increasing backward in time but instead peaks ca. 1962, is skewed and is not even symmetric, pushing the estimated wood age back further is unrealistic as it results in pre-bomb background 14 CO 2 being stored in the biomass. Therefore we need to consider other possibilities as part of the mixture of contributing sources in the Mexico City region. These sources must be "hotter" than what we can reasonably expect from aged wood and are discussed in Sect. 3.3.

Large-scale distribution
The large-scale distribution of 14 C during MILAGRO and the INTEX-B Pacific transit flights is presented in Fig. 1. These data are shown as a function of pressure altitude since the MCMA lies in an elevated basin 2240 m a.s.l. Any anthropogenic emissions emitted from this megacity are therefore introduced into the atmosphere at altitudes considered to be in the free troposphere. An interesting feature of the MI-LAGRO 14 C data set is that 76% (N=16) of the radiocarbon values were above the assigned 57±2‰ background, with surprisingly few samples (N=5) indicating a fossil-fuel influence (<57‰). An anomalously high 14 C content (>112‰) was found in three of the collected samples and these results are examined in more detail in Sect. 3.3.
Canister sample collection locations and their associated radiocarbon values are illustrated in Fig. 2. This figure depicts the degree of spatial coverage within the region and provides a detailed look at the 14 C distribution. The flights on 9 and 16 March had different science objectives, sampling emissions over the Gulf of Mexico and Mexico City, respectively. We use these two flights to explore sources contributing to the total measured CO 2 signal in Sect. 3.1 and 3.2.

Mexico City
An objective of the 16 March (Thursday) science flight was the sampling of fresh Mexico City (MC) pollution during weekday conditions. The flight track, altitude, and data for a portion of this flight are presented in Fig. 3. The DC-8 approached MC from the south, flew a south-north transect over the city along the same line of longitude, and then conducted a terrain-following path just north of the city. Collection sites for the isotope samples are indicated along the DC-8 flight track (blue diamonds). Inset above, in situ CO 2 and CO measurements are plotted with the 14 C observations. The radiocarbon values fell within the range of 20‰ to 120‰ with the lowest values observed directly over MC where the highest CO 2 and CO mixing ratios were measured. Using a 14 C depletion of 2.8‰ per ppm of fossil fuel CO 2 added to the current atmosphere (Turnbull et al., 2006), we estimate from our mixing model the minimum fossil fuel contribution to the total measured CO 2 signal to range between 4 to 15 ppm in the three samples lowest in 14 C value.
Collection of the isotope samples in clusters permitted partitioning of the flight segment into distinct regions labeled as A, B, C (Fig. 3). Regional correlations between CO 2 and CO are illustrated in Fig. 4. Following the method outlined in Yokelson et al. (1999) emission ratios (ERs) were obtained from the slope of the least-squares line, with the intercept forced to zero, in a plot of one set of excess mixing ratios ver-sus another. Background values of 382.61 ppm and 112 ppb were used for CO 2 and CO, respectively based on the median determined from the four DC-8 flights. For comparison, median CO 2 and CO for Mauna Loa (19 • N, 155 • W, 3397 m a.s.l.) over the MILAGRO sampling window were 382.87 ppm and 105 ppb, respectively. Each cluster shows a distinct relationship, indicating a mixture of different sources that is also evident in the radiocarbon data.
Sampling over MC occurred during the January-June dry season. The level of fire activity in March 2006 was above normal and more typical of that seen in April. The Mexico City basin was influenced by emissions from forest fires in the pine-savannas that dominate the mountains surrounding the city (Yokelson et al., 2007). Both ground-based (E. Alvarado, personal communication, 2007) and airborne observers (DC-8 pilots, instrumentalists) reported reduced visibility from smoke generated by widespread wildfires and agricultural burning in South-Central Mexico that, at times, obscured the ground. The radiocarbon samples from south of Mexico City [Region A] were collected over the study area [19-20 • Yokelson et al. (2007) who used a Twin Otter research aircraft to make various measurements of fire emissions during MILAGRO. We find that our CO:CO 2 molar emission ratio derived for Region A (0.024) compares well with Yokelson's (0.028) which was determined from measurements conducted the following day over a prescribed burn. Several correlate source tracer species were examined to elucidate influences on the radiocarbon distribution: CH 3 CN (also acetonitrile or ACN) and HCN (biomass burning); C 2 Cl 4 (industrial emissions); isopentane (vehicular emissions); isoprene (biogenic emissions); and C 2 H 6 (combustion from fossil fuels and biomass burning). Table 1 provides the mixing ratios, representative background values, and respective lifetimes for these tracers. For the short lived chemical tracers isopentane and isoprene (τ ≤1 h), any enhancements in these gases above their detection limits (DL) means that the air has been in very recent contact with a source region. In Fig. 5, the radiocarbon data are plotted along with several source tracers and reveal enhancements in ACN, isopentane, and C 2 Cl 4 over all three regions; significant enrichment in C 2 Cl 4 and isopentane over MC [Region B]; north of the city [Region C] isoprene mixing ratios >200 pptv were observed (not shown).
The pronounced influence of biomass burning on regional air quality during and T-1 (19.703 • N, 98.982 • W, 2273 m a.s.l.) (Fig. 2) re-Fig. 4. Regional correlations between CO 2 and CO for radiocarbon sample collection intervals. Excess CO 2 and CO mixing ratios are plotted versus each other to obtain emission ratios by the method described in the text.
vealed that significant contributions (45-78%) to the organic and elemental fractions of the carbonaceous aerosols arose from biomass burning sources in MC and the surrounding region (Marley et al., 2009;Gaffney et al., 2008). Pyrogenic emissions greatly impacted the peripheral site T-1 as determined from 13 C in aerosol carbon observations that showed a number of plumes highly enriched in 13 C due to the pervasive burning of C4 grasses (Marley et al., 2009). Similarly, Stone et al. (2008) found that the aerosol generated by biomass burning on the perimeter of the city (T-1) was highly correlated with vegetative detritus and chemically different from the point-source wood-burning events that polluted the downtown area (T-0).
Fires emit significant amounts of CO 2 to the atmosphere with emission patterns that are highly variable in space and time (Wiedinmyer and Neff, 2007). Our data indicate that the ubiquitous biomass burning emissions during MILAGRO are a likely contributor to the mass excess in our 14 C observations via the re-suspension of radionuclides from contaminated biomass by fire. Using the variability in these observations to partition our two-point source mixing curve, we surmised that the age of carbon released to account for the excess mass would need to be from vegetation and soil organic matter that had fixed atmospheric CO 2 during the past several decades for a plausible fit. The fluxes from such events are important short-term influences on regional C patterns (Amiro et al., 2001) representing a challenge to the establishment of accurate C source and sink accounting based on atmospheric CO 2 observations (Wiedinmyer and Neff, 2007) alone.

Gulf of Mexico
On 9 March, the meteorological conditions were favorable for the sampling of outflow from Mexico City to the Gulf of Mexico. At takeoff the DC-8 climbed to 8.25 km, and then began descending while heading south over the Gulf in preparation for a boundary layer (BL) run. At this lowest level (∼200 m a.s.l.) the air masses had an unusual composition, containing high amounts of CH 4 (2.1 ppm) and CO 2 (400 ppm). CO 2 , CO, and CH 4 observations, averaged to sampling time of the whole air samples, are presented in Fig. 6. The radiocarbon content of the four whole air samples collected during this southbound BL leg (triangular symbols in Fig. 6) show a decreasing trend as the DC-8 progressed further south (Fig. 2). As the CO 2 , CH 4 , and CO mixing ratios increase in these samples, the 14 C values become more depleted (Fig. 6). We were able to fit the data corresponding to these four 14 C samples with linear correlations that produced 0.285 ppmv/ppbv CO+349.63, R 2 =0.98; and 0.068 ppmv/ppbv CH 4 +262.15, R 2 =0.98. These tight correlations likely reflect co-located sources. Examination of other source tracer species revealed 14 C correlated with ACN, HCN however, exhibiting the opposite relationship with isopentane and C 2 H 6 ( Table 1).
The largest natural gas and principal oil fields in Mexico are located around the Gulf of Mexico states of Veracruz, Tabasco, and Campeche. We use the Lagrangian particle dispersion model FLEXPART (Stohl et al., 1998) to explore the characteristics associated with meteorological transport processes in the region. Details of the model simulations are given by Stohl et al. (2002). Briefly, backward simulations are done from along the flight track and include full turbulence and convection parameterizations. Every simulation consists of 40 000 particles released in the volume of air sampled whenever the DC-8 covered a distance of 0.18 • or changed its altitude by 8 hPa. As emissions originate at the surface, we focus on the residence times (ns/kg) in the FLEXPART regional footprint product, which shows the emission sensitivity averaged over the lowest 100 m adjacent to the ground. We also examined the Retroplume summary product to establish the fraction of particles in the BL as a function of time. The air mass sampled in the BL during the southbound leg originated over northern Mexico, then flowed in a clockwise direction over the southern United States, the Gulf of Mexico, portions of Central America, and southern Mexico prior to interception over the Gulf (not shown). Inspection of Fig. 7 shows an initially narrow (Fig. 7a) influence footprint that broadens (Fig. 7b) with increasing proximity to the southern Mexico oil and gas center. A day prior to sampling, FLEXPART estimates an air mass fraction ≥60% traveling within the atmospheric boundary layer over the petroleum industry center as well as over fires in Guatemala and the Mexican states of Chiapas and Oaxaca. The corresponding decreasing trend of ACN, HCN, isoprene and 14 C while approaching the Mexican coastline with a commensurate increase in CO 2 , CO, CH 4 , C 2 H 6 , and isopentane suggests a biomass burning influence initially dominating the radiocarbon signal that became increasingly obscured by the influence of anthropogenic emissions associated with the petroleum industry. These results are supported by the FLEXPART emission sensitivity products (Fig. 7).
There are some notable differences in the composition of the two remaining BL (∼200 m a.s.l.) samples (square symbols in Fig. 6) collected in the western Gulf later that same day. Where 14 C=53.3‰ or corresponding to a minimum of 7.5 ppm of added fossil fuel CO 2 from our mixing model, Fig. 7c shows a high emission sensitivity from both MC and an oil and gas producing field in northern Veracruz coupled with a widespread biomass burning influence. HCN (582 pptv), ACN (182 pptv), C 2 Cl 4 (7 pptv), and C 2 H 6 (5371 pptv) mixing ratios are higher than the other 9th March samples, reflecting the proximity to and different relative contributions from the varied emission sources of the two regions. Though these two samples (square symbols Fig. 6) exhibit similar CH 4 mixing ratios, respective CO and CO 2 relationships differ, reflecting interception of an air mass that passed over the petroleum center in SE Mexico then Veracruz (not shown).
The gas released when crude oil is brought to the surface is known as associated gas. Drilling companies routinely flare or vent this material for safety reasons or where no infrastructure exists to bring it to market. Flaring is where the gas is burned and emitted as CO 2 , and venting, where the gas is simply released to the atmosphere as CH 4 (Christen, 2004). After CH 4 , ethane is the second most abundant constituent of natural gas and both chemical tracers have some common sources: fossil fuel production, biofuel combustion, and biomass burning (Xiao et al., 2008). We examined the CH 4 :C 2 H 6 molar ratios for these atmospheric observations conducted downwind of these major source regions for source attribution and found CH 4 :C 2 H 6 ratios (34-114) typical of those for "dry" natural gas fields (i.e. absence of condensate or liquid hydrocarbons) (Xiao et al., 2008). These findings point to the sampling of a complex mixture of emissions attributable to both the venting and flaring of associated gas, fugitive emissions associated with the oil and gas industry, biomass burning, and fossil fuel combustion (based on the isopentane signal) as influences on the radiocarbon content of the atmospheric 14 C observations.

"Hot sources"
Anomalously enriched 14 C values (>112‰) were measured in three of the whole air samples, nearly double the radiocarbon in the background atmosphere. The excess levels of 14 C cannot be attributed to the release of old bomb carbon from biomass burning as a plausible result using our two-point mixing curve for determination of the relative age of carbon was not realized. The associated CO, CH 4 , and CO 2 mixing ratios also do not support a stratospheric influence. The 120‰ value was measured in a sample collected south of MC (Fig. 2) at 2.6 km (i.e. near the altitude of MC or 0.75 km radar altitude) in an atmospheric layer rich in combustion and urban related trace gases The other two highly enriched samples (113‰ and 132‰) were augmented with the biomass burning tracers ACN and HCN yet devoid of an elevated industrial and vehicular emission signature (i.e. C 2 Cl 4 or i-pentane), likely a result of sampling on a Sunday (19 March) within a long weekend. These observations were made during a low pass over Monterrey in Nuevo León, and while on approach to MC over the state of Hidalgo, respectively (Fig. 2).
The possible causes of these perturbations in 14 C were investigated and a number of potential sources of the emitted radiocarbon identified: production of 14 CO 2 from the boiling water reactors (BWRs) of the Laguna Verde nuclear power plant located in Veracruz; and/or medical and hazardous waste incineration. Mexico has two hazardous waste landfills, 13 fuel blending plants, 11 private hazardous waste incinerators, and 22 medical waste incinerators (Reed et al., 2000). Most of the medical waste incinerators are located in the Mexico City metropolitan area, although one has been permitted in Monterrey, Nuevo León (Reed et al., 2000). Moreover, Mexico's cement industry has embraced the incineration of hazardous wastes in their cement kilns as an energy recycling strategy. This incineration has been led by the two largest cement producers in Mexico, which have joint ventures with US transnational companies to blend hazardous waste into fuels before burning the waste in their kilns (Reed and Gonzalez, 1997). Twenty one of 30 cement plants throughout Mexico have received either test-burn permits or annual authorizations to burn a variety of hazardous wastes with twelve of these plants located in central Mexico (Reed and Gonzalez, 1997). The locations of some of these potential "hot" point sources are shown in Fig. 2.
The nuclear-waste problem involves not just what existing nuclear power plants produce, but also the low-level radioactive material found in hospitals, universities and in other industries. Most of the 14 C released into the environment by incinerators and BWRs is in the form of gaseous emissions (Magnusson et al., 2004) mainly via the stack gas, with CO 2 as the main carrier of 14 C. This radiocarbon or "hot source" released into the atmosphere is then available to enter plant carbon pools through photosynthesis or leads to an increase in the local 14 C concentration. Previous studies have reported high radiocarbon concentrations (97-184 pMC) in vegetation in areas more than 100 km distant from a nuclear power plant (Mikhajlov et al., 2004) and 14 CO 2 values 190-6000‰ over background in oak leaves sampled near two hazardous waste incinerators .
We examined 5-day kinematic backward trajectories and several FLEXPART data products calculated along the DC-8 flight tracks for each sampling location and time (not shown) which reveal local influences from the northern and southern sectors of the MCMA for the highest 14 C observations, 132‰ and 120‰, respectively. The substantial 14 CO 2 enrichment (113‰) observed near Monterrey also appears to have a local influence south-southwest of the city. From Fig. 2, a close correspondence between these elevated radiocarbon values and local cement kilns is indicated. The transport history of the radiocarbon enriched air parcels consistently reflects an air mass fraction >40% in the ABL 24 h prior to sampling. The numerous and varied potential "hot" point sources located throughout the region complicate source attribution. The observational influence could be attributable to sampling a direct smokestack emission; or from the re-suspension of 14 C in contaminated biomass by the pervasive biomass burning occurring during March 2006. The possibility of either scenario exists in our MILAGRO data set and offers a plausible explanation for the anomalously high 14 C values observed.

Conclusions
Only 23% of the radiocarbon measurements from MILA-GRO indicate a distinct fossil fuel influence, an unexpected result for the MCMA sampling environment. Inspection of correlate source tracer species and air mass transport histories suggests that the fossil fuel detection capability of the radiocarbon method was obfuscated by contributions from other local and regional CO 2 sources emitting higher 14 C than the background atmosphere. By 2025, the number of megacities is predicted to reach twenty seven, with the majority located in the developing world, each with their own unique emission signatures, all likely major sources of atmospheric CO 2 . Observations during MILAGRO have shown the complexity of radiocarbon cycling in the megacity environment within a developing country and indicate adding detailed simultaneous measurements of other chemical tracers to the isotopic marker 14 C are highly desirable for better evaluation of the various sources contributing to the total measured CO 2 signal.