CMDL Aerosol Monitoring Program
J. Ogren (editor), S. Anthony, J. Barnes, M. Bergin, W. Huang, L. McInnes, C. Myers, S. Thaxton, J. Wendell
Aerosol particles affect the radiative balance of Earth both directly, by scattering and absorbing solar and terrestrial radiation, and indirectly, through their action as cloud condensation nuclei (CCN) with subsequent effects on the microphysical and optical properties of clouds. Evaluation of the climate forcing by aerosols, defined here as the perturbation of the Earth's radiation budget induced by the presence of airborne particles, requires knowledge of the spatial distribution of the particles, their optical and cloud-nucleating properties, and suitable models of radiative transfer and cloud physics. Obtaining a predictive relationship between the aerosol forcing and the physical and chemical sources of the particles additionally requires regional and global-scale chemical process, physical transformation and transport models for calculating the spatial distributions of the major chemical species that control the optical and cloud-nucleating properties of the particles. Developing and validating these various models requires a diverse suite of in-situ and remote observations of the aerosol particles on a wide range of spatial and temporal scales.
Aerosol measurements began at the CMDL baseline observatories in the mid-1970's as part of the Geophysical Monitoring for Climate Change (GMCC) program. The objective of these "baseline" measurements was
Since the inception of the program, scientific understanding of the behavior of atmospheric aerosols has improved considerably. One lesson learned is that residence times of tropospheric aerosols are generally less than one week, and that human activities primarily influence aerosols on regional/continental scales rather than global scales. In response to this increased understanding, and to more recent findings that anthropogenic aerosols create a significant perturbation in the Earth's radiative balance on regional scales (Charlson et al., 1992; NRC, 1996), CMDL expanded its aerosol research program to include regional aerosol monitoring stations. The goals of this regional-scale monitoring program are:
A primary hypothesis to be tested by NOAA's aerosol research program is that the climate forcing by anthropogenic sulfate will change in response to future changes in sulfur emissions. The forcing is expected to decrease in and downwind of the U.S. as a result of emission controls mandated by the Clean Air Act, while continued economic development in China and other developing countries is expected to lead to an increased forcing in and downwind of those areas. Testing this hypothesis will require a coordinated research program involving modeling, monitoring, process, and closure studies. This report describes the observations that CMDL is conducting towards this goal.
No single approach to observing the atmospheric aerosol can provide the necessary data for monitoring all the relevant dimensions and spatial/temporal scales needed to evaluate the climate forcing by anthropogenic aerosols. In situ observations from fixed surface sites (like CMDL's), ships, balloons, and aircraft can provide very detailed characterizations of the atmospheric aerosol, but on limited spatial scales. Remote sensing methods, from satellites, aircraft, or from the surface, can determine a limited set of aerosol properties from local to global spatial scales, but cannot provide the chemical information needed for linkage with global chemical models. Fixed ground stations are suitable for continuous observations over extended time periods, but lack vertical resolution. Aircraft and balloons can provide the vertical dimension, but not continuously. Only when systematically combined, can these various types of observations produce a data set where point measurements can be extrapolated with models to large geographical scales, where satellite measurements can be compared with the results of large-scale models, and where process studies have a context for drawing general conclusions from experiments conducted under specific conditions.
Measurements of atmospheric aerosols are used in three fundamentally different ways for aerosol/climate research: algorithm development for models and remote-sensing retrievals, parameter characterization, and model validation. Laboratory and field process studies guide the development of parameterization schemes and choice of parameter values for chemical transport models that describe the relationship between emissions and concentration fields of aerosol species. Systematic surveys and monitoring programs provide characteristic values of aerosol properties that are used in radiative transfer models for calculating the radiative effects of the aerosols, and for retrieving aerosol properties from satellites and other remote sensing platforms. And finally, monitoring programs provide spatial and temporal distributions of aerosol properties that are compared with model results to validate the models. Each of these three modes of interaction between applications and measurements require different types of data, and entail different measurement strategies. Ogren (1995) applied the thermodynamic concept of intensive and extensive properties of a system to emphasize the relationship between measurement approach and applications of aerosol observations.
Intensive properties do not depend on the amount of aerosol present,and are used as parameters in chemical transport and radiative transfer models (e.g., atmospheric residence time, single-scattering albedo). Extensive properties vary strongly in response to mixing and removal processes, and are most commonly used for model validation (e.g., mass concentration, optical depth). Intensive properties are more difficult and expensive to measure than extensive properties, as they generally are defined as the ratio of two extensive properties. As a result, different measurement strategies are needed for meeting the data needs of the various applications. Measurements of a few, carefully chosen extensive properties, of which aerosol optical depth and species mass concentrations are prime candidates, are needed in many locations to test the ability of the models to predict spatial and temporal variations on regional to global scales, and to detect changes in aerosol concentrations resulting from changes in aerosol sources. The higher cost of determining intensive properties suggests a strategy of using a limited number of highly-instrumented sites to characterize means and variabilities of intensive properties for different regions or aerosol types, supplemented with surveys by aircraft and ships to characterize the spatial variability of these parameters. CMDL's regional aerosol monitoring program is primarily focussed on characterizing intensive properties.
CMDL's measurements provide ground truth for satellite measurements and global models, as well as key aerosol parameters for global-scale models (e.g., scattering efficiency of sulfate particles, hemispheric backscattering fraction). An important aspect of this strategy is that the chemical measurements are linked to the physical measurements through simultaneous, size-selective sampling, which allows the observed aerosol properties to be connected to the atmospheric cycles of specific chemical species.
Extensive aerosol properties monitored by CMDL include the total particle number concentration (Ntot), aerosol optical depth (d), and components of the aerosol extinction coefficient at one or more wavelengths (total scattering ssp, backwards hemispheric scattering sbsp, and absorption sap). At the regional sites, size-resolved impactor and filter samples (submicrometer and supermicrometer size fractions) are obtained for gravimetric and chemical (ion chromatograph) analyses. All size-selective sampling, as well as the measurements of the components of the aerosol extinction coefficient at the regional stations, is performed at a low, controlled relative humidity (<40%) to eliminate confounding effects because of changes in ambient relative humidity. Data from the continuous sensors are screened to eliminate contamination from local pollution sources. At the regional stations, the screening algorithms use measured wind speed, direction, and total particle number concentration in real-time to prevent contamination of the chemical samples. Algorithms for the baseline stations use measured wind speed and direction to exclude data that are likely to have been locally contaminated.
Prior to 1995, data from the baseline stations were manually edited to remove spikes from local contamination. For the 1995 baseline data an automatic editing algorithm was applied, in addition to manual editing of local contamination spikes. For the baseline stations (Barrow, Mauna Loa, South Pole, and Samoa), as well as Sable Island, data are automatically removed when the wind direction is from local sources of pollution (such as generators and buildings) as well as when the wind speed is less than a threshold value (0.5-1 m s-1). In addition, Mauna Loa data for upslope conditions are excluded since the airmasses do not represent background free tropospheric air for this case. A summary of the data editing criteria is given in Table 1.
Integrating nephelometers are used to determine the light scattering coefficient of the aerosol. These instruments operate by illuminating a fixed sample volume from the side, and observing the amount of light that is scattered by particles and gas molecules in the direction of a photomultiplier tube. The instruments integrates over scattering angles of 7-170°. Depending on the station, the measurements are performed at one, three, or four wavelengths in the visible and near-infrared. Newer instruments allow determination of the hemispheric backscattering coefficient by using a shutter to prevent illumination of the portion of the instrument that yields scattering angles less than 90°. A particle filter is inserted periodically into the sample stream to measure the light scattered by gas molecules, which is subtracted from the total scattered signal to determine the contribution from the particles alone. The instruments are calibrated by filling the sample volume with a gas of scattering coefficient; CO2 is used for high sensitivity instruments, while CFC-12 is used for the few single-wavelength, lower sensitivity instruments still in use.
The aerosol light absorption coefficient is determined with a continuous light absorption photometer. This instrument continuously measures the amount of light transmitted through a quartz filter, while particles are being deposited on the filter. The rate of decrease of transmissivity, divided by the sample flowrate, is directly proportional to the light absorption coefficient of the particles. Newer instruments have been calibrated in terms of the difference of light extinction and scattering in a long-path extinction cell, for laboratory test aerosols. Instruments at the baseline stations ("aethalometers") have been calibrated by the manufacturer in terms of the equivalent amount of black carbon, from which the light absorption coefficient is calculated assuming a mass absorption efficiency of the calibration aerosols of 10m2g-1.
Particle number concentration is determined with a condensation nucleus (CN) counter, that exposes the particles to a high supersaturation of butanol vapor. This causes the particles to grow to a size where they can be detected optically and counted. The instruments in use have lower particle-size detection limits of 10-20 nm diameter.
Summaries of the extensive measurements obtained at each site are given in Table 2 and Table 3. Table 4 lists the intensive aerosol properties that can be determined from the directly-measured extensive properties. These properties are used in chemical transport models to determine the radiative effects of the aerosol concentrations calcuated by the models. Inversely, these properties are used by algorithms for interpreting satellite remote-sensing data to determine aerosol amounts based on measurements of the radiative effects of the aerosol.
Information and data from the aerosol group at CMDL are available on the Internet via FTP and World Wide Web servers. Recently processed data, file format specifications, documents summarizing data processing and flow, and clean processed data presented in hourly averaged files for all years of station operation are available via anonymous FTP to ftp://ftp.cmdl.noaa.gov , directory "aerosol". In addition to the above, the CMDL WWW server at http://www.cmdl.noaa.gov supplies online plots of recently processed aerosol data and hypertext links to various related documents (including this one).
The annual cycles of selected extensive and intensive properties are illustrated in Figures 1a-f.
The data are presented in the form of box-whisker plots that summarize the distribution of values: the box ranges from the lower to upper quartiles, with a central bar at the median value, while the whiskers extend to the 5th and 95th percentiles. The statistics are based on hourly averages of each parameter for each month of the year, as well as for the entire year ("ANN").
In general, changes in long-range transport patterns dominate the annual cycles. For BRW, the highest values of ssp are observed during the spring, becuase of the long-range transport of pollution from lower latitudes ("Arctic haze"). The BRW CN record shows a more variable semiannual cycle, with a maximum that usually coincides with the maximum in ssp and a secondary maximum in late summer or early fall. The secondary maximum in late summer is thought to be caused by local oceanic emissions of DMS gas that are eventually converted to sulfate aerosol [Radke et al., 1990]. For MLO, the highest ssp values occur in the springtime, caused by the the long-range transport of pollution and mineral dust from Asia. Little seasonality is seen in CN concentrations at MLO, however, indicating that the smallest particles ssp levels observed in the late winter are due to the long-range transport of sea salt in the upper troposphere from stormy regions near the Antarctic coast to the interior of the continent.
Previous reports describing the baseline aerosol data sets include:
The annual cycles for the regional stations are very uncertain, being based on only 2-3 years of measurements, and it is premature to discuss the causes of the observed variability. The proximity of the regional sites to North American pollution sources is apparent in the results, however, with monthly median values that are over two orders of magnitude higher compared to values from the baseline stations.
Long term trends in CN concentration, ssp, and Ångström exponent are plotted in Fig. 2a, Fig. 2b,and Fig. 2c for the baseline stations. The trends are plotted for the annual geometric average, as well as for the geometric averages for the months with the lowest and highest median values observed in the annual cycle plots. Interpretation of the results are complicated by two changes in instrumentation: replacement of the nephelometer at MLO in 1985, and replacement of the CN counters with butanol-based instruments in 1988 at MLO, 1989 at SPO, 1990 at BRW, and 1992 at SMO. The two types of CN counters have different lower size detection limits, which means that any change in the long-term record will depend on the presence of particles not detected by one of the counters. This is the likely cause for the fact that obvious step changes in CN concentration were observed at MLO and SPO, but no apparent change was seen at BRW and SMO.
As discussed in the 1988 Summary Report [Elkins and Rosson, 1989], ssp values at MLO were generally higher since the installation of the new nephelometer in 1985 and have not reached the low values previously observed in winter. The increasing trend in ssp at MLO is obviously caused by higher winter values in the latter part of the record and the reason is believed to be instrumental. A modern, high-sensitivity three-wavelength nephelometer was deployed at MLO in 1994, and future comparison of the results from the two nephelometers is expected to quantify any biases introduced by the older, less-sensitive instrument. All data reported here are from the older instrument, however.
Daily Mean Values of Aerosol Properties
Figures 3a-g show the daily mean values at each monitoring station for total number concentration (CN), aerosol scattering coefficient at 550 nm (ssp) and the Ångström exponent for the 550/700 nm wavelength pair.
The data are plotted for the time period beginning on January 1, 1994 and ending on December 31, 1995. Show significant day-to-day variability in CN concentration, aerosol scattering coefficient, and Ångström exponent is seen. The daily variability of these parameters is due to several factors including changes in local meteorology, aerosol sources, transport time from source regions, and processing of aerosols during transport. It is worthwhile to point out that the data editing procedure for 1995, as seen in the CN plots from the edited stations, results in a more stringent acceptance of data. This can be clearly seen in the Barrow CN plot which shows significantly more breaks in the 1995 data due to the rejection of a greater amount of data compared to the previous year. The more rigorous approach to data screening for 1995 and after generally results in less day-to-day variability in the CN concentrations, which is likely due to the fact that data resulting from local pollution are more efficiently excluded.
Figure 4 shows box/whisker plots of the variability in the daily averages of three different intensive properties measured at Sable Island: the Ångström exponent (550/700 nm wavelength pair) for submicrometer particles, the fraction of scattering caused by submicrometer particles, and the fraction of the light that is scattered into the backwards hemisphere. The data have been classified into three cases: "clean" conditions when both Ntot and ssp (550 nm, Dp <1 µm) are below the lower quartile for the entire data set, "dirty" conditions in which Ntot and ssp are above the upper quartile and all "other" periods that do not meet the previously defined criteria (for example periods with low Ntot and high ssp values). For comparison, the fine/total scattering fraction is plotted for Bondville (data for the other intensive properties are not availavle at BND prior to 1996). It can be seen that the values of the Ångström exponent increase for more polluted periods, suggesting that the submicrometer aerosol shifts systematically towards smaller particles as the degree of pollution increases. This is also reflected in size-segregated measurements of light scattering, which show a larger fraction of the total scatter is due to the submicrometer aerosol as the air becomes more polluted, reaching a median value of 84% at Bondville. In the cleanest cases at Sable Island, only 28% (median value) of the ligh scattering is caused by submicrometer particles; the remainder is presumably caused by larger sea salt particles. Submicrometer particles contribute a larger fraction of the total for each quartile at Bondville, suggesting the continental aerosol is always heavily influenced by fine aerosol pollution. Aerosol number concentrations and values of ssp for the submicrometer aerosol are consistently higher at Bondville than at Sable Island (Fig 1). The backscatter fraction, on the other hand, exhibits relatively little dependence on the degree of pollution.
Linear regression of the Ångström exponent with the fraction of submicrometer scattering (Figure 5) demonstrates that the two variables are highly correlated, suggesting that most of the variance in the Ångström exponent is controlled by the ralative abundance of submicrometer particles. This chanllenges the traditional notion that the Ångström exponent can be interpreted as the slope of a power-law aerosol size distribution, and better supports a bimodal model of the size distribution where the Ångström exponent is a measure of the relative amounts of material in the two modes.
A special version of the CMDL aerosol instrumentation package used at the regional aerosol monitoring sites was developed for use on research aircraft. This effort was undertaken to extend our measurement capability into the vertical dimension, and to greatly increase geographic coverage as well. This airborne aerosol package includes a three-wavelength nephelometer with backscatter shutter, a light absorption photometer, a condensation nucleus counter, and a multi-filter sampler, all interfaced to a laptop computer for instrument control and data logging. As is done on the ground, the sample air is heated as necessary to maintain a relative humidity below 40%, and a multijet impactors are used to restrict the size-range of particles sampled (on the aircraft, only particles smaller than 1 µm aerodynamic diameter are sampled). Additionally, wing-mounted probes permit the determination of aerosol size distributions. These instruments constitute a comprehensive airborne aerosol measurement platform capable of determining an extensive suite of aerosol chemical, optical and microphysical properties.
Measurements of the optical properties of submicrometer aerosol particles were measured from the NOAA WP-3D Orion research aircraft during the summer 1995 Southern Oxidants Study. The majority of the flights were in the midwest and southeastern United States at altitudes below 5 km and provide a survey of the vertical and horizontal variability of the aerosols that dominate the direct aerosol radiative forcing of climate. Some flights were conducted over Colorado, allowing comparison of these aerosol properties between the humid East and arid West. Fig. 6 shows the vertical profiles measured over Colorado of å, ω, b, and ssp (denoted by Bsp in the figure legend). The data in Fig. 6 were obtained over a large area of the state, and some of the variations are due to horizontal inhomogeneity. Nevertheless, the results show fairly constant values of &ohacgr;, b, and å throughout the lower troposphere. The increased variability above 5 km results from the very low (and hence imprecise) values of the primary measured variables, leading to large variations in parameters that are defined as ratios of the primary variables. The single-scattering albedo varies in the range 0.88-0.95, and the hemispheric backscattering fraction is 0.15-0.18.
A similar vertical profile over the southeast U.S. is seen in Fig. 7, in spite of much higher values of light scattering (note the scale change for ssp) than were observed over Colorado. Once again, values at the higher altitudes are much less reliable because of the low values of the scattering and absorption coefficients. In the boundary layer, the single-scattering albedo is 0.95 and the hemispheric backscattering fraction is 0.11. These values are somewhat different from the values obtained over Colorado, suggesting systematic differences in aerosol composition and size distribution in the two regions. However, the differences may also be due to day-to-day variations in the aerosol. Fig. 8 shows the horizontal variability observed in the boundary layer on the transit flight from Colorado to Tennessee, where the values of &ohacgr; and b are identical to the boundary layer values shown in Fig. 7. As was the case for the vertical dimension, the derived parameters &ohacgr;, b, å) are relatively constant in spite of large changes in the primary measured variables.
Finally, Fig. 9 shows the latitudinal variability that was observed in the boundary layer over the midwest U.S. (Tennessee-Indiana). Slightly more variability is seen in the single-scattering albedo (0.89-0.96), but the hemispheric backscattering fraction is once again nearly constant (0.12). In all four cases, the Ångström exponent stays in the range 2.0-2.5. Although instrumental noise is a limiting factor, the observed variability in å may be due to variations in the aerosol size distribution, with the larger values of å corresponding to cases with smaller particle sizes.
Taken as a whole, the results of this study yield values for the single-scattering albedo in the range 0.88-0.96, with more variation observed from day-to-day than from place-to-place (horizontally or vertically). Similar conclusions can be drawn for the hemispheric backscattering fraction (0.11-0.18) and the Ångström exponent (2.0-2.5), although b in the boundary layer was always below 0.13 except for the one vertical profile over Colorado. Although it is difficult to draw general conclusions from a one-month study, the results suggest that ground-based measurements of the light scattering and absorption coefficients of submicrometer, continental particles can be used to derive values of the single-scattering albedo, hemispheric backscattering fraction, and Ångström exponent representative of the dry aerosol throughout the lower troposphere.
Vertical profiles of tropospheric and stratospheric aerosols are regularly determined at MLO with two different lidar system. Section 1.1.2 (Aerosol Monitoring) describes the instruments and analysis techniques, along with the new Nd:YAG lidar.
The integrated aerosol backscatter (IABS) data at 694 nm in Figure 10 show that no volcanic eruptions injected large amounts of aerosols that were detectable in the stratosphere at MLO latitudes in 1994-1995. The decay of Mt. Pinatubo's aerosols continued, and by the end of 1995 the lowest levels of IABS in the past 16 years were in evidence. A small increase and decay in stratospheric aerosols just prior to the Mt. Pinatubo eruption may have been related to the eruption of Kelut, which was observed by the SAGE instrument. A similar small increase in the IABS data in the fall of 1994 may be observed in Figure 3.11 at 532 nm and 694 nm. The 532-nm data are from the new Nd:YAG lidar. This increase coincides with an eruption of Rabaul in New Guinea. The increase abruptly disappears in December 1994 coincident with the air mass above the observatory switching abruptly from tropical to midlatitude air.
In September 1994 the Lidar In-space Technology Experiment (LITE) was flown in the Space Shuttle. The lidar made aerosol measurements at 1064, 532, and 355 nm from the upper stratosphere into the troposphere. LITE observing times were concentrated over the Atlantic and Europe, but during two overflights of MLO, correlative measurements were made at 532 nm. For the first overflight (September 14), the profiles agree well throughout the stratosphere. On the second flight (September 16) the profiles agree (within calculated error) below 23 km. However, the MLO lidar observed higher aerosol backscatter at elevations from 23 to 33 km although the MLO lidar exhibited the same general features in the profile as the satellite instruments.
The MLO lidars, as part of the Network for the Detection of Stratospheric Change (NDSC), participated in an aerosol analysis intercomparison (August 1995) conducted within the lidar group of NDSC to validate the analysis methods used by NDSC lidars. In the study, raw signals and radiosonde data were provided to participants to be used in their respective analysis routines. Preliminary results show good agreement between MLO analysis and the benchmark data.
Atmospheric temperature profiles have been measured with the MLO lidar over altitudes from 33 to 70 km beginning in July 1994. A blind intercomparison of temperature profiles made between the NOAA lidar and three other NASA lidars during the MLO3 campaign was undertaken in August 1995; the results and analysis have not been released to date.
Charlson, R.J., S.E. Schwartz, J.M. Hales, R.D. Cess, J.A. Coakley, Jr., J.E. Hansen, and D.J. Hofmann, Climate forcing by anthropogenic aerosols, Science 255, 423-430, 1992. National Research Council (1996) Aerosol Radiative Forcing and Climate Change. Panel on Aerosol Radiative Forcing and Climate Change, Board on Atmospheric Sciences and Climate, Commission on Geosciences, Environment, and Resources. National Academy Press, Washington, D.C., 161 pp.
National Research Council (1996) Aerosol Radiative Forcing and Climate Change. Panel on Aerosol Radiative Forcing and Climate Change, Board on Atmospheric Sciences and Climate, Commission on Geosciences, Environment, and Resources. National Academy Press, Washington, D.C., 161 pp. Ogren, J.A. (1995) A systematic approach to in situ observations of aerosol properties. In Aerosol Forcing of Climate, edited by R. J. Charlson and J. Heintzenberg, John Wiley & Sons, Ltd., 215-226. Penner, J.E., Charlson, R.J., Hales, J.M., Laulainen, N., Leifer, R., Novakov, T., Ogren, J., Radke, L.F., Schwartz, S.E. and Travis, L. 1994. Quantifying and minimizing uncertainty of climate forcing by anthropogenic aerosols. Bull. Amer. Meteor. Soc. 75, 375. Schwartz, S.E., Arnold, F., Blanchet, J.-P., Durkee, P.A., Hofmann, D.J., Hoppel, W.A., King, M.D., Lacis, A.A., Nakajima, T., Ogren, J.A., Toon, O.B., and Wendisch, M. 1995. Group Report: Connections between aerosol properties and forcing of climate. In Aerosol Forcing of Climate, edited by R.J. Charlson and J. Heintzenberg, John Wiley & Sons, Ltd., 251-280.