1. Introduction
In this paper, we describe the use of multiwavelength Langley plots to compare on-site measurements of aerosol optical depth (AOD) with those made at the same site using the same technique nearly 35 years ago by King et al. (1980). Since then, the city of Tucson, Arizona, has tripled in size to a population of approximately a million people, and it is of interest to examine the effects of urban growth on aerosol optical depth over the long term. We also compare changes in the retrieved ozone column abundance over the past 35 years.
It is well known that aerosols can play an important role in regional and global climate, both directly by scattering and absorbing solar radiation and indirectly by changing cloud microphysical properties (e.g., Twomey 1974; Russell et al. 1997; Forster et al. 2007). Aerosols can also have detrimental effects on human health and welfare (Dockery and Pope 1994; Bascom et al. 1996) and are, therefore, regulated under the Clean Air Act. Because of their short atmospheric residence times and strong local sources, aerosols are often not uniformly mixed in the atmosphere, making them highly variable both temporally and spatially.
On a cloud-free day, the extinction of solar radiation observed at the surface is due to absorption and scattering by aerosols and gases, including water vapor. Relative measurements of direct normal irradiance (DNI) in carefully selected spectral regions with no molecular absorption can be used to extract AOD, since Rayleigh scattering by air molecules can easily be calculated for a known surface pressure. This procedure is complicated by the broad ozone absorption in the visible region (Chappuis band), however. Multiwavelength sun photometry has a long history of being used as a tool to infer optical characteristics of an air column (e.g., Shaw et al. 1973; Shaw 1983; Cachorro et al. 1987; Holben et al. 1998; Goering et al. 2005), and several methods have been proposed to determine AOD using measurements of DNI in the visible portion of the electromagnetic spectrum (e.g., King and Byrne 1976; Flittner et al. 1993). This study adopts the method described by King and Byrne (1976) and King et al. (1980) to simultaneously retrieve AOD and ozone column abundance over Tucson over a 29-month period from 1975 to 1977. Tucson has undergone marked changes over this time, with its population almost tripling. On the one hand, an increase in population causes an increase in energy consumption, which translates into higher aerosol emissions (both vehicular and industrial) and, thus, an increase in aerosol concentration. On the other hand, paving of roads, improved industrial practices, and emission controls are expected to reduce the concentration of aerosols. The goal of this study is to evaluate whether and how these competing factors have affected the AOD and ozone column abundance over Tucson over the last 35 years. This goal is met by comparison of our recent measurements with the work done by King et al. (1980) using the same techniques.
2. Instrumentation
Ground-based measurements of spectral DNI were made with a high-resolution spectrophotometer (Avantes BV models AvaSpec-ULS2048 and AvaSpec-NIR256) at The University of Arizona in Tucson (32.13°N, 110.57°W) on the roof of a 6-story building from March 2010 to August 2012. The same type of instrument is used by the U.S. Department of Energy in the Atmospheric Radiation Measurement Program. A detailed description of their instrument [Shortwave Array Spectroradiometer–Zenith (SASZE)] can be found online (http://www.arm.gov/instruments/sasze). A taller building west of the measurement site obstructed the view in the late afternoons, which corresponded to solar zenith angles of greater than ~75° from October to April. To ensure that we measured only the direct component of the solar radiation, a collimator with a field of view of 5° was mounted on a single-axis solar tracker. The solar tracker was aligned to the sun with a pinhole on a pyrheliometer that was comounted on the tracker with the collimator. The collimator was then aligned parallel to the pyrheliometer. Since the field of view of the collimator is much larger than the solid angle of the sun (0.5°), a diffuser was incorporated in the collimator to eliminate any angular sensitivity in the measurements. The spectrometer covers the spectral region 300–2500 nm with two detectors—a silicon detector (model AvaSpec-2048 USB2-RM) and an indium–gallium–arsenide (InGaAs) detector (model Avaspec-NIR256 2.5-RM). In this paper we report only on 300–1100 nm, where the silicon detector is sensitive (with a temperature coefficient of resistance of −0.07 per degree Celsius), and at a resolution of 2.4 nm. The spectrophotometer was located in an air-conditioned room, and it was connected to the collimator with a 6-m-long optical fiber. Here we do not report data that were acquired with the InGaAs charge-coupled device (CCD) detector, but, since it is known to be temperature sensitive, we placed the instrument on an aluminum water-cooled plate and enclosed the entire setup in an insulated box, resulting in a regulated temperature of 24.2° ± 1.1°C. The solar spectrum was typically measured at 5-min intervals from 0900 to 1800 (local time), yielding a total of ~110 data points for each Langley plot each day. The silicon detector had an integration time of 1.1 ms and averaged 1200 scans per data point. For this paper, only measurements made on cloud-free days were analyzed. The selection of data is based on the linearity of the Langley plot over each observation period and visually observed clear-sky conditions, as described in the next section. A total of 60 days were analyzed over 30 months. By contrast, King et al. (1980) were able to analyze 133 cloud-free days over a 29-month period from 1975 to 1977.
3. Method
a. Determination of total optical depth
As long as the sky remains cloud free and atmospheric conditions remain constant, a plot of natural log of the spectrophotometer signal at wavelength λ versus the air mass should be linear, with the slope being equal to the negative of an average value of the total optical depth for the observation period and the y intercept being equal to the natural logarithm of the signal corresponding to the solar irradiance at that wavelength at the top of the atmosphere, that is, at zero air mass. Figure 1 shows an example of a Langley plot (lnF vs 1/μ) for data at two wavelengths measured in Tucson on a cloud-free day. A linear least squares fit yields a slope of −τt. The y intercept yields lnF0. Because this instrument has not been calibrated for absolute irradiance, however, the intercept cannot be used to infer the top-of-atmosphere absolute solar flux.
The linear relationship between the natural logarithm of the solar irradiance (or the arbitrary spectrophotometer signal) and atmospheric air mass for two wavelengths: 521.5 nm (slope = −0.139 ± 0.002, intercept = 8.549 ± 0.005, and correlation coefficient squared r2 = 0.990) and 440.2 nm (slope = −0.241 ± 0.002, intercept = 7.547 ± 0.005, and r2 = 0.990) on 16 Nov 2010 in Tucson. The inset shows the wavelength dependence of AOD.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-14-0001.1
Only days with visibly cloud-free sky were selected. Satellite imagery from the Geostationary Operational Environmental Satellite 13 (GOES-13) infrared 10.8-μm channel was used, but, because it is difficult to be certain that thin cirrus clouds were absent by visual inspection or satellite imagery, additional criteria were developed (Table 1). The King et al. (1980) study, conducted at the same location from 1975 to 1977, used only afternoon measurements with stable periods of observation. Hence, in this study also only the afternoon data have been analyzed. In reality, even on cloudless days, it is difficult to find long periods of time with constant optical depth in an urban location because of potential effects of human activity. Even a slow, monotonic increase/decrease in optical depth will yield a “good” Langley plot (Marenco 2007) as shown in the inset of Fig. 2, which is a plot of an artificially generated signal given an assumption of a monotonically increasing total optical depth. Figure 2 shows the actual Langley plot on a cloudless day when the total optical depth was increasing. Examination of the figure demonstrates that in such cases the data measured around local noon are most sensitive to temporal changes in the atmosphere, as revealed by a “hook” shape. Below an air mass of 2 the rate of change of air mass is small, creating a greater opportunity for atmospheric changes to affect the linear regression. Therefore, it has been suggested that the data around noon should be ignored and that only measurements between 2 and 6 air masses should be analyzed (Harrison and Michalsky 1994; McArthur 2005). This is a second reason for analyzing only afternoon measurements between 2 and 5.7 air masses (which correspond to approximately 1600–1800 LT) to determine the average total optical depth for a day.
Criteria for identifying cloud-free days without atmospheric disturbances on the basis of statistical parameters of the Langley fit.
The Langley plot for data collected on a cloudless day (12 Oct 2011) in Tucson with a steadily increasing TOD. The inset shows an artificial Langley plot for a monotonically varying TOD using the intercept calculated for the 2 and 4 air masses’ linear fit of the observations. The “hook” shape observed near solar noon is due to increasing optical depth.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-14-0001.1
b. Simultaneous retrieval of aerosol optical depth and ozone column abundance
Absorption coefficients (cm2 molecule−1) for (a) ozone (295 K), (b) nitrogen dioxide (293 K), (c) oxygen (293 K), and (d) water vapor (290 K). Note the different scales on the y axes (Keller-Rudek et al. 2013; available online at http://www.uv-vis-spectral-atlas-mainz.org).
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-14-0001.1
Narrow absorption features of gases like oxygen and water vapor can be eliminated by careful selection of wavelengths, but broad absorption features such as that of ozone cannot be avoided in determining aerosol optical depth using Eq. (2). Ozone absorbs in the visible Chappuis-band region from 440 to 880 nm, centered near 600 nm (Fig. 3).
So, the best estimates of the constants b0, b1, and b2 are those values for which the probability is maximized or χ2 is minimized. The χ2 statistic is calculated for different ozone column abundances η. The value of η for which χ2 attains a minimum value yields the ozone column abundance (m), which is converted to Dobson units. The values of b0, b1, and b2 corresponding to this ozone value are then used in Eq. (6) to retrieve the aerosol optical depth.
c. Selection of wavelengths
The wavelengths were selected so as to capture both aerosol and ozone extinctions while eliminating absorption effects from other gaseous species such as water vapor and carbon dioxide. Wavelengths closely matching those used in the King et al. (1980) study were selected for direct comparison. Spectrophotometer measurements centered at 440.2, 521.5, 611.8, 864.8, and 871.5 nm were used for the analysis. In the King et al. (1980) study, the longer wavelength was centered at 779.7 nm. Although this wavelength lies in a window region, it was avoided in the current analysis because of its proximity to two water vapor absorption features. The King et al. (1980) study used a filter radiometer with a resolution of 12 nm that measured a signal that was smoothed over the 12-nm window and thus was not strongly affected by the absorption band peaking at 780 nm. At the much higher resolution (2.4 nm) of the grating used in the current study, however, the water vapor absorption at 780 nm affected the retrievals, resulting in a higher total optical depth at that wavelength. To avoid the water vapor absorption band, a wavelength of 864.8 nm was used instead of 779.7 nm. The AOD at 864.8 nm was approximately the same as that at 871.5 nm, however (Fig. 1 inset).
4. Results
a. Total optical depth
The technique described above was used to retrieve ozone column abundance and AOD at The University of Arizona site. Between March of 2010 and August of 2012, a total of 60 cloudless days of data were analyzed and compared with the results of King et al. (1980) (who analyzed 133 cloud-free days from 1975 to 1977). Figure 4 shows the monthly variation in total optical depth (TOD) at each of the five wavelengths (440.2, 521.5, 611.8, 864.8, and 871.5 nm). The error bars represent 1 standard deviation in the determination of τt(λ) using the Langley method.
Variation in monthly-mean TOD between March 2010 and August 2012 at different wavelengths. The error bars indicate the standard deviation in determination of TOD.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-14-0001.1
The total optical depth includes Rayleigh scattering, which is proportional to λ−4 and therefore leads to higher TOD at shorter wavelengths, as observed. The error is largest at the longest wavelengths because the actual signal received by the instrument is lowest here and most sensitive to noise. The maximum TOD was observed in July and August (onset of the monsoon season), similar to the results of King et al. (1980). The minimum TOD can be seen to occur during the winter months of November, December, and January at all wavelengths, which was first reported by King et al. (1980).
b. Aerosol optical depth
The monthly-mean aerosol optical depth retrieved from TOD using Eq. (5) is shown in Fig. 5. The AODs were averaged over all observation days in the month, thereby smoothing out any anomalous events such as residual aerosols from dust storms and wildfires in the region. The fact that clouds restricted the observations to only a few days each month, however, might contribute to a bias in the averaged result. The minimum AOD is found to occur during the winter from November to January, and the maximum AOD occurred during July, similar to the behavior of the TOD. The winter minimum is in agreement with Multiangle Imaging SpectroRadiometer (MISR), Moderate Resolution Imaging Spectroradiometer (MODIS), Aerosol Robotic Network (AERONET), and Goddard Ozone Chemistry Aerosol Radiation and Transport (GOCART) data for Tucson (Sorooshian et al. 2011), which is associated with lower concentrations of particulate matter (both coarse and fine) and lower water vapor content. AOD in Tucson has been found to reach a maximum during the North American monsoon season (Sorooshian et al. 2011), which typically lasts from July to September. Hygroscopic particle growth associated with increased humidity might lead to increased scattering, which is a sensitive function of particle diameter. Wildfires that occur just prior to the monsoon might also contribute to the increased AOD. The monthly AOD retrievals presented here were found to be about 19% higher than the monthly AODs retrieved by King et al. (1980), as shown in Fig. 6.
Monthly variation in the monthly averages of AOD between March 2010 and August 2012 at the different wavelengths. The numbers in parentheses indicate the number of days included in the computation of the average.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-14-0001.1
The ratio of the current retrieved AOD to that from the original King et al. (1980) study at each of the four wavelengths. The annual average of this ratio, for all wavelengths, is 1.19, representing an approximately 19% increase in AOD over the 35-yr period.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-14-0001.1
Our study suggests that the AOD increased at a rate of approximately 3.9 × 10−4 yr−1 between the two observation periods that are 35 years apart. Other studies in the region, however, have reported a decline in the local aerosol concentrations in the past decade. The annual mean concentration of particulate matter with effective aerodynamic diameter ≤ 2.5 μm (PM2.5) measured at the Orange Grove site in Tucson showed a decline of 0.34 μg m−3 yr−1 from 1999 to 2010 (Pima County Department of Environmental Quality 2011). Measurements of PM2.5 at a nearby Interagency Monitoring of Protected Visual Environments (IMPROVE) site, Saguaro West (32.24°N, 111.21°W), also dropped over the period of the current study by up to 0.66 μg m−3 yr−1 between 2003 and 2009, depending on the season (Sorooshian et al. 2011). On the other hand, a decade-long campaign on Mount Lemmon (2971 m above sea level), a forested area immediately north of Tucson, showed no statistically significant trend in PM2.0 extinction coefficient from 1992 to 2002 (Matichuk et al. 2006). It must be noted here that, while they are all from the same study region, these comparative measurements are all point measurements made at the surface and are not necessarily representative of the total aerosol loading in the atmosphere over the Tucson valley. The time frames also differ.
c. Ozone column abundance
Figure 7 shows a plot of retrieved daily ozone column abundances. The sinusoidal curve is added to represent a typical annual cycle over the midlatitudes in the Northern Hemisphere (Hudson et al. 2006), with higher values in summer and lower values in winter. The inferred ozone abundances are within the range of ozone values retrieved in the 1970s by King et al. (1980), indicating no significant change in ozone column abundance over Tucson over the past 35 years. The ozone retrievals in the current study were found to be on average 5.5% lower than the Ozone Monitoring Instrument on board the Aura satellite and 4% higher than King et al. (1980). We note here that the ozone absorption cross sections used in the two studies may be different [King et al. (1980) do not reference their cross-section source], which may lead to a systematic bias in the current results relative to the past ozone retrievals.
Daily ozone column abundances over Tucson inferred from the variation of TOD. The dashed line is the least squares sinusoidal fit to the King et al. ozone retrievals, of the form η(x) = 275 + 45 sin(2πx − 0.72), where x is the fractional time of year.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-14-0001.1
Although the cloud-free criteria in Table 1 were satisfied, a close examination of the Langley plots for June revealed a slight upward curvature in the data toward the end of the day, indicating that the optical depth was lower in the late-afternoon hours, probably as a consequence of a change in air mass. The air mass in the afternoon is largely influenced by the thinly populated desert region west of the measurement site. This air mass is expected to have lower aerosol loading relative to that in the morning, which is influenced mostly by the urban emissions closer to the site. The curvature in the Langley plot causes the slope of the linear fit to be smaller, which in turn results in lower AOD and ozone values, as demonstrated in Fig. 8. Hence, the Langley plot was repeated using the entire afternoon data (without the 2–5.7 airmass restriction) to minimize the effect of the curvature. This approach yielded higher TODs and hence higher AODs and ozone column abundances in June but did not make any significant changes in the other monthly retrievals.
Langley plot at 440.2 nm on a cloudless day (10 Jun 2011 in Tucson) showing linear fits over two airmass ranges. The slope decreases by 6.3% if the observations are restricted to 2–5.7 air masses owing to a slight curvature in the data, caused by a change in TOD.
Citation: Journal of Applied Meteorology and Climatology 53, 8; 10.1175/JAMC-D-14-0001.1
Just prior to the observations in June of 2011, there were two wildfires that could potentially have affected our observations. The Horseshoe 2 fire, 150 km southeast of Tucson, lasted from 8 May to 25 June 2011 and burned approximately 900 km2. The Wallow fire, 250 km northeast of Tucson, lasted from 29 May to 8 July 2011 and burned 2200 km2, thereby becoming Arizona’s largest wildfire. Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) back trajectories run for these periods show the 500- and 1000-m winds to be southerly to southwesterly, however, and thus smoke from the wildfires is unlikely to have been advected into the valley.
5. Conclusions
The population in the Tucson valley area has grown threefold from 1970 to the present. With an increase in population we expect an increase in energy consumption and therefore fuel combustion. A developing urban area implies a rise in vehicular activity and construction. With an increase in these activities associated with urban growth, we anticipate an increase in atmospheric aerosols. This study shows that, in 35 years, the aerosol optical depth in Tucson has increased by approximately 19% despite increased road paving, closure of regional mines and smelters, widespread use of three-way catalytic converters in automobiles, and more stringent air-quality regulations.
Acknowledgments
We acknowledge the support of Science Foundation Arizona (SRG 0375-08).
REFERENCES
Alexandrov, M. D., A. A. Lacis, B. E. Carlson, and B. Cairns, 2008: Characterization of atmospheric aerosols using MFRSR measurements. J. Geophys. Res., 113, D08204, doi:10.1029/2007JD009388.
Bascom, R., and Coauthors, 1996: Health effects of outdoor air pollution. Committee of the Environmental and Occupational Health Assembly of the American Thoracic Society. Amer. J. Respir. Crit. Care Med., 153, 3–50, doi:10.1164/ajrccm.153.1.8542133.
Bucholtz, A., 1995: Rayleigh-scattering calculations for the terrestrial atmosphere. Appl. Opt., 34, 2765–2773, doi:10.1364/AO.34.002765.
Cachorro V. E., A. M. Defrutos, and J. L. Casanova, 1987: Determination of the Angstrom turbidity parameters. Appl. Opt., 26, 3069–3076, doi:10.1364/AO.26.003069.
Dockery, D. W., and C. A. Pope, 1994: Acute respiratory effects of particulate air pollution. Annu. Rev. Public Health, 15, 107–132, doi:10.1146/annurev.pu.15.050194.000543.
Flittner, D. E., B. M. Herman, K. J. Thome, J. M. Simpson, and J. A. Reagan, 1993: Total ozone and aerosol optical depths inferred from radiometric measurements in the Chappuis absorption band. J. Atmos. Sci., 50, 1113–1121, doi:10.1175/1520-0469(1993)050<1113:TOAAOD>2.0.CO;2.
Forster, P., and Coauthors, 2007: Changes in atmospheric constituents and in radiative forcing. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 129–234. [Available online at http://www.ipcc.ch/pdf/assessment-report/ar4/wg1/ar4-wg1-chapter2.pdf.]
Goering, C. D., T. S. L’Ecuyer, G. L. Stephens, J. R. Slusser, G. Scott, J. Davis, J. C. Barnard, and S. Madronich, 2005: Simultaneous retrievals of column ozone and aerosol optical properties from direct and diffuse solar irradiance. J. Geophys. Res., 110, D05204, doi:10.1029/2004JD005330.
Harrison, L., and J. Michalsky, 1994: Objective algorithms for the retrieval of optical depths from ground-based measurements. Appl. Opt., 33, 5126–5132, doi:10.1364/AO.33.005126.
Holben, B. N., and Coauthors, 1998: AERONET—A federated instrument network and data archive for aerosol characterization. Remote Sens. Environ., 66, 1–16, doi:10.1016/S0034-4257(98)00031-5.
Hudson, R. D., M. F. Andrade, M. B. Follette, and A. D. Frolov, 2006: The total ozone field separated into meteorological regimes—Part II: Northern Hemisphere mid-latitude total ozone trends. Atmos. Chem. Phys., 6, 5183–5191, doi:10.5194/acp-6-5183-2006.
Keller-Rudek, H., G. K. Moortgat, R. Sander, and R. Sörensen, 2013: The MPI-Mainz UV/VIS spectral atlas of gaseous molecules of atmospheric interest. Earth Syst. Sci. Data, 5, 365–373, doi:10.5194/essd-5-365-2013.
King, M. D., and D. M. Byrne, 1976: A method for inferring total ozone content from the spectral variation of total optical depth obtained with a solar radiometer. J. Atmos. Sci., 33, 2242–2251, doi:10.1175/1520-0469(1976)033<2242:AMFITO>2.0.CO;2.
King, M. D., D. M. Byrne, J. A. Reagan, and B. M. Herman, 1980: Spectral variation of optical depth at Tucson, Arizona between August 1975 and December 1977. J. Appl. Meteor., 19, 723–732, doi:10.1175/1520-0450(1980)019<0723:SVOODA>2.0.CO;2.
Marenco, F., 2007: On Langley plots in the presence of a systematic diurnal aerosol cycle centered at noon: A comment on recently proposed methodologies. J. Geophys. Res., 112, D06205, doi:10.1029/2006JD007248.
Matichuk, R., B. Barbaris, E. A. Betterton, M. Hori, N. Murao, S. Ohta, and D. Ward, 2006: A decade of aerosol and gas precursor chemical characterization at Mt. Lemmon, Arizona (1992 to 2002). J. Meteor. Soc. Japan, 84, 653–670, doi:10.2151/jmsj.84.653.
McArthur, L. J. B., 2005: Baseline Surface Radiation Network (BSRN) operations manual, version 2.1. World Climate Research Programme Rep. WCRP-121/WMO Tech. Doc. WMO/TD-1274, 176 pp. [Available online at http://epic.awi.de/30644/1/McArthur_2005.pdf.]
Pima County Department of Environmental Quality, 2011: 2010 Air quality summary report for Pima County, Arizona, Pima County Dept. of Environmental Quality Rep. AQ-376, 29 pp. [Available online at http://www.rosemonteis.us/files/references/pcdeq_2011.pdf.]
Rozenburg, G. V., 1966: Twilight: A Study in Atmospheric Optics. Plenum, 358 pp.
Russell, P. B., S. A. Kinne, and R. W. Bergstrom, 1997: Aerosol climate effects: Local radiative forcing and column closure experiments. J. Geophys. Res., 102, 9397–9407, doi:10.1029/97JD00112.
Shaw, G. E., 1983: Sun photometry. Bull. Amer. Meteor. Soc., 64, 4–10, doi:10.1175/1520-0477(1983)064<0004:SP>2.0.CO;2.
Shaw, G. E., J. A. Regan, and B. M. Herman, 1973: Investigations of atmospheric extinctions using direct solar radiation measurements made with a multiple wavelength radiometer. J. Appl. Meteor., 12, 374–380, doi:10.1175/1520-0450(1973)012<0374:IOAEUD>2.0.CO;2.
Sorooshian, A., A. Wonaschütz, E. G. Jarjour, B. I. Hashimoto, B. A. Schichtel, and E. A. Betterton, 2011: An aerosol climatology for a rapidly growing arid region (southern Arizona): Major aerosol species and remotely sensed aerosol properties. J. Geophys. Res., 116, D19205, doi:10.1029/2011JD016197.
Twomey, S., 1974: Pollution and the planetary albedo. Atmos. Environ., 8, 1251–1256, doi:10.1016/0004-6981(74)90004-3.
