Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
5 Water-Energy- Vegetation Interactions OVERVIEW Over the last decade, considerable progress has been made in under- standing and modeling both the climate system in general and certain re- lated components of the terrestrial biosphere (e.g., biophysical-atmospheric exchanges such as radiation and water and heat fluxes, ecosystem dynam- ics, and Face gas exchange and biogeochem~cal cycles). As yet, however, there have been few successful efforts, either in modeling or in data acqui- siiion, to link the activities of the physical climate and the terrestrial bio- sphere so as to further understand and improve capabilities to predict global change. This chapter focuses on the interactions between the vegetated land sur- face and the atmosphere, particularly on the exchanges of energy, water, sensible heat, and carbon dioxide between the two. The aim of the research strategy discussed here Is not merely to describe such exchanges but to fully understand them so that predictive modeling can be used to explore possible future states of the earth system. To do this, it will be necessary to make comprehensive, biophysically based models of the atmosphere and land bio- sphere with measurable state parameters as prognostic variables, e.g., tem- perature and humidity for the atmosphere, and albedo, leaf area index, and This chapter was prepared for the Committee on Global Change from the contri- butions of Piers J. Sellers, University of Maryland, Chair; John Bredehoft, U.S. Geological Survey; Christoper Field, Carnegie Institute of Washington; Inez Fung, NASAJGoddard Institute for Space Studies; Alan Hope, San Diego State University; Gordon McBean, University of British Columbia; and William Reiners, University of Wyoming. 131
132 RESEARCH STRATEGIES FOR THE USGCRP photosynthetic capacity for the land. To calibrate, initialize, and validate these models, it will be necessary to acquire a broad range of data covering the space-time domain, from continuous global-scale monitoring to intense, high-resolution field observations. Briefly stated, the goals of the research strategy are as follows: · To develop models that realistically describe the interaction between the land biota and the atmosphere with particular reference to the exchanges of energy, water, heat, and carbon dioxide. Ultimately, these models should be adequate for exploring the consequences of global change in terms of perturbations to the climate system and terrestrial ecosystems. The models will have to cover a wide range of spatial scales (millimeters to global) and time scales (seconds to millennia). · To collect data that can be used to initialize, validate, and prescribe boundary conditions for the models described above. Additionally, the data are to be used for monitoring the global environment and testing new hypotheses and for diagnostic or retrospective studies. · To conduct manipulative experiments, field campaigns, and process studies to improve our understanding of the processes controlling the trans- fer of energy, water, heat, and carbon dioxide between the land surface and the atmosphere at appropriate scales and to develop better methods for quantifying the controls. Figure 5.1 shows the relationship between the modeling, the process studies (experiments and laboratory work), and the large-scale data acquisition program proposed here, and the interdependence and coordination needed among these activities. In general, the modeling studies are intended to distill the results of process studies to provide a realistic predictive capability. The model predictions can be compared with field experiment observations on local scales and with data from the long-term monitoring effort and global data sets on regional and global scales, respectively. The flow of information should be two-way, as the results from the modeling activities will determine which variables should be observed at which time and space scales. To achieve these goals, two things must be done. First, the existing research activities must be adjusted and focused to promote the coordinated, interdisciplinary studies necessary for the collection of specialized data sets and the construction of a new generation of models. Second, efforts must be undertaken to address the most crucial resource as of now, which is not hardware or data, but trained scientists. The remainder of the chapter addresses the particulars of data needs, modeling, infrastructure, and resources, with particular emphasis on those areas that are weakly supported at present (see Tables 5.1, 5.2, and 5.3 for activities in monitoring, manipulative experiments, and field experiments,
WATER -ENERGY-VEGETATION INTERACTIONS 1 1 Model Construction and Testing Manipulative ~Field ~Process 10 Expenment s Cn m paigns S tudies . Fine spatial scaled discontinuous J _I Integer ated Expenmental Ef Sorts | | Hydrology, Ecology, Biogeochemistry Lon~term Monitoring Studies Global Data Set Acquisition l 133 Monitonng pLarge Spatial Scale, l continuous J FIClURE 5.1 Relationship between modeling, process studies, and over data acqui- sinon activities. The numbers on the right-hand side of the figure denote die ap- proximate number of field sites worldwide. respectively). The sections on data needs and modeling follow the structure outlined in Figure 5.1 by first reviewing global data set needs and then reviewing data needs at progressively finer spatial and temporal scales. It should be remembered throughout that the proposed strategy was conceived as a whole and that coordination of the component activities represents a significant challenge by itself. DATA NEEDS AND EXPERIMENTS A wide range of data needs to be gathered, processed, and integrated to provide an information base for modeling and diagnostic studies of the earth system. These include satellite, atmospheric, and in situ observations operating on an extensive and more-or-less continuous basis to provide "monitoring" information, and focused, coordinated observations the product of field and in vitro experiments to provide the insight necessary for model and algorithm development. These two kinds of data sets should be regarded as complementary, as the monitoring data sets will partially determine the list of items to be addressed by experiment and the experimental results should lead to improvement of the data processing methodologies applied to the monitoring data set. It should be remembered that all types of land cover need to be ad
134 TABLE 5.1 Monitoring Programs (Ongoing) RESEARCH STRATEGIES FOR THE USGCRP Activity Program Agency/Coun~y Meteorological data Climate and global change NOAA, national program meteorological agencies Satellite data products National Environmental NOAA, national Satellite, Data, and meteorological Information Service and agencies weather service Earth radiation budget Cloud climatology Earth Radiation Budget NASA Experiment International Satellite Cloud NASA Climatology Project Vegetation index (AVHRR) GIMMS; National NASA, NOAA Environmental Satellite, Data, and Information Service; and other national programs Retrospective land cover IRAP/Intemational Satellite NASA studies Land Surface Climatology Project Soil moisture, vegetation Goddard Space Flight Center NASA cover Snow and ice Goddard Space Flight Center NASA, NOAA Runoff ? WCRP Carbon dioxide Background Air Pollution NOAA Monitoring Network TABLE 5.2 Manipulation Experiments/Gradient Studies (Ongoing) Activity Program Agency Carbon dioxide enrichment, small crops Plastic ecosystem response, remote sensing Effect of land use changes on hydrology (forests) Effect of land use changes on hydrology (forests) and nutrient treatments Florida Area Cumulus DOE/USDA Experiment Experimental lakes area . · . universities Experimental lakes area Valdai, RSFSR Hubbard Brook, Coweeta Canada USSR USA Crop and pasture WorldwideAgricultural fertilization agencies
WA TER -EVER G Y- VEGETA TI ON INTERA C TI ONS TABLE 5.3a Field Campaigns (Ongoing) 135 Lead Agency/ Activity Date Program Location Country Large-scale 1986 HAPEX France France hydrometeorology (100 km)2 Energy balance, 1987, 1989 FIFE Kansas, USA NASA biophysics, 1988 on KUREX Kursk, USSR USSR meteorology, remote sensing (15 km)2 TABLE 5.3b Field Campaigns (Planned) Lead Agency/ Activity Date Program Location County Global Energy Water Cycle Experiment Arid zone, energy, water cycle, remote sensing (100 km)2 Arid zone, energy, water cycle, remote sensing (15 km)2 Tropical forest, energy, water cycle, remote l990s Global Energy and Water Cycle Experiment 1992 HAPEX-II 1991 IFEDA Global WMO Niger France Spain European Community 1990s ABRACOS Brazil U.K., Brazil sensing Boreal forest, 1994 BOREAS energy, water cycle, tropospheric chemistry, remote sensing Canada NASA, Canada
136 RESEARCH STRATEGIES FOR THE USGCRP dressed by the data acquisition effort and experiments, not just "natural, undisturbed" ecosystems. In particular, agricultural systems require a high priority as targets for studies at all levels monitoring, process studies, and modeling. Integration with national agricultural research programs will be essential. Particular needs for each kind of data set are discussed below. Global Data Needs Global data are needed for the specification of boundary conditions for global models as well as for the framework for analyzing and detecting global change. The data would come from extensive ground and aircraft surveys as well as from satellites. It should be stated at the outset that for satellite data to be useful for detecting and monitoring interannual and longer- term changes (1) calibration of satellite data must be of the highest priority, (2) rigorous correction for atmospheric effects and for viewing geometry must be performed, and (3) the calibration and correction procedures must be carried out by data centers and information systems to make the products available to the scientific user community in a timely fashion. In terms of looking for global change indicators, it should be remembered that a large archive of satellite data (20 years' worth) already exists. More resources should be made available to study these records to determine whether changes or the effects of changes can be detected at a usable level of accuracy. The following data are required: . Satellite data. The most obvious need is for the integration of existing techniques into information systems that can deliver calibrated, geometri- cally corrected, atmospherically corrected, and registered data products to the scientific user community in a timely fashion. Until this is done, remotely sensed data will continue to be underused, if not unused, as research scien- tists are forced to complete the whole task themselves from satellite sen- sor counts all the way through to derived products and only then toward their own scientific goals. Better means must be found. It is proposed that more than one effort be initiated to address this task. To be effective, individuals involved will have to form a wide pool of talents and be drawn from a diverse population instrument engineers, atmo- spheric physicists, scientific user groups, and information scientists and will also have to work across agencies. They will have to develop a means for selecting and implementing satellite data algorithms for operational processing of the data stream. The development of these information systems is probably the highest-priority task facing the community at present. The research on satellite data sets involves the continuing improvement
WATER -ENERGY- VEGETATION INTERACTIONS 137 of techniques for obtaining area-averaged parameters from coarse-resolu- tion data. More research needs to be done to apply scaling methods so that knowledge gained from infrequent high-resolution data (e.g., Landsat) can be applied to the continuously acquired global coverage data sets (e.g., Advanced Very High Resolution Radar (AVHRR)~. · Land transformation data. Any kind of large-scale land transformation needs to be documented in a uniform way as part of the monitoring program. If possible, this should be connected with the satellite data acquisition effort. Topographic data. At present there are no reliable high-resolution topographic data sets for the globe. The resolution of available global data sets for the land surface is either 5 or 10 minutes, depending on the continent. Present data are inadequate for many land surface studies. There is a need (1) to compile, archive, and make available additional existing topographic data; (2) to prepare new topographic maps from space-borne (e.g., Systeme Probatoire d'Observation de la Terre (SPOT)) data; and (3) ultimately to fly a dedicated mission (or missions) to acquire a coherent set of topographic data that can be made available to earth scientists in readily usable forms. Vegetation data. There are several digital data sets of vegetation for the globe. The spatial resolution ranges from 0.5° x 0.5° to much coarser for the globe. The resolution of vegetation information is much coarser, and ranges from about 10 blames to more than 150 vegetation types. Some of these vegetation data sets represent potential or climax vegetation, and others include land use and modification. The accuracy of the data sets has to be improved. The feasibility of land surface classification using satellite data has now been demonstrated. Improvement of this new technique in conjunction with ground surveys is critical, not just for mapping vegetation, but also for developing the capability for detecting changes in the structure and function of vegetation. It should be recognized that the appropriate grouping of vegetation types for evapotranspiration and water cycling may be different from the appropriate grouping for, say, surface energy balance. Research into appropriate classification schemes is needed. Vegetation function data. The phonology of vegetation is important for determining the timing and amount of water released through the land surface. There is limited tabular information on leaf area indices for different vegetation types. The normalized difference vegetation index (NDVI), available from polar orbiting satellites, gives information on the seasonal march of vegetation greenness globally and should be applied to the water studies. In order that the NDVI be used for detecting interannual and longer-term changes in phonology and water cycling, calibration of satellite instruments and rigorous correction for atmospheric effects must be of the highest priority. These studies should be carried out in conjunction with surface validation efforts (see following sections).
138 RESEARCH STRATEGIES FOR THE USGCRP · Soil data. Most global soil data sets are digital versions of the Food and Agriculture Organization (FAO) soil maps. Improvements to soil data sets are on the research agenda of several international groups (e.g., Soil and Terrain Data Base (SOTER) of the International Soil Science Society). Soil units, texture, and other parameters in the FAG soil maps are important for hydrologic studies, as are hydraulic potential and rooting depth of vegetation. Some information about these parameters has been gleaned from the litera- ture, but an extensive survey of different vegetation-soil complexes should be carried out. · Soil moisture data. The temporal and spatial variations of soil moisture are not well known. In most models, the soil-water-holding capacity asso- ciated with different vegetation types is specified, but there is no observed climatology of the field capacity. Field capacity climatologies derived as residuals from budget calculations suffer from inaccuracies in the other terms in the water budget, such as rainfall and evapotranspiration. The capability for measuring soil moisture from space, especially in the presence of vegetation, must be developed as far as possible. Meanwhile, because it is impossible to map soil moisture distributions globally, surface-based studies should be carried out to understand the vegetation and soil characteristics that determine soil moisture amount and its spatial variations (see the section "Remote Sensing" below). · Meteorological data. Near-surface atmospheric humidity, rainfall, near- surface air temperature, and wind speed are the driving forces for atmosphere- surface fluxes of water vapor. At present, the weather station network is the primary source for such data. Such networks must be maintained at mid-latitudes and expanded in polar and tropical regions. Technologies for automating these weather stations exist and should be further explored and developed, including such technologies as telemetric tipping rain gauges and ceilometers. Commonly, these data are used by the National Meteoro- logical Center and then discarded. Efforts should be made to integrate some of these data into a research data base. · Rainfall data. Rainfall varies on small scales. It also exhibits a distinct diurnal cycle, which varies geographically. Quantifying and understanding this variability are important to the scaling-up of local site studies to the gross resolution of global models. Thermal infrared radiation at the top of the atmosphere has been shown to be a useful index of convective precipitation, which is a source of small-scale variability. Existing satellite data (e.g., from AVHRR) should be analyzed to investigate the spatial and temporal variability of precipitation and to determine the appropriate precipitation statistics to describe rainfall variability on a global scale. Microwave measurements of precipitation from satellites offer a real hope for providing a global rainfall climatology. More efforts should be committed to research and validation in this area.
WA TER -ENER G Y- VEGETA T10N INTERA C TI ONS 139 · Runoff data. River runoff determines the freshwater inputs to the world oceans. Broecker (1989) has suggested that changes in runoff may trigger changes in deepwater formation and consequently changes in cli- mate. The USGS has an extensive gauge network that measures, among other things, daily flow rates of the rivers and tributaries in the United States. Similar gauge data have yielded a long-term record of the flow of the Amazon River. Such networks must be maintained, and if possible, expanded to other major rivers of the world. For closed drainage basins (see the section "Integrated Monitoring and Process Studies" below), river flow data must be analyzed in conjunction with contemporaneous precipitation and other weather data to develop and test hydrological models and to determine the frequency and~accuracy of river flow data for global change. · Surrogate and corroborative data. As there are several sources of water vapor to the atmosphere, the unique signatures of each water exchange process will provide information to validate models and hypotheses of the terrestrial hydrological cycle. These signatures include oxygen and hydro- gen isotopes of water. Also, for C3 plants, the simultaneous exchange of hydrogen oxide and carbon dioxide through stomata makes the temporal variation of carbon dioxide and its stable isotopes carbon-13 and oxygen-18 critical cross-checks for evapotranspiration. · Photometric data. Standardization, coordination, and augmentation of sun photometry measurements are needed to allow the collection of atmospheric optical thickness data. These data are required for the routine atmospheric correction of satellite data. Long-Term Monitoring A global network of minimally instrumented sites is needed to provide data for diagnosing the effects of climatic change over long periods of time, · testing basic models and hypotheses, and · anchor stations to perform satellite algorithm inversion. The first objective, diagnosing the effects of climatic change, will re- quire different measurement strategies in different parts of the world. These effects may manifest themselves as changes in characteristics such as vegetation cover or composition, mass and heat fluxes, or the hydrological balance. Sites need to be identified that can act primarily as "barometers" that will reflect the impact of climatic shifts on different ecosystems while also pro- viding data for the second objective, the long-term validation of models. This large, diffuse network of stations should total several hundred sites worldwide, and therefore optimal use will need to be made of existing
140 RESEARCH STRATEGIES FOR THE USGCRP resources and monitoring programs such as the Long-Term Ecological Re- search (LTER) network or the proposed observatories of IGBP and WCRP. The range of sites should encompass the world's major vegetation types and biomes, including agricultural systems. The long-term monitoring stations should be located at the centers and edges of ecosystems. This arrangement would provide data that may indi- cate the resilience of ecosystems to changes as well as the shifts on the borders where initial changes could be expected. The full benefit of the long-term data sets will be realized only if sufficient resources are set aside for basic research that will analyze the data sets. Furthermore, it will be essential to provide adequate funds for technical personnel to maintain good quality data collection programs over an extended period. Long-term monitoring of selected hydrological, climatological, and chemical variables has been conducted by a variety of agencies in the past. A reexamination of these monitoring efforts is required with a view to extending and coordinating the monitoring activities. Some of the automated data collection devices may be modified to record additional variables. At minimum, these sites should be committed to collecting meteorologi- cal data and conducting periodic surveys of the vegetation and soils in their locale. Integrated Monitoring and Process Studies Out of the larger set of long-term monitoring sites, a few (fewer than one hundred worldwide) special sites should be selected for integrated studies. There should be a significant and integrated scientific commitment to these sites in the areas of hydrology, biogeochemical cycling, ecology, and satel- lite monitoring. Facilities at these sites serve three purposes. First, they interface with operations at the next higher level of intensity (field campaigns and integrated ecosystem experiments) by providing starting points with infrastructure and background information sufficient to ensure high returns. Second, for fa- cilities at the next lower level of intensity (long-term monitoring sites), studies at the higher-intensity sites will identify variables to monitor, define appropriate sampling intervals, and validate integrated models tailored to the specifics of each biome. Third, these facilities for high-intensity moni- toring will be the primary barometers for changes in subtle parameters that affect aspects of ecosystem function without immediate or profound effects on structure. It is anticipated that the energy, water, and carbon balance models to be tested using data sets from these sites will utilize meteorological data (e.g., temperature, humidity, precipitation, wind speed, and radiation) to provide
WATER-ENERGY-VEGETAT10N INTERACTIONS 141 the forcings and satellite data to provide the slowly changing surface boundary conditions (e.g., photosynthetic capacity and soil moisture estimates). An ultimate goal for this measurement-modeling exercise is to develop the models that can be driven by the combined remote sensing and meteoro- logical data sets to calculate fields of surface fluxes and associated forcings on the important ecosystem processes over the whole globe. Accordingly, the proposed criteria for site selection are as follows: · located within important biome centers or on biome transition zones, · existing long-term research archive, and if possible, a paleoecological record (see chapter 3), . nearby research institute for site support, "feasible" topography for flux measurement, · presence of gauged or gaugeable watersheds of a reasonable size (5 to 20 km2), and away from excessive anthropogenic impacts, e.g., large-scale air pollution, . logistics: airfields nearby and road access. To satisfy the modeling requirements, the stations would routinely acquire the following kinds of data: . hydrological, meteorological (including radiation balance), ecosystem structure and productivity, land use and soil information, atmospheric optical depth, large-area surface fluxes (tower), trace gas concentrations, and occasionally fluxes, and selected satellite data. Most of the data should be taken within a concentrated area of roughly 20 x 20 km, which allows a reasonable area for the sampling of satellite observations and airborne flux measurement. However, this core site should be located within a larger similar zone (100 x 100 km), which could be used for studies of the spatial and temporal variability of some of the parameters. Hydrological data. Various components of the hydrological cycle at local and regional scales are expected to respond to climatic shifts. There- fore studies of snow pack dynamics should include analyses of snowline retreat patterns over time and space. The timing of runoff originating from snow packs may be a significant indicator of general atmospheric warming at higher elevations. Reservoir levels (minus consumptive use) can be taken as a long-term integrated measure of large-scale (regional) water surplus or deficit. Reser
142 RESEARClI STRATEGIES FOR THE USGCRP voir evaporation records also need to be examined as a measure of atmo- spheric evaporative demand. At present these activities may be undertaken in many areas, but it will be necessary to coordinate the data sets to provide larger regional analyses. River discharge records, particularly from watersheds that have been monitored for long periods of time (>30 years), provide the basis for many hydrologi- cal investigations and need to be considered when locating the integrated monitoring sites. These records are valuable for model calibration and for determining the hydrological response of systems to climatic fluctuations. It will be necessary, however, to develop an understanding of the role of land use in the rainfall-runoff process of the watersheds to be examined so that these effects can be accounted for in assessing climatic impacts. Wa- tershed-based studies will have the advantage of providing an integrated view of hydrological processes. Hydrological and other balances (e.g., nutrient) can be examined using the watershed as the unit of study. Furthermore, long-term (months to a year) evaporative losses can be estimated by resolving the various components of the watershed water balance and serve as a check on short-term monitoring approaches. The small (<150 km2) watershed ap- proach may be also be valuable for lumped soil moisture modeling studies, which can be integrated with remote-sensing-based procedures for estimat- ing this variable. Particular attention needs to be given to procedures for making large- scale estimates of soil moisture in a timely and cost-effective manner. In view of the critical role that soil moisture plays in controlling evaporative and discharge losses from watersheds, this monitoring effort should be regarded as a major priority. Therefore research into appropriate sampling schemes, new measuring devices (e.g., time domain reflectometry), and empirical prediction techniques will be required. Since precipitation is a major determinant of moisture fields controlling biological activity and processes, the integrated monitoring sites should have carefully designed precipitation gauging networks. · Data on ecosystem dynamics and biogeochemistry. Intensive monitoring at the sites should address ecosystem function at several levels, including (1) exchange of materials with the atmosphere, (2) exchange of energy with the atmosphere, (3) transport of materials to other ecosystems, (4) storage of biomass, nutrients, and energy, and (5) structural and functional species composition and diversity to all trophic levels. Since the biosphere obser- vatories are intended to provide the most detailed long-term data available for any biome, exchanges of mass and energy should be monitored with an intensity sufficient to support accurate annual budgets of key elements (carbon, nitrogen, and phosphorus), atmospherically and biologically important chemical species (hydrogen oxide, carbon dioxide, ozone, methane, isoprene, nitrous
WATER -ENERGY-VEGETATION INTERACTIONS 143 oxide, nitric oxide, nitrogen oxides, and acidity), and forms of energy ex- change (absorbed radiation, net radiation, sensible heat, latent heat, soil heat flux, and momentum transfer). The appropriate technologies for monitoring these aspects of ecosystem function will vary somewhat from ecosystem to ecosystem but should be standardized as far as possible to allow interbiome comparisons. Exchanges with the atmosphere should be continuously measured with meteorological techniques. Eddy correlation is the most suitable existing technology for measuring fluxes of many of these components, although sensors for some (e.g., methane) need to be improved. The NCAR Atmosphere-Surface Turbulent Exchange Research (ASTER) facility provides one possible model for the kind of advanced eddy correlation systems that should be established in the sites. Other instruments like the Fourier Transformed Infrared (FTIR) spec- troscopy combined with LIDAR also offer potential that should be explored. Transport to other ecosystems should include hydrological as well as airborne transport. Rainwater and stream chemistry are critical components to monitor. The sites, as noted above, should include watersheds for which complete water and nutrient budgets can be obtained. Storage monitoring should emphasize water, carbon above and below ground, and nutrients (nitrogen, phosphorus, potassium, calcium, and sulfur) in biomass and soil. Information on stored material is much more useful when it is accompanied by chemical data indicating resistance to decomposition (especially lignin content) and when it is partitioned into components of differing biological activity (e.g., below-ground carbon should be partitioned into root, microbial, and one or more soil fractions). Whenever possible, the storage terms should be measured directly and not estimated by difference from the flux studies. The sites should also be positioned to serve as primary centers for monitoring diversity. The assessment and study of biodiversity should be based on regular censuses of all major Tanya, including insects and microbes. In addition to an accounting of the number of species present and more sophisticated indices of evenness, the diversity monitoring should report structural diversity, including forest and soil stratification, and phonology. · Satellite data. A goal for all of the studies on these sites is to enhance the use of satellite data for describing earth system processes, in the hope that the developed techniques can be applied globally over the sites. All available satellite data should be collated and archived together with ancil- lary observations, including photometric measurements, mesometeorological network observations and analyses, and radio soundings. These data sets should be continuously available to the scientific teams working in the area. A commitment must be made to regularly process subsets of these data to the point where they are directly useful to the scientific teams. This means
144 RESEARCH STRATEGIES FOR THE USGCRP that atmospheric and geometric corrections should be applied and in some cases the resulting fields processed into useful biophysical parameters (e.g., APAR). Experiments Experiments will be essential to understand the important processes controlling the interactions be- tween the vegetated land surface, the atmosphere, and the hydrosphere, specifically, fluxes of energy, water, heat, carbon, and trace constituents; · understand how these fluxes are associated with state variables associated with important ecosystems (e.g., community composition and physiognomy); and · develop the links between the fluxes and state variables described above and those parameters amenable to remote sensing. Two main kinds of experiments will be necessary to achieve these tasks: integrated ecosystem experiments and field campaigns. Integrated Ecosystem Experiments Integrated ecosystem experiments, manipulations that expose whole eco- systems to simulated climatic change or components of climatic change, will be critical for two reasons. First, they will provide the most concrete and complete evidence of the response of ecosystems to climatic change, and they can potentially provide it in a way that facilitates quantification of the mechanisms driving the responses. Second, they will provide the solid data for model development and testing. Manipulative experiments will provide essential tools for the study of climatic change, but they will also be complex and expensive. These major efforts must be carefully designed to maximize their utility. Three principles can facilitate the choice of systems for study and the design of experiments. First, while there is a need to know the response of every ecosystem type to climatic change, it will not be possible to subject every ecosystem to one or more manipulative experiments. The likely time course of climatic change demands that we rapidly establish experiments in a broad range of ecosystems. Given the need for breadth, it is also important to emphasize ecosystems that will yield the greatest return on the resources invested. Important criteria to consider include the following: · Experimental tractability. Ecosystems with small, short-lived domi- nates and rapid cycling of carbon and nutrients (e.g., grasslands) are inher- ently more accessible than ecosystems with large dominates and slow element cycling. It is critical that the ecosystems developed as model systems re
WATER -ENERGY-VEGFI ATI ON INTERACTI ONS 145 spend quickly enough that experiments proceed faster than climatic change. Rapid time responses and small stature also facilitate replication and facto- rial exposure to the full suite of components of climatic change, enhanced carbon dioxide, temperature, precipitation, and anthropogenic pollutants. In some ecosystems the suite of meaningful manipulations may include factors other than components of climatic change. Altering soil fertility, removing or augmenting populations at one trophic level, and changing light may be efficient ways to change ecosystem processes in a manner that contributes directly to understanding the mechanisms of responses to climatic change. Intensive study of carefully selected model systems, in combination with process studies (see the section "Fundamental Research and Laboratory Work" h~.~nwN will ~rrentiv facilitate the design of the necessary experiments on ecosystems with long-lived, large-stature dominates and with slow carbon and nutrient cycling. · Experimental relevance. It is critical that the ecosystems chosen for emphasis are subjected to experimental designs consistent with quantifying the mechanisms driving the responses to the manipulation. An emphasis on driving mechanisms is the most likely approach to making results of one experiment generalizable to a range of ecosystem types. · Ecosystem sensitivity. Process studies and historical observation have identified several ecosystems likely to be greatly affected by or especially sensitive to climatic change (e.g., coastal marshes and arctic tundra, respectively). These systems merit high priority for study. · Ecosystem importance to humans. Ecosystems that play a critical role in sustaining human populations, especially agroecosystems, deserve spe- cial priority for experimentation. ~ ' ' ~ ~ '--- D-- J Second, natural and unintentional experiments provide excellent oppor- tunities for addressing some questions. Some natural experiments can be based on the responses of ecosystems to unusual climatic conditions, like E1 Nino events. Others can utilize unique geological features, for example, sulfur dioxide and carbon dioxide from volcanic vents or carbon dioxide from natural springs. Areas of natural sea level subsidence may provide useful model systems. Unintentional experiments vary in scale from the entire globe to very local phenomena, for example, pollutant plumes from industrial sources to artificial range extensions from experimental agrofores~y. Third, the dynamics of climatic change and ecosystem responses warrant independent study. At this point, we have little information on the responses of ecosystems to the rate of climatic change. Most past experiments have exposed ecosystems to a step change in carbon dioxide, temperature, or precipitation. While this kind of manipulation may yield results comparable to gradual change, it may also cause different responses, and the differences may be ecosystem specific, depending on the time constants of the compo- nent processes. Study of these dynamics can be pursued with models, with
146 RESEARCH STRATEGIES FOR TlIE USGCRP experiments on model systems, and with process studies on the time con- stants of ecosystem components. Field Campaigns Field campaigns will be essential to · study the biophysical controls exerted by the terrestrial vegetation on the transfer of energy, water, heat, and trace gases between the land surface and the atmosphere at appropriate time and space scales, and · develop better methods for quantifying these controls and exchanges using satellite and standard meteorological data to permit global observation. Again, the design, execution, and costs of these efforts will vary widely. Only their common elements will be discussed here: · Monitoring studies. Field experiments should be sited within long- term monitoring sites, and measurement should be integrated with the long- term effort as far as possible. · Theoretical framework. The design of these experiments should make allowance for both the correlative "top-down" and the biophysically based "bottom-up" approaches. While most researchers believe that the ultimately the bottom-up (reductionist) approach will provide the necessary informa- tion, it is clear that many remote sensing successes in the scientific arena are based on simple correlative techniques. · Multisensor, multiangle, multiscale measurement strategies. It is almost certain that no single remote sensing measurement is going to provide the requisite information in isolation from other data. Likewise, to fully test theories and models, measurements have to be made simultaneously across several spatial scales, preferably using a nested sampling strategy. · Redundancy and checks. Efforts should be made to allow independent estimates of key validation components, e.g., surface fluxes, using different measurement methodologies. Besides being a reasonable precaution, this may allow the validation of measurement technologies across different spa- tial scales. For example, a surface flux network could be used to test the validity of an airborne measurement technique, which could then be applied to a much wider area. Likewise, chemical analyses on outflows from whole watersheds could be compared with the aggregation of point samples taken within the area. These large-scale field experiments are expensive, resource-intensive, and very demanding in terms of the necessary investment of scientists' time. As an example, the First ISLSCP Field Experiment (FIFE) (Sellers et al., 1988) involved the commitment of about 100 scientists, 50 support personnel, several million dollars worth of field equipment, and six aircraft.
WATER-ENERGY-VEGFIATION INTERACTIONS 147 However, these experiments offer unique opportunities for large-scale stud- ies and are necessary for improving integration techniques for scaling from smaller to larger scales. In practice, it is hoped that the national science resource base will be able to support at least one such effort in the field per year without absorbing the energies of most of the interested community. The field experiments would be "nomadic," being focused on one integrated monitoring and process study site (see previous section) at a time and moving on to another site (different blame or different set of problems) on completion. As far as possible, the field campaigns should address both disciplinary and interdisciplinary problems. To the extent that ecological, hydrological, and biogeochemical processes are closely interlinked, this need to address a range of problems will be essential to the planning and execution of such experiments. Field campaigns offer the best opportunity to check theories that extrapolate the results of leaf- or plant-scale phenomena up to the scales relevant to surface-atmosphere exchange and routine satellite remote sensing (i.e., several kilometers). The strategy developed in FIFE provides one template for this problem: surface studies are distributed within an area of around 20 km on a side that is also routinely observed using airborne eddy correlation, remote sensing equipment, and satellites. This design allows for the direct comparison of experimental results across most of the scales relevant to model develop- ment for GCMs (i.e., plant, site, and watershed scales) (see Figure 5.2~. Experiments at much larger scales necessarily involve a dilution of the surface observation network and therefore cannot address the cross-scale problems directly. Large-scale experiments should be viewed as checks on model parameterizations at the scale of application (e.g., a GCM conformation of grid squares), about 1000 km on a side. Field experiments also benefit the community by forcing scientists from different disciplines to collaborate in order to design, execute, and produce a concrete result. As catalysts for changing research policies and stimulat- ing new interdisciplinary research, such experiments can be very useful tools to both the researchers and the agency personnel involved. Simulated and Actual Impact Studies A number of opportunities exist for conducting simulated or actual impact studies, including impacts of tropical deforestation, acid rain, and glacier retreat. Areas where there are large-scale ongoing impacts, e.g., the Amazon tropical forest, should have a high priority for combined long-term process studies and field campaign investigations. Such studies should address changes in surface energy and water balance, soil hydrological processes, soil physics and chemistry, species succession, and dynamics (socioeconomic) of land transformation. Where possible, paired sites should be used.
148 RESEARCH STRATEGIES FOR THE USGCRP SPOTS ~ ~ LANDSAT C-130 lo, ~ - / / \ \ it/ N_ - ~- - \_ FIGURE 5.2 Satellite, airborne, arid surface data acquisition at midday, August 4, 1989, in the middle of the FIFE-89 Intensive Field Campaign (Source: Sellers et al., 1990). Fundamental Research and Laboratory Work Fundamental research needs to be supported in a number of key areas, including remote sensing techniques, detailed ecological and hydrological process studies, and the development of improved measurement techniques. Remote Sensing Satellite remote sensing represents a feasible means of collecting appro- priate, consistent, and temporally frequent data over the entire biosphere at
WA TER -ENER G Y- VEGETA Tl ON INTERA C TI ONS 149 reasonable cost. Its potential should therefore be exploited to the maximum extent possible. In addition to the execution of field campaigns and the analysis of satel- lite data described in previous sections, research needs to be done to place remote sensing science on a more physical, less correlative basis. Model development and testing will be an integral part of this effort, and the development and testing of new instruments and sensor combinations will be essential. Particular emphasis should be placed on the following: . Atmospheric, radiometric, geometric correction algorithms. Almost no satellite data are routinely corrected for atmospheric effects, partly be- cause sufficiently robust algorithms are not available. Much more funding should be directed toward the development, selection, and implementation of such algorithms. Surface and spectral radiation interactions. To date, only correlative studies have been performed to relate surface spectral signatures with canopy properties (e.g., lignin/nitrogen ratios in foliage). More intensive modeling work (e.g., on foliar spectral properties) must be done to determine the links between plant biogeochemistry or ecosystem status and spectral signature. · Microwave (active and passive) remote sensing. Microwave remote sensing offers some real promise in determining vegetation characteristics, soil moisture, and precipitation. More investment must be made in the development of instruments and theoretical treatments. Novel technologies. Atmospheric scientists have discussed the use of satellite-based or airborne lasers to determine aerosol optical properties, atmospheric boundary layer heights, and so on. The utility of such instruments and other novel sensors for surface and near-surface remote sensing should be investigated. Ecological Process Studies Process studies in ecology play two critical roles in global change research. First, they help identify the variables that need to be considered, set opera- tional conditions, and simplify the design of integrated field experiments. Second, they provide a powerful, unambiguous arena for quantifying mechanisms and for testing the structure and parameterization of models that focus on component processes. · Ecosystem processes. The challenge of developing ecosystem models competent to predict ecosystem responses to climatic change has several aspects that need to be supported with process studies. Some of these process studies (e.g., species change) fall within the traditional realm of population or community ecology, but others involve emergent properties at
150 RESEARCH STRATEGIES FOR THE USGCRP the ecosystem level (e.g., effects of climatic change on the species composi- tion of the soil fauna and on the decomposition of existing organic matter) and indirect consequences of species effects (e.g., changes in the quality of soil organic matter resulting from changes in plant blame allocation or species composition). Ecophysiology. Four areas in plant ecophysiology clearly warrant increased priority. First, a better understanding is needed of certain processes that play a role in controlling plant responses to climatic change. For example, biomass allocation, phonology, and respiration are critical determinants of plant performance, but none is understood at a level suitable for develop- ing a mechanistically based model. Second, more work is needed on plant responses to simultaneous, interacting stresses and alterations in resource availability. Much of the past work in plant ecophysiology has focused on the response of one process (e.g., pho- tosynthesis) to one stress (e.g., drought). In order to maximize the contribution to global change research, the focus should shift to the responses of multiple plant processes (e.g., photosynthesis, allocation, and nutrient uptake) to combinations of stresses and altered resources. Treatment factors should include biotic stresses like herbivory as well as abiotic factors like drought, altered temperature, and elevated carbon dioxide. Third, plant ecophysiology is much better developed at the level of the leaf or the individual plant than at the level of the canopy or the ecosystem. Techniques for scaling mechanistic responses at the leaf level to canopy and ecosystem responses must be developed and evaluated. Fourth, the last few years have seen advances in ecological models that predict plant processes (e.g., growth, photosynthesis, and biomass allocation) on the basis of resource availability rather than physiological or anatomical criteria. To the extent that they are generalizable, these models have the potential to dramatically simplify the task of predicting ecosystem responses to climatic change. On the other hand, it is also increasingly clear that potentially co-occurring plant species may differ in access to resources, efficiency of uptake, and efficiency of utilization. The interface between the indications of interspecific generality and species individuality is a critical area to resolve, both because of the potential simplification that results from a species-independent perspective and because the ecosystem consequences of species change are central components of ecosystem responses to climatic change. · Community ecology. Community-level phenomena pose some of the greatest challenges for developing predictive models of the response of ecosystems to global change. Studies in community colony are replete with examples of singular events (e.g., arrival of a pest, pathogen, or predator) that dramatically alter the course of ecosystem development. The challenge for process studies is to assist in specifying the envelope of possible re
WATER-ENERGY-VEG~ATION INTERACTIONS 151 spouses and quantifying the consequences of each for ecosystem develop- ment. It is especially important to relate ecosystem development to resource availability and biotic and abiotic stress. · Population biology. Ecosystem responses to global change will almost certainly result from the combined effects of (1) plastic responses in existing genotypes, (2) migration of new genotypes, and (3) selection on new and existing genotypes. The critical challenge for population biology is to estimate the quantitative importance of each process, in a range of ecosystems. Hydrology Hydrological studies, particularly on scale effects, need to be supported aggressively. Some of this work can be supported under the field experiment umbrella (see the section "Experiments" above), but there is a need for fundamental research at the small scale. Important issues to be addressed are listed below. Soil physics and water relations. The enormous spatial variability in soil physics and microtopography makes a simple arithmetic averaging of surface characteristics for larger-scale applications almost meaningless. More experimental and theoretical work must be done to investigate small-scale variations within bulk treatment in a physically meaningful way. This may involve intensive work on small subcatchments coupled with distributed hydrological modeling efforts. · Soil water and energy exchange. The flux of water from the soil to the atmosphere, whether through plants or the soil surface, is not well understood. A number of small-scale experiments should be conducted to address this issue. These experiments will necessarily involve ecological expertise and the application of scaling theorems as discussed above. · Plant and water relations. Much progress has been made in the area of photosynthesis-transpiration relations, but ideally a great deal more needs to be done. Laboratory and small-scale field research represents the best ~ . · . avenues tor 1nltla . progress. Instrument Development In addition to the remote sensing instruments described in the section "Remote Sensing" (above), there is a widespread need for improved instru- mentation for both laboratory and field studies. Efforts should be supported to improve existing techniques or develop new ones for the following kinds of measurements: · Scalar and isotope concentrations. Here the need is for more measure- ments in addition to new methods.
152 RESEARCH STRATEGIES FOR THE USGCRP · Eddy correlation, lidar sounding, and long-path-length measurements. All of these must be improved to allow more accurate and reliable estimation of surface fluxes. Precipitation, radiation, and so on. Methods for obtaining estimates of the large-scale values of these variables are essential for studies of the hydrological cycle and the surface-aunosphere energy balance. · Hydrological state variables. Sensor development for estimating snow cover and depth and soil moisture distribution at scales ranging from in situ (local scale) techniques to satellite-borne instrumentation should be supported. MODELING Figure 5.3 shows the relationships between the (biophysical) land surface parameterizations (LSPs), the atmospheric general circulation models (GCMs), and the ecosystems dynamics models (EDMs). Currently, there are few biophysically based LSPs in existence that have been successfully combined with GCMs (Dickinson and Henderson-Sellers, 1988; Sato et al., 1989~. These models cover the relatively short-term biophysically controlled interactions between the vegetated surface and the aunosphere: radiative transfer (albedo), momentum transfer (roughness length), General Circulation Mnr1~.1s Global <, Regional_ ct up a, cat ~Local ... ~&, ~ \ l // Forcing ~ ~\` Module J Land S _ e Parameterizations ~ \ \ / \ \/Aggregation) \Module I Seconds Days Years Centuries Ecosystem Dynamics | I Models l Time ~ FIGURE 5.3 Relationships between land surface parameterizations, atmospheric general circulation models, and ecosystem dynamics models show disparity in time and space scales. For more effective global change research, two new "communica- tion" techniques or "modules" must be developed to interpret results between model classes.
WATER -ENERGY-VEGETATION INTERACTIONS 153 and exchanges of sensible and latent heat and carbon dioxide (biophysical control of evapotranspiration and photosynthesis). Essentially, the LSPs assume a static ecosystem structure and a prescribed phonology, which in turn define the albedo and roughness length characteristics of a given area and the evapotranspiration response as a function of soil moisture. Gener- ally, the surface vegetation type, and hence albedo and roughness length, is prescribed from data, and the soil moisture field is initialized from offline climatological studies. As a result, these models in their current state have a limited utility for the study of global change because they merely represent an improvement over the abiotic "bucket" models described by Budyko (1974) and Eagleson (1982~. General circulation models of the atmosphere have improved considerably over 30 years of development to the point where they are the preferred tools for weather prediction and the study of climate. However, it is clear that these models will be subject to certain limitations for the foreseeable future. Most importantly, the models will be limited in terms of spatial resolution, representation of small-scale (subgrid scale) processes, and duration of run. It should also be remembered that each model possesses its own climatology, which differs from reality, and that an adequate description of the model climatology normally requires an extended series of runs. The normal GCM time step is on the order of 10 to 30 minutes (times much longer than this can lead to serious systematic error in the description of dynamical or physical processes), and this effectively limits the number or the length of runs that can be executed and analyzed by a research group. Ecosystem dynamics models operate in an entirely different time and space domain from the LSP-GCM combinations, generally working on small spatial scales (meters to kilometers) and long time scales, integrating over centuries or millennia with time steps of up to one month. (For the time being, the discussion will exclude the "biogeographical" type of model, which describes the continental or global distribution of vegetation forma- tions on the basis of "mean" climatology. These descriptions are not dynamic, as they operate as direct single-solution transforms of imposed climatic fields.) It is likely that global change in the real world will affect elements of all three systems, as shown in Figure 5.4. The sequence of changes could be as follows: 1. A change in the physical climate system would bring about a direct biophysical response from the biota. For example, near-surface temperature or humidity changes would have a direct impact on photosynthesis and evapotranspiration rates (fluxes). 2. Changes in the surface biophysical response would directly affect the near-surface climate. The resulting feedback could be positive, neutral, or negative, depending on the circumstances.
154 RESEARCH STRATEGIES FOR THE USGCRP /~ _ ~ PHYSICAL CLIMATE SYSTEM ~ Heat, Mass | B | Fluxes BIOPHYSICS Albedo Roughness Hydrology ~ CO2, CH4 Release IBI BIOCHEMISTRY ~Nutrient, \ / Carbon \r) Pools Tem rature, Beater l l Relations JO ~C~ ECOSYSTEM STATE Community Composition, Structure, Pedology FIGURE 5.4 Important interactions between the vegetated land surface arid Me atmosphere with respect to global change. (a) Influence of charges in the physical climate system on the biophysical characteristics and ecology of the biome. (b) Changes in nutrient cycling rates; release of carbon dioxide arid methane from soil carbon pool back to We atmosphere. (c) Ecological change in species composition results in changes in land surface characteristics of albedo, roughness, and soil moisture availability with possible feedbacks on near-surface climatology. 3. Changes in the near-surface climate or the surface biophysical re- sponse would have a direct impact on the forcing functions acting on ecosystem dynamics. Obviously, changes in ecosystem processes would be the first manifestation (e.g., rates of decomposition and mineralization), but these could be followed by changes in the gross structure (e.g., species composi- tion and standing biomass). The degree, if any, of such changes is again highly variable and dependent on the locale and site history.
WA TER -ENER G Y- VEGETA TI ON INTERA C TI ONS 155 4. Changes in ecosystem structure and function can be expected to feed back onto the physical and possibly the chemical climate system. As yet, the tools to investigate these phenomena have not been integrated in a scientifically defensible way. Although there has been some progress in coupling LSPs with GCMs, it is highly unlikely that either will be directly coupled with EDMs in the near future because of the gross disparity in time and space scales. It can be convincingly argued that direct coupling is highly undesirable in any case. To force EDMs, many of which incorporate stochastic descriptions of ecological processes, good representations of the mean and variability of a region's climate must be applied: thus there is a need to repetitively apply many variations of a climatology to an EDM before a credible ensemble of results can be collected. In addition, the spatial scale of most EDMs is not consistent with that of GCMs: the answer, of course, is not to increase GCM spatial resolution to finer and finer scales, as this would result in problems similar to those discussed regarding temporal scales. For both of the above reasons it is clear that direct links between LSP-GCMs and EDMs are both impracticable and undesirable. However, before significant progress can be made in the area of medium- to long-term atmosphere-biosphere interactions, it will be necessary to construct more rigorous linkages be- tween the LSP-GCMs and EDMs. It is proposed that this be done by constructing "forcing modules," to convert GCM output into "forcing" climatologies for EDMs, and "aggregation modules" to aggregate the effects of ecosystem dynamics changes into representative LSP parameter sets. In spite of the general state of modeling described above, every effort should be made to support model development in every direction~CMs, LSPs, and EDMs-as these represent the greatest opportunities for predict- ing the mechanisms and effects of global change. Some specific needs that merit special attention or that have not been addressed in previous reports are listed below. Intermodel Transfer Packages As discussed above, two kinds of intermodal transfer packages (ITPs) are needed the first to allow communication from LSP-GCMs to EDMs and the second to allow communication in the other direction. The first, the forcing module, would accept GCM output and generate the requisite "climate" for EDM applications. The module should take into account the following: · Biases due to GCM climatology. · Effects of GCM resolution and the parameterization of subgrid-scale processes.
156 RESEARCH STRATEGIES FOR THE USGCRP · The likely range of microclimates produced by variations in topogra- phy, soil moisture, and pedological effects. · The frequency and type of "extreme" events associated with a clima- tology in addition to the description of the mean condition. . Results obtained from process studies and monitoring data sets. Forcing modules would necessarily have to incorporate some background knowledge of the GCM's structure and performance. In this sense, they would be far more than simple extensions of quasi-stochastic "weather generators." Aggregation modules would be used to analyze the results of EDM runs and would generate the requisite grid-scale parameters for LSP-GCMs. To a degree, the aggregation module is an inversion of the forcing module in that it attempts to generalize and integrate the specific and different results of EDM runs. The modules should take account of the following points: · Integration of physiological characteristics probably cannot be done in a linear, arithmetic fashion. The "importance" of the contributions of dif- ferent organisms to various fluxes, and so on, must be taken into account. · The impact of spatially varying soil moisture should also be integrated to take account of nonlinear effects on the surface fluxes. . Where appropriate, the effects of landscape pattern, e.g., repeating topographic units, should be integrated using ensemble averaging techniques. Phenological Descriptions for LSPs As discussed above, it is impracticable to place full EDMs within GCMs. However, some elements of vegetation phonology could be formalized and placed within LSPs. In particular, the following physiological phenomena should be described as functions of GCM prognostic variables (temperature, humidity, soil moisture, radiation, and so on): time series of green leaf area index; rooting depth; and roughness length, albedo, photosynthetic capacity, and maximum canopy conductance if these are not functions of leaf area index and rooting depth. The models should be able to describe the seasonal course of vegetation attributes and provide a crude response to large interannual variations in precipitation. Hydrological Models Land hydrological modeling is currently split into two effectively noncommunicating camps: · "Wet" hydrology, an extension of the traditional hydrology, which had its roots in engineering applications (e.g., channel routing and storm flow response). Many recent research efforts have been focused on small
WATER -ENERGY- VEGETATION INTERACTIONS 157 or regional-scale catchment models, with spatially distributed descriptions of rainfall interception, overland flow, and infiltration. Usually, these mod- els have fairly simple evapotranspiration descriptions. · "Green" hydrology, mainly concerned with the study of the biophysics of the evapotranspiration process using soil-plant-atmosphere models. The same basic models have been applied to describe processes on small-scale (agricultural) sites up to the scale of GCM grid areas. Clearly some linkage between the different kinds of models is required. In particular, greater efforts must be expended to make the biophysical models into better descriptions of spatially heterogeneous surfaces. Also, some distillation of the wet hydrology models must be introduced into the LSPs: currently, all the LSPs use simple one-dimensional descriptions of the infiltration process with very simplistic representations of overland flow and spatial heterogeneity. The goal should be to provide good descriptions of total runoff losses as integrated over a month or more, rather than the extremely difficult objective of reproducing the correct timing of runoff losses. Surface/Planetary Boundary Layer Models Many of the problems in describing realistic feedback mechanisms be- tween the surfaces and troposphere are associated with the description of mixing processes in the planetary boundary layer. Correct description of these is also important for some modeling inversion techniques driven by satellite remote sensing (e.g., determination of surface heat fluxes using meteorological and satellite data). Modeling efforts to address both goals should be encouraged. Ecosystem Structure Models The models discussed in previous sections can describe the effects of changes in land surface biophysical parameters (e.g., albedo, roughness, and moisture availability) and biogeochemical properties (e.g., nutrient cy- cling rates) as they are defined by a given ecosystem status (e.g., species composition, soil microecology, vegetation health, and leaf area index). Modeling techniques for describing ecosystem structures must be developed in parallel with the models discussed in previous sections. These models should address the issues of alteration of ecosystem structure due to changes in (1) the physical climate system, (2) atmospheric chemistry, and (3) land use change. Depending on the intensity and type of change imposed in a given re- gion, the ecosystem structure may be altered slightly (e.g., by adjustments in carbon dioxide exchange rates) or drastically (e.g., with changes in species
158 RESEARCH STRATEGIES FOR THE USGCRP composition or leaf area index). All types of changes may feed back onto the physical or chemical climate system (see Figure 5.4). Prediction of terrestrial ecosystem change can be approached by using biogeographical methods as well as by using "full-up" ecosystem simula- tion models. It is strongly recommended that research efforts be encouraged on a broad front: the preliminary biogeographical models will provide us with the means to explore the possible sensitivity of land-surface- atmosphere interactions to changes in surface conditions. This approach is represented by Loop I in Figure 5.5, where a climatic change scenario leads to a simple definition of a new (steady state) distribution of biomes, which is then fed back through the climate modeling process to test for second-order effects due to the induced change in surface cover. Results from this kind of study will indicate which regions and blames are important in terms of inducing second-order effects and thus merit further studies and interaction using kinetic ecosystem structure models (see Loops II and III in Figure 5.5~. In this respect, none of the models should be regarded as an ultimate replace- ment for the rest; they have different roles depending on the level of detail required by either the climate modeling effort or the interests of biologists working on the effects of global change. In all cases the intermodel transfer packages discussed in the section above will have to be used as communication models. Radiative Transfer/Plant Physiology Models A number of modeling efforts have been partially successful in retrieving vegetation attributes from remotely sensed optical data. For the most part, these have concentrated on obtaining values for biometric properties such as leaf area index or biomass. More recently, efforts have been made to calculate physiological properties from observed radiances, including canopy (area-averaged) photosynthetic capacity and stomata! conductance. Research continues in the use of radar and passive microwave instruments for interpreting vegetation properties, but so far these efforts have focused on the retrieval of biometric properties and the classification of landscapes into different cover types. The field of remote sensing of biospheric functioning is on the verge of providing invaluable information for the study of global change. The key to progress is clearly in the development of satellite data algorithms that can calculate appropriate states and rates associated with the terrestrial vegetation along with estimates of the uncertainty attached to each derived value. Such algorithms will have to address the following problems: satellite sensor calibration, sun-target-sensor geometry and the effect of atmospheric scat- tering, radiation transfer within the vegetation canopy and soil, and the
WA TER -ENER G Y- VEGETA TI ON INTERA C TI ONS ( GIS ~ o ._ CD in 3 o - CtS in - C\5 in Cot on to - Ct - 159 (a DESIGNATE MODEL REGION (I) Define initial and boundary conditions Select scenario - ~Large scale ~ (correlative modelJ (O Define new climate or other changes on environmental template l (O Create 1 st approximation of new vegetation (no kinetics) Smallscale ~ simulation modelJ Create 2nd approximation of corrected new vegetation (1 st order kinetics) ~ 1 Application of knowledge of animal soil spersal landuse effected - 111 Create 3rd approximation (finer kinetics) Output to ecosystem function models FIGURE 5.5 Proposed approach for modeling vegetation (ecosystem) charge sub- ject to large changes in forcing functions.
160 RESEARCH STRATEGIES FOR THE USGC8P relationship between canopy scattering properties and relevant vegetation properties. Although parts of the above problem can best be addressed by special- ized researchers working in a loose confederation, the final goal of an integrated radiance-to-surface parameter algorithm must be kept firmly in mind. The achievement of the goal will require coordination of the talents of scientists working in several different disciplines: remote sensing technology, atmospheric physics, radiative transfer, plant physiology, and modeling. Ultimately, the effort could provide scientists with the means to calculate carbon, water, and energy fluxes over the land surface from satellite data. Soil Genesis Models Many of the models discussed above require some basic information about soil properties for their operation-soil physics properties in the case of GCMs and soil optical and nutrient properties in the case of the canopy radiative transfer/physiology models. Some of these properties can be derived using soil genesis models, which require data on the parent material, climatological regime, and vegetation cover as input. While these models cannot provide definitive production in most cases, they could have potential in terms of filling the gaps between reliable observations. Sensitivity Analyses All of the above sections have addressed the need to advance the realism and sophistication of different modeling efforts, in essence forming a "broad front" approach to the component parts of the issue of terrestrial biosphere- atmosphere interactions. Another important task must be coordinated with all of these a sensitivity study on the effect of errors or uncertainties in the input data set of each model on the calculated product. Ultimately, it is hoped that all the individual models will provide products that can be used as input or validation for other models. For example, the products from the remote sensing algorithms could be used to prescribe surface conditions for GCM studies. To make the best use of research resources, including money, time, equipment, and personnel, it is important that the sensitivity of each class of models to variations in their input parameter set be well understood. For example, if analyses indicate that GCM-simulated climates for continental interiors are sensitive to the successional stage of the vegetation cover there, it would be highly desirable to increase the flow of resources for improving the ecosystem dynamics models, which could then provide more realistic boundary condi- tions for the GCM. For this and similar problems, a gradualist approach to the sensitivity problem should be used. Simple schemes or parameter prescriptions
WATER-ENERGY-VEG~ATION I=ERACTlONS 161 should be used to determine the sensitivity of the model in question to variations in the input parameter set. The results of these basic tests should determine whether or not to invest heavily in more sophisticated approaches. Summary The above analysis has called out the need to vigorously promote re- search efforts in a few areas of obvious weakness. However, it should be reemphasized Hat modeling efforts in all the relevant areas should be supported to a much greater extent than they are now; these include the ecosystem dynamics models, land surface parameterizations, and atmospheric general circulation models. The descendants of current models will be the tools for understanding and predicting global change. INFRASTRUCTURE The modeling and data gathering tasks discussed in previous sections will not contribute to the overall goal of understanding and predicting global change unless there is a continuing effort to coordinate the activities. De- fining the form of a governing coordinating body is beyond the scope of this document, but describing its purview is a necessity. Operational Observations There is no question that the array of operationally acquired meteorologi- cal, oceanographic, space-based, and other data is invaluable for earth sys- tem studies. However, most of the existing networks are not suitably configured for this work (e.g., aviation forecasting dominates many meteorological activities), and insufficient resources are dedicated to storing the data. All operational systems need to be considered as possible contributors to the earth system science effort, and a means of prioritizing and storing important data types needs to be formalized. Assembling a self-consistent long-term record of variables important for earth system science will require the implementation of new measurement networks, new information systems, and a high degree of collaboration among agencies. Satellite Data Processing The need to produce useful satellite data products for the scientific com- munity has been emphasized above. A mechanism is needed to specify the list of desired products, with associated accuracy and precision requirements, and transmit this to the agencies so that resources, facilities, and personnel
162 RESEARCH STRATEGIES FOR THE USGCRP are dedicated to the task. This need is as urgent as the need for the devel- opment and launch of a new generation of instruments. Centers for Research and Monitoring Centers for research and monitoring should be the sites where more in- tensive, coordinated experiments and scientist training, as well as continu- ous monitoring-type observations, take place. An international effort should be made to establish such centers so that these research tasks are directly addressed. The centers should be sited in areas that are representative of a large and important vegetation formation, a "sensitive" area (e.g., a transition zone), or a benchmark area where there is a long research history and archive. To address the goals of ensuring a continuous, high-quality moni- toring effort while suitable for intensive field experiments of the scale of FIFE or larger, there should be permanent research staff attached to each center. These staff, in cooperation with visiting scientists, should also carry out an educative and training function. Education As noted in other chapters, there is currently a critical shortage of trained researchers to carry out the task of earth system science research. A coordinated approach is required to recruit good students into the field and to train them to be able to participate in interdisciplinary research. This will take money, effort, and organization. Interagency and International Coordination There is a need to integrate the planning and implementation of measure- ment networks, modeling efforts, experiments, and education at the interagency and ~nterna~aonal levels. This requires He interlocldng of experienced bureaucrats · · · . ant ~ practicing scientists. Coordination among most large-scale experiments is usually fairly haphazard, and thus over-redundancy and gaps continue to plague their operational implementation. A central clearinghouse, or at least an information exchange, would be useful. Such clearinghouses lead to better coordination in the use of instruments and personnel. REFERENCES Antarctic Ozone Hole Special Issue. 1987. Geophys. Res. Lett. 13~12~:1191-1362. Broecker, W.S., and G.H. Denton. 1989. The role of ocean-atmosphere reorganiza- tions in glacial cycles. Geochim. Cosmochim. Acta 53:2465-2501.
WATER -ENERGY- VEGETATION INTERACTIONS 163 Budyko, M.I. 1974. Climate and Life. Academic Press. 50B pp. Dickinson, R.E., and A. Henderson-Sellers. 1988. Modeling tropical deforestation: A study of GCM land-surface parameterizations. Quart. J. Roy. Meteorol. Soc. 114:439-462. Eagleson, P.S. 1982. Land Surface Processes in Atmospheric General Circulation Models. Cambridge University Press. 560 pp. Hansen, J., D. Johnson, A. Lacis, S. Lebedeff, P. Lee, D. Rind, and G. Russell. 1981. Climate impact of increasing atmospheric carbon dioxide. Science 213:957-966. McElroy, M.B., and S.C. Wofsey. 1986. Tropical forests: Interactions with the atmosphere. Pp. 33-60 in G.T. Prance (ed.), Tropical Rain Forests and the World Atmosphere. Westview Press, Boulder, Cola. Mooney, H.A., P.M. Vitousek, and P.A. Matson. 1987. Exchange of materials between terrestrial ecosystems and the atmosphere. Science 238:926-932. National Research Council. 1984. Global Tropospheric Chemistry: A Plan for Action. National Academy Press, Washington, D.C. National Research Council. 1988. Report on Global Change. National Academy Press, Washington, D.C. Ramanathan, V. 1988. The greenhouse theory of climate change: A test by inad- vertent global experiment. Science 240:293-299. Rotty, R.M. 1983. Distribution of and changes in industrial carbondioxideproduc- tion. J. Geophys. Res. 88: 1301-1308. Sato, N., P.J. Sellers, D.A. Randall, E.K. Schneider, J. Shukla, J.L. Kinter III, Y.-T. Hou, and E. Albertazzi. 1989. Effects of implementing the simple biosphere model in a general circulation model. J. Atmos. Sci. 46~18~:2757-2782. Sellers, P.J., F.G. Hall, G. Asrar, D.E. Strebel, and R.E. Murphy. 1988. The First ISLSCP Field Experiment (FIFE). Bull. Am. Meteorol. Soc. 69~1~:22-27. Sellers, P.J., et al. 1990. Experiment, design, and operations. Pp. 1-5 in American Meteorological Society Symposium on FIFE, February 1990, Anaheim, Calif. American Meteorological Society, Boston, Mass. TrabaLka, J.R. (ed.~. 1985. Atmospheric Carbon Dioxide and the Global Carbon Cycle. U.S. Department of Energy, Washington, D.C. (Available as NTIS DOE/E/R-0239 from National Technical Information Service, 5285 Port Royal Road, Springfield, VA 22161~.
