Scientific Imperatives and Recommendations for the Decades
To accelerate progress in the first decades of the twenty-first century, the atmospheric sciences must advance understanding of processes through which the atmosphere evolves. To do so requires organizing and optimizing interactions between conceptual and technological advances through a comprehensive integration of observational, modeling, and other research efforts.
Part II of this report presents Assessments of five disciplines within the atmospheric sciences:
1. Atmospheric physics
2. Atmospheric chemistry
3.Atmospheric dynamics and weather forecasting
4. Upper atmosphere and near-Earth space
5. Climate and climate change
These Disciplinary Assessments, prepared by three standing committees and two ad hoc groups of the Board on Atmospheric Sciences and Climate (BASC), emphasize understanding and predicting atmospheric phenomena and processes. For the near future, they emphasize forecasts of atmospheric phenomena with significant societal impacts, including seasonal climate variability, chemical processes and air quality, and space weather events. For the longer term, they emphasize the resolution of climate variability on the scale of decades to centuries and the possibility of projecting climate variations. Each assessment analyzes critical scientific issues, identifies the major opportunities and initiatives
Imperatives for Atmospheric Research
1. Optimize and integrate atmospheric and other Earth observation, analysis, and modeling systems.
2. Develop new observation capabilities for resolving critical variables on time and space scales relevant to forecasts of significant atmospheric phenomena.
Recommendations for Atmospheric Resarch
• Resolve and model interactions at the boundaries between the atmosphere and other Earth system components, and interactions within the atmosphere among phenomena of different scales;
• Apply the discipline of forecasting in atmospheric chemistry, climate, and space weather research in order to advance knowledge, capabilities for prediction, and service to society.
• Develop collaborative studies of three emerging issues: (1) climate, weather, and health; (2) climate change implications for water resource management; and (3) implications of rapidly increasing atmospheric emissions.
for the discipline, and recommends a scientific and programmatic agenda for the decade or two ahead based on an evaluation of priorities within the discipline.
Each of these Assessments cites the critical role of comprehensive and integrated observations in research and applications, leading to BASC's identification of two imperatives concerning observations as the highest-priority endeavors for the atmospheric sciences and services. In addition, three recommendations for research also emerge from these studies. They are stated in brief form in Box I.3.1.
This section discusses the highest-priority imperatives and recommendations for atmospheric research in greater detail, first stating the imperative or recommendation and then providing its justification. Section 4 of Part I addresses some of the leadership and management issues that should be considered in shaping the financial and infrastructural agendas for atmospheric sciences.
Atmospheric Science Imperative
Optimize and Integrate Observation Capabilities
The atmospheric science community and relevant federal agencies should develop a specific plan for optimizing global observations of the atmosphere, oceans, and land. This plan should take into account requirements for monitoring weather, climate, and air quality and for providing the information needed to improve predictive numerical mod-
els used for weather, climate, atmospheric chemistry, air quality, and near-Earth space physics activities. The process should involve a continuous interaction between the research and operational communities and should delineate critical scientific and engineering issues. Proposed configurations of the national and international observing system should be examined with the aid of observing system simulation experiments.
Atmospheric observation, modeling, and prediction systems are increasingly interdependent; thus, the components of the atmospheric information system should be optimized as part of an end-to-end system. Moreover, there is an increasing need in both research and applications to integrate observations of the atmosphere, ocean, and land surface.
As a consequence of the increasing synergy of observation and modeling systems, four-dimensional1 data bases portraying the evolution of actual and predicted conditions are replacing the traditional synoptic snapshots of atmospheric conditions. Made possible by new capabilities in observational, computational, and communications technology, these four-dimensional data bases containing predictions of traditional variables will be the input to distributed computer procedures that prepare application-specific forecasts formulated in terms of key impact variables,2 decision aids, and recommendations for action.
Somewhat different data formats and strategies may also be required. Thus, the effort to optimize and improve national and global observing systems should examine commonalities and differences among the needs of weather prediction, climate monitoring and projection, atmospheric chemistry and air quality prediction, near-Earth space physics, and other environmental disciplines.
New Observing Opportunities
New opportunities for acquiring atmospheric observations suggest that the observing system of the future may be dramatically different from that of today. Three examples illustrate this point:
1. Commercial aircraft observations: Sensors carried by commercial air transport aircraft are now producing thousands of observations of winds and
1 A four-dimensional data base is one in which data are stored according to four independent variables, or coordinates: longitude, latitude, height above mean sea level, and time.
2 Impact variables are descriptions of atmospheric conditions that directly affect a given operation, in contrast to the meteorological variables used in theory and modeling, such as pressure and temperature. For aviation, as an example, such impact variables include cloud height, visibility, and intensity of icing or turbulence.
temperatures daily over the contiguous United States; the use of humidity sensors is currently being evaluated. The ascent and descent of these aircraft at major air terminals can provide a dense set of atmospheric soundings in addition to enroute observations (NRC, 1994a; Fleming, 1996).
2. Global positioning system: Signals from the Global Positioning System (GPS) satellites intended for use in navigation can also, through occultation techniques, produce profiles of the atmospheric index of refraction (Ware et al., 1996) which can be used to derive temperature and humidity profiles or can be assimilated directly into models. With these GPS measurements, global fields of important variables could be measured relatively inexpensively and continuously for weather and climate applications.
3. Adaptive strategies: High-resolution measurements in specific locations may sometimes enhance prediction of significant phenomena, suggesting that some observations can be made adaptively in response to empirical evidence or modeling techniques that identify sources of errors. For example, operational models predicting severe weather or tropical storms might be configured to specify locations where higher-resolution initial conditions would enhance accuracy (Burpee et al., 1996). Then, observations could be obtained in these regions from satellites or remotely piloted scientific aircraft.
Requirements for Optimizing and Integrating Observing Systems
The integrated portrayal of atmospheric evolution and interactions in four-dimensional data bases containing both existing and some critical new observations is essential for the advances envisioned in this report. The following requirements must be met to achieve this objective:
• Integration with modeling efforts: Efforts to optimize observing systems must take into account the analysis and prediction models that will assimilate the data, whether they are weather, climate, upper-atmosphere, or chemical models. Thus, rigorous end-to-end analysesfrom observations through models to accuracy of predictionsare required to assess the benefits and costs of overall strategies and specific observation schemes. For example, issues such as whether radiosonde observations can be replaced with satellite data must take account of many considerations, including forecast accuracy, operational costs, and especially the integrity of the climate record.
• Increases in computing power: Advances in computing power will be necessary to conduct fully coupled simulations of the entire Earth systemincluding the atmosphere, oceans, land surface, ice, and chemical and biological subsystemsin order to study climate and its variability on various scales. High-resolution weather forecasts, along with ensemble approaches to assessing confidence in the prediction will similarly require greater computer power. Both activities may impose changing requirements on optimization criteria.
• Assimilation of new forms of data: Innovative assimilation schemes for incorporating new forms of data into analysis and prediction models may enable significant advances. For example, recent research has shown that using model variables to calculate the model radiance fields and then iterating the model to achieve agreement with radiance fields as observed by satellites is superior to attempting to convert satellite observations into traditional variables such as temperature and humidity. Significant benefits from other new forms of data, including those from commercial aircraft and air quality networks, may derive from similar innovation and analysis. Earth Observing System (EOS) satellites will provide global observations focused on understanding climate variability and evolution. In addition, EOS and other research satellite efforts will provide new streams of data and new instruments that may be adapted for operational use.
• Multiple uses of data bases: As data from surface and upper-air observations, operational radar, geostationary and polar-orbiting satellites, and other sources such as air quality networks are incorporated into the data base, they can be shared by the weather, climate, and applications communities. Because resources are limited, it will be critical to strive for the maximum benefit for all from the incremental investments required to optimize and integrate the observational and analysis systems. Indeed, the goal should be to integrate atmospheric data bases with those developed by other disciplines to represent ocean and land surface conditions.
• Information organizing systems: A requirement in forecasting is to identify the limited subset of data that is critical for the accuracy of a prediction. Machine-learning concepts, statistical models relating key variables or decision parameters to computer forecasts, expert systems, and the concepts of fuzzy logic all have potential for organizing observed data and computer simulations in ways that strengthen atmospheric information services.
• International collaboration: Four-dimensional data bases will transcend national boundaries. It is essential that the long tradition of sharing weather and climate data globally be maintained. The proposed integration of satellite observations between the United States and European space and meteorological agencies is an important harbinger of progress; in contrast, attempts to limit the availability of local data are serious impediments to the cooperation required.
Many of the issues involved in improved observations and improved weather services are being addressed in the federal interagency effort to create a future North American Atmospheric Observing System (NOAA, 1996) as a composite observing system for the twenty-first century. Until it has been established rigorously that new approaches are fully adequate alternatives, it is essential to retain the full capabilities of the worldwide radiosonde network because it is the foundation of the present atmospheric observation and prediction system.
Observing System Simulation Experiments
Getting the highest return from investments in atmospheric information and forecasts is the clear motivation for integrating observing systems. Assessing alternatives is critical as automated systems replace human observers in an attempt to reduce ongoing operational costs by making capital investments in systems that have lower overall life-cycle costs.
Observing system simulation experiments enabled by contemporary computers and numerical models can help to determine the optimum configurations of observing system components and numerical models relative to forecast accuracy and overall costs, including capital investments, operational costs, and the costs of disseminating and archiving data. Such experiments allow rigorous and quantitative examination of a wide range of strategies and can indicate whether new resources will produce demonstrable benefits. Simulations of observing and prediction systems can thus be a critical mechanism for managing the advance of the discipline and its service to society.
Atmospheric Science Imperative
Develop New Observation Capabilities
The federal agencies involved in atmospheric science should commit to a strategy, priorities, and a program for developing new capabilities for observing critical variables, including water in all its phases, wind, aerosols and chemical constituents, and variables related to phenomena in near-Earth space, all on spatial and temporal scales relevant to forecasts and applications. The possibilities for obtaining such observations should be considered in studying the optimum observing systems of Imperative 1.
Contemporary numerical atmospheric computer models are sufficiently varied and powerful that they can predict or simulate a range of phenomena such as climate change or air pollution episodes, as well as the course of the weather. However, observations of critical variables on time and space scales relevant to forecasts are essential to improving such numerical simulations and predictions. Some of the required observations present significant technological challenges.
In some cases, higher-resolution observations of traditional meteorological variables such as water vapor are required to portray important processes accurately. In others, both research and applications require some new observations, for example, of aerosols and trace chemicals and their vertical transports in both clear and mixed-phase environments.
The required improvements will arise in part from new instruments and new
technology, as in new remote sensing techniques, and in part from new observational strategies. For example, automation, adaptive observations, and international cooperation will all contribute to improved observations, as demonstrated by the Tropical Ocean Global Atmosphere-Tropical Atmospheric Ocean array, a network of buoys stretched across the equatorial Pacific.
Water in the Atmosphere
Water in all its phases and changes in these phases have a pervasive role in weather, climate, and atmospheric chemistry; thus, observing the phases and amounts of water in the atmosphere is important to understanding atmospheric events on all scales. The energy released by changes in the phase of water substance is of major importance in driving global wind systems.
On the storm scale, prediction of convective precipitation is limited by uncertainty in the distribution of water vapor in the atmosphere and the amount of water in the soil. On the longest time scale, water vapor is the most important greenhouse gas, and significant uncertainty in climate models can be traced to inadequate understanding of the water budget and water-induced radiation feedback. Water vapor in the atmosphere varies markedly over small scales vertically and horizontally; thus, its variability is significant at scales not resolvable by the radiosonde network.
Today, water vapor profiling technology and its integration with other measurements and models to improve resolution are evolving rapidly. Notable advances are being made with remote sensing techniques. In situ water vapor measurements are essential to resolve fine structure and to constrain remotely sensed data. Radiosonde observations satisfy these needs only partially because of large spatial and temporal gaps. Greater coverage along commercial air routes and more soundings near airports will be made possible through the Commercial Aviation Sensing Humidity (CASH) program [sponsored by the Federal Aviation Administration (FAA) and the National Oceanic and Atmospheric Administration (NOAA) Office of Global Programs], which is installing humidity sensors on air transport aircraft; these sensors will have sufficient resolution to obtain profiles during aircraft ascent and descent. However, improvements are needed because current devices cannot measure humidity accurately on an aircraft flying through clouds.
Clouds are the visible evidence of a host of complex processes involved with the change of water vapor to both liquid and solid forms. Many critical aspects of
the cloud processes involved in the formation of precipitation and in interactions between water drops and ice particles are still not well understood and require detailed observational and modeling studies.
These processes also have important impacts on chemical reactions in aqueous and mixed-phase environments and must be better understood in order to improve models of local and global chemical cycles.
As described later, clouds are an important control on the planetary energy budget, inducing significant modifications of the incoming streams of solar radiation and the outgoing streams of infrared radiation. Moreover, latent heat released during the formation of liquid water is a significant source of energy for both severe weather and large-scale synoptic systems.
The global distribution of precipitation is tied to the dynamics and energetics of the atmosphere, the coupling of atmosphere to ocean, and ocean currents themselves. Precipitation observations thus reflect both regional and local conditions and are critically needed for climate studies, verification of global models, and forecasts of severe weather and flood conditions. Truly global observations can be made effectively only by satellites; progress with precipitation observations is being made with the Special Sensor Microwave/Imager (SSM/I). Another satellite system, the Tropical Rainfall Measurement Mission (TRMM), launched on November 27, 1997, promises to improve precipitation estimates and estimates of energy released over the tropics, using a combination of high-resolution radar, a passive microwave radiometer, and measurements in the visible to infrared. Variability in the location of tropical rainfall has been shown to be a critical link in the changes leading to El Niño events.
Precipitation estimates over the United States have been greatly enhanced by deployment of NOAA's WSR-88D radars, and these improvements are being incorporated into river basin flood models. Further development is necessary to (1) improve the accuracy of rainfall estimates by upgrading WSR-88D radars to include multiparameter capability, (2) to improve rain gauge reliability, and (3) to strengthen the ability to estimate rainfall from satellites in space. As in the case of water vapor, improved precipitation-estimation algorithms must be based on combinations of relevant data. An expert system might be developed to produce precipitation estimates from input information including radar, satellite, and rain gauge data, as well as season of the year and type of cloud system.
Owing to modem technology, the density of wind observations has significantly increased over the United States and some other continental areas in recent years. In the central United States, radar wind profilers, sensitive to the motions
of refractive index fluctuations in clear air, are now demonstrating their usefulness as forecast model input and research tools. WSR-88D Doppler radars, although used mainly to document precipitating weather systems, can measure winds in clear air to altitudes of several kilometers over land during the day. Some civilian transport aircraft provide wind observations along air routes and during ascent and descent.
Wind information over the open ocean is sparse. Radiosondes are launched from islands and from some ships, but many of the remote sites are being abandoned or becoming unreliable owing to fiscal pressures in other nations. Winds inferred from satellite observations of cloud motion can be useful but are sometimes inaccurate.
More accurate and comprehensive wind observations are needed for nearly all the endeavors of atmospheric science. Adaptive and episodic observations, obtained on demand in data-sparse regions where information is critical, could help. Presently, aircraft are deployed to gather data in and around tropical cyclones through penetrations and dropsondes; future strategies could also incorporate remotely piloted scientific aircraft (Holland et al., 1992) and drifting balloons that rise or fall in response to remote control. Satellite-borne lidar systems for measuring wind could become a reality in the next decade, with several promising techniques now under development. Satellite scatterometer measurements, such as those from the European Research Satellite ERS-1, and microwave brightness from SSM/I, when constrained by models or independent observations, should prove useful for estimates of surface winds over the oceans. New airborne radars developed and operated by the National Center for Atmospheric Research and the NOAA Aircraft Operations Center are providing unprecedented detail on the fine-scale structure of weather systems, including severe storms and hurricanes.
Observations in the Stratosphere
The stratosphere affects the world's weather through interactions and exchanges with the troposphere and is an integral part of the global climate and chemical system. Sampling meteorological variables, trace gases, and aerosols in the stratosphere will require a combination of ground-based balloon and remote sensing, satellite measurements, and piloted and remotely piloted aircraftall blended with numerical models. Satellite systems should overlap in time, to provide continuity of record as well as to provide confidence in comparison of studies separated by several years. In situ aircraft measurements will also be used to evaluate new remote sensing devices. Remotely piloted aircraft may become the platform of choice because of their lower cost and high-altitude capability, provided that safety issues are resolved and the technology proves to be operationally feasible. Both space-borne and airborne measurements will benefit from miniaturization of remote sensing and in situ instruments.
Observations in Near-Earth Space
Phenomena created in the near-Earth environment by charged particles and varying solar magnetic fields are referred to collectively as ''space weather'' and have a variety of significant effects on human activities, as discussed later. The prediction of space weather phenomena requires observations that can be obtained in part through present and planned satellite programs. For example, the GPS radio occultation technique (Ware et al., 1996) gives vertical profiles of electron density and total electron content in the ionosphere.
Because space weather phenomena are forced by solar variability, three sets of solar observations should be emphasized: (1) coronal mass ejections, (2) magnetic fields including those in the corona, and (3) near-Sun solar wind properties. Moreover, measurements of the interaction of solar particles and the solar magnetic field with the Earth's magnetosphere and ionosphere are important for basic understanding and development of numerical models.
Atmospheric Research Recommendation 1: Resolve Interactions at Atmospheric Boundaries and Among Different Scales of Flow
The major weather, climate, and global observation programs supported by the federal government and international agencies should put high priority on improved understanding of interactions of the atmosphere with other components of the Earth system and of interactions between atmospheric phenomena of different scales. These programs, including the U.S. Weather Research Program, the U.S. Global Change Research Program, and other mechanisms for supporting atmospheric research, require observational, theoretical, and modeling studies of such interactions.
Atmospheric studies are shaped today by the recognition that interactions with neighboring systems are critical to improved understanding and prediction. It is no longer sufficient to study the individual components of the Earth system in isolationattention must now turn to understanding, modeling, and predicting them as coupled systems. In many cases the relevant coupled systems are the actual physical subsystems; in others, the interacting subsystems are more abstractfor example, when small-scale eddies interact with the largest scales of atmospheric flow. Critical scientific questions focus on the exchanges of energy, momentum, and chemical constituents among the atmosphere and the surface below as well as the layers above leading to near-Earth space.
In the initial study of a dynamical system, interactions with the exterior environment can sometimes be ignored. This approach served the numerical
weather forecasting community well in its initial efforts. However, as the length of the forecast period increased, the need to account for the flows of dynamically significant quantities across system interfaces became evident. In climate dynamics, knowledge of this interaction between the components of the Earth system is essential to progress. For example, it is now recognized that the El Niño-Southern Oscillation (ENSO) is driven by interactions within the combined tropical ocean-atmosphere system.
Processes occurring at the boundary between the atmosphere and the land and sea are vital elements in weather, climate, atmospheric chemistry, and global change. Atmospheric interfaces with the ocean and the land are similar because both represent the intersection of an atmospheric boundary layer with a boundary layer on the surface below. The ocean interface is better understood except under high winds; the land interface merits greater emphasis.
Surface fluxes over the ocean are important in early cyclogenesis (Kuo et al., 1991), and the churning up of cold ocean water by hurricanes (Shay et al., 1992) helps determine their evolution and motion (Bender et al., 1993). The weather and climate are particularly sensitive to air-sea interaction over the tropical South Pacific, where the warmest open-ocean surface temperatures coincide with Earth's greatest annual precipitation (Webster and Lukas, 1992).
Surface temperature, moisture, and fluxes are as strongly linked to land surface properties as they are to atmospheric variables. Satellites provide estimates of surface temperature, vegetation, and moisture characteristics that can be converted into estimates of fluxes but require further calibration against in situ measurements of local characteristics (Kogan, 1995).
Methods of estimating fluxes over the ocean using satellites and buoys are perhaps at a more advanced state than over land, with the major uncertainties associated with light and strong winds, wave spectra, and precipitating convection. The TOGA-TAO array is providing a long time series of atmosphere and ocean data useful for understanding sea-air interaction in the equatorial Pacific. Continued developments in remote sensing of surface wind, temperature, and radiative fluxes are needed, along with adaptive sampling techniques for in situ and remotely sensed data in cyclones.
Long-Term Interactions with the Oceans
On decadal to centennial time scales, the interaction between the upper and the deeper parts of the world ocean is believed to be a primary control on the natural variability of surface temperature on the planet. Moreover, the flow of heat between ocean layers sets the time scale for significant response to forcing such as the greenhouse effect. Better quantification and understanding of these
interactions will help in assessing the difference between natural fluctuations of the large-scale surface temperature and the response to external forcing.
Vertical transport of heat in the oceans is mediated by currents that are often of very small horizontal scale compared to basin-wide gyres. Flows connecting the ocean surface to deep water are thought to occur in only a few locations on the planet. For example, North Atlantic deep water is formed in very small regions in the Norwegian Sea. Local phenomena such as the intermittent formation of sea ice can have significant effects in modulating flow from the cold briny surface waters to the very bottom of the world ocean.
Surface interactions are also important. Accurate simulation of the exchange of heat and fresh water at the ocean surface is a demanding contemporary challenge for coupled ocean-atmosphere climate models. Sea ice variation also is only poorly understood, although it is clearly important in long-term climate change because of its high reflectivity and its inhibiting effects on thermal exchange between the atmosphere and ocean.
The coupling of the atmosphere and ocean in the tropics gives rise to the ENSO cycle of alternating extremes of warm and cold sea surface temperatures with periods of three to seven years that affect the global climate system. The information between system components is communicated through wind stress, sea surface temperature, radiation, and precipitation. Although current models exhibit solutions with ENSO-like oscillations that suggest useful predictive capability, they are not always realistic. Improvement in computation of the fluxes across the tropical atmosphere-ocean boundary should lead to notable improvements in understanding and practical application.
Clouds and Their Consequences
Clouds and cloud systems are relatively small but numerous and, in sum, play a key role in shaping weather and climate. For example, the reflection of sunlight by the planet is determined largely by the cloud cover, and a significant fraction of the infrared radiation flowing from the surface is absorbed by clouds and reradiated to the surface as part of the greenhouse effect. Hence, clouds are strong controls on the amount of solar energy absorbed, the planetary energy budget, and the surface temperature. Success in modeling a changing climate will require correct specification of the effects of clouds and the way in which these effects will themselves change as climate evolves.
The effects of clouds on radiation streams are determined by their horizontal and vertical distribution and the type, size, and concentration of their constituent particles. The properties of cloud particles are a consequence of the initial concentrations of aerosols and thermodynamic properties, subsequent vertical motion and mixing, and encounters with particles having different properties.
Deep, precipitating convection interacts with motions of larger scales in other important ways as well. For decades, theoretical studies have explored the
effects of convective heating (the latent heating associated with convective rain) on the growth of synoptic- to global-scale waves in the tropics. More recently, it has been demonstrated that the effects of ENSO on weather in other parts of the world are determined largely by the atmospheric response to heating by deep convection over the equatorial Pacific, which is in turn modulated by the sea surface temperature distribution.
Today, numerical models that can resolve properties of individual clouds and cloud ensembles are being used to study the response of clouds to larger-scale forcing and to develop convective parameterization schemes for climate models whose resolution is too coarse to represent clouds explicitly. Toward this end, the results of cloud-resolving models are presently being compared to available observations, and cloud-resolving models are being coupled with ocean models to increase the realism of the simulation.
Aerosols and Atmospheric Chemistry
Suspensions in the atmosphere of tiny liquid or solid particles are known as aerosols and can cause macroscopic changes in air quality, atmospheric heating, and air chemistry. Aerosols have both short-term and long-term effects in the atmosphere: cloud particles form on them; they absorb and scatter radiation passing through the atmosphere, thus altering the local heat balance; and they are sites of chemical reactions with atmospheric trace gases. Aerosols should be the focus of much research in the coming decades, including gathering more data on aerosol chemistry and its evolution in time and space, and improved understanding of the light-scattering proportions of different types of aerosols (NRC, 1996a).
Trace gas concentrations in the atmosphere depend on sources and sinks are usually at the surface of the Earth and on processes within the atmosphere. Atmospheric transport and modification of chemical constituents in mesoscale flow include reaction, advection and diffusion, and interactions with water in its various phases. Improved understanding is clearly important for air quality modeling and prediction.
The Fundamental Problem of Nonlinearity
Many of the key challenges in atmospheric sciences embody the fundamental problem of all geophysical fluid flowthe nonlinear interaction between phenomena with various length and time scales in the flow and phenomena of different length and time scales in the boundary conditions, external forcing, and within the flow itself.
Most geophysical flow problems involve interactions with a boundary and thus develop turbulent boundary layers. Energy, momentum, and other properties are passed across these boundary layers from the fixed surfaces to the larger-scale flows. Geophysical problems rarely allow simple closures; some interac-
tions must usually be retained, even if in somewhat simplified form. In fact, it is typical in atmospheric boundary layers for scales varying as much as eight orders of magnitude in both space and time to be involved in fluxes from the surface. Crude treatment of these interactions limits the accuracy of long-range forecasts.
Yet even without significant boundary effects or internal interactions with water processes, nonlinear interactions between flow components of different scales has been a continuing challenge for more than a century. Significant progress has been made, as reflected by contemporary understanding of turbulence, predictability, and the phenomena of chaos. Nevertheless, we have yet to develop ways to describe chaos mathematically by finding analytic methods for specifying the nature of attractors and the statistical characteristics of resulting flows (Dutton, 1992). Of particular interest is the extent to which long-term climate simulations can determine global or local statistics accurately, even though the numerical solutions are surely not deterministic.
Until the difficulties created by nonlinearity are dramatically reduced, they will remain pre-eminent challenges in the attempt to understand geophysical flows and predict them quantitatively.
Atmospheric Research Recommendation 2: Extend a Disciplined Forecast Process to New Areas
A strategy and implementation plan for initiating experimental forecasts and taking advantage of a disciplined forecasting process should be developed by appropriate agencies and the scientific community for climate variations, key chemical constituents and air quality, and space weather events.
Much of the effort in the atmospheric sciences is aimed, either explicitly or implicitly, at extending the range and improving the accuracy of forecasts of atmospheric phenomenaweather and air quality on the short term, climate variation on the longer term. As practiced for more than a century, weather prediction involves the traditional steps of the scientific method:
• Collect and analyze observations of present conditions.
• Develop and use subjective or quantitative methods and models to infer future conditions from these observations.
• Assess the accuracy of the forecast with observations of actual conditions.
• Analyze forecast results to determine how methods and models can be improved.
Through this process, weather prediction imposes a demanding discipline on both individual forecasters and all of the atmospheric sciences. The accuracy of the forecast every day, in a wide variety of locations, is a continuing measure of the
success and progress in this fieldand of the integration of theory, technology, and practical methods.
The impact of this discipline has intensified with the advent of operational numerical weather prediction because precise quantitative comparisons between forecasts and actual observations can be made easily. The inclusion of computers in the forecast cycle also facilitates improvement because proposed modifications in the computer model can be applied to difficult cases retrospectively.
As shown in Part II, several of the atmospheric sciences are now developing capabilities for making quantitative forecasts and will benefit substantially by establishing experimental operational forecast procedures and thus engaging the discipline of forecasting.
Climate forecasting presents one such opportunity, on both the seasonal and the decadal scales. The TOGA program has stimulated considerable progress in monitoring and predicting seasonal to interannual climate fluctuations such as El Niño. In the early 1980s, observations in the tropical Pacific Ocean were so limited that the 1982-1983 El Niño was in progress for several months before its magnitude and implications were known. The Tropical Atmosphere Ocean observing system is now operational and includes 70 moored ocean buoys to observe surface meteorological properties and the upper-ocean thermal structure, surface drifting buoys that measure sea surface temperature, expendable bathy-thermographs from ships of opportunity that probe the ocean temperature field to 700 m, and island tide gauges that monitor sea level variability. Research and operational satellites complement the in situ measurements. This system allowed the evolution of the extreme El Niño event of 1997-1998 to be monitored on a day-to-day basis. Moreover, these observations have supported the development of coupled ocean-atmosphere models of the tropical Pacific Ocean that can predict tropical Pacific sea surface temperatures with confidence months to a year in advance. These developments in the observing system, together with related advances in coupled forecast models, have led to the implementation at NOAA's National Centers for Environmental Prediction of a routine, model-based, short-term climate forecast system for the United States. Similarly, the newly formed International Research Institute for Climate Prediction, created by NOAA, has taken on the complementary responsibility for the rest of the world.
Successful forecasts of climate variations over a decade or more would be of great value to a variety of industrial and government activities. These forecasts would be statistical in nature and thus focused perhaps on seasonal departures from means; as statistics, they might include an estimate of expected accuracy. Verifying such climate forecasts poses special problems because of the length of time necessary to accumulate a meaningful ensemble of cases.
Another opportunity is created by the notable advances in atmospheric chemistry over the past 20 years. The rapidity with which the scientific community established the cause of the Antarctic ozone hole only a few years after its
discovery and impressive advances in both global and mesoscale chemical modeling demonstrate that the necessary links between dynamics and chemical processes are increasingly well understood and that operational models are feasible. The discipline of forecasting applied to air quality modeling will lead to improved understanding of the factors that produce fluctuations in chemical constituent concentrations. As the forecasting process matures, it will provide continuing benefits in managing the environment of urban areas and assessing the consequences of chemical perturbations induced by pollutants, fires, volcanoes, and other environmentally significant events.
A third opportunity arises with space weatherthe collection of phenomena associated with emissions of energy and mass from the Sun and instabilities in the Earth's magnetosphere. Space weather events can produce considerable disturbances in the ionosphere, which in turn induce communications disruptions and produce transient induced currents that lead to failures of electrical power networks. The streams of high-energy radiation can also result in satellite malfunction and may be lethal for astronauts in space without sufficient protection. Today, the increasing societal dependence on more than 250 satellites in geosyn-chronous and low-Earth orbits for communication, navigation, and Earth observations creates a new urgency for understanding and predicting space weather. This urgency will increase as hundreds of new satellites are placed in orbit for global cellular phone and other information transfer services.
Space weather phenomena have been predicted for many years using statistical methods, primarily by the military. Today, advances in observations of solar phenomena and the solar wind can be coupled with increasingly quantitative research models of the solar-terrestrial system to develop an operational space weather forecasting system.
In these three examples, as with prediction of the weather a few days in advance, the relentless discipline of forecasting can be expected to stimulate the interplay between improvements in observation, theory, and practice required to develop predictive capabilities of broad value to society.
Atmospheric Research Recommendation
Initiate Studies of Emerging Issues
The research community and appropriate federal agencies should institute interdisciplinary studies of emerging issues related to (1) climate, weather, and health; (2) management of water resources in a changing climate; and (3) rapidly increasing emissions to the atmosphere.
A number of interdisciplinary scientific issues are emerging as important candidates for attention by atmospheric scientists.
Climate, Weather, and Health
Both threats to human health or other biological systems and infectious diseases have been linked to climate variability (Shope, 1991; IPCC, 1996; Patz et al., 1996). Temperature, rain, sunshine, wind, humidity, and soil moisture may affect the emergence and spread of infectious diseases (Landsberg, 1969; Colwell and Huq, 1994; Epstein, 1995; Morse, 1995; Patz et al., 1996). Climatic factors provide limiting conditions for the distribution of vector-borne diseases; weather events can determine the timing, outbreak, and spread of disease. For example, air temperature controls the latitude and altitude distribution of mosquitoes that are vectors for encephalitis, meningitis, dengue, and yellow fever. In the tropics, rainfall controls the emergence of the anopheles mosquito, the vector for malaria.
The increasing frequency of water-borne diseases, such as typhoid, hepatitis, and bacillary dysentery, is associated with flooding. A direct threat to human well-being arises because the Earth's protective stratospheric ozone shield has been decreasing (WMO, 1995), leading to detectable increases of ultraviolet (UV-B) radiation at the Earth's surface (Herman et al., 1996). Thus, health officials expect a consequent increase in skin cancer, weakened immune systems, and other health-related concerns (Taylor et al., 1988; Cooper et al., 1992; International Agency for Research on Cancer, 1992; Johnson and Tinning, 1995).
Understanding the links between weather, climate, and various diseases and their vectors will require collaboration among a number of disciplines, particularly epidemiology and entomology. As this understanding increases, multi-disciplinary observation technologies, sophisticated numerical modeling techniques, timely weather and climate analyses and predictions, and international cooperation in environmental research will combine to offer new tools in combating disease.
The design and operation of water resource management systems have long taken account of expected extremes in both weather and climate conditions. Now, however, population expansion coupled with the possibility of climate change mandates that systems for water supply or flood control and mitigation be designed for evolving, rather than stationary, weather and climate conditions. The effects of possible climate change on the spatial and temporal distribution of floods and droughts, as well as on patterns of precipitation, temperature, and wind, must be better understood to improve water management designs and strategies. Thus, the stability of climate and possible changes in the statistical structure of extreme weather events are important new issues for water resource management.
Moreover, the operation of water management systems is evidently linked to both weather and climate forecasts; thus, issues of predictability become impor-
tant. Forecast lead times and operational flexibility in managing water systems interact to determine the range of possible actions and the possible benefits.
Atmospheric involvement in water resources extends to certain aspects of the nation's ground water supply as well. Emerging research problems involve aspects of multiphase ground water flow and coupled mass-biochemical contaminant transport that come into play as rainwater is absorbed by the ground and transported laterally through underground rock strata. Robust description and understanding of these complex phenomena involve laboratory and field experiments and the development of mathematical and numerical models capable of simulating unsteady processes with time constants varying over many orders of magnitude.
With the atmosphere providing the rapid transport portion of the hydrological cycle, it is evident that atmospheric science, in collaboration with hydrology and soil science, must focus more sharply on the complex issues associated with understanding, predicting, and managing the fluxes of water on which society depends.
Part of the World Climate Research Program, the Global Energy and Water Cycle Experiment (GEWEX), was initiated in 1988 to observe and model the hydrological cycle and energy fluxes in the atmosphere, at the land surfaces and in the upper oceans. GEWEX will significantly increases our understanding of the water-energy cycle and thus provide the basis for a more sophisticated water management system.
Rapidly Increasing Emissions to the Atmosphere
Beginning with the Industrial Revolution, environmental quality has been increasingly threatened. With the expansion of worldwide industry, new threats to environmental quality pose new research questions. One example is the economic development along the western rim of the Pacific Ocean and the concomitant increase in the emission of gases produced by combustion and industrial processes. The rapid expansion of emissions in this region is now, or will soon be, mirrored elsewhere around the world during the twenty-first century, a trend that seems likely to result in chemical and climate impacts on both regional and global scales.
These increasing emissions pose new questions concerning the regional, hemispheric, and global consequences of rapidly intensifying sources of pollution. What will the fate of these materials be as they move across the globe? Will they reach the Arctic? Will they wash out and fertilize specific regions of the ocean? How and at what rate will gaseous sulfur convert to particulates? Will dispersion and conversion rates differ in El Niño years?
The potential impacts of these emissions, both globally and locally, merit considerable study. To be effective, this study should be carried out in a highly collaborative mode, with social scientists, economists, and others.