Aerosol measurements began at the ESRL/GMD baseline observatories in the mid-1970's as part of the Geophysical Monitoring for Climate Change (GMCC) program. The original objective of these "baseline" measurements was:
- to detect a response, or lack of response, of atmospheric aerosols to changing conditions on a global scale.
Since the inception of the program, scientific understanding of the behavior of atmospheric aerosols has improved considerably. One lesson learned is 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, IPCC,2001), ESRL/GMD expanded its aerosol research program starting in 1992 to include regional aerosol monitoring stations at Sable Island (Nova Scotia, Canada), Bondville (IL), Cheeka Peak (WA), and K'puszta (Kecskemet, Hungary). In 1997, ESRL/GMD took responsibility for the in-situ aerosol sampling system at the U.S. Department of Energy's ARM (Atmospheric Radiation Measurement) Southern Great Plains site in Lamont (OK), effectively adding another site to the regional network. Since 2000, the measurements at the surface Southern Great Plains site have been augmented by a light, instrumented aircraft flying profiles over the site 2-3 times per week. More recently, a mobile system for measuring aerosol optical properties has been designed and was deployed in Kosan (Korea) for one year prior to being moved to a new observatory site at Trinidad Head (CA).
The goals of this regional-scale monitoring program are:
- to characterize means, variabilities, and trends of climate-forcing properties of different types of aerosols, and
- to understand the factors that control these properties.
ESRL/GMD's measurements also provide ground-truth for satellite measurements and global models, as well as key aerosol parameters for global-scale models. An important aspect of the sampling strategy is that 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.
Aerosol properties monitored by ESRL/GMD include the particle number concentration larger than ~15 nm diameter (Ntot), aerosol optical depth ( d), and components of the light 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. At the regional stations, all size-selective sampling, as well as the measurements of the components of the aerosol extinction coefficient, is performed at reduced relative humidity (<40%) to eliminate confounding effects due to 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. Additional details of the aerosol measurement systems, along with an overview of the observations, are included in the latest ESRL/GMD Summary Report.
Table 1 lists the aerosol radiative properties that can be derived from the directly-measured quantities. These properties are used in chemical transport models to determine the radiative effects of the aerosol concentrations calculated 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.
Table 1: Aerosol radiative properties derived from the ESRL/GMD network.
|å||The Ångström exponent, defined by the power-law ssp µ l-å, describes the wavelength-dependence of scattered light. In the figures here, å is calculated from measurements at 550 and 700 nm wavelength. Situations where the scattering is dominated by submicrometer particles typically have values around 2, while values close to 0 occur when the scattering is dominated by particles larger than a few micrometers in diameter.|
|wo||The aerosol single-scattering albedo, defined as ssp/(sap+ssp), describes the relative contributions of scattering and absorption to the total light extinction. Purely scattering aerosols (e.g., sulfuric acid) have values of 1, while very strong absorbers (e.g., elemental carbon) have values around 0.3.|
|g, b, b||Radiative transfer models commonly require an integral property of the angular distribution of scattered light (phase function): the asymmetry factor g, the hemispheric backscatter fraction b, or the upscatter fraction b. The asymmetry factor is the cosine-weighted average of the phase function, ranging from a value of -1 for entirely backscattered light to +1 for entirely forward-scattered light. The hemispheric backscatter fraction b is sbsp/ssp. The upscatter fraction depends on the solar zenith angle, and is equal to b when the sun is directly overhead|
|ai||The mass scattering efficiency for species i, which can be defined as the partial derivative of ssp with respect to the mass concentration of a chemical species, is used in chemical transport models to evaluate the radiative effects of each chemical species prognosed by the model. This parameter has typical units of m2 g-1.|
Highlights of results for the baseline stations, regional stations, and special field projects are given below.
Bodhaine and Dutton (1993) reported a long-term decrease in Arctic haze at Barrow (AK) from 1982-1992. The observed trend was attributed to decreased anthropogenic pollution emissions in Europe and the former Soviet Union. Measurements since then (Fig. 1) show that Arctic haze levels continue to be lower than the values observed in the early 1980's, but that the reported trend has not persisted more recently. Research efforts supported by the NOAA Arctic Research Initiative are currently underway to better understand the observed change in Arctic haze.
Fig. 1: Time series of light scattering coefficient observed at BRW for Arctic haze season (points are monthly average of measurements during April for each year).
Chemical signals embedded in ice cores are believed to contain a record of past climate-altering events (volcanic eruptions, El Niño episodes), as well as anthropogenic influences. Both the atmospheric and ice core records at the South Pole contain a seasonal signal associated with winter sea salt peaks and summer sulfate peaks. Summertime peaks in the ice core sulfate to sodium mass concentration ratio correspond to peaks in the aerosol Ångström exponent (Fig. 2). This suggests that ice cores at the South Pole record aerosol chemical composition information on a seasonal basis. In addition, peaks in the aerosol scattering coefficient are seen for the volcanic eruptions of El Chichón (1982) and Pinatubo (1991). These peaks correspond to elevated sulfate concentrations in the ice. Ice core peaks of methanesulfonic acid are significantly higher during El Niño periods, although there is no obvious change in the aerosol properties. Overall, the results suggest that linking ESRL/GMD's aerosol observations to ice core records may eventually allow a quantitative estimate of atmospheric aerosol properties over the past 1000 years on a seasonal basis.
Fig. 2: Seasonal variations in ice core composition at the South Pole (1981-91) are associated with systematic changes in the wavelength dependence of aerosol light scattering.
A primary application of the observations at the regional aerosol monitoring sites is evaluation of uncertainties in radiative forcing assessments by the Intergovernmental Panel on Climate Change (IPCC). IPCC assessments have not been based on observed variabilities of aerosol radiative properties, but rather on model predictions that use estimates of average values of these properties. The aerosol radiative forcing per unit optical depth (adapted from eqn. 3 of Haywood and Shine, 1995) is controlled by aerosol single-scattering albedo and upscatter fraction (in addition to non-aerosol properties such as surface reflectance). The IPCC (1996) detection and attribution studies considered only aerosol light scattering (not absorption) and used a constant value for this parameter of 25 W m-2, i.e., an annual average optical depth of 0.1 would lead to a forcing of -2.5 W m-2. Fig. 3 presents the results of about one year of observations of submicrometer aerosol light scattering and absorption at Bondville in terms of the diurnally-averaged forcing per unit optical depth for a surface reflectance of 0.15. Measurements were made at low relative humidity; the magnitude of the forcing would be higher at ambient relative humidity, but the relative variability would be similar. These results demonstrate that observed variations in aerosol upscatter fraction alone can lead to about ± 15% variations in aerosol forcing, and that aerosol absorption further increases the variability of the forcing and reduces its magnitude by about 30%.
Fig. 3: Frequency distribution of aerosol radiative forcing per unit optical depth (W m-2) observed at Bondville (IL). The dashed curve is based on scattering measurements only, to facilitate comparison with IPCC (1996), while the solid curve is based on measurements of scattering and absorption.
ESRL/GMD aerosol scientists have been involved in a number of field projects designed to characterize the spatial distributions of aerosol properties and to understand the processes that control aerosol forcing of climate. Recent field campaigns include the Indian Ocean Experiment (INDOEX) in 1999, the Aerosol Characterization Experiment in Asia (ACE-ASIA) in 2001, and Intercontinental Transport and Chemical Transformation (ITCT) study in 2002.
The hygroscopicity of the aerosol particles is related to the aerosol composition. Back trajectories and aerosol properties such as size (indicated by the Angstrom exponent) and blackness (indicated by aerosol single scattering albedo) can be used to separate airmasses into categories such as 'clean','polluted', 'smoke-influenced', and 'dust-influenced'. For clean marine cases at Sable Island, roughly 90% of the light scattering of dry submicrometer particles was from sea salt, and the f(RH) was 2.7. For the polluted continental case, the contribution of sea salt to the light scattering coefficient of submicrometer particles was 45%, with contributions of roughly 20% each by non-seasalt sulfate, carbonaceous, and mineral aerosols. In addition, the hygroscopic growth factor for the polluted continental case was 1.6. Fig. 4a shows the measured ratio of the light scattering coefficient at a given RH to that at 40% RH, fRH(ssp), for the clean marine and polluted continental cases. The results suggest that for the polluted case the contribution of carbonaceous aerosol to the light scattering budget is similar to that of non-seasalt sulfate, and that carbonaceous species may be responsible for the observed suppression of hygroscopic growth. At the Southern Great Plains site airmasses were categorized based on aerosol optical properties into smoke- and dust- influenced aerosol. The hygroscopicity of the smoke and dust influenced aerosol was much lower than that of the overall aerosol sampled at the site (see Fig. 4b). A similar result was also found at Bondville, IL.
Fig. 4: (a) Hygroscopic growth for clean marine (open circle) and polluted continental (filled squares) submicrometer aerosols at Sable Island. (b) Probability distribution of f(RH) values at Southern Great Plains site for different aerosol compositions.
Measurements from aircraft have been perfomed in order to extend the measurements into the vertical dimension and to increase geographical coverage. The airborne aerosol package includes the similar instruments for determining aerosol optical properties as at the regional surface sites. As is done on the ground, the sample air is heated as necessary to maintain a relative humidity below 40%, and multi-jet impactors restrict the size range of particles sampled (on the aircraft, only particles smaller than 1 um aerodynamic diameter are sampled). These instruments constitute a comprehensive airborne aerosol measurement platform capable of determining a wide suite of aerosol chemical, optical and microphysical properties. The example in Fig. 5a shows a vertical profile measured over Colorado of å, wo, b, and ssp. The data 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 wo, b, and å throughout the lower troposphere, in spite of large variations in ssp. A longer term data set based on ~200 profiles flown over the Southern Great Plains site in 2000 and 2001 suggests a similar result. Fig. 5b shows that while extinction (the sum of ssp and s ap) decreases with altitude, the single scattering albedo appears to be constant with altitude. The increased variability at higher altitudes 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 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.
Fig. 5a: Vertical profile of aerosol properties over Colorado, 14 June 1995
Fig. 5b: Vertical profile of aerosol properties over Southern Great Plains, 2000-2001. Purple data is from surface station, yellow data is aircraft measurements. The box-whisker plot gives 5-25-50-75-95 percentiles.