PAPERBACK
\$24.95

• #### INDEX 189-194

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001

Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 94
The Role of Modeling in Unclerstanding 6 Tropospheric Chemical Processes BY R. DICKINSON AND S. LTU PRINCIPLES OF MODELING We have just discussed the various processes entering into tropospheric chemical cycles. These processes are studied individually to better understand them and to improve our descriptions of the quantitative relation- ships between the different variables entering into a process. These relationships are never exact representa- tions of reality. They are subject to continuing improve- ment using better measurements and new insights. These mathematical relationships between various components of a process are referred to as "process models. " It is useful to distinguish between those variables of a process system that are internal to that system, that is, calculable from the process model, and those that are external, that is, prescribed from observation. In partic- ular, the linkages to other processes are external varia- bles in the formulation of an individual process. If all the linked processes of the tropospheric chemical system are considered together with their linkages in- cluded as internal variables, we have a "system model. " A system model still contains some external parameters, but if the model is sufficiently comprehensive, such pa- rameters can be reliably prescribed for current condi- tions. The system model has two basic functions. First, be- cause it has maximized the number of variables it calcu- lates, it provides a good opportunity to carry out exten 94 sive comparisons between the model calculated versus observed variables of the system; such comparisons help identify weaknesses in the individual process models. Second, it allows projections as to the future state of the tropospheric chemical system as various external pa- rameters change with time. Of special interest in this context are external changes imposed by human activ- ity, but also of interest is long-period natural variability in external conditions. Because of the complexity of system models, they are generally integrated by means of computer programs. One important aspect of such integration is the develop- ment and use of numerically accurate procedures for solving the differential equations that are used to define the process models and, hence, system models. Tropo- spheric chemistry shares with meteorology a concern for a wide range of interacting scales, beginning on the scale of individual microeddies, e.g., within a smoke plume from a power plant, and ending in the global scale. Satisfactory parameterizations of the role of smaller scale processes should be one of the objectives in devel- oping and improving system models of global tropo- spheric chemistry. EXISTING MODELS The generality and detail possible in a complete sys- tem model are limited by difficulties of interpretation as

OCR for page 94
THE ROLE OF MODELING a result of its extreme inherent complexity and the large demands placed on computer and programming re- sources. Up to now, the necessary staff and institutional support to pursue a complete system model of tropo- spheric chemistry have not been developed. Further- more, there have been considerable uncertainties in the individual processes. Consequently, the research tools that have been used in tropospheric chemistry studies for synthesis and interpretation lie between the concept of a process model and a complete system model. We shall refer to these tools simply as tropospheric chemistry models. There has been developed a wide range ofthese models, with their content depending on the interests and objectives of their developers as well as their access to computing resources. Basically, what distinguishes the models now used from a complete system model is that they attempt to model accurately only some of the processes of the system, the others being included only in a simplified or ad hoc fashion. Existing models can be best classified by the issues they address. Usually these models consist of two major parts: the chemistry and the transport. Depending on the characteristics of the subject being studied, each model employs different degrees of sophistication in the treatments of chemistry and transport. At one extreme are meteorologically oriented models that obtain the motions ofthe atmosphere and temperature structure as three-dimensional time-varying fields by solution of the continuum equations of hydrodynamics and therrnody- namics in response to realistic boundary conditions; however, these models have until recently approximated tropospheric chemistry by ignoring all species except water. Some studies are now under way using the winds generated by some regional and global meteorological models to provide transport for simplified chemical models. At the other extreme are the one-dimensional or box chemical models. As a consequence of their ex- tremely complex processes of chemical species transfor- mations, their transport consists essentially of empiri- cally derived time scales for movement of species from one box or level of the model to another. At the current state of development, one ofthe promi- nently distinguishing features of chemical models is their dimensionality. Thus there are zero-, one-, two-, and three-dimensional models. The zero-dimensional box models simulate laboratory chemical reaction mea- surements such as in smog chambers. Reactants in the reaction chamber are assumed to be completely mixed so that transport can be neglected. Chemistry is treated in detail by including all relevant elementary reactions. Usually Gear's code with small integration time steps is used to study the time-dependent behaviors of all reac- tants. Multidimensional models can be viewed as alarge number of zero-dimensional models, coupled together 95 by transport and radiation submodels, and each differ- ent because of different transport source-sinks and dif- fering environmental conditions. In the modeling of the global tropospheric photochemistry, emphasis has been on the problems of 03, CO, CO2 (or carbon cycle), nitrogen cycle, and sulfur cycle. Box models and one-, two-, and three-dimensional models have been devel- oped to study the natural chemistry and possible effects of anthropogenic activities on these species. Because of the computer resource requirements for three-dimen- sional modeling, full-scale chemistry has not been in- cluded. On the other hand, simpler models with full- scale chemistry usually do not successfully parameterize the important transport processes and hydrological cy- cle ofthe atmosphere. Besides dimensionality, treatment of model time structure is also notable. For example, is the model steady state or capable of following transient changes? How does it treat diurnal and seasonal variations? Some models only calculate fast chemical transformations but prescribe as external, slowly changing species. Such models avoid the need for transport submodels because transport is primarily important for determining the distribution of the slowly changing species. Other models prescribe the species with fast chemistry, in par- ticular OH, as external, and concentrate on the interac- lion among source, sinks, and transport of slow species. One important distinction with regard to model objective is the difference between climatological and event models. This distinction arises because ofthe large day-to-day variability of meteorological processes, including transport. Thus a detailed case study of the processes of tropospheric chemistry over several days or less must recognize the actual transport occurring over that interval, either by explicit measurement of it over the interval, or by measuring enough initial meteorolog- ical data to allow integration of a weather forecast model for the time and space domains of interest. On the other hand, if a study is more concerned with the average behavior of the atmosphere as described by means and higher statistical moments, then there is less demand on temporal accuracy in providing the meteorological transport terms. The most effective tools in this instance are the general circulation models that obtain from first principles the statistical properties of the atmosphere by direct numerical simulation. That is, they calculate day- to-day weather variations over a long period oftime that do not correspond to any particular time period but are supposed to have the same statistics as actual weather systems. In other words, they model the climate of the atmosphere system. Much of the work on three-dimensional modeling of tropospheric chemistry up to now has been on the urban and, more recently, regional scale. The chemical models

OCR for page 94
96 developed for these studies should be considered in developing models for global tropospheric chemistry. Dispersion models have been developed to study the dispersion of air pollutants from power plants as well as cities. These models are concerned with urban-scale transport but usually ignore chemical transformations of air pollutants. Models that do include air pollution chemistry but employ simpler transport pararneteriza- tions are called air quality models. Most of these models are developed for metropolitan and suburban areas; occasionally regions as large as the eastern United States have been included. Because air quality models emphasize the oxidant problem, the chemistry usually includes that of 03, NOx, and hydrocarbons. Studies using smog chambers have led to the development of detailed mechanisms for specific hydrocarbons. Unfortunately, the chemistry in these mechanisms is far too extensive to be incorporated into air quality computer models. In order to circum- vent this problem, at least two approaches have been utilized. These involved "lumping" the hydrocarbons by classes and using a generalized reaction mechanism for these classes, or by using a carbon-bond approach, which partitions the chemical species on the basis of the similarity of their chemical bonding. These chemical models have been tested against and tuned to a variety of smog chamber data. Usually good agreement is achieved between measured and predicted concentra- tion-time profiles for all measured species. When these reaction mechanisms are incorporated into air quality models and compared to field measurements, the agree- ment becomes much poorer. Discrepancies could be due to poor transport pararneterizations, but there is little doubt that lack of understanding of the chemistry in the real atmosphere also contributes to the discrepancies. In particular, the chemistry of aged and diluted air pollu- tants may be poorly understood because it cannot be effectively tested in smog chambers. Furthermore, het- erogeneous reactions are either not included or treated by oversimplified parameterizations. Acid deposition models are used to study wet and dry deposition of acid material such as sulfur and nitrogen compounds. Acid deposition models have been devel- oped for Europe and the eastern half of North America. The major objective of these models is to establish the source-receptor relationship of acid deposition. So far, very little chemistry is included in the acid deposition models. Constant, linear conversion rates of SO2 to SO4- and NO2 to NO3 have been used. Wet scavenging and dry deposition are assumed to be independent of cloud types or topography. Most of the modeling effort is focused on the development ofthe meteorological aspect of the model. There is a clear need to incorporate into these models the full-scale fundamental chemistry and PART II ASSESSMENTS OF CURRENT UNDERSTANDING kinetics involved in the transformation of sulfur and nitrogen compounds. In conclusion, as the field of tropospheric chemistry matures, the various kinds of models should tend to converge more toward ideal system models. This occurs, on the one hand, as modelers learn to treat more elaborate model systems and, on the other hand, as they are able to understand the error implied by various con- venient approximations and hencejustify these approxi- mations when their implied error is acceptable. Global modelers and regional modelers should collaborate on those aspects of their models that are of common interest. MODELING IN SUPPORT OF THE PROPOSED RESEARCH PROGRAMS We discuss here the modeling programs required to provide guidance to and help synthesize the results ofthe research programs proposed in Part I ofthis report. The programs in biological sources, photochemical proc- esses, and removal processes require the development of submodels describing the individual processes involved. These would serve three purposes: (1) help to under- stand better the individual process, (2) help to extrapo- late from individual observational sites to regional and global averages, and (3) provide submodels to be used in a comprehensive three-dimensional meteorological model coupled to global tropospheric chemistry. By con- trast, the global distributions and long-range transport program would be used to help validate the overall per- formance of the chemical aspects of comprehensive three-dimensional models of tropospheric chemical processes. As submodels are developed for the various subprograms, they will be incorporated into the com- prehensive models. Biological and Surface Source Models The models required to support the biological and surface source subprogram fall into three categories: (1) global empirical models, (2) mechanistic models of bio- logical processes, and (3) micrometeorological and oce- anic models of surface transport processes. The observational efforts in the biological source sub- prograrn will provide measurements at individual field sites. Initial exploratory efforts will identify the ecologi- cal communities that provide significant emissions, but as a second stage, it will be necessary to obtain sufficient observations to determine annual average emissions at various sites. Variability with environmental parame- ters such as temperature, solar radiation, and moisture will also be obtained. However, due to the great variety and small-scale structure of biological systems, it will

OCR for page 94
THE ROLE OF MODELING always be very difficult to collect sufficient data to permit straightforward numerical averaging to establish regional and global average emissions. Rather, more sophisticated approaches will be required to interpolate and extrapolate the available observations to all the non- sampled areas. Exactly the same problem occurs in summarizing other types of data obtained from ecologi- cal communities. Ecologists, in particular, have been forced to resort to empirical procedures for obtaining such parameters as net primary productivity and bio- mass carbon (e. g., Table 5. 1) from a ratherlimited num- ber of field sites (less than 100~. One systematic proce- dure has been to correlate ecological data with readily available climatic parameters pertaining to the sampled sites, in particular rainfall and temperature. The corre- lations so obtained are used to transform global maps of climatic parameters into global maps of ecological parameters. Such a procedure will be used to develop empirical models (i.e., maps) for the global distributions of the various measured biological emissions. Complementary to the development of global distri- butions of biological emissions will be the development of models of the detailed biological mechanisms and processes responsible for the measured emissions. These will range from models of soil or oceanic biochemical processes to models of whole leaf physiology. Their development will require intensive collaboration with experts in other biological and chemical areas outside the atmospheric chemistry community. These efforts will, however, differ from current and past modeling in these other disciplines in the following aspects. First, they will be focused on the processes responsible for providing atmospheric emissions. Because these emis- sions have for the most part been recognized only recently, or in some cases not yet, the other disciplines have only begun to consider how such emissions could be provided from their existing submodels. Second, this modeling effort will be focused on the whole biological - system (plant-soil-microorganisms, etc.) as it interacts with the atmospheric environment. Because of the great complexity of the processes involved, a model of the whole biological system will undoubtedly require sim- plifications in the descriptions of biochemical processes and the treatments of differences between species of organisms. Modeling the effects of soil microorganisms would require modeling the environment where the processes occur. For example, the question of methane production requires a model of the diffusion of CO2, H2, and CH4 from the production site to the atmosphere to address the question as to whether increasinglevels of CO2 could increase methane production. Boundary layer and surface transport models are required to describe the movement between ocean or 97 land surfaces and the atmosphere. In the case of oceanic processes, such models require consideration of oceanic as well as atmospheric boundary layers and the effects at the ocean interface of wave breaking, i.e., the move- ment of air bubbles on the ocean side and spray droplets on the atmospheric side. The transfer of gases from and to a surface generally involves near-stationary diffusion-like transport proc- esses that are represented in terms of effective resistances or conductances. That is, if ca represents the concentra- tion of a gas in the atmospheric mixed layer, and this gas is maintained at some concentration cs at some surface, then the rate of flux of the species to the mixed layer from the surface is modeled as given by (cs-ca)/r`, where rat is the total resistance of the diffusion processes between the atmospheric mixed layer and the surface. The most thoroughly studied gas transfer process, for example, is that of water vapor from soil and foliage. The water vapor concentration next to the mesophyllic cells inside a leaf is that of saturation at the temperature ofthe leaf. To reach the mixed layer, the water must pass through leaf stomata, the leaf boundary layer, the leaf canopy, and a roughness sublayer above the canopy before reaching the atmospheric mixed layer. Each of these barriers is modeled by one or more resistance in series or parallel. This description in terms of resistance, although somewhat simplistic, provides the maximum level of detail that can be practicably matched to models of atmospheric transport above the mixed layer and validated by micrometeorological observations. In most cases, the surface boundary conditions are not as easily modeled as that of water vapor. Surface boundary conditions for species of interest would be one of the practical outputs of the modeling in the biological source subprogram. For example, rather detailed models are now available for leaf photosynthesis that provide the concentrations of CO2 within the leaf cavity. Plants exert physiological control over water losses through stomata! closure; the stomata! resistance is sig- nificant not only for leaf exchange of water and CO2 but also for SO2, NO2, 03, and NH3, and is modeled in terms of soil moisture and root resistances. Detailed boundary conditions for other gases, in particular SO2, NO2, 03, and NH3, and biologically emitted sources need to be established. Field programs, together with continuation of labora- tory (i.e., wind tunnel) studies should provide the data needed to develop and refine models of oceanic gas transfer processes, in particular for those species where surface boundary layers within the ocean provide addi- tional resistance to their flux between ocean and atmosphere. The micrometeorological processes of gaseous and particulate exchange within complex vegetated cano

OCR for page 94