ABSTRACT
Drought-stricken areas of Central America and Mexico were victimized in 1998 by forest and brush fires that burned out of control during much of the first half of the year. Wind currents at various times during the episode helped transport smoke from these fires over the Gulf of Mexico and into portions of the United States. Visibilities were greatly reduced during favorable flow periods from New Mexico to south Florida and northward to Wisconsin as a result of this smoke and haze. In response to the reduced visibilities and increased pollutants, public health advisories and information statements were issued by various agencies in Gulf Coast states and in Oklahoma.
This event was also detected by a unique array of instrumentation deployed at the U.S. Department of Energy's Atmospheric Radiation Measurement (ARM) program Southern Great Plains Cloud and Radiation Testbed and by sensors of the Oklahoma Department of Environmental Quality/Air Quality Division. Observations from these measurement devices suggest elevated levels of aerosol loading and ozone concentrations during May 1998 when prevailing winds were favorable for the transport of the Central American smoke pall into Oklahoma and Kansas. In particular, aerosol extinction profiles derived from the ARM Raman lidar measurements revealed large variations in the vertical distribution of the smoke.
1. Introduction
A large number of forest and brush fires burned out of control during much of the first half of 1998 in drought-stricken areas of Central America and southern Mexico. Southerly and southwesterly winds at times transported the widespread and lingering smoke pall from Guatemala, Honduras, and southern Mexico over the Gulf of Mexico and into the United States, reaching from New Mexico to southern Florida and sometimes as far north as Wisconsin. Visibilities were routinely and sometimes severely reduced in some of these areas during spring 1998 as a result of this smoke and haze, affecting regional air quality. In Oklahoma, the National Weather Service Forecast Office in Norman issued a public information statement on 13 May describing the haze event (Fig. 1).
Biomass burning, such as forest fires, is a major source of air pollution, producing smoke aerosols that significantly contribute to global radiative forcing (Penner et al. 1992; Hobbs et al. 1997). The smoke aerosols produced by biomass burning contain a significant amount of light-absorbing materials composed of black carbon particles (Martins et al. 1998). Once in the atmosphere, smoke aerosols can undergo a series of transformations including coagulation, outgassing, and condensation and gas-to-particle conversion, which change their physical, chemical, and optical properties (Kotchenruther and Hobbs 1998; Reid and Hobbs 1998; Reid et al. 1998). Depending on their chemical composition and the surface albedo, aerosols produced from biomass burning can have either a warming or a cooling effect on the atmosphere (Bahrmann and Saxena 1998; Iacobellis et al. 1999; Fraser and Kaufman 1985). According to a recent estimate, 104 Tg of smoke aerosols are formed annually from biomass burning (Andreae 1991).
The 1998 Central American smoke event was detected by unique instrumentation deployed in Oklahoma and Kansas at the Southern Great Plains (SGP) Cloud and Radiation Testbed (CART) of the U.S. Department of Energy's Atmospheric Radiation Measurement (ARM) program (Stokes and Schwartz 1994) and by sensors deployed in central Oklahoma by the Oklahoma Department of Environmental Quality (ODEQ) Air Quality Division. The CART site covers a rectangular region of over 140 000 km^sup 2^ from southern Kansas into south-central Oklahoma (Fig. 2). Data collected in Oklahoma and Kansas suggest elevated levels of aerosol loading and ozone concentrations persisted over the CART site during parts of May 1998 when the wind field was favorable for transport of the smoke pall from Central America and southern Mexico. This paper describes the encroachment of the smoke pall into Oklahoma and Kansas as measured by instruments that are unique in their spatial and temporal deployment. Analyses include those from continuously operating ARM instruments such as the aerosol observing system, Raman and micropulse lidars, solar and infrared radiation stations, multifilter rotating shadowband radiometer, and Cimel sunphotometer, and from ODEQ ozone monitors. These instruments were augmented during the study period by the University of Utah polarization diversity lidar and a condensation nuclei counter aboard the University of North Dakota Citation aircraft, which both were at the CART site for a special cloud physics intensive observation period. The Raman lidar analyses shown here demonstrate a new capability for retrieving aerosol extinction profiles (Ferrare et al. 1999).
2. Large-scale environment of the smoke event
While the focus of this paper is to describe the smoke event as captured by the CART site's instrumentation in Oklahoma and Kansas, it is important to view the largescale setting of the event. The smoke event, as it unfolded, was highly documented at various locations on the World Wide Web (WWW). An excellent WWW clearinghouse for information and images on this smoke event can be found at a site developed and maintained by the Center for Air Pollution Impact and Trend Analysis, Washington University in St. Louis, Missouri (http://capita.wustl.edu/CENTRAL-- AMERICA/).
a. Satellite detection of fire locations and smoke pall
Geostationary Operational Environmental Satellite-8 (GOES-8) algorithm products [National Oceanic and Atmospheric Administration/National Environmental Satellite, Data, and Information Service/Office of Research and Applications/ Advanced Satellite Products Team (NOAA/NESDIS/ ORA/ASPT); Cooperative Institute for Meteorological Satellite Studies (CIMSS), University of Wisconsin-Madison] helped identify fire "hot spots" in Central America and southern Mexico. The widespread magnitude of the fire area is evident in such images. Figure 3 (http://cimss.ssec.wisc.edu/goes/ misc/mex98135.gif) shows an example of the GOES8 Automated Biomass Burning Algorithm (ABBA, version 5.6) Experimental Fire Product for 15 May. The fire count as detected at 2215 UTC that day was 677. Enhanced visible and multichannel imagery was able to define the areal extent of the smoke and haze pall on many days. Figure 4 (http://capita.wustl.edu/ Data t are /Datasets/osei/goes8/ FSMmex134A.gif) courtesy of NOAA/NESDIS/ Operational Significant Event Imagery shows multichannel imagery for 1915 UTC 14 May. Smoke areas are identified on the image as a yellow-brown haze. Such images and surface reports indicated that the smoke pall made a number of encroachments into the United States during spring 1998. Prior to these encroachments, stagnation in air patterns in Central America and southern Mexico allowed for a buildup of smoke and haze that was subsequently transported northward when flow conditions permitted. See Prins et al. (1998) for a description of GOES-8 images and algorithm products as they were used during the Smoke, Clouds, and Radiation in Brazil (SCAR-B) experiment in 1995.
b. Satellite detection of aerosols
The Total Ozone Mapping Spectrometer (TOMS) aerosol data product [National Aeronautics and Space Administration Goddard Space Flight Center (NASA/ GSFC); http://toms.gsfc.nasa.gov/] contains measurements of UV-absorbing tropospheric aerosols from Earth Probe TOMS using the 331- and 360-nm wavelength channels. It can be used to generate daily global maps of these particles. The images represent smoke from a variety of ground-based sources, such as biomass burning whether naturally occurring or caused by agriculture, oil industry fires, or industrial smoke. TOMS detects smoke particles regardless of source or season and is reliable both over land and water. Since Earth Probe TOMS does not have full global coverage, images are created by compositing 2 days' worth of data. The largest amount of data comes from the date specified on the image. See Herman et al. (1997) for more details on TOMS detection of UV-absorbing aerosols.
Earth Probe TOMS detected smoke from fires in Central America and southern Mexico during spring 1998. Over the SGP CART, 14 and 19 May were indicated to have the most extensive smoke coverage (Fig. 5), though admittedly much more severe condi tions usually occurred to the south and east along the Gulf of Mexico coast. Coverage was indicated over the eastern half of the CART on 14 May and over all but the northwest corner of the CART on 19 May. The greatest degree of aerosol detection over the center point of the CART rectangle, known as the central facility and where the most extensive collection of ARM instrumentation resides, occurred on 19 May. Other days on which Earth Probe TOMS detected smoke over some portion of the CART (not shown) include 9, 15, 17, and 25 May (extreme southeastern Oklahoma); 13, 18, and 23 May (southern Oklahoma); 20 May (parts of eastern Kansas and Oklahoma); 21 May (parts of the eastern and northwestern Kansas and eastem Oklahoma); and 22 May (northeastern Kansas).
c. Trajectory analysis
To apportion the sources and transportation pathways of the aerosols observed at the ARM SGP CART, 5-day air parcel backward trajectories arriving at the CART site were calculated. The backward trajectories were computed by a three-dimensional trajectory model, Hybrid Single-Particle Lagrangian Integrated Trajectory-version 4 (HYSPLIT4) July 1999 version, developed at the NOAA Air Resources Laboratory (ARL) (see also Draxler and Hess 1998). The Final Run (FNL)-Northern Hemisphere (NH) archive furnished by NOAA ARL was used as the meteorological input to HYSPLIT4. The FNL-NH data come from the "final run" of the National Centers for Environmental Prediction Global Data Assimilation System in the Northern Hemisphere. The FNL-NH archive is at a reduced spatial resolution (about 190 km at 60 deg N) over the Northern Hemisphere from the model's native 1 deg grid. Temporal resolution is 6 h. These data were obtained from the National Climatic Data Center and analyzed by our Oak Ridge National Laboratory coauthors.
Trajectory models are subject to uncertainties arising from interpolations of sparse meteorological data, assumptions of vertical transport, wind field errors, and truncation errors (Stohl 1998). Earlier studies have employed either isobaric or isentropic methods for vertical mixing (Moody and Galloway 1988; Harris 1992; Polissar et al. 1999). These two methods may lead to large trajectory uncertainties under environmental conditions as addressed by Harris and Kahl (1994). Stohl and Seibert (1998) also pointed out that isobaric trajectories are not accurate in both tropospheric and stratospheric transports. In this study, the analyzed vertical wind fields from the FNL-NH data were used to calculate the vertical transport of the air masses; therefore, the trajectory uncertainty caused by vertical transport assumptions in isobaric and isentropic trajectories is eliminated and the total uncertainty is reduced. The overall uncertainty of the trajectory model used in this study is estimated to range from 15% to 30% of the traveling distance (R. R. Draxler 1998, personal communication).
The output of the HYSPLIT4 model is hourly trajectory endpoints indicating the location (latitude-longitude pairs) and height (meters above ground level, AGL) of the air parcel. Backward trajectories were computed every 3 h each day starting at 0000 UTC during April-June 1998. For the 3-month period, 728 backward trajectories were computed. Five hundred meters AGL was used to represent the surface level since more than 90% of the computed trajectories reached ground level when the actual sampling height (10 m AGL) of the ARM aerosol observing system (AOS) was used as the arrival height. The mixing heights during the modeling period ranged from 690 to 2190 m AGL (R. Coulter 1999, personal communication). The 500 rn AGL arrival height is below the tropospheric mixing layer. More than 96% of the trajectories were successfully calculated at the end of 5 days. Absent or incomplete (shorter than 5 day) trajectories were caused by missing FNL data during the period. After rejecting trajectories that hit the ground, approximately 85% of all trajectories computed were available for analysis. Trajectory results, which indicated transport from the Central American source region on many days in spring 1998, are presented in detail in section 3b describing ARM AOS measurements.
3. ARM SGP CART site
observations and discussion
a. Local weather observations
One of the responsibilities of operators at the SGP CART central facility is to observe and record hourly weather information. The observations are made whenever the site is staffed (typically between 0430-0830 and 1030-2230 UTC, Monday-Friday). Observations are also made 24 hours a day, seven days a week, during special experiments known as intensive observation periods (IOPs). Making the weather observation includes observing and recording the following: total cloud coverage; cloud amount, type, and height by quadrant of the sky and in a 30 deg arc directly overhead; current weather condition; wind speed and direction; atmospheric pressure; air temperature; and humidity. Other comments, such as indications of smoke, fire, dew, and frost, are also noted. A strict procedure that defines the method to observe, record, and enter the hourly surface weather observations into the CART site's data system is followed to ensure consistency regardless of who is making the observation. World Meteorological Organization standards are used for making cloud observations. These observations are made available on the WWW in the ARM SGP Meta Data System database (http:// www.db.arm.gov/).
This database was examined for all of May 1998 to determine whether there were observer indications of smoke or haze. Indeed, smoke observations were noted on 14-22 and 28-29 May. Observations for 24-- 25 and 30-31 May were not made because those days were on weekends when the site was not staffed, days on which it is possible that smoke or haze could have been observed. A cloud physics IOP occurred during 27 April-17 May, permitting 24 h day^sup -1^ observations through 18 May. Table 1 was constructed to display the hourly weather observations for the primary 14-- 19 May period.
As can be seen in Table 1, numerous smoke and haze visual observations were made on the days listed except for about a 54-h period after a cold frontal passage early on 15 May. The frontal passage, which cleaned the atmosphere of aerosols above the central facility, is clearly evident beginning at 1000 UTC 15 May. Haze observations began again at 1700 UTC 17 May and continued variously for the rest of the month. As is also seen in Table 1, the severity of the smoky haze varied, with 14 May through early 15 May, and 19 May visually observed to have the thickest haze.
b. Ground-based in situ aerosol sampling
The AOS is the primary ARM SGP CART platform for making in situ aerosol measurements 10 in above the ground surface. The system, located at the SGP CART central facility in northern Oklahoma (36 deg 37'N, 97 deg 30'W), was constructed at the Department of Energy (DOE) Environmental Monitoring Laboratory in New York. The AOS was deployed in April 1996 and has been operated and improved by the NOAA Climate Monitoring and Diagnostics Labora tory in Boulder, Colorado.
Quantities measured by the AOS include total scattering and hemispheric backscattering coefficients, light absorption coefficient, total condensation particle concentration, number concentration for particles of sizes ranging from 0.1- to 10-(mu)m diameter, ozone concentration, and aerosol scattering coefficient as a function of relative humidity. The raw sampling interval of all the AOS measurements is 1 min. Before the light scattering and absorption measurements, each aerosol sample is passed through an inertial impactor with a 50% cutoff particle diameter at either 1 (mu)m (alternate size) or 10 pm (default size). An automated inlet switch operated at a 5-min switching frequency is used to direct the air sample to the 1- or 10-(mu)m impactor to remove large particles. Particles of size 1 um and smaller are analyzed here. The total light scattering and hemispheric backscattering coefficients of aerosols are measured by a TSI Model 3563 three-wavelength nephelometer at 450, 550, and 700 nm. The light absorption coefficient is measured by a Radiance Research particle/soot absorption photometer at 565 nm. The total condensation nucleus concentration for particle sizes from 0.01-3.0 (mu)m is measured by a TSI Model 3010 condensation nuclei counter. All aerosol optical properties were measured at a relative humidity below 40% to minimize the effects of changing relative humidity on the measurement (see, e.g., Charlson et al. 1978). Higher relative humidity has been reported to increase the scattering coefficients and to decrease the hemispheric backscatter ratio of aerosols (Kotchenruther and Hobbs 1998). Volatile species, such as ammonium nitrate or low molecular weight organic compounds, may have evaporated when the sample was heated to reduce the relative humidity to below 40%. Bergin et al. (1997) showed that these losses are small for ammonium nitrate (or less volatile species) under the sampling conditions used in the AOS. However, condensed species that are more volatile than ammonium nitrate could have evaporated completely in the sampling system; no chemical measurements are available to assess these potential losses at the SGP CART site.
To comprehensively address the impacts of aerosols on the atmospheric radiation balance budget, one needs a full suite of measurements of the physical (as described above) and chemical properties of aerosols (e.g., Seinfeld and Pandis 1998). However, no aerosol chemistry measurement capability existed at the SGP CART site during the time of the smoke event. Since the time of the smoke event, the following additional data are being obtained at the site: single-- scattering albedo, the relative humidity dependence of aerosol light scattering, aerosol chemical composition (funded by the NOAA Office of Global Programs), and in situ vertical profiles of aerosol absorption and scattering coefficients. Thus, for this observational study of the 1998 Central American smoke event, we focused on the use of light-scattering and hemispheric backscattering coefficients and total condensation particle concentrations as observed by the SGP CART central facility AOS. The April-June 1998 measurements are compared to the same period from the previous year (1997). The 1998 data are also used in combination with air parcel backward trajectories to identify potential source areas for the smoke particles.
For a comprehensive analysis of smoke from biomass burning, see the special section of the Journal of Geophysical Research (1998, Vol. 103, No. D24) describing the SCAR-B experiment.
The hemispheric backscatter ratios and Angstrom exponents were derived based on the light-scattering and hemispheric backscattering coefficients measured at the three wavelengths. The hemispheric backscatter ratio was calculated as the ratio of hemispheric backscattering to the total scattering coefficients. The Angstrom exponent for AOS was calculated according to the following equation:
AOS data were utilized to help identify the smoke event at the SGP CART by considering the measured aerosol optical properties from both April-June 1997 (baseline) and 1998 and the backward trajectories for April-June 1998. A similar approach was used to identify regional and long-range transport of aerosols measured at Mauna Loa Observatory (Bodhaine et al. 1992) and at Mount Mitchell, North Carolina, in the southeastern United States (Bahrmann and Saxena 1998). Figure 6 shows the time series of 1-h-average data of (top) condensation nuclei count, (center) total scattering coefficient at 550 nm, and (bottom) hemispheric backscattering coefficient at 550 nm, respectively, from April to June of the two years (yearday 91 is 1 April and yearday 181 is 30 June in the figure). The plots of the total scattering and hemispheric backscattering coefficients at other wavelengths (i.e., 450 and 700 nm) are similar to those at 550 nm and therefore are not shown. Generally, the total scattering coefficients at 450 and 700 nm are about 150% and 60%, respectively, of the value at 550 nm. The hemispheric backscattering coefficients at 450 and 700 nm are about 130% and 80%, respectively, of the value at 550 nm. The gaps in the plots represent data either rejected during the quality assurance process or missing from the record due to instrument failure.
In Fig. 6 (top), the condensation nuclei counts (interpreted as the number concentration of particles greater than 10 nm) in 1997 and 1998 are comparable. Most of the particle concentrations range from 200 to 33 000 cm^sup -3^ in both years, with 3-month averages around 5000 cml. Such concentrations are typical of the count values observed in rural areas such as northern Oklahoma. The lack of condensation nuclei signal during the smoke event is indicative of the aged character of the smoke, as typical travel times of 60-84 h were sufficient for significant reduction of the particle number concentration due to coagulation.
In Fig. 6 (center and bottom), the total light scattering coefficients are positively correlated with the hemispheric backscattering coefficients during both years (correlation coefficient is greater than 0.97 in 1998 and greater than 0.95 in 1997; see Table 2). Unusually high total scattering and hemispheric backscattering coefficients from mid- through late May 1998 are present. The May 1998 means of the total and hemispheric backscattering coefficients are 86.81 Mm6sup -1^ and 8.26 Mm^sup -1^, respectively, which are about 3.5 times greater than the means in 1997 (23.31 and 2.98 Mm^sup -1^). Both the total scattering and hemispheric backscattering coefficients peaked on 14 May 1998 (yearday 134), with values of 389 and 30.6 Mm^sup -1^, respectively. The elevated levels of the aerosol light scattering coefficients strongly suggest the occurrence of significant pollution events in northern Oklahoma in May 1998.
Table 2 shows the correlation coefficient (r) matrix of the aerosol measurements during April-June 1997 and 1998. In both years, the condensation nucleus counts correlate poorly with the measured optical properties. Since the particles that arrived at the central facility are likely to be well aged and the condensation nuclei count data did not respond to the smoke event (Fig. 6, top), the poor correlation between condensation nuclei count data and the other AOS data shown in Table 2 is not unexpected. The correlation between the scattering and hemispheric backscattering coefficients is strong for both years (r > 0.95) at all three wavelengths.
Figures 7 and 8 show the aerosol optical properties derived for 1997 and 1998. The hemispheric backscatter ratio at 550 nm (Fig. 7) ranged from 0.09 to 0.22 for the 3-month period in 1998. According to Reid et al. (1998), regional haze dominated by biomass-- burning-produced aerosols has a hemispheric backscatter ratio from 0.10 to 0.14 at 550 nm. It is interesting to note that most of the hemispheric backscatter ratios during mid- through late May 1998 (approximately yeardays 130-150) fell within this range. Figure 8 (top) shows the Angstrom exponent plots in 1998 for 450-550, 550-700, and 450-700 nm, respectively. The Angstrom exponent is an indicator of the size of the particles that contribute most to light scattering. Dust and sea salt aerosols, which are dominated by supermicrometer particles, generally have Angstrom exponents close to zero. Pollution aerosols, where most of the mass is in the 0.1-1.0-(mu)m diameter range, usually have Angstrom exponents of 2-3. Increases in the Angstrom exponent are thus indicative of a shift to smaller particles. As seen in Fig. 8 (top), the expovents in May 1998 (yeardays 121-152) were generally smaller than in the other two months. The low Angstrom exponent events also coincided with high scattering coefficients. This indicates that the aerosols responsible for the elevated optical properties were larger in size. To investigate this further, the exponents were also compared by plotting the blue Angstrom exponents (at 450-550 nm) from 1997 and 1998 (Fig. 8, bottom). As shown in Fig. 8 (bottom), the 1998 values during mid- through late May (approximately yeardays 130-150) were clearly smaller than the 1997 values, suggesting that the particles during the smoke event were indeed larger. The values at other wavelengths are not shown since they follow trends similar to the one shown in Fig. 8 (bottom). It is important to note here that the Angstrom exponents derived from in situ light-scattering measurements such as from the AOS may not be well correlated with exponents derived from column-integrated measurements such as those made by the Cimel sunphotometer (CSPHOT; see Fig. 12). This is because aerosols are dried to less than 40% relative humidity in the AOS system. Further, the AOS measurement represents just the surface point in a vertical column; thus, its data should be compared to column-integrated values only after taking into account other atmospheric variables (e.g., water vapor) in the vertical column.
Plots of the 5-day backward trajectories starting at the central facility during April-June 1998 are shown in Fig. 9. In April (Fig. 9, left), the trajectories encompassed much of the United States. Even though some of the trajectories extended toward the south, most of the trajectories arriving at the central facility did not come from the smoke region. However, in both May and June, the trajectories are dominated more by southerly flows (Fig. 9, center and right, respectively). These trajectories extended well into the Central America regions where the forest fires occurred. Based on these trajectories, the transport of the smoke plume from the fire source regions was most likely in May and June.
To identify transport pathways of the aerosols responsible for the high optical loading in 1998, selective plots were made of the 5-day air parcel backward trajectories corresponding to the light-scattering coefficients at 550 nm ((sigma)^sub sp^) greater than 100 Mm^sup -1^ (Fig. 10). (The dates on which this criterion was met are listed in Table 3.) Any trajectories hitting the ground were rejected and are not shown in Fig. 10. As can be seen in Fig. 10, the trajectories associated with high light scattering coefficients were consistently, and nearly exclusively, from the fire source regions and the western Gulf of Mexico. The smoke appeared to be transported along the east coast of Mexico from the source regions, through Texas, and into the CART site.
Table 3 shows the daily averages of the measured and derived aerosol properties of major pollution events ((sigma)^sub sp^_^sub 550^ >= 100 Mm^sup -1^). The days that had high scattering coefficients were mostly in mid- through late May 1998. The most significant events occurred during the second half of May 1998. Although there were persistent southerly flows in June 1998 (Fig. 9, right), only a few high optical loadings were observed by the AOS then (notably on 8, 16, and 18 June), likely because a majority of the forest fires were gradually extinguished in June. The hemispheric backscattering ratios were generally smaller for the haze events than for the April-June means (0.202 for 1997 and 0.225 for 1998).
c. Remote sensing with lidar
1) RAMAN LIDAR
The CART Raman lidar (CARL; Goldsmith et al. 1998), located at the SGP CART central facility, is an active, ground-based, eye-safe laser remote sensing instrument. Lidar (light detection and ranging) is the optical analog of radar, using pulses of laser radiation to probe the atmosphere. A description of the Raman lidar technique can be found in Whiteman et al.(1992).
The CARL system is a custom instrument that was developed for the ARM program by Sandia National Laboratories. It is fully computer automated and can run unattended for many days following a brief (~5 min) startup period. The system transmits about 400-mJ pulses of 355-nm light at 30 Hz. It collects the backscattered return at three wavelengths: the combined Raleigh-Mie return at the laser's wavelength plus the Raman shifted returns for nitrogen and water vapor molecules at 387 and 408 nm, respectively. The Rayleigh-Mie return is separated such that the polarizations parallel and perpendicular to the polarization of the laser beam can be recorded. The backscattered signals are recorded with 1-min temporal and 39-- rn vertical resolution, although averaging during postprocessing is employed to improve the signal-tonoise ratio.
Primary quantities obtained from these backscatter signals are range-resolved vertical profiles of water vapor mixing ratio (g kg^sup -1^), aerosol scattering ratio (unitless), and backscatter depolarization ratio (%). Aerosol backscattering and extinction profiles are derived using the techniques described by Ferrare et al. (1998a,b). Additional cloud- and aerosol-related measurements can also be derived from the backscatter signals (Ferrare et al. 1999).
CARL observed high values of aerosol extinction during 13-14 May and again on 18-20 May when smoke was present over the SGP site. Vertical distributions of water vapor mixing ratio, relative humidity, aerosol extinction coefficient, linear depolarization ratio, and cloud mask over the SGP site during 13-20 May were analyzed using CARL data (Fig. 11). Relative humidity profiles were computed using the CARL water vapor mixing ratio profiles and temperature profiles derived from a physical retrieval algorithm that uses data from the atmospheric emitted radiance interferometer (AERI; Feltz et al. 1998), which is collocated with CARL, and from GOES. W. Feltz (University of Wisconsin) provided these temperature profiles. The combination of GOES and AERI retrievals of temperature together with the water vapor and aerosol profiles retrieved from CARL provide a complete specifi cation of the atmospheric state in cloud-free atmospheres (Turner et al. 2000). By integrating the aerosol extinction profiles between the surface and 7000 m, 10-min estimates of aerosol optical thickness (AOT) were derived from the CARL data (Fig. 12). Similarly, estimates of precipitable water vapor can be obtained by integrating the CARL profiles of water vapor mixing ratio with altitude (not shown).
As can be seen from the aerosol extinction image (Fig. 11c), CARL observed large values of aerosol extinction in the lowest 2000 in on 13-14 May, followed by very low values during the latter part of 15 May into 16 May (white areas in this image are due to clouds; see cloud mask in Fig. lie). The rapid drop in aerosol extinction measured by CARL matches well the similar decrease in aerosol light scattering measured at the surface by AOS as previously discussed. Beginning on 17 May and continuing through 20 May, the CARL measurements revealed high aerosol extinction values extending through a larger (0-6000 m) region of the troposphere. The trajectory analyses described earlier support the transport of smoke throughout a large region of the troposphere at this time. The change in the vertical distribution of aerosols between 14 and 19 May may have a significant impact on the TOMS aerosol retrievals and is the subject of further investigations. The CARL estimates of AOT (Fig. 12, top) show the large variability of aerosol loading during this period and show that the peak values of AOT occurred on 14 and 19 May. They also compare favorably with AOTs derived from other ARM SGP CART measurement devices [see section 3d(2)].
The CARL data also indicated clouds (Fig. Ilie) were often located at the top of the aerosol layers. The locations of clouds during this period were derived from the CARL aerosol and depolarization data (Fig. 11d). On 14 May, the clouds below 2000 in observed by the lidar were located just above the region of very high aerosol extinction. In slight contrast to this, on 17-20 May, there were fewer boundary layer clouds and the aerosols were spread over a much larger altitude region. The CARL data also show the presence of midand high-level clouds during much of the period. The surface AOS data show good correlations between aerosol scattering and concentrations measured at the surface and the CARL aerosol extinction measurements.
The CARL water vapor data show the presence of relatively moist conditions in the lowest 2000 m until midday on 15 May. Increasing moisture on 17 May, which continued through 20 May, followed the rapid decrease in moisture on 16 May after the cold frontal passage. As in the case of aerosol extinction, the relatively high water vapor amounts were spread over a large altitude region. The water vapor mixing ratio and relative humidity (Figs. 11a and 11b, respectively), as well as the AOT (Fig. 12, top) and precipitable water vapor (not shown), showed the high correlation between water vapor and aerosol extinction that existed throughout this period. This correlation is due both to the airmass properties as well as the hygroscopic nature of the aerosols. The stippled areas in the water vapor mixing ratio and relative humidity images in Fig. 11 are due to reduced CARL water vapor measurement performance during daytime because of background skylight.
The CARL data also indicate that the aerosol optical and, consequently, physical properties varied with altitude during this latter period. Figure 13 shows profiles of aerosol extinction and the aerosol extinction-- backscatter ratio for the period between 2215 UTC 17 May and 0340 UTC 18 May. The aerosol profile shows that the largest aerosol extinction values during this period were associated with the boundary layer (BL) aerosols within the lowest 2000 in. Figure 13 also shows that during this time the smoke aerosols were also beginning to reappear over the site as shown by the extended layer of aerosol extinction above 3000 in. The difference in the aerosol optical properties between the BL and the smoke above is indicated by the change in the aerosol extinction-backscatter ratio profile. The aerosol extinction-backscatter ratio (S^sub alpha^ ) is equal to [omega^sub O^ P^sub alpha^ (180 deg )]^sup -1^ where (omega^sub O^, is the aerosol single-scattering albedo and P^sub alpha^ (180 deg ) is the aerosol single-scattering phase function at 180 deg . The value of S^sub alpha^ depends on the size distribution, composition (i.e., refractive index), and shape of the aerosols; S^sub alpha^ also varies with wavelength. Assuming the aerosols are homogeneous and spherical, a Mie code was used to estimate S^sub alpha^ for two different aerosol types: a water-soluble aerosol and dust type and a biomass-burning aerosol type. The aerosol physical characteristics of the water-soluble and dust model were taken from Tanre et al. (1999) and the biomass-burning aerosols characteristics were taken from Remer et al. (1998). Dubovik et al. (2000) also provide a summary of these aerosol types. The biomass-burning aerosols typically have a higher concentration of particles in the accumulation mode (< 1 (mu)m) and are more absorbing than the water-soluble aerosols. At the CARL wavelength of 355 nm, S^sub alpha^ increased from 55 sr for the watersoluble aerosols to about 90 sr for the biomass-burning aerosols. (The corresponding values at 532 nm were 43 and 69 sr, respectively.) The estimates of S^sub alpha^ for water-soluble aerosols are also consistent with the "continental" aerosol model described by Ackermann (1998). These estimates of S^sub alpha^, along with other CARL S^sub alpha^ measurements (Ferrare et al. 1998c), suggest that the aerosols observed within the BL during this period were more characteristic of the aerosols typically seen at the SGP site. Based on the S^sub alpha^ values above, the aerosols observed above 3000 rn are consistent with smoke from biomass burning. This case shows how S^sub alpha^ profiles derived from CARL measurements may be used to possibly identify aerosol types. It should be noted also that, based on the CARL S^sub alpha^ profile during this period, an average value of Sa (at 355 nm) over the lowest 6000 m is 76 sr. If the S^sub alpha^ values within the profile are weighted by the corresponding value of the aerosol extinction at each altitude, this average value then decreases to 59 sr. Therefore, space- or groundbased methods that integrate over the entire aerosol column may not accurately represent the true aerosol characteristics during periods when aerosol properties vary with altitude. The CARL profiles of aerosol extinction and S^sub alpha^ are being used to try to assess how often such cases occur.
2) MICROPULSE LIDAR AND UNIVERSITY OF UTAH POLARIZATION DIVERSITY LIDAR
The ARM micropulse lidar (MPL) is an autonomous, eye-safe lidar system (523 nm) operated continuously at the CART central facility. Eye safety is achieved by expanding relatively low pulse energies (microjoules, whereas standard lidar systems are routinely orders of magnitude higher) at high pulse repetition frequencies (2500 Hz) through a 0.2-m shared transmitter-receiver telescope (Spinhirne et al. 1995). This feature effectively eliminates the need for supervised operation. Though low powered, MPL is capable of detecting all significant tropospheric cloud and aerosol via a high quantum efficiency photon-counting detector, appreciable pulse summation, and suitable geometric signal compression through the unique optical design. Additionally, a narrow receiver field of view (~100 (mu)rad) limits the effects of ambient solar background (i.e., noise) and minimizes complications from multiple scattering. Aside from basic measurements of cloud- and aerosol-layer boundary heights (to the limit of signal attenuation), MPL data can be processed to yield particle scattering and extinction cross sections, and optical thickness profiles, including those into the stratosphere in nighttime cases (Spinhirne 1993). The single-channel (523 nm) prototype MPL, developed at NASA Goddard Space Flight Center, was first installed at the CART central facility in December 1993. A modified version (unit number 02) replaced the original in April 1996 and was the operational system during the Central American smoke event. Its maximum range resolution was 300 m, and data were recorded to disk in 60-s pulse averages.
The University of Utah polarization diversity lidar system, an advanced cloud and aerosol research lidar developed under the ARM program (Sassen 1994), is a mobile platform that has participated in four SGP CART field campaigns. It is a dual-wavelength (1.06 and 0.532 (mu)m), four-channel system, with separate polarization detectors and telescopes for each color. With simultaneous approximate 0.35 J outputs at a pulse frequency of 10 Hz, and a maximum range resolution of 1.5 in, it is a high-resolution lidar. This device was deployed as part of the cloud physics IOP that was conducted from 27 April to 17 May 1998 at and near the SGP CART central facility. Although this non-eye-safe lidar is fully scanable, the system remained pointing in the zenith direction for the measurements taken in spring 1998.
Near-range corrected raw MPL backscatter for 13-- 20 May (yeardays, 133-140) are displayed in the top image of Fig. 14. MPL aerosol inversion and calibration techniques, discussed by Hlavka et al. (1998), require cloud-free areas within and directly above the boundary layer top to avoid cloud-induced ambiguities. Hourly averaged aerosol extinction values are plotted for qualifying profiles in the lower image of Fig. 14. Corresponding cloud-base heights calculated using the Scott-Spinhirne (Sc-Sp) algorithm (Clothiaux et al. 1998) and extinction-to-backscatter ratios (523 nm) are displayed in Fig. 15 (bottom), with hourly averaged cloud-free AOTs from the MPL shown in Fig. 12 (top). Note that condensation atop the CART site transmitting window hatch inhibited MPL observations at intermittent points within the period. These are denoted as ground-level observations in the cloud-base record image of Fig. 15 (top).
While smoke is prevalent throughout most of the period, two notable events were detected by the MPL on 14-15 and 17-18 May (yeardays 134-135 and 137-- 138). These periods are both marked by moist boundary layer conditions enhanced by 25 kt (12.9 m s^sup -1^) southerly low-level wind flow below a 4 deg C temperature inversion at 850 and 820 kPa (1500 and 1800 in above mean sea level, MSL) respectively. Late afternoon balloon sonde profiles for 14 May (Fig. 16, left) and 17 May (profile not shown) from the central facility were plotted. MPL-derived aerosol extinction maximums are found at and directly below the inversion levels on both days. Five continuous hours of cloud are noted in the local hourly observation database (see Table 1) between 1200 and 1600 UTC on 14 May, although no approximate base height is noted, and the possibility of visual error is appreciable given the higher smoke-induced optical thicknesses. Low cloud was also indicated in the observations from the afternoon of 17 May. Intermittent periods of cloud are evident in the MPL returns from both days. Differentiating between cloud and dense haze in lidar signal is nontrivial, though critical in maintaining the integrity of derived aerosol properties. CARL reports cloud within the boundary layer for most of 14 and 17 May. Taking local observations into account, yet keeping in mind the desire to sample as much of the aerosol as possible, the thresholds of the Sc-Sp algorithm were relaxed such that only the greatest MPL returns were reported as low-level cloud. However, strict restraints were placed on the inversion algorithms to restrict cloud-contaminated profiles. Four such hour-long periods from both days were thus removed.
A detailed view of aerosol conditions as collected by the PDL system on the afternoon of 12 May is given in Fig. 17. Comparable to 14 and 17 May, a temperature inversion (profile not shown) was found by balloon-sonde measurement near 800 kPa (2000 m MSL). Horizontal visibility was sometimes reduced by a few kilometers during these observations. In the height versus time displays, range-corrected backscattering in the 1.06-(mu)m laser channel is shown with the strongest signals appearing white. The top image in Fig. 17 shows that aerosols were present essentially throughout the troposphere and that the boundary layer was particularly turbid at this time. A relatively aerosol-free layer is found near the inversion layer top at 2000 m, and also to a lesser extent between about 4000 and 5000 m MSL.
An expanded view of in the bottom image in Fig. 17 depicts the growth of the smoke aerosol into haze particles in the more humid boundary layer. These "near" clouds occur in connection with convective updrafts near the top of the boundary layer where relative humidity is highest. The 1-s in time and 6-m in height data resolution reveals the fine structure of the convective processes. The laser backscatter lidar depolarization ratios in the growing aerosols (not shown) were near zero, showing them to be spherical particles, but the cells were not optically dense enough to produce noticeable laser pulse attenuation effects. Peak optical thicknesses during similar conditions on 14 and 17 May, as detected by the MPL (see Fig. 12, top), are on the order of 1.0. In other words, the enhanced laser backscattering was produced by growing haze particles in incipient stratocumulus clouds. The lack of significant depolarization from the aerosol in the free troposphere indicates that these particles were also spherical in shape, or at least were too small to induce depolarization at the 1.06-(mu)m laser wavelength. Previous polarization lidar studies of smoke layers indicate that large irregular particles soon precipitate from the smoke cloud (Sassen 2000). While the aerosol on 14 May was confined predominantly below the 2000-m temperature inversion, significant elevated layers are seen beginning on 17 May. By the end of 19 May (Fig. 16, right), the inversion cap disappears leading to a near-adiabatic lapse rate below a new stable layer near 500 kPa (~6000 m). MPL optical thicknesses peaked during this afternoon above 1.0 with appreciable extinction values seen throughout the column (see Fig. 12, top).
Radiometric observations
1) SOLAR AND INFRARED RADIATION STATIONS
ARM'S solar and infrared radiation stations (SIRS) are composed of a number of broadband radiometers. These include uplooking unshaded and shaded pyranometers to measure downwelling solar total and diffuse irradiance, a normal incidence pyrheliometer to measure direct normal solar irradiance, a downlooking pyranometer to measure upwelling solar reflectance, and uplooking and downlooking pyrgeometers to measure downwelling and upwelling infrared radiation.
Dust-related events have been shown to affect the broadband solar radiation by decreasing the direct component of the insolation and increasing the diffuse component. The Asian dust storm of April 1998 was shown to have this effect in the northwestern United States (see http://solardat.uoregon.edu/html/ 98_china_dust_cloud.html). A similar approach was used here to identify the impact of the Central American smoke on the broadband radiation data over the SGP CART.
Clearness indices are generally defined as the ratio of the total solar radiation observed at the ground (Hg) and a clear sky total radiation at that location (Ho; for example, extraterrestrial radiation). The clearness index (K) is simply Hg/Ho (Black et al. 1954). Several variants of the clearness index were computed here to better discriminate between the effects of aerosols and water vapor on the total solar radiation. These include the following: Ho, extraterrestrial estimate; H1, extraterrestrial estimate with an airmass correction; H2, clear sky model with no water vapor; H3, clear sky model with water vapor based on the ARM microwave radiometer (Liljegren 1994); and H4, clear sky model with water vapor based on surface dewpoint. The clear sky model was based on Meyers and Dale (1983). It includes simple treatment of aerosol and water vapor effects on the observed total solar radiation. The clearness index was then calculated using the observed total solar radiation and each of the five estimates above. The indices were only calculated under conditions that were determined to be cloud free by observation of the radiation data (and use of satellite data for some of the data from May 1998). All estimates were based on values near solar noon. The clear sky indices were calculated (when possible) during May 1997 for reference and for May 1998 for the SIRS radiometers at the central facility.
In addition, to assess the change in partitioning of the solar radiation into direct and diffuse components, ratios of the diff-use solar radiation to that of the total solar radiation were calculated using data from the SIRS shaded and unshaded pyranometers. It has been shown that independent measurement of the direct (normal incidence pyrheliometer) and diffuse (shaded pyranometer) components and their subsequent summation in which the direct normal irradiance is multiplied by the cosine of the solar zenith angle can produce a better measurement of total horizontal irradiance than can an unshaded pyranometer (Michalsky et al. 1999).
Figure 18 depicts the variation of the clearness indices during May 1997 and May 1998 at extended facility 13, which is part of the central facility. In general, the clear sky indices showed less attenuation of solar radiation in May 1998 than in May 1997 for comparable periods, which was not anticipated. But, a noticeable dip in the indices is noted on 13 May, followed by a strong jump in the indices by 15 May. Based on corresponding satellite imagery and the trajectory analysis, the fire-related aerosols had advected over the central facility by 13 May via southerly surface winds, but were quickly pushed out of the region early on 15 May by the storm system that introduced a drier, cleaner air mass from the northwest. The fluctuations in the clearness indices are viewed to be largely the result of changes in aerosol content in the atmosphere since each of the indices show the same variations whether or not they included some type of treatment of water vapor effects. Lower index values were also noted later in May 1998, but the observations are sparse enough and fairly late in the haze event that a clear assessment of this later period is not available.
Ratios of diffuse solar radiation to that of total solar radiation using data from the SIRS shaded and unshaded pyranometers at extended facility 13 (not shown) verified the clearness index analysis above. The increase observed in the diffuse component on 13 May from 1997 to 1998, which was by a factor of 2.5, is of the same magnitude observed in the northwestern United States during the Asian dust storm of April 1998. Figure 19 shows the spatial variation of the average diffuse-total ratio over a 10-min period at 1830 UTC (near solar noon) across the CART site on 13 May. Slightly different data times were used for some of the sites since they experienced cloudiness at 1830 UTC (extended facilities 7, 10, 20, 22, and 24). This analysis reveals a northwest to southeast gradient in the ratios, likely correlated with the fire-related aerosol loading. There also appears to be some mesoscale variability in the pattern.
2) MULTIFILTER ROTATING SHADOWBAND
RADIOMETER AND CIMEL SUNPHOTOMETER
The multifilter rotating shadowband radiometer (MFRSR; Harrison et al. 1994) makes spectral measurements of direct normal, diffuse horizontal, and total horizontal solar irradiances. These measurements are made at nominal wavelengths (main trace species measured) of 415 (aerosol), 500 (aerosol and ozone), 610 (aerosol and ozone), 665 (aerosol and ozone), 862 (aerosol), and 940 (water vapor) nm. The measurements are made at a sampling interval of 20 s. From such measurements, one may infer the atmosphere's optical thickness at the wavelengths mentioned above. In turn, these optical thicknesses may be used to derive information about the column abundances of ozone and water vapor (Michalsky et al. 1995), as well as aerosols and other atmospheric constituents (Harrison and Michalsky 1994). The MFRSR employs a silicon detector to measure broadband diffuse and total horizontal surface irradiance, from which the direct normal irradiance is obtained, enabling computation of total transmittance. Application of the Beer-Lambert-Bouguer law allows computation of the total optical thickness from which AOT is obtained by subtraction of optical thickness due to Rayleigh scattering and ozone absorption. A constant value of 300 Dobson units is used for the ozone column abundance.
CSPHOT is a multichannel automatic sun- and sky-scanning narrow field-of-view radiometer/ sunphotometer that measures the direct solar irradiance and sky radiance at the earth's surface (Halthore et al. 1999; Holben et al. 1998). Measurements are made during daylight hours at predetermined discrete wavelengths in the ultraviolet, visible, and near-infrared (at 340, 380, 440, 550, 670, 870, 940, and 1030 nm). As in the case of the MFRSR, direct normal solar irradi ance is analyzed to yield total optical thickness and hence AOT; the accuracy in AOT measurements is +/-0.01 at one air mass. In the 940-ran water vapor absorbing channel, column water vapor or precipitable water vapor (PWV, cm) is measured to a +/-10% accuracy. From the sky radiance measurements along the solar almucantar (solar zenith angle, variable azimuth angles), column-averaged aerosol size distribution, phase function, refractive index, and single scattering albedo are all obtained (Nakajima et al. 1996; Dubovik et al. 2000).
For the period of interest here (13-20 May 1998), AOTs inferred from MFRSR, CSPHOT, CARL, and MPL measurements are plotted in Fig. 12 (top). Data are instantaneous values of AOT for CSPHOT, 10-- min averages for the CARL, 30-min averages for the MFRSR data, and hourly averages for the MPL; the averages are performed for cloud-free periods ascertained by inspection of the temporal variability in the AOT and the Angstrom exponent. The MFRSR AOT data show an overall gradual increase in AOT until 13 May, followed by an abrupt increase on 14 May. AOT decreased to near background levels as a result of the frontal passage on 15 May. Following frontal passage, AOTs were abnormally high for about a week, with a peak on 19 May. Values of AOT returned to normal conditions by 1 June. Values of CSPHOT AOT similarly showed peaks on 14, 19, 21, and 27 May. Also noticeable is the sharp drop experienced on 15-16 May after the frontal passage. The AOTs from these two instruments trend nicely with those derived from CARL and MPL. The high values found in all four sets of data are unusually high and are not typical of northern Oklahoma. The low values recorded are probably representative of background conditions.
An Angstrom exponent (beta in t ~ alphalambda^sup -beta^ where t is the AOT, l is the wavelength, and alpha and beta are constants), obtained from CSPHOT measurements of AOT at 440 and 870 nm and from MFRSR measurements of AOT at 415 and 870 nm, stayed near or above 1.0 for much of the period of interest (Fig. 12, bottom). Because the values are below what one would expect for smoke particles, it appears that there was likely some mixing and coagulation with background aerosols along the path from Central America to the CART site, producing a more aged aerosol species with an exponent near 1. Some MFRSR values near the peak of the smoke event may be contaminated by clouds, as it is difficult to ascertain cloud-free periods at times of high temporal variability in AOT. Temporal variability seen in the PWV appears to follow changes in the presence of two distinct types of air masses: a humid air mass (PWV ~ 2 to 3 cm) associated with the southerly flow carrying smoke particles and a dry air mass (PWV ~ 1 cm) associated with a continental, clean air mass. The contrast between these two air masses is evident on 15 and 16 May.
During the period 13-22 May 1998, 15 observations of sky radiance were successfully inverted to obtain volume size distributions. Eight retrievals during the 13-22 May period were obtained during the smoke event as identified by AOT values at 670 nm (tau^sub alpha^ ^sup 670^) of greater than 0.20. Seven retrievals were obtained during the interim 15-16 May period when background conditions prevailed (after the frontal passage) having tau^sub alpha^ ^sup 670^ less than 0.05. Figure 20 shows seven representative samples of retrieved column-averaged aerosol volume size distribution [in cm^sup 3^ (cm^sup -2^)]. The smoky conditions are characterized by increases in the volume of particles in both the accumulation (fine) mode (radius < 1 (mu)m) and the coarse mode (radius > 1(mu)m) with the increase in the fine mode being larger. Additionally, the fine-mode smoke particles are larger in size than their background counterparts. The volume mean radius (r^sub v^) of the fine mode for smoky conditions is observed to be 0.19 (mu)m +/- 0.01 (mu)m, while for background conditions it is 0.12 (mu)m +/- 0.01 (mu)m. The coarse mode is more variable, with r^sub v^ 3.60 (mu)m +/0.60 (mu)m in smoky conditions and r^sub v^ 2.91 (mu)m +/- 0.48 (mu)m for background conditions.
e. Aircraft in situ aerosol sampling
The University of North Dakota owns and operates a Cessna Citation II aircraft (N77ND) for the purpose of atmospheric research. This aircraft type has a number of design and performance characteristics that make it an ideal platform for a wide range of atmospheric studies, including certification for flight into known icing conditions.
The Citation made overflights of the SGP CART site for the ARM program as part of the cloud physics IOP conducted in April-May 1998. In addition to carrying a wide array of cloud physics instrumentation, the Citation was also equipped with a condensation nuclei (CN) counter. The counter is a TSI 3760, which counts essentially all particles from about 0.01 to 3.0 (mu)m. Flights were made on 29 April and also on 1, 2, 8, and 14 May.
Two summary images have been produced showing data collected by the Citation's CN counter and are shown in Fig. 21. The top image of Fig. 21 shows the vertical profiles of CN concentrations in the lowest 1500 m for all IOP flights, while the bottom image shows profiles through the full altitude range for three IOP flights (two flights on 8 May and another on 14 May). These were obtained during descent back into the Ponca City, Oklahoma, airport, which is located about 40 km northeast of the CART central facility. In the top image of Fig. 21, the aerosols on 14 May saturated the CN counter below 1200 m, which was likely the top of the boundary layer. Occasional saturation can also be seen from the data of several other flights. This condition seemed to cause the CN readings to spike downward toward relatively low values. These spikes were removed from the curve for 14 May to clean up the graph, but were left in for the others. While it is unfortunate that the CN counter did saturate, these data still show how the boundary layer filled with what was apparently the smoke aerosol. They also show the magnitude and structure of the CN distribution for the other flights. Of interest in the bottom image of Fig. 21 is the elevated CN values in the 3000-6000-m layer, which increased dramatically from the start of the first flight on 8 May (the lefthand blue curve), or around 1500 UTC, to the time of the second flight, at about 2000-2300 UTC. Forward trajectories (not shown) calculated by HYSPLIT4 for air parcels starting early on 7 May west of the Yucatan Peninsula crossed into Oklahoma at the 850- and 750-kPa levels on 9-10 May.
f. Ozone monitoring
The ODEQ Air Quality Division, performed a detailed analysis of conditions in central Oklahoma on 11 May 1998. The division was aware of the fires buming in Central America and Mexico, but forecast information for 11 May indicated that the smoke plume would be well south of Oklahoma that day. Thus, no impact was anticipated. Those expectations were correct for most pollutants. However, an increase in haze was perceptible that day in central Oklahoma. Later, it would be determined that the division's particulate monitors recorded higher than normal values, but those values were not elevated enough to cause concern. Satellite imagery on 11 May (not shown) had denoted a barely visible upside-down V-shaped plume over Oklahoma. The abnormal amount of haze observed, and the satellite image, led the division to believe that a shallow smoke-laden air mass moved north from Texas into southern and central Oklahoma on 11 May.
The division did detect high levels of ozone from Oklahoma City area monitors on 11 May. Ozone concentrations, created by reactions between sunlight and aerosol particulates and pollutants, are highly dependent on temperature and wind speed. Thus, given that wind speeds were high and the temperatures were relatively low on 11 May (the opposite of what is normally expected for producing high ozone concentrations), it was unusual for the ozone concentrations observed that day to be as high as they were. Table 4 lists, for each ozone-monitoring site in the Oklahoma City area, the 10 days on which were recorded the highest 8-h averages of ozone in 1998. Average wind speeds for the 8 h, which coincide temporally with the ozone readings and daily high temperature, are also listed. The values for each parameter were averaged for each of the 10 days. On 11 May, the average ozone concentration over four metropolitan area measurement sites (Oklahoma City, Edmond, Moore, and Goldsby) was 0.09 ppm. According to Environmental Protection Agency (EPA) standards toughened in 1997, if air quality readings on a day show ozone concentrations in excess of 0.08 ppm, the day is considered in violation of the EPA standard. It is apparent from the tabled values that wind and temperature conditions accompanying the elevated ozone levels on 11 May (higher wind speeds, lower temperatures) were quite different from those of other days shown (lower wind speeds, higher temperatures). This could be indicative of an ozone intrusion from another area instead of a locally produced effect. The division presented this data to the EPA in an effort to illustrate that ozone conditions in the Oklahoma City area on 11 May constituted an unusual event and should not be included in data for determining attainment status.
4. Summary
Intrusions of the spring 1998 Central American smoke pall into Oklahoma and Kansas were detected by specialized instrumentation deployed for long-term data collection in support of the ARM program. In particular, Raman lidar analyses demonstrated a new capability for retrieving aerosol extinction profiles. These profiles revealed large variations in the vertical distribution of the smoke in May, as the smoke tended to be confined to the lowest 2000 in earlier in the month but extended up to 6000 in after southerly flow was reestablished following a strong cold frontal passage on 15 May. The ARM micropulse lidar and the University of Utah polarization diversity lidar also observed this vertical structure. Aerosol extinction-- backscatter ratios derived from the Raman lidar were able to delineate between aerosols typically observed at the ARM SGP CART and those of smoke associated with biomass burning. Backscattering to total scattering coefficient ratios in May from ARM's aerosol observing system were also found to be typical of biomass burning. The aerosols detected during the smoke event had relatively large total and hemispheric backscattering coefficients compared to aerosols detected during a comparable period in spring 1997. Based on Angstrom exponents derived from data collected by the aerosol observing system, Cimel sunphotometer, and a multifilter rotating shadowband radiometer, the aerosols associated with the smoke event, especially those in May, tended to have larger sizes compared to aerosols measured at other times during the spring. Further research on the character and distribution of the aerosols observed in 1998 is required.
The ARM program represents a long-term initiative of the Department of Energy to improve our understanding of the processes and properties that affect atmospheric radiation, with a particular focus on the influence of clouds and the role of cloud radiative feedback. The program has created a set of observational facilities and associated datasets that are now beginning to allow researchers to focus their efforts on improving parameterizations of clouds and radiation for general circulation models. The instrumentation array has also proven quite useful for documenting special events, such as the Central American smoke pall, that are part of the climatological data stream being collected.
Acknowledgments. The authors gratefully acknowledge the comments and suggestions of the anonymous reviewers of this manuscript. We thank Leon Ashford of the Oklahoma Department of Environmental Quality, Air Quality Division, for providing ozone monitoring data and analyses. We also thank those individuals and organizations that saved and created documents and images on the WWW during the 1998 Central American fire event as it unfolded. In particular, we acknowledge the Center for Air Pollution Impact and Trend Analysis at Washington University in St. Louis, Scott Bachmeier of the Cooperative Institute for Meteorological Satellite Studies at the University of Wisconsin-Madison, George Stephens, head of the NOAA/ NESDIS Operational Significant Event Imagery team, Elaine Prins of the NOAA/NESDIS/ORA/Advanced Satellite Products Team, the NOAA/NWS Norman Forecast Office, the NOAA Air Resources Laboratory, and Dr. Jay Herman, principal investigator for TOMS aerosols studies at the NASA Goddard Space Flight Center. Forrest Hoffmann of the Environmental Sciences Division of Oak Ridge National Laboratory retrieved the FNLNH data from 8-mm tapes. Richard Coulter of Argonne National Laboratory and Richard Cederwall of Lawrence Livermore National Laboratory provided estimates of mixing heights for the SGP CART. Raymond McCord and Thomas Wainman of Oak Ridge National Laboratory provided helpful comments. The authors acknowledge the ARM program and the Environmental Sciences Division of DOE for making studies such as this possible. Work performed at the University of Oklahoma (RAP, CPB) and the University of Utah (MES) was supported by DOE Subcontract 354047-AQ5 from Pacific Northwest National Laboratory. Pacific Northwest National Laboratory (JCB, NSL, DDT) is operated for DOE by Battelle Memorial Institute under Contract DE-AC06-76RLO 1830. Oak Ridge National Laboratory (MDC, CJL) is managed by Lockheed Martin Energy Research Corporation under Contract DE-ACOS-OR22464. Che-Jen Lin was also supported in part by an appointment to the Oak Ridge National Laboratory Postdoctoral Research Associate program administered jointly by the laboratory and the Oak Ridge Institute for Science and Education. Work at Brookhaven National Laboratory (RNH) was supported by DOE Contract DE-AC0298CH 10886. Raman lidar work (RAF, LAH, DDT) was performed as part of the DOE ARM program (DE-AI02-98ER62638) and the NASA EOS Validation program. Micropulse lidar work at NASA GSFC (JRC, DLH, JDS) was supported by DOE Grant DE-AIO I92ER61367, as part of the ARM program. University of Utah lidar participation (KS) was supported through DOE Grant DEF-039ER61747, as part of the ARM program. Aerosol observing system work at NOAA CMDL (JAO) is supported by DOE ITF Agreement 354492. North Dakota Citation flights and subsequent research (MRP) were made possible by DOE Contract DE-FG03-97ER62360.
[Reference]
References
[Reference]
Ackermann, J., 1998: The extinction-to-backscatter ratio of tropospheric aerosol: A numerical study. J. Atmos. Oceanic Technol., 15, 1043-1050.
Andreae, M. O., 1991: Biomass burning: Its history, use and distribution and its impact on environmental quality and global climate. Global Biomass Burning, J. S. Levine, Ed., MIT Press, 3-21.
Bahrmann, C. P., and V. K. Saxena, 1998: Influence of air mass history on black carbon concentration and regional climate forcing in southeastern United States. J. Geophys. Res., 103, 23 153-23 161.
Bergin, M. H., J. A. Ogren, S. E. Schwartz, and L. M. McInnes, 1997: Evaporation of ammonium nitrate aerosol in a heated nephelometer: Implications for field measurements. Environ. Sci Technol., 31, 2878-2883.
Black, J. N., C. W. Bonython, and J. A. Prescott, 1954: Solar radiation and the duration of sunshine. Quart. J. Roy. Meteor. Soc., 80, 231-235.
[Reference]
Bodhaine, B. A., J. M. Harris, J. A. Ogren, and D. J. Hofmann, 1992: Aerosol optical properties at Mauna Loa Observatory: Long-range transport from Kuwait? Geophys. Res. Lett., 19, 581-584.
Charlson, R. J., D. S. Covert, T. V. Larson, and A. P. Waggoner, 1978: Chemical properties of tropospheric sulfur aerosols. Atmos. Environ., 12, 39-53.
Clothiaux, E. E., G. G. Mace, T. P. Ackerman, T. J. Kane, J. D. Spinhirne, and V. S. Scott, 1998: An automated algorithm for detection of hydrometeor returns in micropulse lidar data. J. Atmos. Oceanic Technol., 15, 1035-1042.
Draxler, R. R., and G. D. Hess, 1998: An overview of the HYSPLIT-4 modelling system for trajectories, dispersion and deposition. Aust. Meteor. Mag., 47, 295-308.
Dubovik, O., A. Smirnov, B. N. Holben, M. D. King, Y. J. Kaufman, T. F. Eck, and I. Slutsker, 2000: Accuracy assessments of aerosol optical properties retrieved from Aerosol Robotic Network (AERONET) sun and sky radiance measurements. J. Geophys. Res., 105, 9791-9806.
Feltz, W. F., W. L. Smith, R. 0. Knuteson, H. E. Revercomb, H. M. Woolf, and H. B. Howell, 1998: Meteorological applications
[Reference]
of temperature and water vapor retrievals from the groundbased atmospheric emitted radiance interferometer (AERI). J. Appl. Meteor., 37, 857-875.
Ferrare, R. A., S. H. Melfi, D. N. Whiteman, K. D. Evans, and R. Leifer, 1998a: Raman lidar measurements of aerosol extinction and backscattering: 1. Methods and comparisons. J. Geophys. Res., 103, 19 663-19 672.
M. Poellot, and Y. J. Kaufman, 1998b: Raman lidar measurements of aerosol extinction and backscattering: 2. Derivation of aerosol real refractive index, single-- scattering albedo, and humidification factor using Raman lidar and aircraft size distribution measurements. J. Geophys. Res., 103, 19 673-19 689.
-, D. D. Turner, S. H. Melfi, D. N. Whiteman, G. Schwemmer, K. D. Evans, J. E. M. Goldsmith, and T. Tooman, 1998c: CART and GSFC Raman lidar measurements of atmospheric aerosol backscattering and extinction profiles for EOS validation and ARM radiation studies. Proc. 19th Int. Laser Radar Conf., Annapolis, MD, NASA/CP-1998-207671, 207-210.
L. A. Heilman, and W. F. Feltz, 1999: CART Raman lidar retrievals of aerosol extinction and relative humidity profiles. Proc. Ninth Atmospheric Radiation Measurement (ARM) Science Team Meeting, San Antonio, TX, U.S. Dept. of Energy. [Available online at http://www.arm.gov/docs/documents/technical/conf 9903/ferrare-99.pdf.1
[Reference]
Fraser, R. S., and Y. J. Kaufman, 1985: The relative importance of aerosol scattering and absorption in remote-sensing. IEEE Trans. Geosci. Remote Sens., 23, 625-633.
Goldsmith, J. E. M., F. H. Blair, S. E. Bisson, and D. D. Turner, 1998: Turn-key Raman lidar for profiling atmospheric water vapor, clouds, and aerosols. AppL Opt., 37, 4979-4990.
Halthore, R. N., S. E. Schwartz, Y. Liu, P. H. Daum, B. N. Holben, and J. J. Michalsky, 1999: Sun and sky radiometric measurements at the CART ARM SGP site. Proc. Ninth Atmospheric Radiation Measurement (ARM) Science Team Meeting, San Antonio, TX, U.S. Dept. of Energy. [Available online at http:/ /www.arm.gov/docs/documents/technical/conf_9903/ halthore(3)-99.pdf.]
Harris, J. M., 1992: An analysis of 5-day midtropospheric flow patterns for the South Pole: 1985-1989. Tellus, 44B, 409-421. and J. D. W. Kahl, 1994: Analysis of 10-day isentropic flow
patterns for Barrow, Alaska: 1985-1992. J. Geophys. Res., 99, 25 845-25 855.
Harrison, L., and J. Michalsky, 1994: Objective algorithms for the retrieval of optical depths from ground-based measurements. Appl. Opt., 33, 5126-5132.
[Reference]
and J. Berndt, 1994: Automated multifilter rotating shadow-band radiometer: An instrument for optical depth and radiation measurements. Appl. Opt., 33, 5118-5125.
Herman, J. R., P. K. Bhartia, 0. Torres, C. Hsu, C. Seftor, and E. Celarier, 1997: Global distribution of UV-absorbing aerosols from Nimbus 7/TOMS data. J. Geophys. Res., 102, 16 911-- 16 922.
Hlavka, D. L., J. D. Spinhirne, J. R. Campbell, J. A. Reagan, and D. M. Powell, 1998: Aerosol analysis techniques and results from Micro Pulse Lidar. Proc. 19th Int. Laser Radar Conf., Annapolis, MD, NASA/CP-1998-207671, 155-158.
Hobbs, P. V., J. S. Reid, R. A. Kotchenruther, R. J. Ferek, and R. Weiss, 1997: Direct radiative forcing by smoke from biomass burning. Science, 275, 1776-1778.
[Reference]
Holben, B. N., and Coauthors, 1998: AERONET-A federated instrument network and data archive for aerosol characterization. Remote Sens. Environ., 66, 1-16.
Iacobellis, S. F., R. Frouin, and R. C. J. Somerville, 1999: Direct climate forcing by biomass-burning aerosols: Impact of correlations between controlling variables. J. Geophys. Res., 104, 12 031-12 045.
Kotchenruther, R. A., and P. V. Hobbs, 1998: Humidification factors of aerosols from biomass burning in Brazil. J. Geophys. Res., 103, 32 081-32 089.
Liljegren, J. C., 1994: Two-channel microwave radiometer for observations of total column precipitable water vapor and cloud liquid water path. Preprints, Fifth Symp. on Global Change Studies, Nashville, TN, Amer. Meteor. Soc., 262-- 269.
Martins, J. V., P. Artaxo, C. Liousse, J. S. Reid, P. V. Hobbs, and Y. J. Kaufman, 1998: Effects of black carbon content, particle size, and mixing on light absorption by aerosols from biomass burning in Brazil. J. Geophys. Res., 103, 32 041-- 32 050.
[Reference]
Meyers, T. P., and R. F. Dale, 1983: Predicting daily insolation with hourly cloud height and coverage. J. Climate Appl. Meteor., 22, 537-545.
Michalsky, J. J., J. C. Liljegren, and L. C. Harrison, 1995: A comparison of sun photometer derivations of total column water vapor and ozone to standard measures of same at the Southern Great Plains Atmospheric Radiation Measurement site. J. Geophys. Res., 100, 25 995-26 003.
E. Dutton, D. Nelson, M. Rubes, T. Stoffel, M. Wesely, M. Splitt, and J. DeLuisi, 1999: Optimal measurement of surface shortwave irradiance using current instrumentation. J. Atmos. Oceanic Technol., 16, 55-69.
Moody, J. L., and J. N. Galloway, 1988: Quantifying the relationship between atmospheric transport and the chemical composition of precipitation on Bermuda. Tellus, 40B, 463-- 479.
Nakajima, T., T. Glauco, R. Rao, P. Boi, Y. Kaufman, and B. Holben, 1996: Use of sky brightness measurements from ground for remote sensing of particulate polydispersions. Appl. Opt., 35, 2672-2686.
[Reference]
Penner, J. E., R. E. Dickinson, and C. A. O'Neil, 1992: Effects of aerosol from biomass burning on the global radiation budget. Science, 256, 1432-1433.
Polissar, A. V., P. K. Hopke, P. Paatero, Y. J. Kaufmann, D. K. Hall, B. A. Bodhaine, E. G. Dutton, and J. M. Harris, 1999: The aerosol at Barrow, Alaska: Long-term trends and source locations. Atmos. Environ., 33, 2441-2458.
Prins, E. M., J. M. Feltz, W. P. Menzel, and D. E. Ward, 1998: An overview of GOES-8 diurnal fire and smoke results for SCAR-B and 1995 fire season in South America. J. Geophys. Res., 103, 31 821-31 835.
Reid, J. S., and P. V. Hobbs, 1998: Physical and optical properties of young smoke from individual biomass fires in Brazil. J. Geophys. Res., 103, 32 013-32 030.
R. J. Ferek, D. R. Blake, J. V. Martins, M. R. Dunlap, and C. Liousse, 1998: Physical, chemical, and optical properties of regional hazes dominated by smoke in Brazil. J. Geophys. Res., 103, 32 059-32 080.
Remer, L. A., Y. J. Kaufman, B. N. Holben, A. M. Thompson, and D. McNamara, 1998: Biomass burning aerosol size dis
[Reference]
tribution and modeled optical properties. J. Geophys. Res., 103, 31 879-31 891.
Sassen, K., 1994: Advances in polarization diversity lidar for cloud remote sensing. Proc. IEEE, 82, 1907-1914.
2000: Lidar backscatter depolarization technique for cloud and aerosol research. Light Scattering by Nonspherical Particles: Theory, Measurements, and Geophysical Applications, M. L. Mischenko, J. W. Hovenier, and L. D. Travis, Eds., Academic Press, 393-416.
Seinfeld, J. H., and S. N. Pandis, 1998: Atmospheric Chemistry and Physics: From Air Pollution to Climate Change. Wiley, 1326 pp.
Spinhirne, J. D., 1993: Micro Pulse Lidar. IEEE Trans. Geosci. Remote Sens., 31, 48-55.
-, J. A. R. Rall, and V. S. Scott, 1995: Compact eye safe lidar system. Rev. Laser Eng., 23, 112-118.
Stohl, A., 1998: Computation, accuracy and applications of trajectories-A review and bibliography. Atmos. Environ., 32, 947-966.
[Reference]
and P. Seibert, 1998: Accuracy of trajectories as determined from the conservation of meteorological tracers. Quart. J. Roy. Meteor. Soc., 124, 1465-1484.
Stokes, G. M., and S. E. Schwartz, 1994: The Atmospheric Radiation Measurement (ARM) Program: Programmatic background and design of the Cloud and Radiation Testbed. Bull. Amer. Meteor. Soc., 75, 1201-1221.
Tanre, D., L. A. Remer, Y. J. Kaufman, S. Mattoo, P. V. Hobbs, J. M. Livingston, P. B. Russell, and A. Smirnov, 1999: Retrieval of aerosol optical thickness and size distribution over ocean from the MODIS airborne simulator during TARFOX. J. Geophys. Res., 104, 2261-2278.
Turner, D. D., W. F. Feltz, and R. A. Ferrare, 2000: Continuous water vapor profiles from operational ground-based active and passive remote sensors. Bull Amer. Meteor. Soc., 81, 1301-- 1317.
Whiteman, D. N., S. H. Melfi, and R. A. Ferrare, 1992: Raman lidar system for the measurement of water vapor and aerosols in the Earth's atmosphere. Appl. Opt., 31, 3068-3082.
[Author Affiliation]
aCooperative Institute for Mesoscale Meteorological Studies, University of Oklahoma, Norman, Oklahoma.
"Pacific Northwest National Laboratory, Richland, Washington. `Science Systems and Applications, Inc., Lanham, Maryland. dOak Ridge National Laboratory, Oak Ridge, Tennessee. eNASA Langley Research Center, Hampton, Virginia. Brookhaven National Laboratory, Upton, New York.
gScience Applications International Corp./NASA Langley Research Center, Hampton, Virginia.
"Oak Ridge Associated Universities, Oak Ridge, Tennessee. NOAA Climate Monitoring and Diagnostics Laboratory, Boulder, Colorado.
[Author Affiliation]
iUniversity of North Dakota, Grand Forks, North Dakota. kNASA Goddard Space Flight Center, Greenbelt, Maryland. 'University of Utah, Salt Lake City, Utah.
'Cooperative Institute for Regional Prediction, University of Utah, Salt Lake City, Utah.
Corresponding author address: Randy A. Peppler, Cooperative Institute for Mesoscale Meteorological Studies, 100 E. Boyd Street, Room 1110, Norman, OK 73019-1011.
E-mail: rpeppler@ou.edu In final form 18 May 2000.
2000 American Meteorological Society
