B

Figure 4.1. A: Schematic depiction of the traditional approach to investigate biospheric interactions with the physical climate system. B: Schematic of the processes represented in a comprehensive dynamical global vegetation model. See color plate section for a color version.

Marepwan samn meHHwft aBTopctc u m npaeoM

What is the overall effect of vegetation on the climate of the Earth? A recent sensitivity modeling study (KLeidon et al., 1999; Fraedrich et al„ 1999) demonstrates this control. In two experiments with the climate model ECHAM4 (Roeckneret al., 1996), the standard surface parameters in nonglaciatcd regions have been replaced either by values representative of a desert or by an evergreen forest that is optimally adapted to the prevailing climate. The two model experiments explore extremes of future anthropogenic changes in land use: a hypothetical scenario of global desertification and a hypothetical scenario of global afforestation. Apart from the different land surface parameterization*, the two simulation experiments included identical, simple boundary conditions: present-day climatological annual cycles of sea surface temperatures and present-day land ice cover. The model simulations were run for 11 years, and the first year was discarded to avoid spin-up effects. The simulations were performed in T42 horizontal model resolution (corresponding to approximately 2.8 latitude by 2.8 longitude) with 19 model layers in the vertical dimension.

Figure 4.2A show s the difference in the 10-year mean annual 2m temperature, and Figure 4.2B shows the ratio of the 10-year mean annual precipitation of the two simulations. The 2m-temperature difference demonstrates the fundamental role of évapotranspiration in the green planet simulation, lowering the temperature by as much as 8 in the Tropics over land. This happens despite the generally lower land surface albedo in the green planet simulation. The effect of the lower surface reflectance in the green planet simulation is dominant only over parts of the Sahara desert, where limited precipitation doesn't provide enough water for sustained evaporative cooling. Over the northern temperate latitudes, temperatures tend to be higher over the Asian continent, which is related to the vegetation reducing the surface albedo under low snow cover. Because of the prescribed sea surface temperatures, there arc no significant temperature differences over the oceans between the two simulations, Based on Figure 4.2A, one could argue that one sees only a local efTect, caused by the differently specified surface parameterization in the two model runs. However, an inspection of the other meteorological fields in the simulations shows that this is not the case. Indeed, the climate of the green planet simulation is characterized by a much more vigorous w ater cycle, w ith doubled precipitation (Figure 4.2B) and tripled évapotranspiration over land as compared w ith the desert world simulation (Kleidon et al., 1999). The relatively larger enhancement of évapotranspiration versus precipitation implies that global river runoff is reduced in the green planet simulation, and that is caused by the large waterholding capacity of vegetation. The enhanced water cycle in the green planet simulation leads to increased convection in the Tropics and thus a stronger Eiadlev circulation, with higher upper troposphere temperatures and decreased precipitation in the descending branches of the Hadley cell over the Oceans. The green planet simulation also exhibits a more zonal circulation structure in mid-latitudes (Fraedrich et aL, 1999), Overall, the dominant effect of the vegetation is found to he the strongly enhanced évapotranspiration, which tends to turn the Earth almost into a pure aquaplanet.

Clearly, the two climate simulations can demonstrate only a gross picture of the maximal control of the vegetation on the climate of the Earth, because they are subject to at least two major limitations: They do not consider any oceanic feedbacks, w hich ilaiepuan, 3amnmeHHbii?i asTopcKUM npaBOM

Figure 4.2, \: Difference in annual mean surface temperature (2m temperature) between the green planet and the desert world simulation. Colored areas indicate regions with si^nilieant (Student t-tesit, p > 0,9**), and gray areas indicate regions with insignificant, temperature differences. 1J: Ratio of precipitation in the green planet to precipitation in the desert world simulation. Qilored areas indicate regions with significant (Student t-test„ p > 0.99), and gray areas indicate regions with insignificant, precipitation differences. See color plate section for color version.

Marepwan samn meHHwft aBTopcK u m npaeoM

might amplify or lessen the vegetation impact to a vet unknown degree. Furthermore, the prescribed surface vegetation in the green planet simulation might not be ecologically sustainable in every part of the world. N evertheless, the simulations serve to illustrate that indeed, vegetation must he considered an important integral part in any credible Earth System Model.

The drastic changes investigated here arc not likely to occur within the next century. Nevertheless, there exist several more-subtle feedback mechanisms in the vegetation-climate system that might significantly affect the response of the climate system to the anticipated future change in greenhouse gas forcing. For example, a recent simulation study has demonstrated the effects of possible changes in stomata control on évapotranspiration and thus on the surface temperature (Sellers et al, 19%). Plants regulate their stomata conductivity to make maximum use of the available leaf water. The stomata, however, are also the pathway for the CO> that is needed for photosynthesis. From empirical studies it is well established that the conductivity is not only inversely related to water stress but also directly proportional to the rate of leaf photosynthesis (Hall, 1988; Collatz et al., 1991). Hence, photosynthesis of atmospheric CO2 can directly control leaf conductance and thus évapotranspiration, and this process provides a tight coupling of the cycles of carbon and water, The consequences of this tight coupling have only recently been investigated in atmospheric general circulation models that, necessarily, include a description of the terrestrial carbon cycle as part of the land surface package. Using such a model it has been demonstrated (Sellers et al., 1996; Bounoua et al., 1999) that this coupling might considerably enhance the land surface warming under a doubled atmospheric CO2 concentration. If the latter is rising, plants may profit by increased assimilation; however, they may also "decide" to downregulate photosynthesis to present-day levels - for example, by developing fewer stomata per leaf area (Woodward, 1987) or by more intensive but shorter leaf photosynthesis periods (Hattenschwiler and Koerner, 1996; Koerner, 1998). Whether downregulation indeed occurs in a world with enhanced CO? is questionable, but it reflects a possibility. If so, évapotranspiration would be significantly reduced, yielding an additional surface warming beyond the direct COi-induced radiative effect. According to the present simulations (Sellers et al, 1996; Bounoua et al, Î999), this additional warming might be on the order of 30% of the direct response.

4,3 Exchanges of Carbon Dioxide

It is well known that in addition to the ocean, the terrestrial biosphere plays an important role in the global carbon cycle and the regulation of the atmospheric CO2 concentration. There exists the fundamental question of the fate of the excess CO2 that is directly injected into the atmosphere by anthropogenic activities (burning of fossil fuels and from changes in land use). This question traditionally concerns the atmospheric budget of CO> and how this changes with time as a function of the increasing CQ2 concentration (Schimcl et al, 1995, 1996; Heimann, 1997; Prentice et al, 2000; Prentice, Chapter 11, this volume). There exists also a scientifically challenging second question, regarding the climatic feedbacks on the global carbon cycle.

Maiepnan, 3amnmeHHbiw aBTopcKMM npaeoM

Clearly* as witnessed by the changes in CO? during the glacial cycles (Petit et al., 1999) or during the Holocenc (Indermuhle et aL, 1999), there exists considerable scope in the Earth System for climate-induced shifts in carbon storage between the different reservoirs on different time scales.

As a climate forcing agent, only the mean atmospheric CO? concentration is of any significance, because the spatiotemporaJ concentration variations induced by sources and sinks at the surface of the Earth are too small to have any radiative effects. Nevertheless, an accurate observation of these variations all over the globe provides an important means to evaluate the surface exchange fluxes simulated by carbon cycle models. As part of the international Carbon Cycic Model Linkage Project (CCMLP), funded in part by the U.S. Electric Power Research Institute (EPRI) and with support from the Global Analysis, Interpretation and Modelling (GAIM) task force of the I GBP, several global terrestrial carbon cycle models have been scrutinized using this approach (Heimann et al., 1998; Kicklighter et al., 1999).

Most pronounced is the seasonal cycle w ith atmospheric variations of as much as 20 ppmv in high latitudes in the north, which gradually becomes smaller in mid-latitudes and toward the equator, vanishes south of the equator, and reappears in the deep Southern Hemisphere with opposite phase. Figure 4.3 shows the result of a simulation

-HflBM -SILVAN

Figure 4.3. Seasonal cycle of atmospheric COJr Black dots: Observations from the NOAA-CMDL database (Conway ct aL, 1W4), with error bars reflecting interannual variability. Colored solid lines: simulation results from five prognostic terrestrial carbon cycle models (Heimann et al., 1W8). Dashed black line: simulation by a diagnostic terrestrial model driven by remote sensing and climate data (Knurr and I leimann, 1995). See also color plate section.

Maiepwan, 3auiM)MeHHbift auTopcKMM npaBow study, in which five geographically explicit, process-based terrestrial models arc evaluated by injecting the simulated surface exchange fluxes into a global atmospheric transport model to calculate the concentration variations at the monitoring stations (Heimann et aL, 1998). The terrestrial models included in the study are of intermediate complexity. They describe the major processes of the cycling ofcarbon (photosynthesis, leaf development and abscission, allocation of carbon to different parts of the plants, litter fall, soil carbon, heterotrophic respiration) on a global grid of 0.5 latitude and longitude in a prognostic fashion, but with a prescribed, fixed vegetation distribution. For comparison, the results from a sixth model are also shown (Simple Diagnostic Biosphere Model [SDBM], Knorr and Heimann, 1995). This is not a prognostic model but uses satellite remote sensing information ("greenness index" NDYI) to describe the photosynthetic activity of the vegetation.

Shown are the simulated seasonal cycles of atmospheric C(>2 at four stations in the Northern Hemisphere and the Tropics: Alert in northern Canada, Maun a Loa, Hawaii, Barbados in the Antilles, and Ascension Island in the southern Atlantic Ocean. The observations are from the global monitoring network of the Climate Monitoring and Diagnostic Laboratory of the U.S. Oceanographic ant! Atmospheric Administration (NOAA-CMDL) (Conway et al. 1994); the error bars reflect both instrumental errors and interannual variability. The modeled signals also include an oceanic component derived from a global ocean carbon cycle model (Six and Maier-Rcimer, 1996), which, however, is very small at the Northern Hemisphere stations.

The simulations show thai the agreement with the observations is not perfect, indeed, by comparing the simulations of the prognostic models with the diagnostic SDBM, it is seen that the latter matches the atmospheric observations significantly better (Heimann et al, 1998). This indicates that the modeled processes leading to the leaf area still require refinement. The simulation results at Ascension Island also show that all the prognostic models overestimate the observed seasonal cycle, an error that can be traced to defects in the modeled water cycle, in particular to an assumed too shallow rooting depth in the Tropics.

Modeling the seasonal cycle in this \va\ has become a standard test for the evaluation of terrestrial carbon cycle models (Nemry et al., 1999). Clearly, a test such as shown here addresses only particular aspects of a terrestrial model and cannot be conclusive.

An additional atmospheric signature of variability in the carbon cycle is provided by the interannual fluctuations in the grow th rate of atmospheric CO?. Af ter subtraction of the seasonal cycle, the globally averaged CO2 concentration exhibits significant variations that have been shown to correlate w ith large-scale climate fluctuations, such as the El Nino-southern oscillation (Bacastow, 1976; Bacastow et al,, 1980) or the climate anomaly induced by the Mt. Pinatubo volcanic eruption after 1992 (Sarmiento, 1993). These variations are substantial, showing year-to-year variations of as much as 2-4 PgC a~1 or more than 50% of the annual emissions from the burning of fossil fuels. They are larger than the documented interannual changes in the anthropogenic sources and hence clearly represent the response of the global carbon cycle to short-term climatic fluctuations. Unfortunately, both the ocean and the terrestrial carbon cycle contribute to these fluctuations, and although concurrent measurements of the nC/uC isotopic

MaTepwa/i, 3aL4WLneHHbifr aBTopcKMM npaBOM

composition of atmospheric CO7 could in principle be used to discriminate between the two components, the existing isotope ratio observations are contradictory (Keeling et al., 1995; Francey et al., 1995). Recent modeling and observational studies nevertheless point to a relatively smaller contribution of the ocean to the interannual variability of atmospheric CO?.

These interannual variations are a prime target for the assessment of the short-term climate sensitivity of terrestrial carbon cycle models. Within CCMLP a series of model simulations have been carried out in which the carbon evele models have been driven by the observed changes in temperature and precipitation compiled by the Climate Research Unit of the University of East Anglia (updated from Jones, 1994; Hulmc et al., 1998). Interestingly, the terrestrial models were able to reproduce a significant fraction of the observed interannual variability generated in the Tropics, both in magnitude and in phase. They were less successful, however, in the simulation of the extratropical fluctuations, in particular the Mt. Pinatubo anomaly after 1992 (Heimann et al, unpublished). This indicates a too-wrcak modeled climate sensitivity in temperate latitudes. As an example, Figure 4.4 shows the global response of one of the more successful prognostic models compared to the observations. From the latter, the anthropogenic long-term trend has been removed. The agreement between simulation and observations is fair, that is, significant at the 90% confidence level (r2 = ~0.7), but the modeled amplitude of the variations is too large, Also shown, for comparison, is the globally averaged atmospheric growth rate as modeled by an oceanic carbon cycle model driven with the observed surface fluxes of wind stress, temperature, and freshwater fluxes from the European Centre for Medium Range Weather Forecasts

Figure 4.4. Interannual, climate-driven variability in the global carbon cycle. Red line: observed growth rate of globally averaged atmospheric CO2 concentration after subtraction of the anthropogenic trend (Heimann, 1997). Cireen line: growth rate of atmospheric COj simulated by a process-based terrestrial model driven by observed changes in temperature and precipitation (Sitch et al., in preparation). Blue line: growth rate of atmospheric COi simulated by a coupled ocean general circulation-carbon cycle model forced with observed anomalies of surface wind stress,, temperatures, and freshwater fluxes (LeQucre, in press). See also color plate section.

Figure 4.4. Interannual, climate-driven variability in the global carbon cycle. Red line: observed growth rate of globally averaged atmospheric CO2 concentration after subtraction of the anthropogenic trend (Heimann, 1997). Cireen line: growth rate of atmospheric COj simulated by a process-based terrestrial model driven by observed changes in temperature and precipitation (Sitch et al., in preparation). Blue line: growth rate of atmospheric COi simulated by a coupled ocean general circulation-carbon cycle model forced with observed anomalies of surface wind stress,, temperatures, and freshwater fluxes (LeQucre, in press). See also color plate section.

Maiepi/tan, 3aLHwmeHHbit?i aBTopcKUM npaBOM

(ECMWF) analyses over the period 1979-1997 (LeQuere et al., in press). Clearly, the oceanic contribution is dwarfed by the terrestrial fluctuations.

4.4 Exchanges of Oxygen and Its Isotopes

Parallel to the exchanges of water vapor and carbon dioxide, oxygen is also exchanged by the terrestrial biosphere. This exchange proceeds in relatively well-known, fixed stoichiometric ratios of about 1.1 mole of oxygen released for I mole of CO? assimilated during photosynthesis and reversed during oxidation of organic matter. On geological time scales this exchange is crucial for the habitability of the planet. The present amounts of oxygen in the atmosphere, however, are so large that the fluctuations induced by biotic and abiotic processes are small (a few ppmv on a background concentration of200,000 ppmv). As a result, their measurement has only very recently become feasible. Variations in the ratio of molecular oxygen to nitrogen (which is the quantity that is measured) provide a means to separate terrestrial from oceanic carbon exchange fluxes (Keeling and Shertz, 1992; Keeling et al,, 1993; Keeling et al,, 1996; Bender et al., 1996). This is because CO? and molecular 0> exchanges with the ocean exhibit quite different stoichiometric relations compared with the exchanges on land. Because of the relatively well-known, fixed stoichiometric relations for terrestrial exchanges, atmospheric oxygen measurements primarily have been used to evaluate oceanic models, such as through the seasonal cycle of CO2 and O2 (Keeling et al., 1998) or the mean annual meridional gradient in CO? and O2 (Stephens et ah, 1998).

Oxygen consists of three stable isotopes (|60, 170, and ihO) with characteristic iso-topic ratios in the different molecular forms (1MX CO], and ()?). Chemical, physical, and biological transformation processes induce fractionations, which lead to different isotopic ratios in the various reservoirs of these compounds (e.g.. Keeling, 1995). As described below, until now primarily the 1sO/lftO ratio has been of interest in observational and modeling studies. Because almost all fractionation processes are mass-dependent, the behavior of 170 can be predicted from ls() and hence appears to be redundant. However, it has recently been found that exchange processes with the ozone chemistry in the stratosphere induce a mass-independent fractionation that is not yet completely understood {Thiemens et al., 1991; Mauersberger et aL, 1993,1999). This effect creates a l70 anomaly that may be traced through the various exchange processes of CO2 and C>2 within the entire biogeochemical system.

To model these isotopes, the cycles of water, carbon, and oxygen all must be described in a consistent way. This requires an atmospheric general circulation model that includes a full description of the oxygen isotopes in the hvdrological cycle (1 ioffmann et al., 1998). This prov ides the background isotopic composition of precipitation, groundw ater, and water vapor all over the globe, which, together with the energy and water balance of the canopy, determines the isotopic composition of leaf water. This in turn is the anchor point that determines the isotopic composition of molecular oxygen released to the atmosphere during photosynthesis, and it also imprints its isotopic signature to the CO2 that is exchanged through the stomata of the leaves with the atmosphere. Conversely, during respiration the CO? released carries the isotopic signature of the

MaTepwa/i, 3aL4WLneHHbifr aBTopcKMM npaBOM

leaf and soil water, while the strongly fractionated accompanying oxygen consumption induces an isotopic signal in the atmospheric oxygen. The oxygen isotopic composition of atmospheric CO2 and O2 is also influenced by oceanic exchanges and by the (small) impacts from the anthropogenic burning of fossil fuels,

In the case of molecular oxygen, these exchange processes lead to an atmosphere enriched in [S0 compared with mean ocean water, known as the Dole effect (Dole, 1935). Variations in this quantity in the past reflect not only the relative strength of terrestrial and oceanic productivity (Bender et al., 1994) but also the geographical location of terrestrial oxygen exchanges, w hich may change with the displacements of the vegetation zones during the glacial-interglacial cycles. The isotopic composition of atmospheric O2 is also not uniform but is expected to exhibit spatiotemporal variations. These variations are relatively small (^0,02%, Seibt, 1997), but it is expected that they will become measurable within the near future. The fractionation effects associated with the exchanges of CO2, however, generate easily detectable atmospheric signatures (Francey and Tans, 1987).

All the isotopic fractionation processes are relatively well understood and can be described in models (Ciais et aL, 1997a; Ciais et aL, 1997b; Seibt, 1997), In these modeling studies, however, the isotopic tracers are not yet treated in a fully consistent way but are simulated using an off-line model hierarchy. That is, an atmospheric general circulation model computes the climate fields and the isotopic composition of precipitation, groundwater, and near surface water vapor. These output fields arc then used to drive a terrestrial carbon cycle model that computes the exchange* fluxes of CO2 and (>2 and their isotopic compositions. In a third step, these exchange fluxes are given to an atmospheric transport model for the computation of the atmospheric signatures in the concentration and isotopic ratio fields at the observing stations. Clearly, with the advent of comprehensive DGVMs that include the f ull biogeochemical cycles and can be coupled to climate models, simulation st udies of the oxygen isotopic tracers provide a promising approach for the evaluation of the modeled exchanges on a regional and global scale.

4.5 Conclusion

Terrestrial hiospheric exchanges of water, carbon dioxide, and oxygen are controlled essentially by the same set of processes. Hence, a realistic representation in comprehensive biogeochemical models necessitates a consistent treatment of all three species. The existing models have not yet reached the stage depicted in Figure 4.113, in that they still are being used mostly ofT-line in an otl-line mode, that is, driven by the output of climate models. Nevertheless, the process knowledge exists to represent the various exchange processes in a realistic manner as an interactive component within comprehensive Earth System Models, and, as show n here, appropriate techniques exist that allow an evaluation of the model performances on regional and global scales.

Through exchanges of H2O and CO2, the biosphere can significantly modify the climate of the Earth. Hence, a consistent implementation of the terrestrial carbon cycle and its coupling to the hydrological cycle on land represent an immediate next model

Maiepi/tan, 3ainwmeHHbit?i aBTopcKUM npaBOM

development step. Eventually, the coupling to existing comprehensive models of the oceanic carbon cycle will permit a consistent treatment of the carbon cycle in the next-generation atmosphere-ocean climate models. Such models are needed to investigate the history of climate, and they are also an indispensable tool for the assessment of anthropogenically induced climate change in the near future.

As outlined here, the cycle of oxygen and its isotopes is tightly coupled to the carbon cycle. Hence, variations of atmospheric molecular 0> and of the stable oxygen isotope ratios in H2O, CO?, and O? provide an additional tool to evaluate the modeled biospherie exchange processes» Variations of these tracers are being or will be monitored soon, and their temporal variations on longer time scales have been observed in ice cores. It is hoped that the inclusion of the cycles of oxygen and its isotopes in the next generation of coupled climate-biogeochemistry models will receive high priority.

0 0

Post a comment