Clouds are a critical component of the climate system. They both influence and are influenced by climate via a myriad of pathways. Anthropogenic perturbations on clouds have been hypothesized and studied for decades, yet despite significant progress, few have been demonstrated and quantified. The inextricable linkages between clouds and other components of the climate system necessitate careful assessment of the nature and extent of anthropogenic perturbations on clouds. This was the purview of our working group. More specifi cally, we were charged with identifying knowledge gaps that must be bridged in order for progress to be achieved on this important question.

We identified three primary pathways through which anthropogenic influences can perturb clouds; namely, through (a) change in the concentration and composition of aerosols; (b) greenhouse gas warming and consequent changes in the climate system; and (c) land-surface changes, primarily as a result of land use. Because of the expertise in our group, we focused primarily on the more direct means by which aerosols affect clouds, although our discussions did touch on all three types.

It was explicitly recognized that the aerosol-cloud-climate system is tightly coupled; that is, all mechanisms in the system interact. For the sake of simplifying the structure of our discussions, we chose to address separately anthropogenic perturbations in terms of their impact on cloud and aerosol microphysics, radiation, precipitation, atmospheric dynamics, and atmospheric chemistry.

These are two-way interactions. For example, aerosols affect precipitation, but precipitation also affects aerosols (and similarly for the other components). We stress, however, that these pairs of interactions are all tightly coupled and that an anthropogenic perturbation may set off a chain reaction resulting in feedbacks, both positive and negative, to the cloud system and involving all of these categories.

We divided the gaps in understanding into three categories:

1. Conceptual gaps, which represent "big picture" gaps in our understanding of how to proceed.

2. Knowledge/data gaps, which are deficits that could be filled using present-day instruments and data, but for some reason (e.g., lack of resources) have not.

3. Tool gaps or deficits in our ability to make relevant measurements.

We begin with a general discussion related to our current understanding of perturbed clouds and then proceed to each of the specific areas impacted by anthropogenic perturbations. We focus on the three types of gaps as they relate to each of these topics. As with the other reports in this volume, our intent is to highlight our discussions during the Forum. By nature, this report does not comprehensively cover any given topic nor does it encompass all possible topics.

General Discussion on Perturbed Clouds

It was generally felt that although a great deal is known about the physical mechanisms of particle activation, cloud microphysics, and cloud dynamics at the single cloud scale, our level of understanding regarding the nature and the extent of anthropogenic perturbations on cloud systems is relatively poor. This may be because perturbations to clouds, especially from aerosol changes, tend to be small relative to the natural variability of a cloud system, or perhaps because the system is self-regulating (e.g., Feingold and Siebert, this volume) and thus perturbations lead to feedbacks (both positive and negative), which makes their manifestation difficult to identify.

Establishing a causal relationship, rather than a correlative one, between, for example, aerosols and clouds requires us to minimize confounding (i.e., co-varying) factors: we must choose a location and/or regime where meteorological variability is minimal while anthropogenic perturbations are maximal (i.e., an area where the aerosol is ideally uncorrelated with meteorology). The usefulness of time series analysis as an attribution strategy is limited because of these confounding factors (i.e., covariance between aerosols and meteorological factors such as relative humidity; for further discussion, see Stevens and Brenguier, this volume). Other fields of study, such as epidemiology, have experience in addressing such issues, and thus may have tools that are useful to account for such factors.


Remote Sensing

High-quality (well-calibrated) long-term records (both in situ and remotely sensed) are critical to trend detection. A timescale of thirty years is generally regarded as a minimum record length. Satellites provide a critical source of global-scale observations, but are subject to uncertainties that must be accounted for and understood. Satellite overpass times drift over the lifetime of the satellite platform, leading to spurious trends. Moreover, the entire period of satellite observations is not yet even 30 years old. Satellite sensors suffer from degradation over time, lack (in many cases) absolute calibration, and suffer calibration drift. An overlap of sensors to conduct similar measurements is critical to resolve any discrepancies. The lack of an overlap period between ERBE and CERES, along with a significant bias between the two, has led to considerable uncertainty in the radiative budget. Advances in measurement capabilities can be attained through synergistic deployment of instruments (e.g., the A-Train constellation of satellites). Polar-orbiting satellites are constrained to observe any given location once or twice a day, and this greatly limits their usefulness over hourly to daily timescales. Polar-orbiting satellites do not measure clouds at the peak of afternoon convection; for that, we must depend on geostationary satellites with their limited payloads. Geosynchronous satellites have typical resolutions of 1 km and cannot (at the present) constrain many of the critical parameters identified above. They also suffer from edge detection problems. The possibility of a satellite positioned at the Lagrange point was mentioned. Such a satellite would view the entire sunlit hemisphere (with problems near the edge of the field of view) with roughly an 8 km resolution; this is suitable for larger-scale studies, but does not match well with smaller-scale studies.

A lack of co-location of aerosols and clouds viewed from space confounds correlation studies. Whereas subcloud aerosols in updrafts are most pertinent to cloud drop formation, the presence of the cloud generally excludes the ability of most sensors to measure subcloud aerosol properties. This problem can be alleviated using surface remote sensors, but only at limited locations. Aerosols above clouds may also influence cloud properties, but it is difficult to measure aerosols from space accurately above a bright (i.e., cloudy) layer. Finally, because of the complication of surface radiative inhomogeneity, aerosol measurements over ocean locations are for some instruments more reliable than those over land.

Measurement simulators are viewed as a powerful means to compare model output with measurements, including assessment of measurement biases and random error, as well as to optimize the measurement strategy for an observational program, including understanding the added benefit of a given instrument. For example, EarthCARE simulators (Schutgens and Donovan 2004) are being used prior to deployment to optimize the sensor capabilities and exact the most useful dataset. We recommend that future satellite missions include the development of such simulators for distribution to the scientific community. We feel that this is a fruitful direction for future studies to consider.

Integration of Models and Observations

The appropriateness of tools for a given problem depends, among other things, on their spatial and temporal resolution. Figure 18.1 illustrates a conceptual framework for the interaction of different tools (modeling, observations, and laboratory measurements) at various scales. Laboratory measurements provide information on fundamental physical and chemical cloud processes. This information is critical for informing models and field observations. At the smallest length scales, field observations and large eddy simulation (LES) models operate at very similar length and timescales, and resolve aerosol and cloud processes as well as small-scale dynamics. Progress in our understanding at this scale depends on the close interactions between models and observations. One important step is to establish the usefulness of a given model. Generation and testing of new ideas and hypotheses can only be achieved through such cooperation. Further, this strategy is the most effective method to establish the extent to which observed correlations are causal.

Comparison between models and observations should be done to the greatest extent possible using fundamental measured properties (e.g., irradiances rather than albedo; aerosol extinction rather than mass). This will minimize the inconsistencies in assumptions in derived quantities. One example is the comparison between modeled and measured radar reflectivity, which is likely to be much more meaningful if the model actually calculates the sixth moment of the drop size distribution, rather than rely on an empirical relationship between reflectivity and a modeled variable (e.g., liquid water mixing ratio or other empiricisms). Another example would be a comparison between measured and modeled optical depth, which would be greatly facilitated by model calculation of extinction. Confidence in smaller-scale models obtained through this process can feed into larger-scale models in the form of parameterizations. Collaboration between the modeling and observational communities must occur at the full range of scales, if any real progress is to be made.

The quote, "All models are wrong, but some are useful," attributed to George E. P. Box, was raised as an important point for consideration. A suggested modification to this quote is: All models and measurements are wrong, but some are useful. We recognize that all measurements and models are imperfect; still, they can be extremely useful if used and applied appropriately.

Models Observations

Models Observations

Figure 18.1 Conceptual schematic of the relationship among different subdisciplines for addressing questions regarding anthropogenic perturbations to clouds and precipitation. GCM: general circulation model; LES: large eddy simulations.

The main point is that the quality of data and model output, as well as usefulness for addressing the problem at hand must be given careful consideration to evaluate the strengths and weaknesses.


In our discussion of predictability, a number of interesting questions were raised: Is there a limit to the predictably of cloud systems? Does predictability depend on length scale? At the largest scales, for example, precipitation must be constrained by total evaporation and large-scale circulation, which in turn relate to top-of-atmosphere (TOA) and surface energy budgets. These constraints may lead to greater predictability at these scales. At small length and timescales (e.g., single cloud scale), the largest-scale features (such as mean vertical profiles of potential temperature and humidity, subsidence rates, etc.) may be easily predicted because they vary in ways that are fairly well understood (e.g., diurnally with solar heating) on these short timescales. If this exists, such a scenario may grant advantages for prediction at small scales. At intermediate scales, however, it is conceivable that high-frequency variability from smaller scales, as well as large-scale constraints, will play a significant role, thus causing a minimum of predictability. Whether or not this is true remains an open question (cf. Siebesma et al., this volume).


Perturbations on cloud microphysical properties, especially through changes in particle concentration and/or composition, serve as a starting point for discussing many of the subsequent perturbations related to radiation, precipitation, dynamics, and chemistry. The state of cloud condensation nuclei (CCN) measurements and knowledge, as well as anthropogenic perturbations of CCN, has been presented by Kreidenweis et al. (this volume). Another excellent summary is available from McFiggins et al. (2006). Because of this existing body of work and the limited time available, the group did not address anthropogenic perturbations to a number of cloud microphysical processes (e.g., activation, collision-coalescence and evaporation/entrainment) in much detail. This is reflected in the summary below and does not imply an absence of gaps within these important subjects.

Some group members felt that the largest remaining gap in understanding CCN activation is the mass accommodation coefficient of water vapor onto atmospheric aerosol. Recent measurements by Ruehl et al. (2008) suggest that the value is variable, even for the limited number of sampling sites and short periods that were sampled. In contrast, measurements of ice nuclei are limited to a few instruments worldwide and are unable to observe all possible modes of heterogeneous nucleation. This technological gap creates a fundamental difficulty in understanding potential perturbations to glaciated clouds. Bioaerosols are potentially important as ice nuclei, and our understanding of this component remains poor, although some fluorescence-based instruments do exist for detecting a subset of such particles.

Supersaturation (with respect to both water and ice) is not accurately measurable, and thus this important constraint on cloud drop production is not available. This technological gap has been known for a long time, yet it endures in large part because of the difficulty of making such a measurement.

There is a need for long-term, spatially resolved records of CCN spectra. To our knowledge, the only multidecadal record of CCN is from Cape Grim, Tasmania, although a handful of other sites are now instrumented for routine CCN measurement (e.g., Mace Head, Ireland). Thus, we recommend that the creation and maintenance of such a network be considered an important priority for understanding this fundamental perturbation to clouds. The usefulness of CCN proxies, such as the dry-state aerosol accumulation mode number concentration along with some (limited) hygroscopic information, should be considered. A satellite sensor-derived aerosol index (product of aerosol optical depth and Angstrom exponent) has also been proposed as a proxy for CCN. This may be qualitatively useful but its quantitative accuracy is highly questionable, since it cannot account for particle humidification (Kapustin et al. 2006). Finally, we suggest further exploration of possibilities to obtain routine CCN and other aerosol measurements from platforms of opportunity, such as ships and commercial aircraft.

The Extent and Nature of Anthropogenic Perturbations of Clouds 439 Removal Mechanisms

Aerosol-scavenging mechanisms that are associated with clouds and precipitation (also known as wet deposition) dominate aerosol removal. Although generally not as well-studied as other aerosol-cloud interactions, they are important in understanding aerosol budgets at a range of scales (single cloud to global).

The following questions and associated challenges were raised:

1. A perturbation in precipitation by aerosols can lead to a perturbation in aerosol wet deposition. Could this be an important positive feedback? Some boundary layer modeling studies suggest this (e.g., Feingold and Kreidenweis 2002).

2. Model results (e.g., Respondek et al. 1995) show a relationship between precipitation efficiency and scavenging efficiency. Is it possible to make important inferences about precipitation efficiency by measuring scavenging efficiency?

3. Is coalescence scavenging an important mechanism for converting external mixtures of particles to internal mixtures?

4. What are the consequences of aerosol scavenging for the climate system? For example, the scavenging of soot by snow, its deposition to the surface, and associated amplification of snow melt, is an active topic of research.

A number of important tool and data gaps exist. Measurements of precipitation chemistry are routinely made, but are likely inadequate for understanding the lifetime of aerosols and the cycling of organics, dust, and soot. This represents a crucial gap in the understanding of the global aerosol budget. Some existing programs include the European Monitoring Evaluation Programme and the U.S. National Atmospheric Deposition Program. In-situ measurement of cloud drop chemical composition is challenging to conduct on appropriately fast tim-escales, but would be valuable for understanding cloud chemistry. Scavenging efficiency is hard to assess because of volatility of some scavenged species. It was proposed that ice cores represent a potential historical record for deposition, but they do not appear to have been fully exploited.


The globally averaged temperature over the last 10,000 years appears to have been stable to within ~1 K, thus constraining, most likely, the globally averaged albedo to within ~1% throughout this period. The implications of this 1% variability are not negligible since it represents about 3.5 W m-2, or 1% of the incoming solar radiation at TOA. Clouds represent roughly two-thirds of the global planetary albedo. This suggests the presence of feedbacks that maintain global cloud albedo, which remain poorly understood. Siebesma et al. (this volume), however, suggest that this is simply a result of the Law of Large Numbers and does not, therefore, require the presence of such feedback mechanisms.

By directly scattering and absorbing radiation, and through modifying cloud drop size, aerosol particles modify the Earth's radiation budget at the top of and within the atmosphere, as well as at the surface. The interaction of aerosols and clouds with radiation modifies heating profiles with feedbacks, which in turn modify cloudiness. Perturbations in land use, or aerosol-induced perturbations in net surface radiation, can alter surface fluxes of sensible and latent heat and moisture, thus perturbing convection and, therefore, cloud formation. Anthropogenic perturbations in cloud albedo have the potential to impose changes in the planetary radiative balance. Local changes have been demonstrated (e.g., ship tracks). At larger scales, however, studies have not been definitive. There was general agreement in the group that perturbations on cloud fraction, precipitation, and the longevity of clouds have not been convincingly demonstrated (cf. Anderson et al., this volume).

Cloud Fraction

The effect of aerosols on cloud fraction is poorly constrained. Not only the magnitude, but also the sign of the effect (i.e., positive or negative) is unclear and may be dependent on the regime. Differentiation between cloudy air and cloud-free air is somewhat arbitrary, and thus cloud fraction is not a well-defined quantity. The presence of cloud haloes as a result of humidified aerosols and small cloud fragments renders the separation between cloudy and cloud-free atmospheres artificial (Koren et al. 2007; Charlson et al. 2007). Thus, aerosol perturbations on cloud fraction are also ill-defined. Furthermore, measurement of cloud fraction depends both on the spatial resolution and detection sensitivity of the instrument. Studies of the size and area distribution of clouds show that small clouds are more numerous, and they contribute more to the cloud fraction than larger clouds, down to the smallest scales resolved by current instruments (Koren et al. 2008). Thus, we suggest that the definition of cloud fraction is dependent on the application. For example, even though small clouds contribute more strongly to cloud fraction, there may exist a size below which smaller, optically thinner clouds can be neglected for the purposes of determining planetary albedo. This might yield a lower bound on the size of clouds relevant to cloud fraction, and thereby an appropriate way to def ne cloud fraction in this context. This lower bound may be different for other problems, such as longwave radiation. If true, this would imply that the appropriate measurement spatial resolution also depends on the problem under consideration.

The Extent and Nature of Anthropogenic Perturbations of Clouds 441 Radiative Forcing

For detecting trends in radiation, one possible strategy for overcoming uncertainties in satellite retrievals is to utilize regions where models reproduce observations well; that is, regions where the signal is strongest. Such models can then be used to generate a global distribution of radiation, which suggests the following question: Is albedo best determined by measurements alone or through a hybrid of measurements and models? Arguments for both sides were presented.

Surface radiation measurements are useful for deriving surface (as opposed to TOA) forcing. In the absence of surface measurements, surface forcing has to be derived from TOA measurements. This conversion requires assumptions about atmospheric absorption. Surface measurements of net irradiance, combined with satellite TOA measurements, obviate the need for such assumptions. Surface supersites that combine remote sensing from a variety of instruments with in-situ aerosol sampling and thermodynamic profiles provide extremely useful datasets for comparison with satellites. The high temporal resolution allows for process studies not achievable from space. Examples include sites at Southern Great Plains (U.S.A.), Cabauw (Netherlands), Lindenburg (Germany), Gosan (Korea), and Okinawa (Japan). Lidar networks provide information on the vertical distribution of aerosol, cloud base, and cirrus clouds at high vertical and temporal resolution. New technologies, such as high spectral resolution lidar, enable the measurement of extinction profiles and comparison with aerosol optical depth, but they are limited to thin clouds. Combining lidar with radar, or radar with infrared radiometry, enables retrievals of effective ice crystal radius. This is a good example of measurement synergy. Optimal estimation methods are increasingly being applied to yield best-estimates of physical parameters based on observations from multiple instruments, each with their own strengths and weaknesses (Feingold et al. 2006; Hogan 2007).

The AERONET sun photometer network has developed since the late 1990s into an important research tool aimed at characterizing the global aerosol for climate study purposes (see also Kinne, this volume). Although its primary purpose is to validate remote-sensing products, AERONET serves also to constrain atmospheric aerosol modeling in all aspects (e.g., the international AeroCom initiative). A link to other sun photometer measurement networks (e.g., those assembling under the Global Aerosol Watch Program) is most desirable. It would be useful if these and other measurement systems could be combined with AERONET to address questions related to aerosol-cloud interactions. This includes higher measurement frequencies, which should not pose a problem with current computing capacities. Better documentation of the actual sequence of cloudy and clear-sky scenes would allow better proxies for total cloud cover (see Koren et al. 2007). Additional parameters (e.g., rain amount, rain drop size spectra, cloud base height, and broadband radiative fluxes) should be measured at an increasing number of AERONET stations.

The location of new supersites for aerosol and cloud characterization should be coordinated with long existing AERONET sites. Supersites should ensure that more than one sun photometer is operational at any time to avoid any discontinuity of these basic aerosol observational time series.


The CERES and ERBE satellite missions have been critical in advancing our understanding of the Earth's radiation budget. Their lack of overlap raises questions, however, about their apparent disagreement and limits their utility. The fact that no replacement for CERES is planned was unanimously viewed as a serious problem, and we recommend strongly that this be remedied as soon as possible.

It was suggested that a satellite at the Lagrange point L1 would be useful for radiation studies, as it would always be pointed at the sunlit side of Earth, yielding time series (including the diurnal cycle) of global albedo, which requires a radiance-flux calculation, potentially resolved down to ~8 km. Such a satellite could also serve as an important constraint on other satellite sensors.

It was noted that the multi-wavelength radiometric data from MODIS, aboard both the Terra and Aqua platforms, are being used for a number of inverse retrievals; among these are the fractional cloud cover, the effective radius of cloud droplets, and the aerosol optical depth. Because of the importance of these quantities to studies such as those conducted by the IPCC, the specificity and uncertainty of these retrievals are currently being questioned. Although these satellite retrievals are of great importance to global sensing of gross phenomena, such as the annual outbreaks of dust from the Sahara Desert or the smoke from biomass burning in Africa, they may not be fully adequate to quantify accurately the direct and indirect climate forcing by anthropogenic aerosols. Some of the difficulty involved in quantifying the uncertainty of the remote retrievals from MODIS lies in the inherent difficulty of separating clear and cloudy conditions (e.g., over the horizontal extent of a MODIS pixel; Charlson et al. 2007). This open issue remains to be resolved.

Measurements of earthshine (i.e., the sunlight reflected by Earth that illuminates the unlit (by the Sun) side of the Moon) are a relatively inexpensive and simple technique used to derive large-scale albedo of the Earth. Earthshine data from multiple locations could provide a very useful, independent measure of albedo. Such data can also provide temporal information that is not available from polar-orbiting satellites. Comparison of climate model output could be achieved through an "earthshine simulator."

Finally, surface albedo and emissions are fundamental to the production of accurate satellite retrievals of aerosol optical depth. The baseline surface radiation network was also mentioned as a potential source of measurements, but its utility for cloud measurements requires most likely additional calibration and instrumentation effort.

Semi-direct Effect

The semi-direct effect influences clouds over a range of spatial scales. Absorbing aerosol particles generate local heating, and the presence of the semi-direct effect at the appropriate levels can modify the atmospheric stability and suppress vertical motion and cloud formation (Hansen et al. 1997). Alternately, heat added to the planetary boundary layer might raise the temperature and evaporate cloud droplets. We expand our definition to include the fact that the presence of aerosol, particularly absorbing aerosol, is very effective at reducing the downwelling solar radiation at the surface, and consequently, the surface fluxes. The combination of these factors, as well as the (much smaller) microphysi-cal influences on droplet growth, results in significant reduction in cloudiness. This has been observed in biomass burning regions in Brazil (Koren et al. 2004), and satellite imagery strongly suggests effects at the cloud scale as well. Modeling studies suggest that the semi-direct effect might change regional precipitation patterns (Menon et al. 2002). Therefore, an open question is: To what extent do absorbing aerosols modify dynamics, clouds, and precipitation, and might these effects be large enough to influence global circulation?


Clouds are the only entities that deliver precipitation to the Earth's surface. Precipitation is critical to water resources and the global energy cycle. Globally, precipitation must balance global evaporation over sufficiently long timescales. (How long is long enough is unclear, but it may be approximately the residence time of water vapor in the troposphere, which is about one week.) Thus, perturbations in total precipitation amount are constrained by impacts on global evaporation. However, the spatial and temporal distribution of precipitation as well as the distribution of precipitation intensity may change significantly. Greenhouse gas warming is predicted to cause increases in evaporation and precipitation on a global scale, as well as changing precipitation patterns on a regional scale. Currently, however, there is not enough evidence to suggest that changes occur on smaller scales. Aerosol perturbations have been linked to precipitation suppression (Warner 1968), but there is no statistically robust proof of aerosol suppression of surface rainfall at scales large enough to be of importance to the hydrological cycle. Moreover, although questions persist about warm rain initiation, greater unresolved issues remain in terms of the ice phase, especially concerning possible anthropogenic perturbations to ice processes (for further discussion, see Levin and Cotton 2008; Ayers and Levin, this volume; Cotton, this volume).

Aerosol perturbations are spatially inhomogeneous on a number of scales. Because their ability to affect precipitation depends on the co-location of precipitation and aerosol, aerosol-related precipitation perturbations are also expected to be spatially inhomogeneous across a variety of scales. The variability in precipitation associated with natural variability is very large relative to the aerosol perturbation, and thus attribution of changes in precipitation to aerosols is extremely challenging. Presently, the sign of the overall aerosol effect is unclear and may not be the same for all cloud regimes and environments. Single cloud studies show strong changes in precipitation in response to aerosol perturbations in particular places, times, or regimes, but response at larger scales remains elusive. Models can help, but important physical knowledge is lacking (particularly in regards to ice microphysics), thus limiting the usefulness of models that might otherwise be adequate to study this problem.

Numerical weather prediction (NWP) models may be a resource that has not been fully exploited for investigating aerosol-precipitation interactions. Over the last forty years, NWP has improved because data accuracy has improved more than microphysical parameterizations. What are the relative benefits to further improvements in defining model initial and boundary conditions versus a better representation of aerosol? Some members in our group stated that there does not appear to be a difference in skill between clean and polluted aerosol conditions, although this may not necessarily imply that aerosol representation is not needed to improve NWP. The question of how much of the residual variability in NWP is the result of aerosols was raised, but no clear answer was proposed. Use of retrospective (30-yr) data to test model improvements in quantitative precipitation forecasting may allow some of these ideas to be tested. One advantage of NWP is that it is performed daily; assimilation methods establish the errors and biases daily, at relatively low cost.

Because precipitation perturbations that result from aerosols are expected to be small relative to natural variability, it is unclear whether we have tools sufficient to the task of detecting such perturbations. Traditional in-situ measurement of drops relevant to precipitation (especially precipitation embryos) exhibit large uncertainties. It was suggested that there exist aerosol-dependent biases in radar refl ectivities (i.e., the relationship between radar reflectivity Z and rain rate R is aerosol dependent), and therefore they might not be the ideal tool for aerosol-precipitation interactions, although not everyone agreed on this point. Rain gauge data represent an important long-term record with widespread ground stations in populated areas. This dataset is, however, problematic for a variety of reasons (e.g., representativeness, biases). How can this dataset be improved? It is important to note that the rather large number of weather modification studies over the past half-century or more (see Cotton, this volume) has not demonstrated highly significant perturbations in precipitation despite very large local aerosol perturbations.

A number of strategies may be useful in detecting aerosol influences on precipitation. For example, it could be advantageous to capitalize on human-induced variability as a perturbation experiment (e.g., weekly aerosol cycles) wherever possible. Weekly aerosol cycles are observed in some, but not all locations. However, the weekly precipitation cycle is even less clear in the observation record than in the aerosol cycle. Orographic clouds may serve as a useful test bed, with mixed results thus far (for further discussion, see Ayers and Levin, this volume). Single water drainage basin studies are proposed as targets. Consensus was not achieved in terms of the potential for closing the energy and moisture budgets. Some argued that this is the necessary approach, and therefore we should do the best we can, while others stated that a single basin is too large of an area over which energy and moisture budgets can be closed, and thus this strategy cannot succeed at present. This issue remains to be resolved.


Dynamics (i.e., the three-dimensional movement of the atmosphere) are the primary driver of clouds. Anthropogenic perturbations to clouds can affect dynamics through myriad feedbacks on various spatial and temporal scales. The resulting perturbations in dynamics from such feedbacks are, however, poorly understood. Below we present some proposed feedbacks that emanated from our discussions, which we felt were of particular interest and/or importance:

1. Aerosol perturbations can modify precipitation rates. Evaporating precipitation cools and thus stabilizes the subcloud layer.

2. In a field of clouds, perturbations to precipitation can change the strength of downdrafts associated with rain shafts, altering surface convergence and thus subsequent convection. This can occur at a range of scales, from squall lines down to open cell convection in stratocumulus.

3. Suppression of freezing associated with smaller drops formed in polluted clouds results in latent heat release at higher altitudes, leading to cloud invigoration and deepening.

4. Land-use changes can alter the magnitudes of the surface fluxes, as well as the Bowen ratio, with consequences for the strength of convection and cloud base and cloud top heights.

5. Smaller drops associated with polluted clouds may evaporate more readily, promoting increased entrainment mixing and turbulence, with possible ramifi cations for turbulent collection and precipitation formation.

6. Greenhouse gas warming is predicted to change general circulation patterns, with possibly important consequences for the distribution of clouds as well as the distribution and intensity of precipitation.


There was broad discussion about strategies for addressing dynamic feedbacks. It was generally agreed that a team comprising experts in atmospheric dynamics, cloud physics, and aerosols, as well as their models and observations, offers the best hope for success. The ultimate goal should not be merely correlation but causation, for which models appear necessary. We should attempt to look for systems that are naturally simple, but it was unclear which cloud systems best fit this criterion. It was believed that working at the smallest scales plausible for examining dynamic feedbacks held a number of advantages, including the ability to integrate large eddy simulations with observations, as well as in finding the largest signals. However, the minimum spatial and temporal scale at which such feedbacks are observable is unknown. Even with a relatively simple system, it was unclear to us what an appropriate strategy would be, and whether a realistic set of instrumentation would detect any such feedbacks, either because of inherent instrument limitations or because of the lack of an adequate sampling strategy. It was noted that increasing sampling statistics is useful only for reducing random errors, whereas biases (e.g., co-variance of aerosol and meteorology; see Stevens and Brenguier, this volume) cannot be reduced in this way. Therefore, we recommend that future studies should identify and study regimes with minimum biases.


The role of measurements in understanding anthropogenic perturbations to dynamics was viewed primarily as constraints on models (horizontal arrows in Figure 18.1). As the quality and quantity of constraints improve, so too will the degree of our confidence in the model predictions. We had a long discussion on which variables are most important to constrain. The critical ones that we identified fell into both the dynamic and microphysical categories: temperature and relative humidity vertical profiles; large-scale subsidence rates; surface fluxes of moisture, energy, and momentum; winds; cloud liquid water content profiles; cloud liquid water path; cloud base and top height; vertical profiles of rain rate (including below-cloud); cloud fraction; supersaturation with respect to water and ice (particularly in the upper troposphere); and many microphysi-cal properties related to the ice phase. Many of these variables are very dif-fi cult to measure accurately, especially over the spatial and temporal scales that might be needed to observe dynamic feedbacks properly. At large scales, the most successful strategy will likely require combining numerous high-resolution satellite products (e.g., gases, aerosol, cloud hydrometeors, radiation), over large spatial and temporal scales, with appropriate models. A number of problems with existing satellite retrievals were identified that confound studies of the coupling between clouds and atmospheric dynamics. In particular, there is a strong bias between surface- and satellite-derived cloud fraction, and satellite- and in situ-derived drop sizes in broken cloud fields.

The Extent and Nature of Anthropogenic Perturbations of Clouds 447 Chemistry

We considered the broader issue of the interactions of atmospheric chemistry and clouds. The average particle cycles through clouds many times before it is removed from the atmosphere. Therefore, the potential for chemical modification of particles is very high (Hoppel et al. 1994). Most anthropogenic pollutants are reduced relative to the oxidizing atmosphere, and thus atmospheric processing leads to increases in polar and therefore water soluble species. This chemical modification has implications for physical, chemical, and optical particle properties and affects subsequent cloud cycles. Cloud particle chemistry can occur in both aqueous and ice phases. Aqueous phase chemistry is likely to be faster because the components have greater mobility. Cloud chemistry also affects gas phase chemistry, partitioning soluble components to the aqueous phase. Aqueous phase chemistry in sea-salt haze may be significant despite its small volume, as a result of spatial and temporal persistence (Sievering et al. 2004). Processing of hydrophobic particles (e.g., dust, soot) in clouds may be an efficient way to increase hygroscopicity. For dust, this could be a mechanism for producing giant CCN. In general, it was thought that quantitative understanding of cloud processing remains a knowledge gap, but one that may be filled with current instrumentation and existing sampling strategies.

Clouds are an important means of transporting boundary layer constituents (trace gases, aerosol) to the free troposphere and lower stratosphere. Perturbed clouds may exhibit changes in their depth of convection as a result of changes in, for example, precipitation or latent heating, which will affect the altitude to which these constituents are transported.

It is well known that the sulfur cycle is strongly perturbed by anthropogenic pollutants. Sulfur is a key component of the atmospheric aerosol and therefore CCN. Oceanic seawater sulfate is highly abundant and thus substrate availability does not limit the production of volatile sulfur compounds (e.g., dimethyl sulfide, carbonyl sulfide). Instead, biology appears to be the determining factor in producing such compounds. Advances have been made in understanding the specific factors that govern such production, but many gaps in knowledge remain. Perturbations to the surface ocean (e.g., temperature, pH, salinity) may affect biological production of such compounds to a degree sufficient to impact the atmospheric sulfur cycle signifi cantly. Production of volatile sulfur compounds is sensitive to sea surface temperature as well as meteorology, setting up potential feedbacks.

New particle formation in the vicinity of clouds is poorly understood and is hard to measure, yet it may represent an important contribution to the global particle number concentration budget. In-situ measurements may constitute the only method for constraining this problem because Aitken mode particles are extremely difficult to detect optically. Existing techniques are often insufficiently fast for aircraft measurements, which may make this a measurement gap. However, slower airborne platforms do exist (e.g., dirigibles) and may permit present instrumentation to address this question (see Feingold and Siebert, this volume.)

Organic compounds are an important component of atmospheric aerosol. Models underestimate strongly the amount of secondary organic aerosol relative to measurements. Clouds may contribute to this missing source (e.g., Ervens et al. 2008), which we suggest is a potentially fruitful direction to explore to help resolve this issue. On a global scale, biogenic organic particles appear to be more abundant than anthropogenic particles. Perturbations to precipitation, temperature, sunlight, and thus to the land ecosystem have the potential to alter this source strength (for a recent review, see Möhler et al. 2007).

The deposition of inorganic ions is documented by a number of networks, such as the European Monitoring and Evaluation Programme and the U.S. National Atmospheric Deposition Program. Long-term monitoring on large spatial scales of the organic and black carbon concentrations in rain water is lacking, yet it is important for closing their budgets. The techniques necessary to conduct these measurements exist, so this gap could be filled given appropriate resources.

Finally, if anthropogenic perturbations alter convection and therefore lightning production, this may alter NOx production. The potential significance ofthis to photochemistry was unclear to the group and remains to be illuminated.


Charlson, R. J., A. S. Ackerman, F. A. M. Bender, T. L. Anderson, and Z. Liu. 2007. On the climate forcing consequences of the albedo continuum between cloudy and clear air. Tellus B 59(4):715-727. Ervens, B., A. G. Carlton, B. J. Turpin et al. 2008. Secondary organic aerosol yields from cloud-processing of isoprene oxidation products. Geophys. Res. Lett. 35:L02816 Feingold, G., R. Furrer, P. Pilewskie et al. 2006. Aerosol indirect effect studies at Southern Great Plains during the May 2003 Intensive Operations Period. J. Geophys. Res. 111(D5):S14.

Feingold, G., and S. M. Kreidenweis. 2002. Cloud processing of aerosol as modeled by a large eddy simulation with coupled microphysics and aqueous chemistry. J. Geophys. Res. 107:D23. Hansen, J., M. Sato, and R. Ruedy. 1997. Radiative forcing and climate response. J.

Geophys. Res. 102(D6):6831-6864. Hogan, R. J. 2007. A variational scheme for retrieving rainfall rate and hail reflectivity fraction from polarization radar. J. Appl. Meteor. 46(10):1544-1564. Hoppel, W. A., G. M. Frick, J. Fitzgerald, and R. E. Larson. 1994. Marine boundary-layer measurements of new particle formation and the effects nonprecipitating clouds have on aerosol-size distribution. J. Geophys. Res. 99(D7):14,443-14,459. Kapustin, V. N., A. D. Clarke, Y. Shinozuka et al. A. 2006. On the determination of a cloud condensation nuclei from satellite: Challenges and possibilities. J. Geophys. Res. 111(D4):202.

Koren, I., Y. J. Kaufman, L. A. Remer, and J. V. Martins. 2004. Measurement of the effect of Amazon smoke on inhibition of cloud formation. Science 303(5662):1342-1345.

Koren, I., L. Oreopoulos, G. Feingold, L. A. Remer, and O. Altaratz. 2008. How small is a small cloud? Atmos. Chem. Phys. Discuss. 8:6379-6407.

Koren, I., L. A. Remer, Y. J. Kaufman, Y. Rudich, and J. V. Martins. 2007. On the twilight zone between clouds and aerosols. Geophys. Res. Lett. 34(8):L08805.

Levin, Z., and W. R. Cotton, eds. 2008. Aerosol pollution impact on precipitaton: A scientific review. The WMO/IUGG International Aerosol Precipitation Science Assessment Group. New York: Springer, in press.

McFiggans, G., P. Artaxo, U. Baltensperger et al. 2006. The effect of physical and chemical aerosol properties on warm cloud droplet activation. Atmos. Chem. Phys. 6:2593-2649.

Menon, S., J. Hansen, L. Nazarenko, and Y. F. Luo. 2002. Climate effects of black carbon aerosols in China and India. Science 297(5590):2250-2253.

Möhler, O., P. J. DeMott, G. Vali, and Z. Levin. 2007. Microbiology and atmospheric processes: The role of biological particles in cloud physics. Biogeosciences 4:10591071.

Respondek, P. S., A. I. Flossmann, R. R. Alheit, and H. R. Pruppacher. 1995. A theoretical study of the wet removal of atmospheric pollutants: The uptake, redistribution, and deposition of (NH4)2SO4 by a convective cloud containing ice. J. Atmos. Sci. 52:2121-2132.

Ruehl, C. R., P. Y. Chuang, and A. Nenes. 2008. How quickly do cloud droplets form on atmospheric particles? Atmos. Chem. Phys. 8:1043-1055.

Sievering, H., J. Cainey, M. Harvey et al. 2004. Aerosol non-sea-salt sulfate in the remote marine boundary layer under clear-sky and normal cloudiness conditions: Ocean-derived biogenic alkalinity enhances sea-salt sulfate production by ozone oxidation. J. Geophys. Res. 109:D19317.

Warner, J. 1968. A reduction in rainfall associated with smoke from sugar-cane fires: An inadvertent weather modification? J. Appl. Meteor. 7:247-251.

0 0

Post a comment