SCOPE 35 - Scales and Global Change

ABSTRACT

Climate models, particularly General Circulation Models (GCMs), could serve as a basis for the construction of models for biosphere-geosphere interactions. Such an approach requires the development of methods for how to describe ecosystem processes on the scale of 100 x 100 km, i.e. how the processes on a smaller scale could be parameterized. This fundamental problem in model building has been addressed in the development of parameterization schemes for physical processes in climate models. Present methods used for  describing convection and cloud formation and for considering surface boundary layer processes are summarized. The problem of how to incorporate the role of terrestrial biota for the exchange of heat, moisture, and chemical compounds between the atmosphere and the ground in global models is discussed on the basis of existing ecosystem models that have been designed for analysis of local problems.

INTRODUCTION

The original definition of the term 'biosphere' referred to the part of the earth system in which there is life, i.e. the lower atmosphere, the top layers of the soils, the oceans, and the uppermost layers of the ocean sediments. It is interesting to note that meteorologists, physical oceanographers, and climatologists today call this thin shell around the earth the climatic system. In their pursuit to develop climatic models for this system, chemical and biological processes have been almost completely ignored. Climate models rather describe the thermodynamics of the atmosphere and the oceans as they are being driven by the spatially varying distribution of solar radiation and with due regard to the importance of the rotation of the earth. Climate models describe the physical behaviour of the biosphere.

With the major advances in our understanding of how the motions of the atmosphere and the oceans are being maintained as a background, we now ask ourselves: How do the life processes on land and in the sea, i.e. at the interfaces between the atmosphere and land and between the atmosphere and the sea, modulate the climate on earth? To answer this fundamental question we must broaden our approach and consider the dynamics of the biosphere in its entirety, i.e. the interplay between physical, chemical, and biological processes on earth.

It is well-known that present conditions for life on earth are, to a considerable degree, the result of the evolution of life on earth during hundreds of millions of years. This evolution is a truly global change of the biosphere, but a slow one. On shorter time scales we know about the variations between glacial and interglacial times that have occurred during the last one million years. We know particularly well the change from the last period of glaciation to the present interglacial period that took place 8000 to 15000 years ago. Undoubtedly a close analysis of this major climatic change is of great interest in the study of biosphere dynamics as defined above.

The concept of 'Global Change' now is becoming a major issue in the environmental sciences due to a concern about the present rapid change of the environment. The focus is on the problem of biosphere dynamics on time scales from a decade to a few centuries. Although we certainly need to learn about natural variations of the biosphere in the past to understand the changes that are taking place today, we shall here primarily address the problem of physical, chemical, biological interactions on this short time scale. We shall assume that the basic geological setting is not changing except for possible changes of the topsoil by direct exploitation of natural resources by man (e.g. deforestation and changing land use) and changes due to the natural response of ecosystems to changing abiotic conditions such as climate.

MODEL HIERARCHY AND SUBGRID SCALE PARAMETERIZATION

Some basic principles

O'Neill (1988) has discussed in some detail theories of hierarchies of ecological systems. He emphasizes that it is most important in developing models for ecosystem behaviour to define carefully the particular problem to be addressed and the scales in space and time that will be considered explicitly. Although processes on smaller scales may be essential for the behaviour of the system on the scale of direct concern, it is usually not feasible to proceed by considering the role of these subgrid scale processes by modelling them through successively increasing the scale (decreasing resolution) starting from the smallest observable scale. We should instead attempt a direct formulation of the role of subgrid scale processes for the processes dealt with explicitly in our model. Usually much of what happens at smaller scales is of no importance in this context and should accordingly be disregarded. Our objective must not be to analyse all aspects of the processes on a smaller scale but merely those of direct importance for the problem being considered.

We are concerned with global processes in the present context. To develop models for biosphere dynamics on this scale it is important to make use of the experiences gained in climate modelling. The most advanced models of the climate system (so-called general circulation models, GCMs, cf. e.g. Manabe and Stouffer, 1980) describe processes in the atmosphere and the sea by resolving features on scales larger than a few hundred kilometres in the horizontal directions and one or a few kilometres in the vertical. Altogether some 100 000 gridpoints are used, in the most detailed computations with such models, and even the advent of the next generation of computers will not permit better horizontal resolution than about 100 km in the case of long-term integrations. As a matter of fact, much cruder models of the climatic system are often used, particularly for principle studies aimed at a better understanding of the interplay and relative importance of different fundamental processes. It is accordingly necessary to develop methods for how to consider ecological processes (at the atmosphere-land/ocean interface) as well as the transfer of important chemical compounds within the atmosphere and the sea by using models with a resolution of a few hundred kilometres in the horizontal directions. Climate models resolve, on the other hand, processes in time with much detail. Changes are computed by forecasting the changing weather from day to day but the analysis of the results is restricted to the consideration of the statistical manifestation of weather such as the seasonal or annual mean temperature, precipitation and cloudiness, the characteristic variability of weather phenomena as well as the motions of the sea and their transfer of heat, momentum, and water in the atmosphere and the sea. It is therefore possible in the model to relate statistically temporal variations of ecosystem behaviour to meteorological variations if required.

Our problem thus becomes how to consider statistically the importance of those biological and chemical processes on smaller spatial scales that influence the large-scale features of the system. Before doing so it may be of some interest as an illustration to describe the approach used when parameterizing physical sub-grid scale processes in climate models. We note for example that the formation of clouds and precipitation, particularly in statically unstable air masses where convective clouds are formed, show important features on scales of the order of kilometres. Also, vertical exchange of heat and momentum both in the atmosphere and the sea next to the atmosphere/land and atmosphere/sea interfaces is accomplished by small scale turbulence, with characteristic eddy dimensions of a few hundred metres.

It is first important to recognize that climate models basically treat the flux of energy, momentum, and water. Therefore, when parameterizing the role of sub-grid scale processes we need primarily to consider their role in this regard. If we are successful in formulating models for the large scale behaviour of the biosphere in this way, we may use these deduced features as given when attempting to analyse how small scale characteristics of the biosphere are generated and maintained.

Parameterization of cloud formation and precipitation

Clouds form when the humidity reaches 100%. For precipitation to occur, however, ice nuclei or a broad size spectrum of cloud water droplets with increased probability for coalescence are required. These conditions apply to the microscale and since the moisture field in the atmosphere shows large sub-grid scale variations they cannot be applied directly by using mean values for temperature and humidity on the scale of resolution of the climate model. A crude description of clouds and some conditions for precipitation release is, however, obviously necessary, not only because these phenomena themselves are important meteorologically, but also because cloud formation implies transformation of latent heat into sensible heat, which may mean a drastic change of the vertical static stability of the air. This in turn changes the conditions for vertical transfer of energy, momentum and water, which is of prime importance for the large-scale dynamical behaviour of the system. Let us limit ourselves to considering the formation of convective clouds.

Convection occurs when the vertical temperature gradient (lapse rate) exceeds the temperature decrease that a rising air parcel experiences, which is about 1°C km^-1 in non-saturated air (dry-adiabatic lapse rate) and 0.4-1.0 °c km^-1 in saturated air (moist-adiabatic lapse rate), the value being smaller the higher the temperature is. Such vertical instability is due to the changing temperature distribution brought about by the large-scale motions of the atmosphere, or by heating or cooling due to radiative processes, or heat exchange with the underlying surface. Convection is a rapid process compared with the rate of change of large-scale systems and one may therefore, to a first approximation, assume that the process of vertical equilibration of a convectively unstable air column takes place instantaneously. The new state thereby created is characterized by neutral static stability and complete mixing within the vertical layer , where the instability develops and implies vertical transfer of heat, moisture and momentum when this vertical mixing is accomplished.

Convective clouds have a horizontal dimension of a few kilometres, which is small compared to the resolution of the global model. Only a part of the grid area is therefore usually covered by clouds, and downward motion takes place in the clear areas between the clouds. Obviously the average relative humidity on the scale of the grid is therefore less than100%. In an area with deep convection it may merely be about 80%, while it is closer to saturation in the case of shallow convection. To parameterize the conditions for initiation of convection we must determine empirically the grid scale features (e.g. relative humidity) that should prevail in order for convection to occur. Reasonably satisfactory statistical formulations have been developed in this way and been tested by extensive model experimentation (see e.g. Ogura, 1985).

Atmospheric boundary layer modelling

The role of the atmospheric surface boundary layer for the large-scale motion of the atmosphere is primarily associated with its vertical transfer of momentum, heat, and moisture between the surface of the earth and the free atmosphere due to small scale turbulence. (Transfer of biologically important constituents through the boundary layer also takes place and will be considered later). We are not interested in the detailed structure of the turbulent eddies that bring about the transfer, but merely need to account for the transfer as accurately as possible. One way is, for example, to make use of Monin and Oboukhov similarity theory (cf. Panofsky, 1985) which, on dimensional grounds, permits a reduction of all possible quasi-steady state vertical profiles of wind and temperature to two universal profiles. With the aid of these, knowledge about the characteristic features of the earth's surface (roughness, temperature, and wetness), and wind, temperature and humidity in the atmosphere above the boundary layer, the vertical fluxes can be determined. A simple parameterization of transfer through the boundary layer is possible in this way. Considerably more complex parameterization schemes are, however, often used (cf. e.g. Panofsky, 1985).

The land surface within an area of the size of the grid used for large-scale modelling is seldom homogeneous. Since we are only able to describe the average vertical wind and temperature profiles in a grid and also need to assign average characteristic values for surface roughness, surface temperature and wetness the question arises how these should be determined. Presumably there are not simple arithmetic averages over the area considered but probably depend non-linearly on more detailed features of the surface and the bulk values for wind and temperature in the free atmosphere. Not even the most advanced climate models include any sophisticated treatment of such features. As a matter of fact roughness has often been assumed not to vary at all over land in most experiments with climate models. Similarly soil moisture has been modelled very approximately without much consideration of physical and biological process of importance for evapotranspiration. Still, to a first approximation these models describe the role of boundary layer processes for the large-scale atmospheric motion systems reasonably well. The problem will be further considered in the next section.

A closer comparison between the present distribution of climate on earth and model simulations using the most advanced climate models shows significant discrepancies. It is difficult to determine to what extent these are due to inadequate treatment of the internal dynamics of atmospheric or oceanographic processes or to improper treatment of the transfer processes between the atmosphere and the underlying surface, particularly the irregular land surface. It seems plausible that the latter may be of some significance. We need to analyse present climate models with regard to their sensitivity to the treatment of exchange processes at the earth's surface and presumably develop more realistic models for their description, including the role of biological processes, to find out if exchange processes between the atmosphere and the land surface are adequately treated.

It is important to emphasize that the reverse problem, i.e. the determination of how detailed features of climatic conditions in the boundary layer are created by sub-grid scale irregularities of the land surface, requires a more cumbersome analysis (Gates, 1985). It is, however, obviously necessary that mean conditions on the scale of the grid of the climate model agree with observed climatic conditions to expect that successful simulations of subgrid scale features can be made.

MODELLING BIOTIC PROCESSES IN GLOBAL MODELS OF THE BIOSPHERE

Principal considerations

Let us, in the following, limit ourselves to terrestrial processes. It has already been emphasized that it is not feasible to develop a hierarchy of models starting from the species level and proceeding step-wise to the scale of concern which is required to treat large-scale dynamics of ecosystems. The scientific challenge is, rather, to attempt the formulation of simple hypotheses and try to verify to what extent they may be applicable. Failures in such an approach still means increased insight.

We shall use the experiences gained in developing climatic models as the basis when treating biological and chemical processes. In addition to the consideration of transfer of energy, momentum, and moisture, as now included in climatic models, we also need to consider

• the modulation of water flux between land and the atmosphere due to the presence of terrestrial biota

• the role of soil processes for transfer between the atmosphere and the land surface.

• the role of terrestrial biota for the generation and destruction of chemical constituents in the atmosphere

• the transfer of chemical constituents in the atmosphere

• chemical, often photochemical, interactions within the atmosphere

The prime objective is: How to model the relevant processes and how to develop appropriate parameterization with due regard to the scales we wish to resolve, i.e. those larger than a few hundred kilometres.

The problem of parameterizing fluxes does not only concern the development of methods to average spatially over inhomogeneous land surfaces, but also to capture the importance of long-term changes of biota and soils which often in reality are obscured by the variations that occur on short time scales i.e. those associated with the daily cycle, weather variations, and the annual cycle. The role of such changes of the characteristics of biota and soils for example due to slow changes of climate has seldom been considered. Important feed-back mechanisms may thereby have been excluded in the analysis of long-term changes of climate. In the following sections we shall consider the transfer processes between the atmosphere and land with particular consideration of the role of the biota. A number of microclimatological models dealing with the detailed transfer problem have been developed in recent years. We shall discuss the global modelling problem with these local models as a starting point. It is appropriate first to consider the exchanges of energy and water , which are closely coupled.

Energy and water transfer between the land surface and the atmosphere 

For detailed analyses of energy and water transfer between the land surface and the atmosphere reference is made to Jones (1983), Dickinson (1984), Dickinson and Hanson (1984), Verstraete (1985) and Sellers, Mintz, and Sud (1985). We shall not reproduce their results here, but merely emphasize some principally important points. The fundamental processes to be considered are illustrated in Figure 7.1. The heat balance of a thin surface layer of soil may be written (Verstraete, 1985)

  where

R_n = net energy gain due to radiative flux
H = sensible heat flux from the land surface to the atmosphere
LH = latent heat flux from the land surface to the atmosphere
 S_h = storage of heat in the top soil layer
G_n = heat flux from below into the top soil layer
Similarly we can formulate the moisture balance of this top layer as follows 

 where

P = precipitation rate
E = evaporation rate
R_w = runoff rate
S_w = storage of water in the top soil layer
G_w = water flux from below into the top soil layer

Figure 7.1 The fluxes of water (on the left side) and energy (on the right side) in the atmospheric surface boundary layer and the topsoil as modulated by the presence of a vegetation cover (after Dickinson, 1984)

Within the atmosphere water vapour flux (E) and heat flux (LE + H) are maintained by turbulent processes which are included in present climate models. In the soil, water is transferred by water diffusion, which depends on the soil water potential and which in turn is a function of soil porosity. Heat flux is primarily a molecular diffusion process, but to some extent also dependent on the water transfer in the soil. In addition, however, water is transferred by the plants extending their roots into the ground and their above-soil parts into the atmosphere. If the ground is covered by vegetation most of the water flux (E) is due to evapotranspiration from plants, which obviously also affects the heat transfer between the ground and the atmosphere. The water flux due to the presence of plants depends on plant processes, particularly the stomatal conductance g_s, and the water uptake by the roots. Following Jones (1983) we may write.

where

g_o = minimum conductance
g_r = maximum less minimum conductance
F_R = influence of solar radiation
F_S = specific humidity factor
F_P = water potential factor
F_T= temperature factor
S = the ability of the root system to supply the water that the leaves would be able to evaporate under prevailing conditions.

All the latter five factors are between zero and unity and their magnitudes depend on atmospheric conditions (temperature, humidity, solar radiation) and on soil and biota response to such abiotic conditions. Similarly, we need to formulate the transfer of water from the soils to the roots which depends on the potential for water transfer into the root, and on a root geometry factor which takes into account: root length per unit volume of soil, distance between roots, root radii, and root depth.

Rather than considering these processes in some detail as outlined above most climate models treat the air-land surface exchange as well as transfer processes within the soil in a very simplified manner usually not including biological processes at all. Evapotranspiration is simulated by evaporation from a soil moisture reservoir which has a prescribed maximum storage capacity. Overflow is considered as runoff and the rate of evaporation depends on the degree to which this soil moisture reservoir is filled. Accordingly G_w = G_n = 0 in equations (1) and (2) and no further considerations of soil and biota processes are required.

A more elaborate treatment would be to prescribe characteristics of the soil (e.g. porosity, heat conductivity) and of the biota (e.g. minimum and maximum water conductance, their sensitivity to temperature, humidity etc., root geometry and water uptake capability). A simple model for transfer processes in the soil below the top soil layer is then required to describe the exchange of heat and moisture with deeper soil layers by molecular diffusion.

Microclimatological models of this kind have been developed (cf. Verstraete, 1985). Considerably more computing capacity is, however, required include such a description in a general circulation model and we may legitimately ask if soil and biota characteristics are known well enough to justify such an effort. It should, however, be done for exploratory purposes. The most advanced model of this kind, which has been developed so far is due to Dickinson (1984).

The role of ecosystem processes in global models

The crucial question to address concerns the mutual interplay between abiotic factors that describe atmospheric conditions, i.e. weather and climate (solar radiation, temperature, humidity, cloudiness, precipitation, etc.) and the inorganic substrate (soil conditions) on one hand, and the characteristics of the biota including the organic component of the soil on the other. Ecological models of the kind described briefly above have been reasonably successful for the local scale. Dickinson (1984), Dickinson and Hanson (1984) and Sellers, Mintz, and Sud (1985) now propose to use similar models for the scale of the General Circulation Model grid. We should then consider them as conceptual models, but cannot expect that functional relationships and rate coefficients that have been derived for small-scale processes can be directly adopted when concerned with the scales of the order of 10⁴ km² (100 x 100 km). This difference between local models and means for parameterization of such processes for the scale of a grid square in a global model is in principle similar to our earlier observation that cloud formation and precipitation formation occur at a mean relative humidity in a grid squre well below 100% .

A serious problem then arises: How do we test our models? Testing against real data is necessary, but it is immediately obvious that such attempts constitute major observational efforts particularly because we need to consider spatially very heterogeneous land surfaces. The planning of an International Geosphere Biosphere Programme (IGBP) should attempt to define the observational design that would be optimal. Both satellite and ground observations are obviously required.

It is important to recognize, however, that it would also be desirable to develop testing procedures of the global model directly related to their use diagnostically, including biological and chemical processes as sketched above. We may assume as a first approximation that the present (or preferably the preindustrial) ecosystem distribution is (or was) in equilibrium with prevailing climate and ecosystem distribution. How well are we able to simulate these distributions when including the consideration of biotic processes? How sensitive are our results to the choice of functional relationships and rate constants that describe the ecosystem behaviour in the model? In this way we are, however, only able to deduce to what extent a given model is compatible with observed conditions but not to exclude the possibility that other model formulations might be superior. The approach is one of trial and error, but the testing of models in this way would undoubtedly increase our understanding of large-scale ecosystem dynamics.

Such an experimental approach must proceed step wise. It is advisable to start with very simple models, perhaps even simpler than the one outlined in the previous section. Some such analyses have been attempted and might be of interest in the present context.

(i) Dependence of surface temperature maximum on stomatal dynamics

It is well known that the daily variations of atmospheric temperature and moisture largely determine the rhythmn of stomata opening and closure. Since evapotranspiration thereby is modified, there is a direct feedback from the biota on moisture and temperature variations in the atmosphere. The weather forecasting model, developed at the European Centre for Medium Range Weather Forecasts, has been tested using different formulations of the process of evapotranspiration (L. Bengtsson, private communication). In one case a simple 'bucket' formulation was used in which availability of water for evaporation remained essentially the same throughout the day and the rate of evaporation was determined by the flux of water vapour through the atmospheric boundary layer. In another case an attempt was made to prescribe roughly the daily variation of the stomatal conductance being comparatively large early during the day and decreasing as the temperature increases an humidity decreases towards noon and during the afternoon. Since the time step of integration of a general circulation model is considerably less than an hour, the daily cycle of meteorological variables is obtained with good resolution. Experiments show that the maximum daily temperature is significantly higher in the latter case, because of decreasing evapotranspiration during the day caused by decreasing stomatal conductance. Because the ecosystem model is part of a large scale three-dimensional forecast model, the horizont, advection of air and other changes of weather is implicitly accounted for. If w accept that the daily variation of temperature in addition to being influenced by changing weather conditions is primarily dependant on the variations of stomatal conductance we can determine an optimum formulation for the dependence of stomata on atmospheric temperature and moisture by trying to find out how the best possible agreement can be obtained between daily temperature variations in the model and in reality.

(ii) The dependance of the monthly mean temperature on availability of water for evapotranspiration

Also, the monthly mean temperature in a region is greatly influenced by availability of soil moisture for evapotranspiration, which in turn, of course depends not only on precipitation but also on the characteristics of the plant cover (e.g. species composition). An ongoing climate change may well change these and thereby significantly modify the ratio of evapotranspiration to runoff .The rapidity with which this takes place depends on the rate of renewal of species in a plant community. A few experiments which have bearing on this problem have been carried out by Shukla and Mintz (1982). With the aid of a general circulation model for climatic studies they have determined the global surface temperature distribution in July that would prevail in the mean for two extreme cases:

a) constantly wet soils, i.e. enough water would be supplied for evaporation even if precipitation were inadequate
b) constantly dry soils, i.e. all precipitation would run off or penetrate into the soil and not be available for evapotranspiration.

Figure 7.2 a and b show the computed July temperatures for these two cases assuming the present distribution of solar radiation. The difference is shown in Figure 7.2c. The modulation of surface temperature due to the presence or absence of water in the soil is very significant, the difference being 20-30 ^oC in the interior of the major continents. At high latitudes present climate is close to that of wet soils, but in the subtropical arid and semi-arid regions of Europe and North Africa the climate is 10-20 ^oC hotter than in the wet case.

With these two extreme cases as starting points it would be most essential to explore in which way simple ecosystem models for different types of biomes would modify the climatic regimes and in this way get a first idea about the response characteristics between abiotic and biotic factors in the climatic system. It is particularly important to relate ecosystem characteristics (e.g. biomass, soil properties) to the regional characteristics of the hydrological cycle. Attempts in this direction have been made although not by combining ecosystem models and climatic models but merely considering the local scale. A comparison with climatic changes during the last 15000 years has also been made (Solomon et al., 1984)

Figure 7.2 July surface temperature (0 °C) as simulated by a General Circulation Model assuming present distribution of solar radiation, ocean surface temperature, and albedo of the earth's surface; (a) the soil is always wet, (b) the soil is always dry and (c) shows the difference between (b) and (a) (after Shukla and Mintz, 1982). (Copyright 1982 by the AAAS)

Biochemical interactions between atmosphere, soil, and biota

The composition of the atmosphere is due to biochemical processes in soil and biota as well as chemical and photochemical processes in the atmosphere. The atmospheric concentration of carbon dioxide is markedly influenced by the terrestrial and marine ecosystems and, in the long term perspective, also by geological processes. Photosynthesis of terrestrial plants, and accordingly CO₂ variations in the atmosphere, are directly related to evapotranspiration in that the water transfer from the plants to the atmosphere is in the opposite direction of the CO₂ flux from the atmosphere into the plant and both are regulated by the opening and closure of the stomata. To a first approximation we may assume that the humidity in the interior of the stomata is close to saturation, while the CO₂ partial pressure usually is small compared with the exterior atmospheric concentrations. We can accordingly relate the fluxes of CO₂ and water between the atmosphere and terrestrial ecosystems.

The transfer of some other gases is similarly related to the process of photosynthesis. Nitric oxide (NO) behaves, however, differently. If atmospheric concentrations are above an equilibrium concentration (compensation point) there is a transfer of NO into the plant, while the opposite is true for low atmospheric concentrations. The transfer processes are also dependent on the availability of fixed nitrogen in the soil and accordingly influenced by the application of fertilizers (cf. Johansson and Granat, 1984).

There is also exchange of sulphur in gaseous form between the atmosphere and plants (Hallgren et al., 1982). SO₂ enters the stomata, but for increasing ambient concentrations the total flux of sulphur into the plant decreases, possibly due to the formation and emission of reduced sulphur compounds.

Another set of gases is emitted from the terrestrial ecosystems because of decomposition processes in the soil. We shall not consider the details of the biochemical processes but wish merely to emphasize the importance of relating them to the abiotic, physical processes that are largely governed by atmospheric processes. We need, for example, to find ways of how to determine the rate of their emissions as dependant on precipitation and soil characteristics. There is a profound difference between conditions of dry and wet soils. In the former case primarily oxidized compounds (e.g. CO₂ are emitted, while in the latter case reduced gases are formed (e.g. CH4 and from basic soils, NH₃). How is the emission of nitrous oxide (N₂O) dependent on the climate? The problem becomes difficult because release and transfer of gases from the soil into the atmosphere is an intermittent process very much governed by the wetness of the soil. We need to establish conditions for how such emissions are related to the large scale parameters of a global model, probably in a satistical sense, in order to be able to analyse the regional and global biogeochemical cycling of elements and compounds. Criteria must be established using the large-scale variables to determine how one or the other process on the small scale is switched on or off .This is not simple and work of this kind has merely begun.

CONCLUDING REMARKS

The global biosphere is a complex system. To gain an understanding of its structure and dynamic features, it is necessary to not only increase our knowledge about the detailed processes, but also to develop models of how global interactions take place. Our attempts to analyse the detailed physical chemical, and biological processes need, in this context, be guided by  an advancement of our understanding of the latter. It is necessary to develop a strategy of data gathering that serves both these purposes. Climate research during the last decade may serve as a useful example of how to approach this difficult problem in a systematic way. While realizing the necessity of a systematic and long lasting effort of observing the atmosphere, the oceans, land, and life on earth, such programmes must remain flexible enough to permit those modifications and even sometimes improvisations that a necessary to maintain are viable programme.

REFERENCES

Dickinson, R. E. (1984). Modeling evapotranspiration for three-dimensional global climate models. In Climate Processes and Climate Sensitivity, pp. 58- 72. Geophys. Monograph 29, Vol 5., Am. Geophys. Soc.

Dickinson, R. E., and Hanson, B. (1984). Vegetation-albedo feedbacks. In Climate Processes and Climate Sensitivity, pp. 180-186. Geophys. Monogr. 29, Vol 5. Amer. Geophys. Soc.

Gates, W. L. (1985). The use of general circulation models in the analysis of ecosystem impacts on climatic change. Climatic Change, 7, 267-284.

Hällgren, J. E., Linder, S., Richter, A., Troeng, E., and Granat, L. (1982). Uptake of SO₂ in shoots of Scots pine: field measurements of net flux of sulphur in relation to stomatal conductance. Plant, Cell and Environment, 5, 75-83.

Johansson, C., and Granat, L. (1984). Emission of nitric oxide from arable land Tellus, 36B, 25-37.

Jones, H. G. (1983). Plants and Microclimate. Cambridge University Press: 323 pages Manabe, S., and Stouffer; R. J. (1980). Sensitivity of a global model to an increase of CO₂ concentrations in the atmosphere: J. geophys. Res. 85 (C10), 5529-5554.

Ogura, Y. (1985). Modeling studies of convection. In Saltzman (Ed.) Issues Atmospheric and
Oceanic Modelling. Advances in Geophysics, 28, pp. 387-421 Academic Press.

O'Neill, R. V. (1988). Hierarchy theory and global change. (Chapter 3, this volume). Panofsky, H. A. (1985). The planetary boundary layer. In Saltzman (Ed.) Issues in Atmospheric and Oceanic Modelling. Advances in Geophysics, 28, pp. 359-385 Academic Press.

Sellers, P. J., Mintz, Y., and Sud, Y. C. (1985). A simple biospheric model (SiB) for use within general circulation models. Report Laboratory for Atmospheres( NASAl Goddard Space Flight Center, Greenbelt, MD 2077.

Shukla, J., and Mintz, Y. C. (1982). Influence of land-surface evapotranspiration on the earth's climate. Science, 215, 1498-1501.

Solomon, A. M., Tharp, M. L., West, D. C., Taylot, G. E., Webb, J. W., and Trimble, J. L. (1984). Response of Unmanaged Forests to CO₂ Induced Climatic Changes: Available Information, Initial Tests and Data Requirements. DOE / NBB-0053. Nat. Techn. Information Serv. U.S. Dept. Comm., Springfield, Virginia.

Verstraete, M. M. (1985). A Soil-Vegetation-Atmosphere Model for Microclimatological Research in Arid Regions. National Center for Atmospheric Research. Cooperative thesis No 88.