There was a problem loading page 62.
Atmospheric Physics Research Entering the Twenty-First Century1
Atmospheric physics seeks to explain atmospheric phenomena that occur on a variety of temporal and spatial scales in terms of physical principles. Areas included in atmospheric physics are atmospheric radiation, aerosol physics, the physics of clouds, atmospheric electricity, the physics of the atmospheric boundary layer, and small-scale atmospheric dynamics.
Major Scientific Goals and Challenges
In each of these areas, we generally have a useful understanding of the physical principles involved at the most fundamental level. However, understanding these physical principles alone does not ensure an adequate understanding of observed atmospheric phenomena because the various realizations of these phenomena are inherently complex and result from complicated interactions among physical processes. Further, these interactions occur across a great range of time and space
1 Report of the Ad Hoc Group on Atmospheric Physics: W.A. Cooper (Chair), National Center for Atmospheric Research; T. Ackerman, Pennsylvania State University; C. Bretherton, University of Washington; S. Cox, Colorado State University; J. Dye, National Center for Atmospheric Research; E. Gossard, Environmental Technology Laboratory, National Oceanic and Atmospheric Administration; D. Lenschow, National Center for Atmospheric Research, V. Ramaswamy, Geophysical Fluid Dynamics Laboratory, National Oceanic and Atmospheric Administration; D. Raymond. New Mexico Institute of Mining and Technology; E. Williams, Massachusetts Institute of Technology.
scales, and many small-scale processes have significant influence on those occurring on larger scales. An understanding must be developed of the collective influence of individual physical processes on larger spatial-scale and longer time-scale phenomena, which are referred to here as organizing principles.
Key Components of the Scientific Strategy
The most critical components of a program to address the scientific issues and challenges are the following:
• To develop and verify an ability to predict the influences of small-scale, atmospheric physical processes, such as turbulence, on large-scale atmospheric phenomena such as thunderstorms.
In many cases, large-scale atmospheric phenomena arise from the collective effects of an ensemble of interactions that occur on much smaller spatial and much shorter temporal scales. Two central impediments to more accurate simulation and prediction of weather and climate, which are comprised of these large-scale events, are the physical understanding of the smaller-scale events and the inability to include them explicitly in models due to computational constraints. The solution lies in developing organizing principles to relate small-scale events to larger-scale phenomena. Progress is being made in this area due to improvements in modeling and computational power, expanded observations from the ground and in situ, and increased communication and collaboration among research scientists.
Considerable attention is being directed toward the use of field observations to verify model results. In the atmospheric sciences, verification is essentially the process of establishing the accuracy of a theory to within some error estimate where the errors include those of both the computational version of the theory (the model) and the observing system. Error estimation is itself often an imprecise quantification in the atmospheric sciences because of a lack of understanding of the propagation of error in nonlinear systems and the inability to create repeatable atmospheric experiments.
• To develop a quantitative description of the processes and interactions that determine the observed distributions of water substance in the atmosphere.
The importance of water, whether vapor, liquid, or solid, in climate and weather processes is self-evident, but there are weaknesses in the current ability to specify the atmospheric water cycle. Among these are poor characterization of upper-troposphere water vapor, uncertainties in surface fluxes and precipitation efficiency, poor representation of ensemble effects of cumulus convection on the transport of water and the characterization of precipitation over oceans, and the
absence of a comprehensive understanding of the links between the atmospheric water cycle and other components of the hydrological cycle.
Fortunately, recent improvements appear able to support a comprehensive new approach to these problems. These improvements include, or will include, in situ and remote sensing methods, characterization of precipitation over the oceans, new modeling capability, and comprehensive international programs to study the hydrological cycle at regional scales.
• To improve the capability of making critical measurements in support of studies in atmospheric physics.
Many areas of atmospheric physics instrumentation lag behind what is technically feasible. Platforms suited to atmospheric physics studies are among the most needed facilities. These include new observing satellites and high-altitude, long flight time aircraft. Measurements of water substance, radiation, and trace gases are of maximum importance.
Initiatives to Support the Strategies
Implementation of these research strategies requires the following disciplinary initiatives:
• Atmospheric Radiation: to understand the interactions between radiation and components of the hydrologic cycle.
• Cloud Physics: to understand water substance interactions and processes, for example, initial droplet formation, cloud chemistry, and the interaction between radiation and clouds, such as is needed in weather and climate research.
• Atmospheric Electricity: to enable reduction of fatalities and economic losses due to lightning discharges, and to determine the usefulness of electrical activity as an observational tool for monitoring severe weather and more typical weather as well as climate.
• Boundary Layer Meteorology: to understand and make use of the knowledge of boundary layer effects on weather, climate, and human activity.
Expected Benefits and Contributions to the National Well-Being
Many components of the research program recommended here address sources of uncertainty in climate prediction. In addition to expediting the reduction of this uncertainty, other benefits will accrue, such as the improved ability to predict regional and local weather, and will aid in the development of polices to mitigate anthropogenic impacts on the environment and the management of the nation's natural resources.
Recommended Atmospheric Physics Research
Recommended Atmospheric Radiation Research
• Develop and/or test the ability of theory and models of radiation transfer to (1) understand water vapor continuum absorption and in-cloud solar absorption; (2) develop the theory of scattering by nonspherical, including irregular, particles; and (3) understand radiative transfer in cloudy atmospheres.
• Develop observational studies and analyses to (1) better utilize satellite and other remote sensor data. (2) represent the four-dimensional distribution of water vapor, and (3) quantify the direct radiative forcing of climate by trace gases and aerosols.
Recommended Cloud Physics Research
• Develop the ability to predict the extent, lifetimes, and microphysical and radiative properties of stratocumulus and cirrus clouds to (1) resolve stratocumulus-related issues as in atmosphere-ocean coupling, (2) resolve the aerosol-stratocumulus albedo feedback effect, and (3) resolve the role of cirrus clouds in global warming or cooling.
• Improve models of atmospheric radiative transfer; test these models using observations to verify radiative transfer models under different atmospheric conditions and improve parameterizations of these effects in general circulation models (GCMs).
• Increase the attention paid to clouds and their interaction with radiation to (1) permit estimation of the coverage and radiative properties of clouds and (2) improve the theoretical understanding of liquid and solid precipitation formation.
• Conduct critical tests of precipitation mechanism theory with increased attention to dynamic consequences for (1) testing models of warm-rain and ice-phase precipitation processes, (2) evaluating the effects of precipitation production and evaporation on the dynamical evolution of storms, and (3) evaluating the importance of precipitation processes on advertent and inadvertent weather modification.
• Develop the ability to predict size distribution of hydrometeors and aerosol populations to (1) determine their joint influence on the Earth's radiation balance, (2) understand their role in sustaining heterogeneous atmospheric chemical reactions and precipitation formation, and (3) evaluate the influences of microphysical processes on cloud models and the influence of clouds on climate models.
• Investigate the interactions among aerosols, trace chemical species, and clouds; develop and improve the characterization of atmospheric aerosols to (1) characterize cloud condensation nuclei (CCN) activity in chemical global models and (2) develop representations of the radiative effects of aerosols.
Recommended Atmospheric Electricity Research
• Determine mechanisms responsible for charge generation and separation in clouds to understand cloud formation mechanisms and elucidate the fundamental physics of lightning.
• Determine the nature and sources of middle-atmospheric discharges to (1) increase knowledge of these recently discovered phenomena and their possible association with severe weather and (2) explore their effects on radio propagation and atmospheric chemistry.
• Quantify the production of oxides of nitrogen (NOx) by lightning to better understand upper-troposphere ozone production or loss.
• Investigate the possibility that the global electrical circuit and global and regional lightning frequency might be an indicator of climate change.
Recommended Boundary Layer Meteorology Research
• Understand the structure of cloudy boundary layers to enable characterization of the effects of boundary layer clouds on climate.
• Improve understanding of turbulence and entrainment in the boundary layer to aid in the parameterizations of boundary layers in numerical models and to improve pollution modeling.
• Improve measurements of exchange of water, heat, and trace constituents at the Earth's surface. This is fundamental information for use in most aspects of tropospheric functioning.
• Understand model interactions of the planetary boundary layer, surface characteristics, and clouds for use in analytical and predictive models of daily temperature cycle, hydrologic studies, and pollution prediction.
• Exploit new boundary layer, remote sensors for obtaining a more complete description of three-dimensional boundary layer flow to use in direct comparisons with boundary layer simulations.
Recommended Small-Scale Dynamics Research
• Develop better representations or parameterizations of physical processes occurring on scales smaller than the grid scale in climate models to improve GCM parameterizations.
• Represent the effects of moist convection in large-scale models to improve models of potentially destructive mesoscale convective thunderstorm supercells.
• Improve the dynamical representation of small-scale features in large-scale models to permit better understanding of local severe weather, large-amplitude gravity waves, clear air turbulence, and stratospheric-tropospheric exchange.
Improvement of Capabilities
• Studies in atmospheric physics require new measurement systems such as Doppler infrared radar, polarized lidar, millimeter radar, microwave radiometers, Doppler wind profilers, and polarimetric Doppler radar.
• Develop new analysis techniques such as (1) multispectral algorithms to infer optical depth, cloud liquid water content, and trace gas concentrations from infrared spectrometers, and (2) techniques from other areas that include pattern recognition, intelligent systems, artificial intelligence, chaos theory, and computer visualization.
Atmospheric physics seeks to explain, in terms of basic physical principles, atmospheric phenomena that occur on a variety of temporal and spatial scales. Thus, atmospheric physics could be interpreted broadly as including all atmospheric phenomena. The field of atmospheric science, however, has traditionally separated large-scale dynamics (meso-, synoptic, and planetary scales) and atmospheric chemistry from atmospheric physics, and this tradition is followed here. Therefore, unavoidable overlap with material in other parts of this document is to be expected.
Areas emphasized here include atmospheric radiation, aerosol physics, the physics of clouds, atmospheric electricity, processes in the planetary boundary layer, and small-scale atmospheric dynamics. In each of these areas, there is generally a useful understanding of the physical principles involved at the most fundamental level, including the laws of motion or electromagnetics. However, understanding these physical principles alone does not ensure an adequate understanding of observed atmospheric phenomena because the various realizations of these phenomena are inherently complex and result from complicated interactions among physical processes. Further, these interactions occur across a great range of time and space scales and many small-scale processes have significant collective influence on large-scale processes. For example, consider the influence that the collective interactions between photons and cloud drops have on cloud life cycle, the influence that cloud life cycle has on synoptic development, and the influence that synoptic development has on climate. Since no large-scale model can hope to include all these processes ab initio, there is a need to develop an understanding of the collective influence of individual physical processes on larger spatial-scale and longer time-scale phenomena, an intellectual process that is called here ''the development of organizing principles.'' Current research in the atmospheric sciences is primarily concerned with finding ways to represent, understand, organize, and predict the results of these complex interactive phenomena.
Major Research Themes and Past Accomplishments
This section reviews some central research themes and recent results in atmospheric physics and identifies as challenges some key problems that impede progress. This review indicates that there have been some notable shifts in the field in the recent past. Most of them are related to the increased attention now being directed to climate studies.
For example, studies of atmospheric radiation have received increased attention as their importance to atmospheric science has become more evident. Growing attention to climate modeling has increased the need to understand radiative transfer in the atmosphere, and recent results demonstrate the influences of radiation on mesoscale weather systems and weather forecasting. The scope of cloud physics has also broadened considerably, with increased attention directed to the roles of clouds in climate, mesoscale meteorology, and atmospheric chemistry. Because of the need to represent small-scale processes in weather and climate models, there has been increased attention to the problem of parameterization, and this is now an active area of research in cloud physics, atmospheric radiation, boundary layer research, and cloud dynamics. Many of the problems being addressed in small-scale dynamics relate to the interactions between this scale and larger scales, and may also involve significant parameterization efforts. Note that "parameterization" is defined here as including two conceptually different processes. It includes using both organizing principles and empirical relationships as a way of incorporating subgrid-scale processes in models. The former implies that the underlying physics of some process is reasonably well understood but cannot be included due to computational constraints so the physics is included via statistical methods or aggregate models. The latter implies that we have a less than adequate physical understanding of the process in question so the process is included by using some appropriate set of observations. Thus, in this context, it is understood that empirical processes include some amount of curve fitting and extrapolation. In practice, most parameterizations include a blend of both of these ideas, which is the view adopted in this document.
Additionally, the simultaneous expansion of computational power and instrument performance in the past two decades has led to an increased focus on the relationship between model output and observational data sets, a process that is referred to as model validation or verification. Validation carries the connotation of establishing the legitimacy of some theory, usually by empirical or logical means. Verification has the connotation of testing the accuracy of some theory, usually by empirical means. Clearly, these concepts are closely allied. Unfortunately, atmospheric phenomena are generally so complex and span so many spatial and temporal scales that logically rigorous validation is impossible; we simply cannot compute or measure all of the relevant quantities.
Verification, as the more limited concept of testing the accuracy of a theoryparticularly if it includes testing only some aspects of the theoryis conceptually
more tractable. Despite this ambiguity in a precise definition of these terms, however, the concept of model verification by observation is deeply ingrained in current atmospheric physics and has been an important intellectual driver of research in recent years. It is likely to continue to be a strong driver. In this document, the term verification is used to denote the process of establishing the accuracy of a theory within some error estimate. It should be noted further that error estimation is itself often an imprecise quantification in atmospheric science because of a lack of understanding of the propagation of error in nonlinear systems and the inability to create repeatable experiments in the atmosphere.
Studies in atmospheric physics are improving our understanding of radiative transfer, precipitation formation, transport, and other fundamental processes in the atmosphere. Beyond this, the principal benefit to society of studies in atmospheric physics is an improved ability to predict the effects of weather and climate. The present extent of human activity has demonstrable effects on our weather and climate, so it has become imperative to understand these effects and determine their consequences. This adds urgency to the current scientific thrust to understand these phenomena. Because the consequences of weather and climate are so important to us individually and collectively, more reliable predictions of weather and climate would be of great societal and economic value.
Perspective for the Future
Possible components of a program to address the challenges listed below, and to take advantage of the opportunities presented by recent advances, are discussed in the section that follows, which concludes by focusing on three particularly important aspects of the research program recommendedaspects that should be pursued with highest priority:
1. Develop and verify an ability to predict the influences of small-scale physical processes on large-scale atmospheric phenomena.
2. Develop a quantitative description of the processes and interactions that determine the observed distributions of water substance in the atmosphere.
3. Improve capabilities to make critical measurements in support of studies in atmospheric physics.
These imperatives span the needs of all areas in atmospheric physics. Accomplishing the first two would be a significant scientific achievement that would also lead to immediate improvement in our ability to predict weather and climate.
Scientific Challenges and Questions
The discipline of atmospheric radiative transfer and remote sensing has matured greatly in the past decade, and its importance to atmospheric sciences has become increasingly evident. A fundamental interaction that must be understood if we are to develop realistic climate projections is the interactions between radiation and the components of the hydrologic cycle, particularly clouds. A number of studies have shown that radiative processes are important in the development and maintenance of convective systems and therefore have important consequences for weather forecasting and mesoscale meteorology. Satellite remote sensing is providing the meteorological community with a wealth of data on the state of the global atmosphere, while ground-based remote sensing is beginning to supply detailed, temporally continuous data that can be used to understand physical processes in the atmosphere.
Despite the maturation of the discipline, radiative transfer and remote sensing continue to present significant research challenges. Foremost is the continued dichotomy between the modeling of cloud-radiation interactions at the scale of climate models and the understanding of these same processes at the observational and cloud-scale level. Merging of these efforts will be needed before valid parameterizations of radiative transfer in a cloudy atmosphere can be developed. Secondly, the resources devoted to the development of high-quality radiation instrumentation, particularly for in situ and ground-based use, have been minuscule until very recently. At the same time, operational satellite instrumentation has been allowed to expire without adequate replacement. As a result, the quality and quantity of radiation data needed to address current problems are simply not available in many cases. Thirdly, support for basic theoretical and computational research on more realistic cloud-radiation problems has been very limited. This has hampered our ability to understand, for example, the nature of three-dimensional radiative transfer, the scattering of radiation by nonspherical hydrometeors, or the recent and past indications of anomalous absorption by clouds.
At present and in the near future, the following actions are likely to be important aspects of the study of atmospheric radiation. The challenge will be to accomplish them, or to make significant progress in these areas.
Radiation Transfer Models and Observations
The accuracy of fluxes and heating rates given by current models of radiative transfer must be verified by rigorous comparisons of computed quantities with observations at the top and bottom of the atmosphere, and with radiative divergence observations within the atmosphere, especially under cloudy conditions. Recent comparisons of terrestrial infrared model calculations with observations
have shown encouraging results for selected cases under clear sky conditions. The verification must be demonstrated for both clear sky and cloudy sky conditions and for both solar and infrared wavelengths. These comparisons are essential in order to address outstanding questions of water vapor continuum absorption and in-cloud solar absorption. A significant part of this challenge is obtaining simultaneous combined data sets characterizing the microphysical and radiative properties of clouds, especially ice clouds, and the distributions of temperature and water vapor.
Radiation Transfer Through a Medium Containing Nonspherical Particles
Although there is a complete and apparently adequate theory for scattering by spherical particles, there is no similar basis for theoretical prediction of the scattering by particles having the complex shapes assumed by ice crystals. Such a theory, spanning the relevant ranges in particle size and wavelength, is needed to assess the radiative effects of cirrus clouds and other clouds containing ice crystals. Given the difficulty of constructing a theory to cover all possible ice crystal shapes, theoretical developments will have to be combined with coupled observations of microphysics and observations to test and extend the theory.
Three-Dimensional Models of Radiative Transfer in Cloudy Atmospheres
Recent research has shown that the macrophysical (three-dimensional) variations in cloud fields may be as important in determining the radiative properties of the cloud field as are the microphysical characteristics of clouds. It is important that these macrophysical effects on radiative transfer be quantified and included in parameterizations of radiation used in weather and climate models.
Innovative Approaches to Analyses of Data From Satellites and other Remote Sensors
The volume of data produced by satellite and ground-based sensors is difficult to handle by current techniques, and the problem will escalate in the next decade. Automated systems and new techniques will be needed for routine analyses of remotely sensed data from both satellite and ground-based systems. These analyses should produce quantitative results with specified error limits that go well beyond our current capabilities in terms of retrieving geophysical parameters. There is also the opportunity to convert a number of remote sensing instruments previously used in research applications to more routine usage. These instruments include sophisticated lidars, millimeter radars, microwave radiometers, and interferometers. Examples of critical parameters they could monitor are cloud location (base and top), surface radiation budget on scales from small
mesoscale to global, profiles of atmospheric radiative heating, hydrometeor size distributions, ice and liquid water contents, and water vapor.
Four-Dimensional Distribution of Water in the Atmosphere
Because radiative transfer in the Earth system is inextricably linked to the components of the hydrologic cycle, understanding the hydrologic cycle and the resulting distribution of water vapor is crucial to a complete understanding of radiative interactions. If climate models are to represent radiative transfer properly, they must incorporate improved representations of the hydrologic cycle on scales ranging from cloud-scale and mesoscale processes to the large-scale circulation. The horizontal and vertical distributions of water vapor play critical roles in determining the radiation fluxes and heating rates, often producing preferred regions for cloud development by their dominating influence on the radiation budget. These distributions may be monitored locally from the surface by Raman scattering or differential absorption lidar (DIAL), infrared interferometers, and rawinsondes; regional and global distributions may be monitored using radiometric or interferometric measurements from satellites or using sensors carried by commercial aircraft.
Improved Understanding of the Roles of Clouds in Climate
Both model simulations and observations have revealed that cloud-radiative interactions play a significant role in climate and climate change. The fundamental physical issues in this regard represent a huge challenge that can be broken down into a number of steps such as (1) developing more physically based cloud-radiation parameterizations; (2) including the explicit treatment of cloud micro-physical and macroscopic properties in climate models; and (3) incorporating the dependence and influence of these properties on the large-scale dynamics and thermodynamics of the models. Upon successful completion of these three steps, an accurate representation of the radiative heating of the atmosphere should emerge.
Explicit efforts should be directed toward using the results of process studies, cloud-scale and mesoscale model simulations, and long-term cloud-scale data, as well as large-scale global data. This will improve our understanding of the roles of clouds in climate and the parameterization of these effects in climate models.
Direct Radiative Forcing of Climate by Trace Gases and Aerosols
The potential importance of changes in concentration and distribution of radiatively active species other than the traditional, well-mixed greenhouse gases and clouds, has become more evident in recent years. The effects of species
having short lifetimes and pronounced spatial and temporal dependencies (e.g., tropospheric and stratospheric aerosols, ozone) should be evaluated accurately in order to understand their effects on climate.
Interactions Between Radiation and other Physical Processes Such as Chemistry, Transport, and Transformation Processes
Some trace gases and aerosols have complex life cycles from the time they enter the atmosphere until they are removed. These cycles may be influenced by photochemical, chemical, and microphysical processes. They may also be influenced by local environmental conditions. It is critical to understand these life cycles if the effects of these constituents are to be reliably simulated in climate models.
Recent research has led to a substantial broadening of the scope of studies in cloud physics. Studies related to initial droplet formation, growth of rain, ice formation and growth, scavenging of particles, and other topics in traditional cloud microphysics are still being investigated, but in the past few years, increased attention has been directed to the roles of clouds in climate, to cloud chemistry, and to studies of the interactions between radiation and clouds.
At present and in the near future, the actions described below are likely to be important aspects of the study of cloud physics.
Coverage and Radiative Properties of Clouds
Cirrus and stratocumulus clouds have received particular attention over the past decade because of their important roles in the radiation balance of the Earth. Both have been the subject of intensive field campaigns and many numerical simulations. Although these studies have led to improved understanding of the nature of such clouds, they still have left important problems to be resolved. Among these are the causes of stratocumulus breakup, the quantitative factors determining entrainment into stratocumulus clouds, the factors determining ice concentrations and size distributions in cirrus clouds, and the detailed interactions with radiation in both cloud types. Additionally, for clouds that undergo substantial entrainment, cloud dynamics could strongly modulate cloud drop number concentration (CDNC), thereby helping to describe the relationship with height between cloud condensation nuclei and cloud drop number concentration. Solution to these problems appears to be feasible in the coming decade. In addition, these studies require extension to clouds of the middle troposphere, which often have a more complicated, mixed-phase structure, and to the cirrus clouds of the tropical troposphere.
Ice Formation in the Atmosphere
The factors controlling ice formation in clouds remain poorly understood despite the importance of ice for almost all aspects of cloud physics and despite much effort over the past decades directed toward understanding this problem. Neither ice nucleation nor secondary ice formation is understood as well as needed to predict ice concentrations observed in clouds from the controlling variables. Many current studies, including efforts to understand radiative effects on clouds and precipitation formation, require improved understanding of these basic processes.
Improved Understanding of Precipitation Formation
Despite good understanding of the basic processes involved in precipitation formation, we do not yet understand them in detail. Significant gaps remain in the chain of events leading to precipitation by all the major mechanisms for precipitation formation (e.g., coalescence of water drops; ice formation followed by accretion of supercooled cloud water; and the classical Bergeron process). Understanding of ice-phase precipitation is incomplete, not only because of the problem of understanding how ice originates, but also because the detailed roles of aggregation, accretion, melting, breakup, and evaporation remain incompletely understood. Although it is thought that the basic processes involved in warm-rain formation are probably understood, it is uncertain if the current knowledge of collection efficiencies and the drop breakup process is adequate. The roles of large and giant particles in warm-rain precipitation formation need resolution. Definitive tests of all aspects of precipitation formation are lacking, but data are now available to test many of these processes.
Predict Size Distributions of Hydrometeors and the Aerosols That Affect Radiative Transfer in the Atmosphere
The important links among cloud structure, aerosols, and trace gases have been the subject of intensive study in connection with acid rain, and attention has been devoted to some aspects of the connection between cloud condensation nuclei and cloud microstructure. However, with a few exceptions, this area of research suffers from the absence of a comprehensive approach that addresses all components of the problem (trace gases, aerosol, cloud structure, and radiation) as opposed to the coupling between subsets.
Aerosols have many controlling but poorly understood roles in the atmosphere. They influence the Earth's radiation balance, sustain heterogeneous chemical transformations, control most precipitation formation, and determine the microscale structure of clouds. Their influence is evident in many current studies, including those of global warming, stratospheric ozone depletion, air-
craft icing, radiative effects of cirrus, and possible effects of aircraft emissions on climate. To understand the influences of aerosols on these processes, we need a better understanding of the role of aerosols in heterogeneous chemical transformation in the atmosphere, improved definition of the role of particles in the ice formation in clouds, and an ability to predict the formation and life cycle of atmospheric particles. Aerosols thus are an important focus for future research because their study involves fundamental scientific problems that have immediate practical relevance.
The radiative effects of clouds are governed by their hydrometeor size distributions. Questions needing resolution include the following: Are there high concentrations of small ice crystals in cirrus clouds that dominate their radiative characteristics, as some remote sensing observations indicate? How important is oceanic production of dimethyl sulfide (DMS) in stratocumulus albedo? What is the source of the discrepancy between observations and theoretical calculations of cloud absorption? What are the factors controlling the concentrations of ice crystals in cirrus clouds? Thus, many of the questions involving radiative effects also involve fundamental questions of cloud-physics. The radiative effects of clouds cannot be understood, in a predictive sense, until the factors determining the size distributions are understood. In the case of cloud droplets, this is initially the CCN population; in the case of ice, it is initially the ice nucleus (IN) population. To understand how either might change as the climate changes, we need to know the sensitivity of the CCN and IN populations to other climate parameters such as global temperature, solar radiation, surface humidity, and soil characteristics. We are currently far from this understanding; there is no good explanation even for the order-of-magnitude concentration of these particles in the atmosphere, much less a predictive capability for how they might change.
Parameterizations of the Subgrid-Scale Influences of Clouds and of Microphysical Processes on Cloud Models
With increased attention directed to climate-related studies, new efforts have been devoted to developing ways to represent or parameterize the effects of clouds in global-scale climate models. These include ways of representing the radiative effects of cloud droplets or ice crystals; parameterizations of the effects of clouds on the momentum, heat, and water budgets; documentation of the connections between cloud coverage and large-scale conditions such as relative humidity; satellite observations to provide global coverage; and the development of new modeling techniques to represent clouds and liquid water in mesoscale and global models.
Global climate models must represent the effects of clouds in terms of large-scale variables, whereas the scales at which important processes occur in clouds are often many orders of magnitude smaller. The effects of clouds on large-scale
transport processes and on radiation must be represented in terms of variables that apply over distances of the order of 100 km. It is not yet clear if this gap can be bridged in a way that represents the essence of the relationships needed for climate prediction. However, there are many promising avenues of approach to this problem. Cloud-resolving models are being used to determine the influences of clouds on scales smaller than those that can be resolved in GCMs. Radiative characteristics of cloud droplets have been determined from observations in ways that provide improved representations of the present climate situation, although probably not adequate for climate prediction or understanding how CCN populations vary. Some parameterizations of ice size distributions have been attempted, but these seem particularly weak at present and are unlikely to be improved until ice-forming processes are better understood. Satellite remote sensing studies have provided global documentation of the radiative effects of clouds, cloud populations, and cloud coverage. They promise to provide still more information on hydrometeor spectra in clouds.
As the population, urbanization, technological sophistication, and economic wealth of the country and world increase, the impact of severe weather events and lightning is likely to be increasingly felt (see Part I, Table I.2.1 for data on the impact of severe weather events). Lightning is a leading cause of weather-related fatalities, and it results in huge economic losses through forest fires. power outages, and damage to computers, communications, and other electronic equipment. Determining the total electrical activity shows promise as a future observational tool for monitoring severe weather such as tornadoes, hail, flash floods, winter storms, and hurricanes.
The field of atmospheric electricity traditionally encompasses six areas of research: (1) lightning, (2) cloud electrification, (3) the global electrical circuit, (4) ion physics and chemistry, (5) ionospheric and magnetospheric currents, and (6) telluric (Earth and oceanic) currents. This assessment focuses primarily on the first three areas of research because they are the ones normally pursued in the domain of atmospheric sciences. The topic of ion physics and chemistry is discussed in the report of the Committee on Atmospheric Chemistry (NRC, 1996a). An earlier summary of research in all of these areas is found in The Earth's Electrical Environment (NRC, 1986). Although the topics have been separated here, there is considerable interdependence among them. For example, knowledge of the global lightning frequency is important for a better understanding of global electrical activity, NOx production by lightning, and perhaps global temperature.
At present and in the near future, the following actions are likely to be important aspects of the study of atmospheric electricity.
Mechanisms of Charge Separation in Clouds
Although there is now considerable laboratory, observational, and modeling evidence showing the importance of ice-graupel collisions for the electrification of clouds, there are important gaps in our understanding. We lack fundamental understanding of the physical mechanisms(s) responsible for the charge transfer, which seems to be intimately linked to the nature of the ice surface, which in turn is determined by temperature, liquid water content, particle size, and other micro-physical parameters. Improved understanding of ice particle formation and growth is important both in cloud electrification and in radiation-climate feedback and should be a high priority (as discussed further in the ''Cloud Physics'' section in Part II). Other charge separation mechanisms may also be acting. Continued observational, laboratory, and model efforts are needed on the electrification of clouds containing ice, as well as warm clouds, over geographically different parts of the globe.
Investigate the Global Electrical Circuit and Lightning as Measures of Stability and Temperature in Climate Change Studies
The global electrical circuit may prove useful in monitoring climate change. Lightning production is known to be sensitive to convective updraft speeds, which are influenced by atmospheric stability; thus, lightning rates may increase with increasing instability. It has therefore been hypothesized that global warming could be manifested by increased instability and lightning frequency. Continued monitoring of ionospheric potential, air-Earth current, and Schumann resonances may detect global trends in tropospheric stability and perhaps surface temperature and moisture as well. Global monitoring of lightning is now also technologically feasible. However, since electrical conductivity and electrification are connected to particulate concentrations, these quantities may require simultaneous measurement in order to understand the relationship between electrical circuit properties and climate.
Nature and Sources of Middle-Atmosphere Discharges
Electrical discharges above storms into the middle atmosphere (stratosphere and mesosphere) have only recently been recognized as occurring quite frequently. At least two different types of events take place, extending into the stratosphere and mesosphere. Observations are needed to understand the nature and mechanism(s) responsible for the discharges and their possible effects on radio propagation and stratospheric chemistry.
Production of NOx by Lightning
The production or loss of ozone in the upper troposphere is strongly dependent on the distribution and strength of NOx (NO + NO2) (i.e., nitric oxide + nitrogen dioxide) sources, and there is ample evidence that lightning is an important and perhaps dominant source. An immediate challenge is to assess the importance of lightning to global and regional NOx concentrations. Before the effects of emissions from present and future fleets of commercial aircraft or anthropogenic surface sources can be assessed, quantification of this natural source of NOx is essential. Proper investigation of this topic requires observational and modeling expertise in lightning physics and morphology, atmospheric chemistry, cloud dynamics, and mesoscale and global dynamics. These studies are interdisciplinary; but instruments, techniques, and models are now becoming available that can address the problem. This constitutes an important challenge with high priority in terms of understanding and protecting our atmospheric environment.
Boundary Layer Meteorology
As the part of the atmosphere in which all of humanity lives, the atmospheric boundary layer has a special relevance to our lives. The boundary layer is defined as the part of the atmosphere that is turbulently coupled to the Earth's surface and includes fields of nonprecipitating shallow cumulus or stratocumulus clouds. The boundary layer plays a central role in weather and climate because it couples processes at the Earth's surface such as evaporation and sensible heat flux with the rest of the free troposphere. This transfer of energy and momentum from the Earth's surface to the atmosphere, although complicated, is crucially important as a determinant of the behavior of the atmosphere.
Structure of Cloudy Boundary Layers
Boundary layer cloud has a climatically crucial radiative effect. It is also inextricably coupled to the turbulent and convective dynamics of the boundary layer in which it is embedded. We understand the vertical thermodynamic structure of boundary layers capped by solid stratocumulus clouds fairly well, and those consisting of shallow cumulus clouds to a lesser degree. However, we are only beginning to understand the dynamics of boundary layers that are transitional between these two types, although transitional conditions cover a substantial fraction of the subtropical and middle-latitude oceans. Over land, there are few integrated studies of boundary layer cloudiness, turbulence, and surface fluxes to compare with models. In the arctic summertime, persistent clouds form in a stable or multilayered boundary layer. One challenge is to model and develop parameterizations that realistically represent the tight coupling among clouds,
microphysics, radiation, and turbulence in these diverse boundary layers. A second challenge is to provide integrated data sets over land, over arctic sea ice, and over the middle-latitude oceans to test these models.
Turbulence and Entrainment
For unsaturated boundary layers driven by convective heating, we have a good understanding of turbulence statistics, turbulent fluxes, and large eddy structure, with good agreement between large eddy simulation (LES) models, one-dimensional models, and observations. Fundamental challenges arise when we consider turbulence and entrainment in other ubiquitous types of boundary layers. The relation between entrainment rate, turbulence characteristics, and cloud-top profiles of temperature and moisture in radiatively driven stratocumulus-capped boundary layers is still controversial. The stably stratified boundary layer is especially challenging because of intermittent turbulence, small length scales, and sensitivity to surface variability. It is also of great importance because of its propensity for accumulating trace constituents released at or near the surface.
Effects of Inhomogeneity and Baroclinicity on the Boundary Layer
We have a good understanding of how the boundary layer behaves over horizontally homogeneous surfaces, but in the real world, surfaces are almost inevitably inhomogeneous. A major challenge is dealing with this heterogeneity, which occurs due to both complex terrain and varying surface characteristics. An important practical application is developing techniques for scaling up flux estimates obtained over nonuniform surfaces to scales applicable to meso- and large-scale models. Boundary layer heterogeneity is also tightly coupled to convection. Localized downdrafts from deep convection interacting with surface fluxes create heterogeneity in the subcloud temperature and moisture field that is important in determining when and where future convection will occur. In middle-latitude cyclonic storm systems, the boundary layer often is highly baroclinic. Large-scale models do not accurately represent the vertical wind profile or surface momentum fluxes in a baroclinic boundary layer. Most boundary layer modeling studies purposely exclude consideration of baroclinicity. The challenge of the next few years is to tightly couple observational strategies, numerical modeling experiments, and parameterizations to deal with inhomogeneous and baroclinic boundary layers.
Measurements of the Exchange of Water, Heat, and Trace Atmospheric Constituents at the Earth's Surface
Surface heat and moisture fluxes are fundamental to the atmospheric heat engine on all length and time scales. Many trace atmospheric constituents are of
climatic importance: ozone, carbon dioxide, methane, and particles. Their atmospheric distribution cannot be understood without knowledge of their surface fluxes. Modeling of these fluxes is empirically based and requires accurate flux measurements. This in turn requires continued development of sensitive and fast-response sensors for direct eddy flux measurements, the development of alternative techniques for measuring fluxes when fast-response sensors are not available, and investigation of techniques for measuring turbulent fluxes remotely.
Interactions of the Planetary Boundary Layer, Surface Characteristics, and Clouds
The boundary layer over land is particularly important as the environment for most human endeavors. Boundary layer processes and boundary layer clouds have important regional impacts on the climate over land and must be considered in tandem with land surface processes that exchange heat and moisture with the atmosphere. These include vegetation and soil moisture models that are currently in rapid evolution. Turbulent processes in the boundary layer redistribute the heat and moisture and help determine the surface temperature and moisture. Clouds also affect surface temperature and evaporation, and produce rainfall. One current forecast problem is to predict the daily cycle of temperature correctly over land directly from a numerical model. This has proved to be surprisingly difficult. Flooding and hydrology are also involved, because storms both replenish soil moisture and feed off evaporated water from the soil. In the Arctic, the surface is a jumble of sea ice, meltpools, leads, seasonal snow, and tundra. The complexity of the surface is matched by a complex boundary layer microphysics, including clouds and suspended ice crystals.
Models of these interactions are interdisciplinary and require regional field experiments for their verification. Past studies in boundary layer meteorology have focused on determining the vertical profiles of wind and turbulence, surface stress, surface heat, and moisture fluxes in the boundary layer and have made substantial progress toward understanding these features. The interaction with clouds and with various land surfaces is the frontier on which boundary layer meteorology should focus in the next decade.
Small-Scale Atmospheric Dynamics
It would be rash to conclude that we have learned everything we need to know about small-scale circulations in isolation from other influences. Nevertheless, it seems clear that some of the most interesting research opportunities relating to small-scale circulations lie in questions of the interactions of such circulations with other processes. The following are some examples: How do the large scales control such processes as tropical cyclogenesis and mesoscale convective system formation? How can gravity wave drag be realistically parameterized
into global circulation models? How do convection and the resulting precipitation at a given location affect the prospects for future convection in this area? The last question seems particularly pertinent to the development of the floods in the upper midwestern United States during the summer of 1993.
Most current research on small- and middle-scale atmospheric dynamics is related to (1) moist convection and mesoscale convective systems, (2) fronts and middle-latitude cyclones, (3) tropical cyclones, and (4) topographic and other surface-induced flows. At present and in the near future, the following actions are likely to be important aspects of the study of small-scale atmospheric dynamics.
Effects of Moist Convection in Large-Scale Models
We have witnessed a long series of field programs focused on the challenge of determining the structure and evolution of convection and the mesoscale systems in which convection is embedded. These have occurred in a wide variety of geographic settings (e.g., the continental high plains, the western mountains, southern U.S. coastal regions, the tropical oceans), and we have a reasonably good idea of the morphology of convection around the world. Numerical modeling of convection saw its first success in helping to understand the supercell thunderstorm. Subsequent work has elucidated the crucial role of the cold pool and gust front in multicell storms. Models are now reaching the stage where agglomerates of thunderstorms, known as mesoscale convective systems (MCSs), can be simulated.
Large-scale models must correctly incorporate the effects of convection using convective parameterizations. One of the major challenges is to incorporate our current understanding of convection into the development of improved parameterizations. Attempts to do so reveal limitations in our understanding that must be filled by a combination of observations and high-resolution cloud modeling. Fundamental questions about issues such as the development, distribution, and evaporation of precipitation; entrainment and detrainment under various environmental conditions; the transport of heat and momentum by convection; and the control of convective initiation and amount by environmental factors have to be answered.
Dynamical Representation of Small-Scale Features in Midlatitude Cyclones
Quasi-geostrophic theory and semigeostrophic theory have been very successful in explaining the gross features of middle-latitude cyclones and associated fronts. Analyses using the operational network and special field programs have verified these ideas. The current focus is on smaller-scale features of these disturbances. The three-dimensional structure of fronts, including the study of
frontal waves and small-scale unbalanced effects, is a topic of current interest. The relative contributions of adiabatic dynamics, surface fluxes of energy and momentum, and latent heat release in fronts and cyclones are also being investigated. The truly intense parts of cyclones are often small in scale and therefore cannot be described adequately by conventional balance models. There is some interest in using more accurate balance schemes, such as nonlinear balance, in describing these systems. The breakdown of balance and the production of gravity waves are topics that are beginning to be addressed. These studies could lead to better understanding of local severe weather, large-amplitude gravity waves, clear air turbulence, and stratospheric-tropospheric exchange in tropopause-fold areas.
Incorporation of Surface-Induced Flows into Large-Scale Models
The challenge of replicating the flow over and around various mountain ranges has been addressed by several observational programs. Theoretical and modeling studies have progressed from highly idealized two-dimensional calculations to three-dimensional numerical simulations taking into account nonlinearities and surface fluxes. Some agreement is often seen in comparisons with observed mountain-wave and lee-wave structures. Progress is also being made in understanding the process of lee cyclogenesis, which often occurs on a relatively small spatial scale. The challenge in all of these cases is to incorporate our knowledge of these phenomena into large-scale models.
Disciplinary Research Challenges
The topics discussed here are suggested components in a research program for the next decade. They address some of the challenges in current research and arise from opportunities presented by recent developments in research and technology. After each is presented briefly, some broader aspects considered of highest priority are discussed at the conclusion of this Disciplinary Assessment.
Most of the research topics discussed here are intrinsically cross-disciplinary. Indeed, many of the opportunities involve studies of the interactions between processes that have conventionally been studied in isolation. This is especially true in cloud physics, where the interactions of clouds with radiation, trace chemical species, atmospheric dynamics, and electrification are among the focal points. Nevertheless, it is important to maintain a balance that advances the traditional bases of these disciplines while developing the new themes. There are significant deficiencies in our understanding of the traditional topics that will surely limit our ability to conduct these new cross-disciplinary studies, and such barriers can be removed only if we continue to direct a significant fraction of our attention and resources to the solution of the fundamental problems.
Develop Adequate Representations or Parameterizations for Physical Processes Occurring on Subgrid Scales in Climate Models
General circulation model (GCM) calculations currently use prognostic variables having grid spacings of the order of 100 km. Significant fractions of the fluxes of mass, humidity, energy, and chemical constituents occur at smaller scales, so the effects of these subgrid-scale (SGS) transport processes must be represented parametrically in GCMs. Calculations of radiative transfer must also consider SGS distributions of cloudiness and other inhomogeneities in the atmosphere and must represent the effects of irregular ice crystals in cirrus clouds in terms of the GCM variables. Boundary layer fluxes are dependent on the nature of the land surface, which is often quite variable within the grid box of a GCM. Transformations among the phases of water, with associated heating and cooling effects, also occur primarily on scales smaller than the grid spacing. The parameterization problem is thus to represent these SGS processes as functions of the GCM prognostic variables (and perhaps other available information such as geographic location and season).
Improve, Test, and Verify Models for Radiative Transfer in the Atmosphere Using Observational Data
Over the past 30 years, remarkable progress has been made in our ability to model radiation processes during clear weather. However, there are serious deficiencies in our understanding and modeling of radiative fluxes and heating rates, especially in cloudy atmospheres. This, in turn, affects our understanding and modeling of cloud-radiative interactions and the role of clouds in climate.
The principal reasons for these deficiencies are inadequate representation of the spatial and temporal variabilities of the atmosphere, cloud cover, and cloud microphysical and macrophysical properties. Recently, some innovative work has begun to address aspects of these deficiencies, but a great deal of effort remains. Work must move forward with a combined program of modeling and observation to address these issues simultaneously.
Recent innovations in radiative transfer modeling include a variety of approaches to three-dimensional radiative transfer and improved treatment of combined gaseous absorption and particle scattering. In addition, there is considerable interest in applying new insights gained from fractal mathematics to the study of cloudy environments. These new models suggest that there may be the theoretical tools to treat radiative transfer in actual three-dimensional cloud fields. However, the science is a long way from having adequate models and testing such models in real environments. To test these models, it would be necessary to make simultaneous measurements of both the three-dimensional cloud field and its properties and the resultant radiances and irradiances. A variety of sophisti-
cated observing tools, particularly millimeter-wavelength radar, hold the promise of providing the necessary information.
The highest priority is to establish the validity of both the input quantities and the related three-dimensional model calculations. At the present time, the most pressing problem is a lack of sufficiently sophisticated radiation instrumentation and platforms. Once there is an accurate assessment of the ability of three-dimensional models to compute radiative transfer, these models can be used to assess the performance of the simpler models that must necessarily be employed in weather and climate activities.
Develop an Ability to Predict the Extent, Lifetimes, and Microphysical and Radiative Properties of Stratocumulus and Cirrus Clouds
We are well positioned to improve parameterizations of marine stratocumulus boundary layer cloud from the insights and data gained from past field experiments in the subtropics and a proposed Arctic experiment in 1997. In particular, the connection between the vertical structure of the boundary layer and the type and amount of cloud cover is becoming much better understood. It will be particularly useful to compare new GCM parameterizations against these regional data sets as well as global data sets such as those from the International Satellite Cloud Climatology Project. Large eddy simulations also are beginning to show skill in predicting how cloud and boundary layer properties depend on large-scale variables and may prove useful in parameterization development. Further focused field research is needed on specific issues such as entrainment, where different models are in disagreement and where new instrumentation should allow us to more tightly test model predictions.
Improved boundary layer cloud prediction should help resolve some important climate modeling problems. Major failures of present coupled ocean-atmosphere models can partially be traced to deficiencies in predictions of boundary layer clouds. For instance, "full-physics" coupled models currently cannot maintain realistic E1 Niño oscillations, partially because they do not predict a large expanse of low cloud off the coast of South America that serves as a "refrigerator" for the eastern subtropical Pacific. Over the Arctic, climate models have predicted a large warming that has not been observed. Interactions between persistent boundary layer cloud and sea ice may be responsible.
The indirect climate effect of aerosol through its impact on the microphysics of boundary layer clouds should also continue to be a particularly fruitful research area. Here, small-scale modeling and more observational work are necessary before any climate model can reliably incorporate this aerosol-cloud-albedo feedback effect.
The properties and radiative feedbacks of boundary layer clouds over land have not been sufficiently studied either in the field or in detailed models, and
such studies should provide an important opportunity for progress in the next few years. Here, better observations and more modeling studies on all scales are required.
Cirrus clouds are recognized as important components in the Earth's climate system. Their direct radiative effects can act to either cool or warm the planet, depending on the relative values of solar and infrared optical depths. Recent research has shown that tropical upper-tropospheric cirrus clouds play more subtle, though possibly critically important, roles in determining the vertical distribution of water vapor throughout the tropical atmosphere and the strength of tropical atmospheric circulation systems; these sensitivities are apparently manifest through a modulation of the static stability and resulting increase in convective motions and moistening of the upper troposphere by evaporation of cirrus cloud ice. It is very important to capture the essence of these tropical upper-tropospheric cirrus cloud systems if we are to simulate the climate successfully. We have the opportunity within the next decade to gain significant understanding of the evolution of these systems and their linkages with other processes in the climate system.
Tropical cirrus systems requiring investigation may be grouped into three categories: (1) convective cirrus, which is in close spatial and temporal proximity to the convective systems that produce it; (2) detached anvil cirrus, which can be identified with its convective source but has become spatially detached from the convection and takes on an evolution of its own; and (3) subtropopause cirrus, a spatially pervasive layer of optically thin aerosol detectable a high percentage of the time at tropical and subtropical latitudes. Each of these three cirrus systems has distinct evolutionary cycles that must be defined, described, evaluated in terms of climate sensitivity, and incorporated into climate models where appropriate. Advances in in situ and remote sensing instrumentation from aircraft, ground, satellites, and aircraft platforms scheduled to become available in the late 1990s, coupled with models capable of simulating convective-scale systems, provide the opportunity for progress in describing and understanding these climatically important systems.
Investigate the Interactions Among Aerosols, Trace Chemical Species, and Clouds; Develop and Improve Characterization of Atmospheric Aerosols
There are a number of feasible research objectives to scope the relationships between aerosols and their interactions with trace chemical species and with clouds. One would be the development of a predictive capacity of the concentrations of soluble aerosols that are active as cloud condensation nuclei. This capability would then be applied in an atmospheric global model that contains an atmospheric chemistry module and accounts for the effects of other aerosol processes as well as the aerosol effect on radiative forcing.
Another research objective would be to develop representations of the radia-
tive effects of aerosols, suitable for incorporation into a climate model, that are interactive with the atmospheric chemistry in the model. This would require information on CCN (above) as well as the radiative properties (particularly absorption) of the aerosols.
Additional research objectives would include modeling and documenting the effects of heterogeneous reactions in the major atmospheric chemical cycles and determining how these reactions are influenced by aerosol populations and concentrations.
Increasing international sensitivity to the results of extensive biomass burning suggests that research attention should be directed toward determination of the magnitude and contribution of biomass burning to the global aerosol population and to CCN populations. Additionally, it is necessary to document the characteristics and lifetimes of aerosols in the upper atmosphere.
Other field studies have been outlined by the International Global Atmospheric Chemistry (IGAC) Project and in the document A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change (NRC, 1996a).
Determine the Sources of Ice in the Atmosphere
Despite the importance of understanding ice formation in the atmosphere, relatively little recent effort has been devoted to studies of ice nucleation. This is a problem ready for a fresh approach. Nucleation processes responsible for cirrus formation are largely unstudied because of the unavailability of suitable research platforms and instruments, but this situation promises to change with the future availability of new high-altitude research aircraft. Global aerosol models offer the possibility that the contributions of desert dust aerosols to ice nucleation might be assessed. Perhaps the foremost need is for development of suitable instrumentation to support these studies in the laboratory as well as in field experiments. A practical goal would be to document the role of aerosol particles in ice formation in some of the simpler cloud systems, including widespread cirrus and upslope stratiform clouds, and to learn the origins of particles responsible for nucleating the formation of ice in conditions, including both homogeneous and heterogeneous nucleation.
Quantify and Parameterize Surface Effects on Atmospheric Dynamics
The boundary layer transfers heat, moisture, and momentum between the surface and the free troposphere, acting as a valve and a reservoir for these quantities. Energy and momentum fluxes from the surface are crucial to many processes in the atmosphere. For many purposes, existing bulk flux formulations valid at moderate wind speeds yield adequate treatment of surface fluxes for large-scale models over the ocean. Recent work has shown how surface heat, moisture, and momentum fluxes behave at low speeds over the ocean. However,
major uncertainties still exist for the high wind speeds experienced in tropical storms. The character of these systems depends critically on the relative magnitude of the moisture and momentum exchange coefficients. Over land, the wetness of the soil, terrain variability, and the nature and distribution of the underlying vegetation play major roles in determining surface fluxes. These fluxes are crucial to large-scale dynamics but may play a role on the mesoscale as well.
Feedback by convection onto the boundary layer in the form of moist downdrafts is a crucial process. These downdrafts temporarily suppress further convection, but also enhance the heat and moisture fluxes from the surface and introduce horizontal variability in the fluxes. In this way it appears that the boundary layer plays an important role in the control of deep convection in the atmosphere.
Exploit New Remote Sensors to Broaden the Scope of Boundary Layer Studies
New developments in lidar and radar technology have given us the prospect of remote measurements of velocity and scalar fields from ground-based and mobile platforms. Doppler lidar can resolve mean and turbulent fluctuations in the radial velocity throughout the clear boundary layer. Doppler radar can provide detailed radial velocity, velocity variance, and reflectivity fields in boundary layers, which can be used to examine turbulence throughout clear and cloudy boundary layers, including the entrainment zone at the top of the boundary layer. Analysis of the velocity spectrum of short-wavelength radars (e.g., 8 mm wavelength) can provide the height profiles of drop size distributions for those drops having substantial settling velocity. This remote sensing technology could be used to provide a more complete description of three-dimensional boundary layer flow, including vertical profiles of momentum flux, velocity variance and higher-order moments, length scales, and vertical coherence of turbulent eddies in both cloudy and clear boundary layers. These data sets can then be used for more direct comparisons with numerical simulations than are possible with in situ measurements. However, this opportunity presents challenges. New strategies for digesting such large and complex data sets have to be developed so that they can be used efficiently to address such problems as comparisons with, and validation of, numerical simulations and visualizing in quantitative ways the three-dimensional morphology of turbulent structures.
The development of lidar techniques for estimating concentrations of trace gases in the boundary layer (e.g., by differential absorption lidar or Raman scattering) offers the opportunity to obtain vertical cross sections of, for example, water vapor or ozone, which could be used to study how these trace species diffuse, especially in cases of horizontal inhomogeneity. When combined with Doppler lidar, it may be possible to measure vertical flux profiles either from the ground or from aircraft, and thus to estimate surface exchange, entrainment rates,
and in the case of chemically active species such as ozone, the photochemical source-sink term in the species budget.
Another opportunity is provided by the development of boundary layer wind profilers, which may be used in networks to measure boundary layer height and vertical profiles of wind, temperature, and possibly fluxes of heat and momentum. This may be particularly applicable to addressing problems of horizontally inhomogeneous flow on scales larger than those that can be addressed by a single ground-based scanning lidar or radar. Airborne lidars and millimeter Doppler radars also provide new opportunities for studying heterogeneity in clear and cloudy boundary layers.
Utilization of the global positioning system (GPS) for improved aircraft navigational accuracy can be combined with a Doppler laser to provide a major enhancement in the accuracy of wind measurement from aircraft. This has the potential to provide measurements of flow divergence and of mean flow perturbations associated with mesoscale phenomena such as land or sea breezes and flow over variable terrain, more accurate eddy flux and coherent eddy structure measurements, and measurements of wind shear over lengths of the order of ten meters. GPS also may have other important applications. For example, it can be used as a basis for airborne air motion systems that are smaller and less expensive (although less accurate) than inertial navigation-based systems, so that more systems can be available for multiplatform experiments.
Although the preponderance of development currently seems directed toward remote sensing, there are many possible developments in direct sensing that are important. An ever-increasing number of trace constituents can be measured with sufficient sensitivity and time response that they can be used for direct eddy flux measurement. Advances in this area include radiatively important species such as methane and ozone. Several of the species for which sensor technology has advanced to where eddy flux measurements are becoming possible have application as tracers for atmospheric processes such as entrainment and diffusion. At the same time, alternative flux-measuring techniques that do not require as fast a response as eddy correlation are being developed. These include devices that control the flow or accumulation of air in a reservoir according to the vertical air velocity. In this way, the requirement for fast response is placed on the collection strategy and not on the sensor, which broadens the class of species whose flux can be measured.
Investigate the Interactions of Small-Scale Circulations with Larger-Scale Processes
It would be rash to conclude that we have learned everything we need to know about small-scale circulations in isolation from other influences. Nevertheless, it seems clear that some of the most interesting research opportunities relating to small-scale circulations lie in questions of the interactions of such circula-
tions with other processes. Some examples follow: How do the large scales control such processes as tropical cyclogenesis and the formation of mesoscale convective systems? How do cloud physical processes affect the structure and evolution of convective systems? How can gravity wave drag be realistically parameterized into global circulation models? How do convection and the resulting precipitation at a given location affect the prospects for future convection in the area? The last question seems particularly pertinent to the development of the floods in the upper midwestern United States during the summer of 1993.
Large-scale motions are nearly balanced. Whether such motions can be characterized by quasi-geostrophic theory or whether they require something more complex such as semigeostrophic or nonlinear balance, potential vorticity dynamics and the invertibility principle apply. In this picture the prognostic nature of the theory is encapsulated in the advection and nonadvective changes in the potential vorticity and surface potential temperature fields. A knowledge of these fields is then sufficient to obtain all other fields of dynamical interest by the inversion process. Given this picture, the lasting effects of small-scale circulations on large-scale motion are limited to the changes they induce in the potential vorticity and surface potential temperature fieldsall other changes are transient and, therefore, of considerably less interest. Recent theoretical work has shown how diabatic heating, friction, and the turbulent transfers of heat and momentum generate nonadvective fluxes of potential vorticity. This work provides a useful framework for viewing the action of smaller-scale processes on the large scale. In the free atmosphere, two types of small-scale phenomena can cause significant nonadvective transport of potential vorticitygravity waves (and possibly associated shear instability) and moist convection. We now discuss these processes and other effects of convection, such as moisture transport, that have indirect dynamical significance.
Gravity waves are produced in the atmosphere by flow over terrain, by convection, by shear instability, and possibly by geostrophic adjustment. These processes are reasonably well understood with the exception of geostrophic adjustment. Once produced, gravity waves transport momentum through the atmosphere without depositing it until they dissipate. Thus, gravity waves are agents of ''action at a distance'' in the atmosphere. The propagation of gravity waves is far from simple; multiple reflection and refraction take place. Wave breaking and dissipation are complicated, so the momentum transports are complex and subtle. Unfortunately, although very challenging, this area does not seem to offer much opportunity for great progress in parameterizing the effects of gravity waves on large-scale circulations. Although we should continue to look for fundamental advances in this area, it is perhaps most realistic to try to set bounds on the effects of gravity waves on the large scale. Potential vorticity dynamics should be useful in this respect, because it shows how isolated gravity wave breaking and the resulting deposition of momentum affect the large-scale flow.
The opportunities for producing a reasonably accurate moist convective parameterization look somewhat brighter than they do for gravity waves, mainly
because convection acts locally. (This, of course, ignores the generation of gravity waves by the convection itself.) The convective parameterization problem splits naturally into two partsthe control of convection by the large-scale flow and the action of this convection back on the large scale. Many schemes have been proposed for resolving the control aspect of the problem. This issue remains controversial, but it is one with many opportunities for progress in the next decade. An opportunity to resolve the lively debate between the "convergence causes convection" school and the "instability causes convection" school may lie in examination of the results of existing field experiments such as the Tropical Ocean Global Atmosphere (TOGA) Coupled Ocean-Atmosphere Response Experiment (COARE). Also important to this effort is increased understanding of the types of convection that result from particular kinds of large-scale situations. For instance, what effects do varying degrees of environmental shear and midlevel relative humidity have on the character of the convection? The compilation of individual case studies, as well as composite analyses and cloud models, are all required to fill in this picture.
Convective ensemble simulations of convection should also be helpful. Such simulations differ from the usual type of calculation in that the convection is allowed to develop naturally from large-scale forcing, rather than being initiated by a buoyant bubble or some other imposed feature. They are computationally expensive, but the increasing power of computers over the next decade may make them more feasible. Understanding the action of convection on the large scale requires a reasonably accurate model of how convection works. The traditional picture of convection as an ensemble of entraining plumes has been challenged by models in which air moves both up and down under the influence of condensation and evaporation of cloud particles. More work is needed to clarify this picture, particularly since the study of convection has been hampered by inadequate observations.
A particularly difficult aspect of convective action on large scales is in the realm of convective momentum transfer. It has been shown that convective systems transfer momentum up the gradient and sometimes down. The net effect is uncertain. A great deal of work has been done to document the stratiform rain areas that are often associated with agglomerations of convection. The updrafts and downdrafts in these systems are very different from those of the convection itself. The relative strengths of the convective and stratiform parts of MCSs are known in the average sense for particular geographical areas, but variations in this ratio with environmental conditions have to be better understood.
Moisture has indirect dynamical significance due to the energy transformations associated with phase transitions between vapor, liquid, and ice. However, in other respects it acts like other trace constituents of the atmosphere in that convective motions distribute it vertically. (It also differs from other constituents in that precipitation processes remove it from the atmosphere.) Convective transport and precipitation of water are particularly important to large-scale dynamics
because of the diabatic processes with which they are associated. The convective fluxes of water substance are poorly understood. This situation occurs partly because the dynamics of clouds are still somewhat uncertain, but it also stems from our lack of understanding of cloud microphysical processes in the complicated context of convection.
Conduct Tests of Current Understanding of Major Precipitation Mechanisms and Their Dynamic Consequences; Improve Ways of Representing These Processes in Parameterized Form
Cloud modeling and observational capabilities have progressed to the point that critical comparisons of predictions with theory are possible. The sensitivity and availability of polarization diversity radars, when combined with trajectory calculations based on measured, modeled, and retrieved fields of cloud properties and on current and emerging capabilities for detailed microphysical modeling, coupled with realistic cloud dynamics, make such a critical comparison possible. Some attempts at determining if the speed of warm-rain formation is consistent with current theoretical predictions are now planned, and further investigations directed at similar comparisons in other locations and for other precipitation mechanisms are now possible. Such comparisons are necessary to develop confidence in current theory and modeling or to learn where improvements are needed. They also contribute to developing improved parameterizations for the precipitation process suitable for use in situations where complete simulation of the microphysical details of this process is not practical.
The speed with which precipitation forms not only determines the precipitation efficiency of many cloud systems but also influences dynamical and radiative processes by affecting the distribution of condensate, the lifetime of the systems, and the size distribution of hydrometeors. The two principal mechanisms for rain formation, the warm-rain and ice-phase precipitation processes, are both understood only partially, and hence difficult to parameterize from current knowledge of the fundamental processes. There is a need for critical observational data against which to test key aspects of currently accepted theory. For example, the collision efficiency for collisions among water droplets determines the speed of precipitation formation via the warm-rain process, and rates of secondary ice production influence the concentration and sizes of thunderstorm ice crystals that enter anvil regions. One realization of this goal would be to verify the predictions from cloud models of the warm-rain process or of anvil generation.
The production and evaporation of precipitation influence the dynamic evolution of precipitation storms. The factors that control the partitioning of precipitation into convective and stratiform components are not well understood. This partitioning is important because convective and stratiform precipitation are subject to very different fates. The former generally falls out with less evaporation,
whereas the latter is often carried tens or hundreds of kilometers from the generating cloud and is subject to a great deal of evaporation from the freezing level downward. In addition, a fraction of the stratiform condensed water is in the form of small ice crystals that are important to the radiation balance of the atmosphere. The evaporation of rain is of extreme dynamical importance, and its rate depends on the size distribution of the raindrops, which is quite variable.
Developing a verified ability to predict rain production is also an important step toward successful weather modification and toward understanding the inadvertent weather modification that results from anthropogenic emissions. Weather modification research appears to have stagnated. Demonstration of a better understanding of the natural precipitation process, in a verified quantitative model, could revive this area of research by providing a basis for developing and testing hypotheses and for assessing the likely effects of modification programs. One additional consequence of this effort would be an updating of parameterizations of precipitation formation, which in most cases continue to be based on the Kessler parameterization or the parameterization determined by Berry and Reinhardt by integration of the stochastic coalescence equation. Neither has been verified by comparison to experimental data, and both are based on theoretical formulations of the coalescence process that are now outdated. Because various forms of these parameterizations enter many climate calculations as well, verifying or improving them would have widespread applicability in the modeling of precipitation-producing systems.
Determine the Utility of Lightning Observations and Measurements of the Global Electrical Circuit as Proxy Atmospheric Data
The lightning location and detection systems that have become operational in the past decade have been an invaluable aid to forecasters in tracking the motion and intensification or decay of storms, particularly in the western United States where radar coverage is incomplete. However, case study research suggests that much additional information about storm behavior and the likelihood of severe weather events, such as tornadoes, flash floods, winter storms, and hail, might be obtained by combining radar and meteorological data with continuous measurements of both intracloud and cloud-to-ground lightning flash rates, the former being more difficult to measure. To be of most benefit for "nowcasting" and warnings, the combined meteorological-electrical behavior of a statistically significant number of different types of storms should be determined. The needs for forest fire forecasting in the West are very different from the needs for tornado forecasting in the Midwest, or for protection of the power network in the East. A significant number of agencies and industries could benefit from increased use and availability of lightning data.
Another recent development suggests that lightning systems operating at very low frequencies can provide lightning detection over a much longer distance
than the present National Lightning Detection Network. Such systems may provide information over much of the globe from a few remote sites. For example, the intense deep convection associated with tropical cyclogenesis over the eastern Atlantic Ocean could be detected by a couple of widely spaced stations in North America. This detection could precede by several days the arrival of a full-fledged hurricane at the Florida coast.
Lightning frequency and the electrical circuit of the globe may also prove valuable as indicators of climate change. There has been great concern and debate over the possibility of global warming due to increases in greenhouse gases, but it is very difficult to obtain reliable measures of changes in global temperature. The global electric circuit, an old and well-established concept, affords a new approach to the issue of global change by virtue of its natural integration of the electrical contributions of weather worldwide. The empirical argument for global circuit sensitivity is that increases in surface air temperature create changes in buoyancy [Convective Available Potential Energy (CAPE)], that enhance cloud updrafts, leading to increased electrical activity. The global electric circuit thus provides a mechanism for monitoring the entire planet from a single, or a few, measurement sites. Available measurements suggest that the global circuit is indeed positively correlated with temperature. However, additional investigations are needed to demonstrate the utility of using global circuit measurements. Another potential link between global climate and electrical activity is the possibility that if ice production and charge separation are associated, cirrus cloudiness and electrical activity may be correlated. Clearly, opportunities for progress exist and further studies are warranted.
Determine the Mechanisms of Charge Generation in Clouds, Middle-Atmosphere Discharges, and the Propagation of Lightning
Important progress toward understanding cloud electrification and lightning propagation has been made in the past decade through observational, laboratory, and modeling studies. In regard to charge separation in clouds, these studies continue to show the importance of the development of ice and particularly of graupel, which requires stronger updrafts for growth, to the electrification process. Laboratory studies have also demonstrated the importance of the collisions of smaller ice particles with simulated graupel in separating significant charge. Despite this progress, fundamental questions remain with regard to the basic physics of the charge transfer process, and progress has been limited by the need for more and better observations of the electrical (particularly particle charge) and simultaneous microphysical and dynamical structure in a wide variety of clouds.
In situ measurements of electric fields and particles, lightning location systems, networks of electric field meters to identify lightning discharge locations and characteristics, and polarization Doppler radar measurementsall provide
valuable opportunities to obtain information on cloud electrification and micro-physics. Even some simultaneous measurements of charge, size, and particle-type characteristics are beginning to become available, which is an important parameter in testing different theories of electrification. Although each of these observations requires care and scrutiny, they are now possible. It is time to coordinate these different measurements and to make the measurements in a larger variety of clouds in different geographical regions of the world. In view of the importance of tropical thunderstorms to the global lightning budget and climate change, it is of particular importance to give added emphasis to clouds in the tropics.
In the past decade, there has been considerable progress toward characterizing the peak current, rise time, electric fields, maximum voltages, and other physical properties of the lightning discharge itself. The results of this research show that the rise times and currents are even shorter than expected and that they create quite large electromagnetic disturbances. One of the difficulties encountered in the course of this research has been in simulating the lightning stroke. The combination of high-current and high-voltage discharges over long distances has not been adequately duplicated.
Although there have been significant advances in understanding the properties of lightning, there has been relatively little work on the fundamental physics. The reasons for the initiation point of discharges in clouds are at best poorly understood. The geometrical development of discharges is largely unknown. The radiation patterns, microscale processes in the channel, chemistry in and near the channel, and attachment processes that are responsible for damage are all areas in which opportunities exist for important advances. New initiatives in the development of physical models are needed to help interpret the observations that are available.
Ordinary lightning is a large-scale electrical discharge commonly found within and beneath thunderclouds and confined to the troposphere. In recent years, at least two exceptionally large-scale discharges occurring above the top of large thunderstorms have received considerable attention. One of them, which has been called a stratospheric discharge, is initiated within intense thunderstorms and propagates upward into the stratosphere. The other phenomenona weak, luminous discharge called a spritehas a reddish hue with the greatest intensity in the mesosphere in the vicinity of 50 to 80 km height. This discharge seems to be associated with the mature phase of large intense convective complexes when positive cloud-to-ground lightning is present, as in the storms responsible for the flooding of the Mississippi during 1993. The behavior and morphology of both of these phenomena are poorly understood. Since the impact on the atmosphere is virtually unknown, they offer both a challenge and a singular opportunity to increase our knowledge of a newly discovered fundamental phenomenon. Although the expected temperature increases are not large, the affected volumes of the sprites are large, and the potential impact on atmospheric
chemistry via NO production or alteration of the mixing ratios of other species requires investigation. Likewise, investigation of the spectral characteristics of these discharges is needed.
A better understanding of cloud electrification could lead to a much better ability to predict when and where clouds are likely to become electrified. One example of the potential economic benefit of better understanding of cloud electrification is in the launch of space vehicles, which is often postponed because of lightning, at the expense of about a million dollars per day. Lightning threats to aircraft also result in significant course deviations and costs for commercial airlines. Low-voltage devices are sensitive to low-level transients, and there is concern about the use of composite materials in commercial airplanes because of their susceptibility to lightning damage. Thus, better understanding of electrical phenomena in the atmosphere would have widespread benefits, not only through reduced hazards to humans but also through direct economic payoffs.
Determine the Rate of Generation of NOx by Lightning
The concentration of NOx (NO + NO2) is one of the major factors in determining whether there is a net in situ loss or production of ozone in the troposphere. Ample field and laboratory evidence shows that lightning is an important source of NOx, but the impact on a global scale remains uncertain` Current estimates of the global production rate of NOx by lightning range between 2 and 200 Tg N/yr. However, models have not accepted the higher estimates and usually restrict the source strength to 3 to 5 Tg N/yr. Even these lower estimates are comparable to, or larger than, source strengths in the middle and upper troposphere from stratospheric exchange and subsonic aircraft, and the upper estimates are comparable to anthropogenic sources in the boundary layer. Likewise, lightning is reported to produce other chemical species in lesser concentrations, and further studies are needed here as well. The newly recognized phenomena of middle-atmosphere discharges, discussed above, may also have an impact on the chemistry of the middle atmosphere.
Determining the importance of lightning as a source of NOx from observations is difficult, because it requires not only measurements of NOx within or near storm systems but also characterization of the number and type of lightning flashes within a given area and, for some regions, discharges. It also requires knowledge of the transport and evolving chemistry of the species from the cloud to the regional and global scales. Although this may be a difficult problem, it is one that is tractable and provides an opportunity for real advance. Most of the technical and modeling capabilities are in place, and relatively minor improvements to others would be necessary. Chemistry instruments to measure the proper species, airborne platforms from which measurements can be made, Doppler radars for cloud motion measurements, lightning interferometers to identify location and type of lightning in conjunction with surface networks of electric
field sensors, and chemistry and dynamical models on different scales to examine the transport, transformation, and redistribution of the species are now available. One area in which measurement capability is lacking involves the global frequency and location of lightning, which has been discussed above. The main challenge of this problem is to combine and focus the necessary expertise in several interdisciplinary areas.
Improve Observational Capabilities in Support of Studies in Atmospheric Physics
Although some studies of the atmosphere use advanced, modern, state-of-the-art instrumentation to make reliable measurements, there are many glaring weaknesses in our overall ability to measure atmospheric characteristics in support of the studies outlined above. We should learn from the great amount of time wasted in trying to interpret inadequate measurements that it would be more efficient to devote a larger fraction of the community's resources to instrument development, characterization, and improvement. There are numerous examples where current measurement capabilities are inadequate but technical solutions are possible. Yet the atmospheric science community devotes only a very small fraction of its effort to instrument development. Universities are hampered by lack of the long-term funding commitments needed to conduct such research. Private companies do not fill the gap because of the small market. The National Center for Atmospheric Research and other national facilities are always pressed to support a high level of deployment, but have limited personnel and resources to devote to instrument development. The result is seriously inadequate attention to this problem by the atmospheric physics community, perhaps contributing to a community shift toward computer simulations in which technological progress is impressive and new frontiers open every few years. This problem strikes at the infrastructure of the sciencethe ability to test theoretical understanding with observations. A substantial increase in the overall level of effort devoted to instrument development is needed, especially in the National Science Foundation-supported community. This is a clear "imperative" and is discussed further below.
Develop New Analysis Techniques for Meteorological Data
There is an opportunity to make an initial assault on many of the research challenges identified in this report by using the large data sets previously collected by satellite sensors and in-process experiments. We must seek new ways to express and visualize these data. Examples include multispectral algorithms to infer optical depth, cloud liquid water content, and trace gas concentration from infrared spectrometers. A second opportunity stems from advances made in research applications of remote sensing from the ground. These research tools
could be configured to provide routine data on variables important to the parameterization and validation of various physical processes simulated in regional and climate models. Candidate instrumentation includes Doppler infrared lidar, polarized lidar, millimeter radar, microwave radiometers, infrared and solar interferometers, Doppler wind profilers, and 3 cm and 10 cm polarimetric Doppler radars.
At the same time that we seek to optimize the utility of existing data and instrumentation, we must continue an ambitious program of development of new instrumentation if we are to meet the challenges of the next century. The advent of the Department of Energy's (DOE's) Atmospheric Radiation Measurement (ARM) program, the planned launching of the National Aeronautics and Space Administration's (NASA's) Earth Observing System (EOS) platform in 1998, and the new interest in unmanned aerospace vehicles (UAVs) hold the promise of a wealth of new data on radiation and the hydrologic cycle. Some of these data streams will be similar to those being acquired by current systems, but many will be substantially different. There is and will be a tremendous opportunity to bring new and different analysis techniques to bear on these data and associated atmospheric problems. This field can benefit by importing and adapting techniques from areas such as signal processing, pattern recognition, intelligent systems and artificial intelligence, chaos theory, and computer visualization. Such techniques may also have considerable applicability to analyses of model results. As the complexity and resolution of atmospheric models continue to grow, we have to improve our ability to extract meaningful information from them. A number of atmospheric scientists are working in these areas, but more support, particularly for innovative and perhaps risky research, is needed. Unfortunately, in times of static or diminishing support, it is precisely this type of research that is most likely not to be funded.
Exploit Rapidly Increasing Computational Power
As the price of computer cycles continues to fall, new modeling approaches can be used both to circumvent the need for parameterizations and to tackle new issues. In the next decade we may expect numerical modeling of time-dependent problems spanning three orders of magnitude in spatial scale to become feasible. For example, one fundamental problem for which this may provide important new understanding is the organization of deep tropical convection into clusters, superclusters, and synoptic-scale waves. It should be feasible to simulate the motions explicitly in individual cells within a synoptic-scale region of the tropical ocean or over the midwestern United States, avoiding the need for cumulus parameterization. Increasingly sophisticated techniques for placing areas of increased grid resolution adaptively within a flow, so as to cover only the convective regions and not the large expanses of stably stratified nonturbulent flow in between, are a promising way to improve the efficiency of such large computa-
tions. These calculations can contribute to fundamental insights into self-organizing mechanisms for moist convection and help in the development of convective parameterizations. As a second example, higher-resolution large eddy simulations of the inversion-capped cloudy atmospheric boundary layer should be able to properly resolve turbulent eddies near the entrainment interface and the surface layer, thereby providing a valuable tool for better understanding and parameterizing entrainment and surface-layer structure.
Increased computer power also allows new and often more fundamental modeling approaches to be used. In radiation, the exploration of three-dimensional radiative transfer in realistic cloud fields and nonspherical scattering problems are becoming feasible and should be fostered. Large eddy simulation models, including explicit drop and aerosol size distributions and even simple chemistry, are a promising tool for exploring climatically important feedbacks such as those between clouds, aerosols, and radiation, as well as for understanding drizzle processes in boundary layer clouds. The effects of orography in climate and forecast models can be represented much more faithfully with increased spatial resolution, while parameterizations of the boundary layer benefit from increased vertical resolution. These are only a few examples of problems in which increased computer power can make a fundamental contribution. However, experience has taught us that increased computer power and model resolution are not substitutes for improvements in the underlying basic physics of the model.
All of the preceding topics are opportunities that should be pursued for their scientific and societal value. However, some stand out because of their particular scientific importance or their likely societal impact. We suggest that the following three topics are of particular importance and should be special foci for research in the coming decade.
Focus 1: Develop and Verify an Ability to Predict the Influences of Small-Scale Physical Processes on Large-Scale Atmospheric Phenomena
The atmosphere is an interactive system with mutually dependent processes that span scales from microscopic to global. In most cases, the important influences among processes arise from the collective effects of an ensemble of interactions. A central impediment to our ability to predict weather and climate is the fact that even when the fundamental physical laws governing individual processes are understood, their collective influences on other phenomena cannot be predicted because of the complexity and number of interactions involved. This problem is particularly evident in climate, weather, and cloud models, where many small-scale processes have collective influences on the largest-scale phenomena. Examples are the interaction of solar radiation with cloud droplets, the effects of cumulus convection and turbulence on the transport of mass and momentum, chemical interactions that influence cloud microstructure and hence
radiative properties on a global scale, and the generation of lightning and the global electric field from many small transfers of electrical charge during collisions of hydrometeors. To model all of the processes involved in the atmosphere would require representing processes that occur with scale sizes from at least 0.0001 cm to 10,000 km, a range of 1013. Most modern computer models can only handle scales that cover a range of about 103, so we are far from circumventing this problem even with the most optimistic future increases in computing power.
The solution lies in developing organizing principles that make it possible to understand the collective influence of these small-scale processes on larger-scale phenomena in the atmosphere. We need to develop a better understanding not only of the interactions between phenomena that occur with different scales but also of the interactions among processes historically studied by different groups of scientists. In many cases, the linkage between these processes is the essence of the outstanding problem. For example, cloud physicists understand how soluble particles in the atmosphere influence the sizes and number of cloud droplets, and chemists understand the fundamental chemical reactions involved in producing soluble particles. Nevertheless, this understanding has not yet led to a predictive capability for the number of soluble particles in the atmosphere and hence for cloud droplet concentrations and sizes, because these processes interact with each other and with radiation, chemical cycles in the atmosphere, global circulation patterns, and the hydrological cycle in ways that are beyond our current understanding.
We suggest that increased effort directed to this class of problems will provide substantial dividends in the next decade. Our optimism is based on the following considerations:
1. In many cases the treatment of physical processes in models is inconsistent with current understanding of the phenomena. This has resulted partly from a failure to bridge the communications gap between those involved in fundamental studies and those involved in modeling the effects of these studies. An initial basis for improvement in the representation of processes such as turbulent transport, entrainment, boundary layer influences, and precipitation formation already exists.
2. Modeling and computing capabilities are improving rapidly. Model simulations of phenomena on small scales can be used to develop representations of these processes on larger scales, and the techniques and capabilities for this approach are now available but are only beginning to be exploited.
3. New observational capabilities are appearing or expected that will aid in these studies. New satellites for global monitoring will provide extensive new data sources; new weather observing networks are now available; improved atmospheric soundings may become available from new uses of commercial aircraft soundings and remote sensors; new long-range and high-altitude aircraft are
available to support this research; and new fixed sites for global monitoring that are being established or are in operation will provide crucial data.
4. There have been recent satisfying advances suggesting that we are on the threshold of significant progress toward improved understanding of these interactions. Success in predicting radiative transfer for clear skies, new understanding of the structure and lifetime of stratocumulus and cirrus clouds, new appreciation of the roles of aerosols in the atmosphere, the success of cloud-resolving models in representing ensemble influences of clouds, increasingly realistic models of the atmospheric boundary layer, and the development of techniques for inferring hydrometeor and cloud characteristics from satellite observations are all important steps that can support continued and enhanced focus on these and related problems.
5. There has been a significant shift in perspective among scientists capable of addressing these problemsfor example, cloud physicists, who traditionally were oriented more toward microscale studies, have developed increasingly global perspectives. This shift has to continue, for example, through field programs that seek to make observations with characteristic scales similar to the grid scale in climate models. Planned programs are formulating better ways of studying the interactive and large-scale consequences of the phenomena being studied.
A key consequence of an improved ability to predict the consequences of small-scale processes will be the development of improved representations or parameterizations for physical processes that occur on scales smaller than the grid scale in models. This will permit the effects of subgrid-scale processes to be represented in simulations of climate, weather, and clouds. Two complementary approaches to parameterization are needed. First, the fundamental interactive processes (e.g., linking radiative properties to hydrometeor spectra and hence to aerosols and cloud dynamics) require clarification. Second, existing knowledge and existing or new data can be used to develop representations that are consistent with current understanding. Such parameterizations, although sometimes unsatisfying because they fall short of representing true physical relationships, are still useful and necessary to support studies of the complex interactions in the atmosphere.
Currently, a wide gulf exists between the parameterization approaches used in climate models and data, and process model studies associated with field programs. The parameterization issue cannot be resolved without bringing to bear the accumulated knowledge from process studies. At the same time, narrowly focused process studies that exhaustively study single cases cannot supply the information needed to improve model parameterizations. Until a much better representation of reality is introduced into parameterizations of processes such as the effects of clouds on radiation, there is little hope of improving confidence in climate and climatic change simulations. Emphasis should be placed on the acquisition of data sets in which cloud and hydrometeor properties, fluxes of
radiation, aerosol size distributions, and concentrations of significant trace gases are measured simultaneously. Such data sets will also have to characterize different cloud regimes and geographical areas.
This is an extremely difficult problem, justifying a multifaceted approach. This approach should include increased attention to the acquisition of long-term, temporally and spatially consistent data sets characterizing the properties of clouds, radiation, water vapor, and trace gases. It must also include continued efforts to develop models that explicitly link these quantities. In addition, a concerted effort must be made to link the observational, data analysis, and modeling communities to focus on this problem. Promising approaches include the following:
• using cloud-resolving models and nested models to determine interactions with large-scale variables;
• using cloud models with explicit microphysics to develop parameterizations for microphysical processes suitable for simulations of weather and climate;
• developing parameterizations exclusively from process study observations, then using satellite observations to generalize and extrapolate to the global scale;
• isolating a limited number of empirical parameters that have a physical basis, then fitting to a set of observations to determine the best values of these parameters; or
• using operational models as sources of finer-scale ''data'' from which to develop parameterizations.
The inability to represent interactions among phenomena, especially those that occur on small scales, is the primary weakness in current models of the climate and weather, so we regard the need for these results as "imperative" and offer this as a key challenge for the next decade. Although a central justification for such studies is the need to improve representations of these processes in models, we nevertheless argue that the challengeand opportunityis to understand the linkages among various physical and chemical processes. A short-range approach is inappropriate because we are still at an early stage in understanding these interactions. Significant effort must be directed toward improving the foundations that will lead to adequate representations of these processes.
Focus 2: Develop a Quantitative Description of the Processes and Interactions That Determine the Observed Distributions of Water Substance in the Atmosphere
Precipitation is the source of essentially all fresh water on Earth, so the hydrological cycle is truly "vital" to humans and to most plant and animal life on
land. Although precipitation and cloudiness are the most evident results of the cycling of water through the atmosphere, water also has many other effects on weather and climate. Water vapor is the most important greenhouse gas, and variations in cloudiness and ice cover are the primary sources of variability in the albedo of the Earth. Clouds scavenge particles and trace gases from the atmosphere, and thunderstorms maintain the Earth's electric field. The latent heat released or absorbed as water changes phase is the source of energy that drives hurricanes and other severe weather systems. Thus, we cannot understand weather and climate without a good understanding of the distribution of water substance in the atmosphere.
Weaknesses in current specifications of the atmospheric water cycle include poor characterization of upper-tropospheric water vapor, uncertainties in surface fluxes, poor understanding of the factors controlling precipitation efficiency, inability to represent the ensemble effects of cumulus convection on the transport of water, poor characterization of rainfall over the oceans, and the absence of a comprehensive understanding of the links between the atmospheric cycle and other components of the hydrological cycle. Emerging technologies and recent developments now can support a comprehensive new approach to these problems. Improved characterization of water vapor in the atmosphere may become possible from sensors on commercial aircraft, remote sensors employing radiometric or GPS technology, and improved research instruments capable of accurate measurements at low humidity. Satellite characterization of precipitation over the oceans will be possible in the coming decade. Existing and new modeling capabilities and data sets can be used to characterize the effects of cumulus convection, and improved models and understanding of boundary layer fluxes are emerging. Comprehensive international programs to study the hydrological cycle over regional scales appear feasible and are planned.
These expected new results should provide an opportunity to characterize the distribution of water in the atmosphere with new confidence and to relate this distribution to the underlying interactions with the global hydrological cycle. Such an improved characterization is needed for accurate determination of radiative transfer in the atmosphere, for climate predictions of precipitation amounts and global temperature, and for improved weather forecasts. These studies thus provide a good match between the opportunities presented by new research capabilities and the needs of current research. However, a comprehensive and systematic approach is required if the many factors and processes entering the atmospheric hydrological cycle are to be understood.
Focus 3: Improve Capabilities to Make Critical Measurements in Support of Studies in Atmospheric Physics
Although impressive advances in instrumentation have been made in some government laboratories, especially in the development of high-technology re-
mote sensing instruments, other areas have fallen far behind what is technologically feasible. For many measurements, research aircraft still employ inadequate, decades-old instruments for lack of better alternatives. Critical measurements of humidity in the upper atmosphere cannot be made because of deficiencies in the humidity sensors used for routine soundings. Our ability to make critical measurements is also lagging far behind current needs in studies of hydrometeor size distributions or radiation. Numerous examples exist in all areas of atmospheric physics.2
In most cases there are good candidate techniques for making the needed measurements, but they await implementation. The instrumentation in current use is seriously out of date and is not taking advantage of modern knowledge and modern technology. There is an opportunity for a concerted effort in instrumentation to have a large impact. New experimental instruments to measure temperature and wind from aircraft are beginning to answer crucial questions about cumulus convection that have been posed for 50 years or more. Similar advances can be expected from other improvements in the basic instruments used in atmospheric science research.
There are special needs for new equipment and instrumentation to study the issues raised in the first two imperatives. Foremost is the need for platforms suited to the studies that must be conducted, including new observing satellites, high-altitude research aircraft, and aircraft with long range and flight duration. Remotely piloted vehicles will provide new opportunities to collect measurements at high altitude and over long periods. Examples of missing observations that would make critical contributions include ice mass (especially in anvils), ice crystal radiative characterization (especially near the tops of cirrus and other clouds), wind fields with improved coverage and resolution, upper-troposphere humidity, and cloud fractional coverage as a function of cloud area and altitude. Instrumentation presently available is particularly weak in the following areas:
2 For example, measuring the temperature inside a cloud from aircraft is a problem that has apparently been solved only recently by new radiometric thermometers, and these are not yet thoroughly evaluated or in widespread use. Measurements of equivalent potential temperature in rain are not reliable, so important dynamical influences of precipitation cannot be evaluated. Measurements of cloud-condensed water still do not provide the accuracy required for most studies. Measurement of low concentrations of large precipitation particles. and the sizes and shapes of small ice particles, have not been possible until quite recently, and the new instruments that can make these measurements are still experimental. Unless corrected by GPS measurements, the accuracy of measured horizontal winds from research aircraft is only a few meters per second due to the character of inertial navigation systems, and the degradation of GPS accuracy in nonmilitary applications still prevents removal of these errors in many cases. Broadband radiometers used on research aircraft are not sufficiently accurate and do not respond fast enough to provide the needed measurements of flux divergence in clouds.
• the characterization of hydrometeor size distributions and shapes;
• measurements of total water content and ice water content;
• water vapor mixing ratio, especially in the upper troposphere;
• cloud condensation nuclei, where instruments have to be more widely available and suited to use at high altitude;
• ice nucleus measurements;
• other radiation instrumentation for the characterization of irradiance, radiance, and net flux as a function of wavelength, optical depth, and volume absorption coefficient, and for determining the optical properties of ensembles of ice crystals;
•chemistry instrumentation, including trace gas detectors;
• equipment for the detection of organic constituents, especially aerosol instrumentation;
• airborne remote sensing equipment such as lidars, short-wavelength radars, and near-field profilers for temperature and humidity that can provide some extension away from an airborne platform and thus increase the representative-ness of its sample; and
• measurement of mean vertical motions to better than 10 cm/s, averaged over areas ranging from cloud-scale to mesoscale.
There are some particular needs for instrumentation to support studies of atmospheric radiation:
• Produce an inexpensive observing system for radiative fluxes that is capable of making continuous, accurate, and reliable observations. Such a system is crucially needed to increase the data base of surface radiative fluxes.
• Make a concerted effort to upgrade all radiometers, particularly broadband flux radiometers used on aircraft. The typical thermopile devices in use are not sufficiently accurate, and their response times are too slow to supply the required in situ data.
• Develop simultaneous multiwavelength, active and passive systems to probe the atmosphere from both space and the surface and to retrieve hydrologic parameters. These systems must be robust and well calibrated. Examples are combinations of passive microwave radiometers and radar, or Doppler lidar and radar.
• Develop observational techniques to sample routinely the regional-scale hydrological cycle. Required observations beyond our current capabilities include water vapor profiles in the upper troposphere and lower stratosphere, ice water path and content, and precipitation.
Although major additional resources are needed for improvements in instrumentation, a modest program, perhaps combining the efforts of university scien-
tists, government laboratories, and research support facilities, could lead to significant improvements in our ability to make critical measurements in the atmosphere.
Contributions to National Goals
Human activity is having documented effects on the weather and climate. For example, carbon dioxide and sulfate concentrations in the atmosphere have been changed dramatically by human activity, and the effects of these changes are beginning to appear in the climate of the Earth. In some major cities, pollution advisories warning of hazards to health and leading to restrictions on activities are becoming increasingly common. Biomass burning by humans is the predominant source of particulates and of some chemicals in some regions, and perhaps worldwide. As the global population continues to increase and to industrialize, these problems will become more urgent.
Many of the components of the research program recommended here address sources of uncertainty in climate prediction. Other benefits will result from improved abilities to predict regional climate and weather. When policies to mitigate anthropogenic effects on climate and weather or to manage water and other natural resources are debated, the results of the research proposed here will provide the critical basis for making sound decisions. The difficult choices are likely to involve accepting damage to the economy or to health, penalizing developing or industrialized countries, and burdening current or future generations. In this context, the importance to society of the scientific information that accrues from the research recommended here is indisputable.