A tunable diode laser absorption spectrometer (TDLAS) was deployed during the PMTACS-NY Supersite winter 2004 intensive field campaign at Queens College in New York City to measure the ambient gaseous ammonia. For the characterization of ammonia emissions from the mobile sources, a LI-7000 CO2/H2O analyzer was also collocated with the TDLAS to measure ambient CO2 and H2O vapor. The field measurements and laboratory calibration with certified ammonia standard have been used to evaluate the performance of the TDLAS system. High time resolved TDLAS ambient ammonia measurements performed at Queens College from 10 January to 6 February showed high variability, with NH3 concentrations ranging from below the detection limit (0.1 ppb) to maxima of 197.4 ppb and a mean value of 0.8 ppb over the entire campaign. Many high-frequency NH3 spikes spanning over a less than 1-min duration were observed during the high traffic periods. The occurrence of the NH3 spikes was closely correlated with observed CO2 spikes, a good marker of traffic exhaust. This correlation yielded an NH3 emission ratio of 0.12 ppbv/ppmv, which can be used to estimate an NH3 emission factor of 35.5 mg/km. The [NH3]/[CO2] ratios over the entire field study was also obtained and added into the best NH3 emission estimates. On a snowy day, no obvious drop of NH3 and CO2 concentrations was measured as the ambient H2O vapor increased. The observed dramatic decrease in the ambient NH3 and CO2 concentrations on a rainy day resulted from a quick air mass switch. Two similar bimodal diurnal patterns associated with the rush hour traffic were observed during school holidays and school days of Queens College, New York. More NH3 emissions from cold start vehicles might contribute to a higher peak in the late afternoon hours. Such observations suggest that the NH3 emissions from the traffic exhaust could be a major source of the ambient NH3 in urban areas.
If you can't find a tool you're looking for, please click the link at the top of the page to go "Back to old version". We'll be adding more features regularly and your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Ammonia is the most gaseous base in the atmosphere. It fundamentally determines the overall acidity of cloud, precipitation and atmospheric aerosols [Seinfeld and Pandis, 1998]. An important role of ammonia in the atmosphere is neutralizing acidic compounds such as nitric and sulfuric acids to form ammonium nitrate and sulfate, which contribute significantly to fine particle mass [Finlayson-Pitts and Pitts, 1999]. In addition to its role as an important nutrient for plant growth, the deposition of atmospheric ammonia also results in the eutrophication of ecosystems and soil acidification by nitrification [Pearl, 1991; Olivier et al., 1998].
 Ambient gaseous ammonia is emitted by a large number of sources. Major sources of ammonia emissions include volatilization from animal waste and synthetic fertilizers, biomass burning, losses from soils under native vegetation and agricultural crops. Most internal combustion processes also produce ammonia, even though the amount of ammonia produced is typically small compared to the other sources mentioned above [Bouwman et al., 1997; Asman et al., 1998]. Recently, a few studies have demonstrated that road vehicles emit ammonia into ambient air [Kean et al., 2000; Huai et al., 2003; Baum et al., 2001; Herndon et al., 2005]. These results suggested that on-road vehicle ammonia emissions increased significantly following the introduction of three-way catalytic converters. So far, the ammonia emissions from road traffic have not been considered as a significant source.
 Atmospheric ammonia concentrations are not routinely measured by governmental air quality monitoring networks. This has led to significant uncertainties in the calculations of total dry deposition of nitrogen [Krupa and Moncrief, 2002]. To date, most of ammonia measurements have been carried out in rural areas, especially close to the major sources: agricultural areas and cattle feedlot operations [Krupa, 2003, and references therein.]. The employed techniques are mainly based on passive sampling methods [Krupa and Legge, 1999; Tang et al., 2001; Pul et al., 2004], diffusion denuder systems [Wyers et al., 1993; Baek and Aneja, 2004], and nebulizing-reflux chamber [Lefer et al., 1999]. The known setbacks with these methods are low time resolution and certain artifacts. The difficulty for accurate ammonia measurements is mainly due to the ammonia multiform presence in gaseous (NH3), particulate (e.g., NH4NO3 and (NH4)2SO4) and liquid (NH4OH in water droplets) phases as well as its tendency to form strong hydrogen bonds with water. To overcome these difficulties, over the years a continuous effort has been made in the development of sensitive ammonia measurement techniques. Several intercomparison studies [Williams et al., 1992; Fehsenfeld et al., 2002] showed that a photofragmentation/laser-induced fluorescence technique [Van Dijk et al., 1989; Schendel et al., 1990] and a chemical ionization mass spectrometer (CIMS) [Fehsenfeld et al., 1998; Huey et al., 1998] demonstrated more reliable performance among a number of newly developed techniques involved. The measurement time resolutions of both techniques were substantially improved to 1 min and 10 s, respectively. An application of the tunable diode laser absorption spectrometer (TDLAS) to the ammonia measurements has been reported by two recent studies. Huai et al.  performed a research on the effect of fuel sulfur on NH3 emissions from 2000–2001 model year vehicles. In their studies, the detection limit (0.5 ppmv) of their particular TDLAS system limited the ammonia measurements to rather high concentration levels over 2-s time intervals. Famulari et al.  made field measurements of NH3 fluxes at intensively managed grassland in Southern Scotland. These studies showed that the high sensitivity, together with high selectivity and fast time response has made low-detection limit TDLAS systems well suited for accurate, reliable measurements of ambient ammonia even at a very low level.
 In this work, the measurements of the ambient gaseous ammonia in an urban area were performed using a tunable diode laser absorption spectrometer (TDLAS) during winter 2004 New York City intensive field campaign. In order to characterize the emission features of mobile sources, a LI-7000 CO2/H2O Analyzer was also collocated with the TDLAS system to simultaneously determine ambient CO2 and H2O vapor. Here, we will briefly describe the TDLAS system including the experimental parameters and the instrument performance. Subsequently, the results of the NH3, CO2 and H2O measurements in the field study, together with the effect of the precipitation on the ambient gaseous NH3 and CO2 in two events and two bimodal diurnal patterns believed to be associated with traffic exhaust are presented, followed by a brief discussion of the experimental results. The principal objective of this paper is to present an application of a TDLAS system for the accurate measurements of ambient NH3 concentrations as well as the characterization of NH3 emissions in an urban area.
 An optical module and an electronic module compose the TDLAS system. The optical module is built on a 0.61 by 1.22 meter aluminum optical table and contains one liquid-nitrogen-cooled Dewar for temperature control of tunable diode lasers and detectors, optics for laser beam collection and transport, and one reduced pressure multipass absorption cell. The purpose of the optical module is to form the light from each of the infrared laser diodes into a pair of beams (main beam and reference beam), to direct the main beam into the multipass absorption cell that provides a long path length of 153.50 m in a volume of 5 liter, and then to direct the light leaving the cell back to a detector. The reference beam is sent through a short reference cell containing a high concentration of the gas of interest (in this experiment, NH3), in order to give a high-contrast spectral signal for line position locking. A key element of the optical module is the low-volume, long-path length Astigmatic Herriott multipass absorption cell [McManus et al., 1995]. Depending on the pumping speed, a fast time response (>1 Hz) can be obtained because of the small volume of the Astigmatic Herriott multipass absorption cell. In order to reduce the pressure broadening of the absorption lines, the ambient air is pumped into the absorption cell from atmospheric pressure to 25 Torr (a typical sampling pressure in the field measurements) across the inlet orifice.
 The electronics module comprises a fast computer with two data acquisition boards (National Instruments), and a dual laser control unit (Laser Components GMBH, Olching, Germany). The electronics module controls the laser diode frequency, and processes the detected absorption signals to return trace gas concentrations. Both of these functions are controlled via a Windows 95/98 based TDL Wintel data acquisition program. The computer sends commands to the laser controller, which in turn adjusts the laser diode temperature and average current and provides a fast ramp sweeping the laser frequency across the trace gas absorption feature. The laser light exiting from the absorption cell or the reference cell is detected and converted to electrical signals digitalized by a fast data acquisition board. As the laser frequency is swept across the spectral feature of interest, TDL Wintel program calculates the change in absorption by fitting to a calculated line shape on the basis of the tabulated spectral parameters, the measured sample temperature and pressure, and then yields an absolute trace gas concentration.
2.2. Calibration of the TDLAS System
 The TDLAS system employed in this study is an absolute measurement technique, which does not require the calibration, eliminating the need for calibration gas mixtures in the field. To ensure data quality, laboratory calibration were performed using a certified NH3 standard purchased from Matheson. The certified 10 ppmv NH3 standard in nitrogen is stored in an aluminum cylinder. In the calibration, four lower-concentration mixtures (obtained by diluting the certified source with zero air) were measured with the TDLAS system under the same conditions as those in the field campaign. The measurement showed a linear response up to 35 ppbv. A slope of 0.86 obtained from the least regression fit suggested a low bias by 14%, relative to the certified value. We believe that the low bias was due to the ammonia decay in the storage cylinder since an independent calibration showed the same low bias, consistent with TDLAS measurements. The independent calibration involved the direct bubbling of the NH3 source gas through an acid solution (pH = 1), followed by analysis by a chemical derivative technique coupled with a HPLC detection system. The concentrations measured by the independent technique showed that it agreed with TDLAS within 3%.
2.3. Absorption Feature and Measurement Selectivity
 The strong NH3 absorption feature at 1065.5654 cm−1 was employed to measure NH3 concentrations in the field campaign. In this region, three strong absorption lines at 1065.5654 cm−1, 1065.5817 cm−1 and 1065.5943 cm−1 compose a distinctive triplet. The strongest line at 1065.5654 cm−1 has an integrated cross section of 2.60 × 10−19 cm2 molecule−1 cm−1. Figure 1 depicts a NH3 spectrum acquired from sampling a diluted mixture of the certified NH3 source. In the data analysis, all three lines were included in the fitting to yield the NH3 concentration value. Compared to monitoring only one single absorption line, fitting multiple lines makes the retrieved concentrations less susceptible to potential interferences from other species, and can also enhances the specificity and sensitivity.
 In the HITRAN spectroscopic database, O3 is the only species with spectral lines near the NH3 absorption feature. The nearest strong O3 line is 1065.6369 cm−1, and its line strength is weaker than the NH3 absorption feature by two orders of magnitude. In the spectrum fitting procedure, the O3 absorption feature was also fit as a second species to eliminate its potential interference to the NH3 measurement. When the ambient O3 concentration increased up to a higher level (for instance, 50 ppbv), its presence did not affect the accurate measurements of the ambient NH3. For example, Figure 2 shows the fitting of NH3 and O3 in a retrieved 1-s spectrum of the ambient air during the field campaign.
2.4. Background Subtraction and Instrument Performance
 To minimize the wall effect and improve the instrument performance, which was discussed in detail by Li et al. , a rapid background subtraction was employed. The laboratory tests of the inlet system revealed that NH3 came into equilibrium with the walls with a characteristic time constant of approximately 4 s, shown in Figure 3. In the background subtraction cycle repeated every min, 12 s are used for the cell flushing (before and after the background spectrum acquisition), 16 s for background acquisition, and 20 s for ambient data acquisition. Such a background cycle allowed us not only to remove the wall effect but also to be able to measure high fluctuation of ambient NH3.
 During the field measurements, the replicate precisions (1 standard deviation) for consecutive 60 s cycles were obtained by sampling relatively constant ambient NH3 for 15 min. The 60 s replicate precision based on numerous such time periods is 100 ppt (1σ), consistent with the value obtained in the laboratory test by sampling a constant NH3 mixture (4.7 ppb).
 The absolute accuracy of concentration measurements performed using tunable diode laser differential absorption spectroscopy is fundamentally determined by how well the line strengths used in the spectral fitting procedure are known. Additional factors that contribute to a systematic error in these measurements are as follows: the path length, pressure, temperature, line shape model used in the fitting procedure and the diode mode purity. These “additional” factors for the overall instrument will be discussed first, and then the specific line strength uncertainty will be considered. The uncertainty in the path length for the 174 spot/153.5 meter astigmatic Herriot cell is estimated to be less than 0.1% or 15 cm [McManus et al., 1995; Herndon et al., 2004]. The derived quantity from the spectral fitting is a number density measurement, however for convenience, this is immediately converted to a mixing ratio, which requires an accurate measurement of both pressure and temperature. The systematic uncertainty in the pressure and temperature measurements directly contribute to the accuracy of the measurement, however both pressure and temperature play a subtler role in the spectral fitting. A Voigt line shape model [Humlicek, 1979; Armstrong, 1967] is used to fit the spectral data and extract concentrations for each of the species specified in a HITRAN style [Rothman et al., 1998] input file. With a “well behaved” diode the uncertainty associated with the fitting procedure can be shown to be less than 1% [Herndon et al., 2004]. For this level of certainty one must demonstrate that the diode characteristics are within the following two specifications: apparent laser line width <0.005 cm−1 and high mode purity >98%. These specifications were met by the NH3 laser diode in the laser characterization experiment. When the diode is operating outside of these typical limits, the potential systematic error in the fit increases. However, these arguments also assume that the concentration and path length are such that the minimum transmission is greater than 50% or that “optically thin” conditions apply. The line strengths present in the HITRAN database for the 10.5 μm band of NH3 ν1 band come largely from the 1986 version HITRAN database [Rothman et al., 1987]. The uncertainties of the NH3 line strengths in the employed range are not reported or unavailable. Nevertheless, we can make a good estimation on the basis of the calibration results described earlier. The accuracy of the NH3 measurements is better than 14%.
2.5. Data Dropouts in the Field Campaign
 In the field study, the liquid nitrogen refill was automatically implemented every 6 hours. As discussed in our previous study [Li et al., 2004], the liquid nitrogen refill can cause a dramatic temperature fluctuation of the optical table, resulting in a quick movement of the laser frequency around the lock position and subsequent larger error of the fitting results. Such an unstable period usually lasts about 10 to 15 min so that the data taken during this period are normally dropped. Beside the liquid nitrogen refill, the mirror cleaning also caused some missing data. The design for the TDLAS sampling system ensures fast time response and a minimization of wall effects. One disadvantage of the design is the lack of the mirror protection from ambient contaminants (aerosols mainly) requiring the mirrors to be cleaned every three or four days under typical urban pollution conditions. On average, it took about 3 hours for the mirror cleaning for full system recovery.
2.6. LI-7000 CO2/H2O Analyzer
 The LI-7000 is a differential, nondispersive, infrared gas analyzer developed by LI-COR, Inc. The CO2 and H2O measurements are based on the difference in absorption of infrared radiation passing through two gas-sampling cells. One cell used as a reference cell contains a gas of known CO2 or H2O concentration, and the other cell (normally the sample cell) is used for a gas of unknown concentration. Infrared radiation is transmitted through both cell paths, and the radiation at the detectors is measured in both cells and used to compute absorption. Data are output with a frequency response up to 20 Hz. Calibration of the LI-7000 CO2/H2O analyzer was performed before the field campaign. The ability to measure the high fluctuation of ambient CO2 concentration to identify source plumes and to characterize emissions of mobile sources is a major attribute and purpose for deploying this system.
3. Instrument Field Deployments and Operation Description
 Field experiments were conducted at parking field #6, Queens College, New York City (40°44′ N, 73°49′ W, ∼ 25 m a. S. L.) from 10 January to 6 February 2004. There are two busy highways nearby: less than 1 km to the north is the Long Island Expressway (I-495), and 1 km to the west is the Van Wyck Expressway (I-678). The meteorology data for the entire field study are summarized in Table 1. As can be seen, the dominant wind direction was northwest and the wind speed ranged from 0.4 m/s to a maximum of 12.1 m/s. The monitored ambient temperature was mostly below 0°C and experienced several super cold days recorded with a minimum temperature of −17.1°C. There were four significant precipitation events over the month of the field campaign. Three snow events were on 15, 18 and 27 January. One rainfall event started on 3 February and lasted for 7 hours.
Table 1. Summary of Meteorological Data From 10 January to 6 February 2004
Wind direction, degree, clockwise
Wind speed, m/s
 The TDLAS system and LI-7000 CO2/H2O analyzer were housed next to each other in the same room sited at the corner of a one story multiple unit shelter. A schematic of the sampling systems for both the TDLAS system and LI-7000 CO2/H2O analyzer is given in Figure 4. To minimize the wall effects and maximize time response, a straight 0.8 m long, 12.7 mm OD fluorinated silane coated glass tube was used for the TDLAS sampling line, which extended through the window with the inlet height about 3 m above the ground. The glass inlet was heated to 50°C to the multipass absorption cell. As shown in Figure 4, a 6.35 mm OD PTFE tube supplying zero air for the background subtraction was also connected to the glass tube with an exit port next to the sampling inlet. The second port on the glass tube near the sampling inlet of TDLAS system was used for the connection to LI-7000 CO2/H2O analyzer with a 6.35 mm OD PTFE tube. Such a design ensured the sampling synchronization for two systems. A residence time of less than 1 s in the sampling cells of both the TDLAS and LI-7000 CO2/H2O analyzer was achieved with flow rates of 10 LPM and 5 LPM, respectively.
4. Experimental Results
 The measurements were performed during the entire field campaign, from 10 January to 5 February. Data coverage for NH3 measurements was greater than 80%. As mentioned in the experiment section, the missing data were due to the data dropouts during the liquid nitrogen refill and the mirror cleaning. The measurements of ambient NH3 and CO2 concentrations were performed in the original 1-s interval. Figure 5 (top) shows typical time series data of NH3 and CO2 at 1-s intervals. As can be seen from Figure 5, the measured ambient NH3 gave a very flat concentration change (NH3 = 0.1 ± 0.2 ppb) from midnight to the earlier morning, and then began to increase, accompanied with some high-frequency spikes (lasting for about 1 min), which can be clearly seen in Figure 5 (bottom), a blown-up plot over 1 hour measurement period. In contrast, CO2 measurements showed many more spikes on the top of baseline over the same measurement period. In the entire field campaign, the measured NH3 varied from below the detection limit (0.1 ppbv) to a maximum of 197.4 ppbv with an average of 0.8 ppbv.
 As mentioned earlier, there were four significant precipitation events during the field campaign. Figure 6 shows the effect of the snowfall from 27 to 28 January on the measurements of NH3 and CO2. The snowfall with an accumulation of 5.3 inches started at 1830 LT, 27 January and ended at 0915 LT, on the following day while the ambient temperature varied from −7.3°C to −2.7°C. As indicated in Figure 6, there was no observed drop of either NH3 or CO2 (as anticipated) concentrations when the snow precipitation started. Meanwhile, the measured H2O vapor increased slowly as the snow started, and then decreased as the ambient temperature dropped during the nighttime periods. Interestingly, some high NH3 spikes occurred in the earlier morning hours. Most likely, these earlier morning hour spikes resulted from the traffic emissions from snowplows clearing the service road and parking lots adjacent to the measurement site. It is noted that each NH3 spike was closely correlated with the corresponding CO2 spike. However, not all CO2 spikes were associated with NH3 spikes. In contrast to a snowy day, Figure 7 appears to demonstrate an apparent rain-scavenging effect of the ambient NH3 and CO2 concentrations. The rain precipitation started at 1155 LT and ended at 2000 LT on 3 February. A rainfall of 1.15 inches was documented while the ambient temperature varied from 1.8°C to 5.2°C. When the rain started, both NH3 and CO2 showed a dramatic drop with the increase of ambient water vapor. However, meteorological measurements indicate that the observed decrease was most likely due to a rapid air mass change as depicted by the abrupt shift in wind direction from the north to the southeast. When the wind direction changed, the warm front was quickly moving through from the Atlantic Ocean. As evident in Figure 7, there were still NH3 spikes observed in the steady rain. Likewise, the observed NH3 spikes were correlated with the occurrence of CO2 spikes.
 A close correlation between NH3 and CO2 measurements was also observed over longer measurement periods. Such an observation is shown in Figure 8 (top), which presents the data comparison of NH3 and CO2 measurements over 24 hours from 5 to 6 February. Clearly, the 10-min time series for both species tracked each other very well. High R2 (0.66) of the corresponding scatterplot in Figure 8 (bottom) of NH3 versus CO2 implies that up to 66% of NH3 variation can be explained by mobile source emissions using CO2 as a marker of traffic exhaust. The yielded slope also reveals an exhaust [NH3]/[CO2] emission ratio of 0.12 ppbv/ppmv.
 In the first half of the field campaign, parking on campus was minimal because of the school winter break. A few days before the start of the spring semester of Queens College (2 February), car traffic increased on campus. Two different concentration patterns were observed during no-school and school days, on the basis of a variety of chemical parameters (such as NOx, CO, VOC) measured during the campaign. Accordingly, the NH3 data set used for the diurnal pattern analysis is divided into two groups, on the basis of the two activity periods. Figure 9 presents the diurnal pattern of NH3 measurements in no-school days from 10 to 29 January. As indicated in Figure 9, the NH3 results were relatively flat with a lower value around 0.5 ppb from the midnight to earlier morning, and then climbed up until the first peak at 1000 LT. The second peak occurred around 1730 LT, and then decreased slowly into the lower value at 2200 LT. Obviously, the second peak during the late afternoon hours is much higher than the first peak in the morning hours. This observation will be reasonably explained in the later discussion. A diurnal pattern obtained from the school day data set is shown in Figure 10. In contrast, a more distinct bimodal diurnal pattern was observed. The NH3 concentration change from the midnight to the earlier morning hours was almost the same as the no school days. One observed difference is that the first peak showed up 1 hour earlier, but higher in magnitude compared with the no-school days. The second peak came up 1 hour later, with a much higher magnitude. Such an observation was consistent with the traffic hours based on the school schedule.
 It has been shown by previous studies [Shelef and Gandhi, 1974; Pingent and De Soete, 1989; Huai et al., 2003] that ammonia can be formed on the catalyst surface of mobile source emissions control systems. The high-frequency NH3 spikes observed typically during rush hour periods were most likely produced by vehicles entering and exiting the parking lot on the service road, next to the sampling site. It is understandable that such an observation by TDLAS system might be missed by other measurement techniques with a slow time response, as the duration of most spikes was less than 1 min. As stated earlier, not all CO2 spikes were associated with NH3 spikes while each NH3 spike was closely correlated with a CO2 spike. This observation suggests that not every vehicle under on-road driving conditions emits gaseous NH3 into ambient air. However, the previous study performed by Huai et al.  showed that to certain extent, all 8 tested vehicles emitted NH3. In order to better understand the nature of NH3 emissions from mobile sources, it is necessary to perform further study.
 For a better understanding of the NH3 emission ratio, we have done the regression analysis for the whole data set from the entire month of study. The analysis generated a NH3 emission ratio of 0.15 ppbv/ppmv and a poor R2 of 0.16. Considering the fact that more near detection limit data measured in the no-school days might result in the poor correlation, the data analysis was then performed for the school-day data set since the ambient NH3 levels were observed to substantially increase during the school activity period. However, the yielded R2 of 0.23 only showed a little improvement. In comparison, two other reasons might play a significant role in the poor correlation. First, the ambient CO2 concentration used for the regression analysis contained contributions not only from traffic exhaust but also other sources such as heating systems. Its resulting impact would be a less correlation with ammonia and the underestimated NH3 emission ratio. Secondly, the NH3 emission ratio varied by vehicle, as is evident in the analysis of individual car plumes shown in Figure 11, where the linear regression of eight plumes sampled on 4 February 2004 indicates the NH3 emission ratios ranging from 0.02 to 0.93 ppbv/ppmv. For a few plumes illustrated in Figure 12, the NH3 emission ratio can be much higher (2.7 ppbv/ppmv). Evidently, each plume shows a very tight correlation between NH3 and CO2 concentrations with R2 ranging from 0.74 to 0.97. The observed difference in the emission ratios for individual plume is likely associated with many parameters including engine/catalyst type, engine state (hot vs. cold), fuel type and ambient conditions. Such emission ratios, [NH3]/[CO2], can be used to estimate vehicular NH3 exhaust emission [Baum et al., 2001]. Considering the variability observed in this ratio over the course of the field experiment, one is inclined to view the emission ratio of 0.12 ppbv/ppmv in Figure 8 as more representative of a fleet average. However, this must be qualified significantly, since we have no data to support that the operating conditions and vehicle population sampled is at all representative of the fleet as a whole. That being said, it is interesting to note, that on the basis of a mean CO2 mass emission rate of 2100 g/L, and a mean fleet fuel efficiency of 7.1 km/L [Baum et al., 2001] and the reported emission ratio, one calculates an NH3 emission factor of 35.5 mg/km, a value within the range of those reported in laboratory vehicle emissions testing studies. [Huai et al., 2003].
 The diurnal patterns for both activity periods showed bimodal features with the afternoon peak higher than the morning peak. In the morning, vehicles entering the parking lot are representative of engines that have reached typical operating temperatures as a result of the commute to the college. Hot engines and catalysts have been shown to emit lower levels of NH3 on average than cold start engines and catalysts [Huai et al., 2003]. Commuters exiting the parking lot with their vehicles in the afternoon will typically be in a cold start mode, that is, the engine and catalytic converter will not have reached operating temperature. Under these conditions, vehicles are known to emit more NH3. The observed higher NH3 concentrations in the late afternoon hours can in part be explained by this enhanced local source of NH3 emissions from the cold start vehicles. It is also estimated that the relatively higher ambient temperatures in the afternoon might contribute to a higher NH3 concentration by 10%, because of a shift in the thermo equilibrium of NH4 NO3 [Tang, 1980].
 As shown in Figure 13, the [NH3]/[CO2] ratios for the entire measurement period also presented a bimodal feature with a higher afternoon peak. This feature further confirms that more NH3 is released from the catalyst as it warms up. The mean value over the entire month of the study is 1.9 × 10−3 (ppbv/ppmv), with a maximum value of 0.042 (ppbv/ppmv). Different from the NH3 emission factor discussed above, the [NH3]/[CO2] ratio is time-dependent. By knowing the ambient CO2 concentration, the NH3 emission would be readily estimated.
 As described in the introduction, traffic exhaust has not been considered a significant source of NH3 emissions. In this work, the diurnal patterns associated with traffic exhaust strongly suggest that the NH3 emissions from mobile sources could be a major source in an urban area, like New York City. It is noted that this field campaign was performed in a cold winter. The NH3 emissions from major sources such as animal wastes, ammonia-based fertilizers, as well as industrial and soil emissions are strongly dependent on ambient temperature, typically low during wintertime high-latitude conditions. It is quite likely that a different diurnal pattern would be observed at this location under summertime conditions.
6. Summary and Conclusion
 During the PMTACS-NY 2004 winter field campaign, the TDLAS together with LI-7000 CO2/H2O analyzer was collocated to measure NH3, CO2 and H2O from 10 January to 6 February 2004. More than 80% data coverage for NH3 measurements was achieved over the entire field campaign. The time series of NH3 measurements showed a high variability with the concentrations ranging from the below the detection limit (0.1 ppbv) to 197.4 ppbv with an average of 0.8 ppbv (1σ).
 Many high-frequency NH3 spikes believed to be associated the NH3 emissions from mobile sources were observed during rush hour periods. The occurrence of the NH3 spikes was closely correlated with the CO2 spikes, a good marker of traffic exhaust. The correlation between two species yielded an NH3 emission ratio of 0.12 ppbv/ppmv, which can be used to estimate a NH3 emission factor. Another avenue to estimate NH3 emission is the NH3 to CO2 ratios, which showed both ambient CO2-dependent and time-dependent. On a snowy day, no obvious drop of NH3 and CO2 concentrations was measured as the ambient H2O vapor increased. The observed dramatic decrease in the ambient NH3 and CO2 concentrations on a rainy day most likely resulted from a quick air mass switch. Two bimodal diurnal patterns associated with the rush hour traffic were observed in no school days and school days. The difference in peak magnitude was due to more NH3 emissions from cold start vehicles in the late afternoon hours than the early morning hours. Such an observation may suggest that the NH3 emissions from the traffic exhaust could be a major source of the ambient NH3 in urban areas.
 This work was supported in part by the New York State Energy Research and Development Authority (NYSERDA), contract 4918ERTERES99; the U.S. Environmental Protection Agency (EPA) cooperative agreement R828060010; New York State Department of Environmental Conservation (NYS DEC), contract C004210; and New York State Science, Technology and Academic Research (NYSTAR) contract 3538479. Although the research described in this article has been funded in part by the U.S. Environmental Protection Agency, it has not been subjected to the Agency's required peer and policy review and therefore does not necessary reflect the views of the Agency and no official endorsement should be inferred. The excellent technical support from Aerodyne Research Inc. and insightful review of Mark S. Zahniser are gratefully acknowledged.