Simulating biogenic volatile organic compound emissions. in the Community Climate System Model - PDF

Please download to get full document.

View again

of 22
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information Report



Views: 6 | Pages: 22

Extension: PDF | Download: 0

Related documents
Simulating biogenic volatile organic compound emissions 1 in the Community Climate System Model Samuel Levis, Christine Wiedinmyer, Gordon B. Bonan, Alex Guenther National Center for Atmospheric Research
Simulating biogenic volatile organic compound emissions 1 in the Community Climate System Model Samuel Levis, Christine Wiedinmyer, Gordon B. Bonan, Alex Guenther National Center for Atmospheric Research PO Box 3000 Boulder, CO Corresponding Author: Samuel Levis National Center for Atmospheric Research PO Box 3000 Boulder, CO Phone: FAX: submitted to JGR- Atmospheres 11/21/02 Abstract 2 The Community Climate System Model (CCSM) calculates terrestrial biogenic volatile organic compound (BVOC) emissions using an algorithm developed from field and laboratory observations. This algorithm is incorporated in CCSM, a coupled atmosphere, ocean, sea-ice, and dynamic vegetation model, as one step towards integrating biogeochemical processes in this model. CCSM is designed to easily incorporate more complex BVOC models in the present framework when such models become available. In CCSM, warm and forested regions emit more BVOC than other regions in agreement with observations. On a global scale, simulated isoprene emissions from land of 507 Tg C per year compare well with other model simulations. As simulated climate and vegetation vary from one year to the next, BVOC emissions respond. The interannual variability of the simulated biogenic emissions can exceed 10% of the estimated annual anthropogenic emissions provided in the IPCC emission scenarios. We include BVOC emissions within the CCSM to ultimately reduce the simulated climate uncertainty due to natural processes in this model. 1. Introduction Vegetation contributes about 90% of the non-methane Volatile Organic Compounds (VOC) emitted globally to the atmosphere [Guenther et al. 1995]. These biogenically-emitted VOC (hereafter BVOC) include isoprene (C 5 H 8 ), monoterpenes (C 10 H 16 ), and other reactive carbon compounds. The amount of carbon in global BVOC emissions may exceed 1 Pg yr -1 [Guenther et al. 1995]. Many BVOC react with ozone and other oxidants in the atmosphere, with effects on local, regional, and global atmospheric chemistry [Fehsenfeld et al. 1992; Houweling et al. 1998]. For example, Poisson et al. [2000] simulated a 15% increase in the tropospheric lifetime of methane due to the presence of BVOC. 3 Secondary organic aerosol formation from BVOC emissions is estimated to range from Tg yr -1 [Griffin et al. 1999] to Tg yr -1 [Andreae and Crutzen, 1997]. This amount of aerosol is similar in magnitude to the estimates of carbonaceous aerosol emitted globally to the atmosphere from fossil fuel and biomass combustion [Liousse et al. 1996]. Aerosols affect the atmosphere's radiative balance. Through their effects on atmospheric chemistry, aerosol concentrations, and the global carbon cycle, BVOC emissions potentially influence global climate [Constable et al. 1999]. Collins et al [2002] show that isoprene has a positive secondary global warming potential. The Intergovernmental Panel on Climate Change (IPCC) includes non-methane VOC emissions in their recommended emission scenarios for climate models. These scenarios estimate about 140 Tg C yr -1 of VOC from anthropogenic sources for the mid-1990s, and from less than 100 to more than 550 Tg C yr -1 for 2100 [Nakicenovic et al. 2000]. BVOC emissions vary by plant species and depend on environmental conditions such as temperature, solar radiation, plant water stress, and ambient ozone and carbon dioxide concentrations [Penuelas and Llusia, 2001]. Algorithms have been developed from field and laboratory research to simulate BVOC emissions [e.g., Guenther et al. 1995] to show, for example, that a 6 K rise in annual temperature may double isoprene emissions in a Texas savanna [Guenther et al. 1999]. To further track BVOC concentrations and their chemical transformations, BVOC emission rates are input into atmospheric transport and chemistry models [e.g., Poisson et al. 2000; Bouchet et al. 1999]. The results from such models have been used to examine BVOC chemistry and to investigate BVOC effects on regional pollution [e.g., Pierce et al. 1998]. As an alternative to the iterative coupling of land-based emissions to atmosphere-based 4 transport and chemistry, we use an integrated framework in which BVOC emissions are calculated within a land surface and dynamic vegetation model. This model simulates ecological and physiological processes, which influence BVOC emissions, and operates as a component of a climate system model, including atmosphere, ocean, and sea-ice components. The atmosphere component may include models of atmospheric chemistry and transport. This way, BVOC emissions, transport, and chemistry occur synchronously in an internally consistent framework. Synchronous coupling of processes is a favored approach when non-linear interactions are involved, and becomes imperative when surface characteristics change, as is the case with vegetation dynamics [Foley et al. 2000]. With this study, we evaluate our model's global scale BVOC emissions. We also assess the model's simulated interannual variability in these emissions. Ultimately, including BVOC emissions within a climate system model serves the purpose of reducing climate uncertainty from natural processes. 2. Methods We have incorporated the BVOC emission model developed by Guenther et al. [1995] into the Community Land Model version 2.0 (CLM) [Bonan et al. 2002a]. The former is an emissions model developed from field and laboratory observations. The latter is a state of the art land surface model which computes all the variables required by the BVOC model. 2.1 Community Land Model The Community Land Model version 2.0 (CLM) [Bonan et al. 2002a] is a component of the Community Climate System Model version 2.0 (CCSM) [Blackmon et al. 2001]. CCSM couples atmosphere, ocean, sea-ice, and land processes synchronously in an internally consistent framework. The land component, CLM, simulates the exchange of water, carbon, energy, and 5 momentum between land surfaces and the atmosphere through a wide range of ground and canopy biogeophysical processes. Vegetation is described in terms of plant functional types which arise from landscape differences observable from satellite (i.e., annual versus perennial vegetation, evergreen versus deciduous, and needleleaf versus broadleaf) [Running et al. 1995]. These remotely sensed differences are further categorized into broad climatic groups (Table 1). CLM has a standard set of input data for the land surface (i.e., percent lake, wetland, glacier, and plant cover, as well as leaf area index (LAI) per plant functional type). The land surface data are satellite derived and are described in detail by Bonan et al. [2002b]. As an alternative to using prescribed vegetation from satellite data, CLM can simulate global vegetation characteristics with the inclusion of a dynamic vegetation model (Bonan et al. 2002c; Sitch et al. 2002). In this case, plant biomass responds to climate variations from year to year; plant phenology (evergreen, summergreen, raingreen) responds to temperature and moisture conditions to determine daily LAI; and plant photosynthesis and respiration are calculated at every twenty minute time step of the land model. The annually summed carbon balance influences the success of each plant functional type against its competitors in a model grid cell. Bonan et al. (2002c) found the vegetation dynamics simulated by this land model satisfactory for use with the fully coupled CCSM. 2.2 Module of BVOC emissions Terrestrial BVOC fluxes are estimated from F = ε D C L C T equation 1 [Guenther et al. 1995], where F is the BVOC flux (µg C m -2 h -1 ), ε (Table 1) is the plant type dependent emission capacity (µg C g -1 dry foliar mass h -1 ) normalized to a photosynthetically active radiation (PAR) flux of 1000 µmol m -2 s -1 and a leaf temperature of K, D is the foliar density (g dry foliar mass m -2 ), and C L and C T are dimensionless empirical terms that 6 modulate the emissions in response to incident PAR (isoprene only) and leaf temperature, respectively [Guenther et al. 1993]. We calculate foliar density, D, by dividing LAI by the specific leaf area (Table 1). C L is calculated separately for sunlit and shaded leaves with the assumption that sunlit leaves receive all the direct-beam PAR. Sunlit and shaded leaves share the diffuse PAR as described by Bonan [1996]. Emission factors have been measured for a wide variety of plant genera and species, which CLM does not represent explicitly. CLM employs the concept of plant functional type to describe vegetation in broad terms, for use within global climate and dynamic vegetation models. CLM's plant functional types are grouped into six emission factor categories (Table 1). Grasses and crops (4 functional types) are assumed to have negligible isoprene emissions and very low monoterpene emissions as recommended by Guenther et al. [2000]. Needleleaf deciduous trees (1 type) are assigned the emission rates recommended for Larix species by Guenther et al. [1994]. Needleleaf evergreen trees tend to have high monoterpene emissions [Guenther et al. 1994]. We have assigned relatively low isoprene emission factors to all needleleaf evergreen trees but have a higher factor for boreal (1 type) than temperate (1 type) trees to account for the higher proportion of isoprene-emitting spruce trees in North American boreal forests. Tropical broadleaf evergreen trees (1 type) are assigned the emission factors recommended by Guenther et al. [1995] for tropical rainforests. The remaining broadleaf trees and all shrubs (7 types) are assigned the global average rates suggested by Guenther et al. [1995]. This model does not account for BVOC fluxes from soils, which are lower than fluxes from vegetation by at least one order of magnitude Model Simulations We show results from two simulations here: (1) A CLM simulation with prescribed NCEP atmospheric data for 1990 [Kalnay et al. 1996] 7 and prescribed vegetation (all data described in detail by Bonan et al. [2002b]). The model is run on a 1 in latitude by 1 in longitude grid for one model year. With this simulation we evaluate the model's ability to calculate BVOC emissions. (2) A present day CCSM simulation in which vegetation characteristics are simulated by the vegetation dynamics component of CLM. The model operates on a T31 gaussian grid (approximately 3.75 in latitude by 3.75 in longitude) long enough for vegetation cover and LAI to be near equilibrium with the climate. We look at output from ten model years. With this simulation we explore the potential for interannual variability in BVOC emissions due to variability simulated in climate and vegetation. Isoprene, monoterpene, Other Reactive VOC (ORVOC), Other VOC (OVOC), and carbon monoxide (CO) emissions are predicted by the model at each twenty minute time step of the simulations. Monthly-averaged outputs are presented here. The currently available atmospheric component of the CCSM does not include chemistry and transport of chemical species. 3. Results and Discussion 3.1 CLM Simulation Simulated BVOC emission patterns for January and July (Figure 1) agree with the global patterns simulated by Guenther et al. [1995]. Higher emissions occur where LAI, PAR, and temperature are higher. The tropics stand out as a source year round, with some variability due to the annual LAI cycle of drought-deciduous trees. The extra-tropics respond to the annual LAI, PAR, and temperature cycles with higher emissions in summer and lower emissions in winter. Local maxima correspond to forests as opposed to deserts, grasslands, and crops. Certain regional maxima can also be attributed to the dominant plant functional type's emission capacity, ε. Broadleaf trees are assigned higher ε for isoprene than needleleaf trees, resulting in the 8 particularly high summer emissions in the eastern half of the United States. Agreement with Guenther et al. [1995] suggests a satisfactory simulation of BVOC emissions by CLM. CLM simulates a global land BVOC emission of 692 Tg C annually, while Guenther et al. [1995] simulate Of that value, 507 Tg C are isoprene emissions and 33 Tg C are monoterpene emissions in CLM. Guenther et al. [1995] report these numbers as 503 and 127 Tg C. Potter et al. [2001] and Wang and Shallcross [2000] quote 559 and 530 Tg C from isoprene in their respective models. CLM's lower estimates compared to Guenther et al. [1995] for compounds other than isoprene are partly due to the use of lower emission capacities for certain plant types and partly due to differences in the land cover and LAI data sets. CLM's LAI values are generally lower than those used by Guenther et al. [1995]. BVOC emissions are directly proportional to the LAI (equation 1). We believe that CLM's vegetation and LAI data, derived entirely from satellite products [Bonan et al. 2002b], improve upon the data used by Guenther et al. [1995]. CLM's global isoprene emission is similar to previous estimates, partly due to two offseting processes in CLM: (1) Only isoprene emission depends on PAR in this model and for lower LAI values more PAR is absorbed as direct beam radiation by the canopy. Direct beam PAR results in higher isoprene emissions than diffuse beam PAR. (2) CLM operates on a 20- minute time step. The same emission algorithms were driven with monthly-average inputs in Guenther et al. [1995]. CLM's higher time resolution can lead to emission increases [Penner et al. 2001]. In a similar, but North America only simulation, we used natural vegetation where the satellite data include crops. We found that BVOC emissions increase significantly due to the higher LAI and emission capacities corresponding to natural vegetation (Table 1). Greatest emission increases were found in the eastern half of North America and especially during the 9 summer months (not shown). This result is consistent with previous studies showing the effects of land use on BVOC emissions [Guenther et al. 1999; Schaab et al. 2000]. 3.2 CCSM Simulation Ten years of output from a CCSM simulation (atmosphere, ocean, sea-ice, and land components all active) are analyzed. The land component, CLM, operates in dynamic vegetation mode. The atmospheric component does not include chemistry and transport of emitted compounds at this time. Here we address interannual variability in the simulated BVOC emissions in response to variability in CCSM's climate and vegetation. Annual simulated BVOC emissions vary by as much as 5% and monthly simulated emissions by as much as 18% from year to year over all land. Three individual grid cells are selected to illustrate the behavior of interannual variability in these emissions. A tropical forest in Amazonia (grid center at W, 1.86 S), a temperate deciduous forest in eastern Asia (grid center at E, N), and a boreal forest in Canada (grid center at W, N). Monthly terrestrial BVOC emissions are shown for the ten simulation years (Figure 2). The emissions vary in response to changes in climate and LAI (Figures 3 and 4). In general, higher LAI, PAR, and temperature contribute to higher BVOC emissions. As a result, the tropical grid cell has higher BVOC emissions than the extra-tropical grid cells year round. Interannual variability is more during the wet season (January through April). Monthly isoprene emissions vary by as much as 1.4 Tg C month -1 in April from year to year, or approximately a factor of 2.5 (not shown). Monthly BVOC emissions vary by up to a factor of 2 and annual emissions by up to 29% from year to year. Simulated emissions integrated over the Amazon basin agree with the emissions reported by Guenther [2002] (near 0.2 Pg C yr -1 ). The 10-year average basin-wide interannual variability is 4% in CCSM compared to 10% simulated by Guenther [2002] in a 14-year simulation. 10 The boreal and temperate grid cells both show a seasonal pattern with maximum emissions during boreal summer. BVOC emissions from these grid cells are smaller than from the tropical grid cell, but the relative variations in the emissions tend to be larger. In particular, summer emissions of isoprene at both sites vary by as much as a factor of 3 from one year to another. Winter emissions are relatively small. Still, interannual variability is high, especially for January, mainly due to the high variability in winter temperature (Figure 3). Maximum LAI at the temperate grid cell doubles over the course of the simulation, because the vegetation is not in equilibrium with the climate in this particular grid cell. BVOC emissions increase with time in response to the LAI. This grid cell consists mainly of broadleaf deciduous trees, which are strong isoprene emitters. As a result, this grid cell is highly sensitive to the variability in PAR (Figure 4). 4. Summary and Conclusions A terrestrial BVOC emissions module has been added to the CLM as one step towards introducing biogeochemical processes in the CCSM. CLM-simulated BVOC emissions and their spatial and temporal patterns generally agree with past modeling estimates. Emissions are sensitive to the choice of land cover, so it is advisable to include human land use when estimating BVOC emissions. In a fully coupled CCSM simulation (all model components active), the interannual variability in BVOC emissions is assessed. Global and point BVOC emissions vary substantially from year to year due to the simulated variability in climate and vegetation and play a role in the net ecosystem flux of carbon. Tian et al. [1998] investigated the interannual variability of CO 2 flux from the Amazon basin from 1981 to 1994 and found that the net ecosystem production varied from 0.7 Pg C yr -1 (a net uptake) to -0.2 Pg C yr -1 (a net release of CO 2 ). CCSM's 11 simulated variability in BVOC emissions from the Amazon basin is of comparable magnitude. The IPCC estimates about 140 Tg C yr -1 of VOC emitted to the atmorsphere from anthropogenic sources for the mid-1990s, and from less than 100 to more than 550 Tg C yr -1 for 2100 [Nakicenovic et al. 2000]. The variability in global BVOC emissions shown in CCSM can exceed 10% of these estimated anthropogenic VOC emissions, suggesting an additional level of uncertainty in the emission scenarios, which has not been accounted for, due to BVOC. By introducing a model of BVOC emissions to the CCSM, our goal is to reduce climate uncertainty due to natural processes. The simulated interannual variability of BVOC emissions found at three sample grid cells also has local to regional implications. Emission changes can affect regional photochemistry. For example, ozone productivity could become NO x -limited as opposed to VOC-limited in years with elevated emissions, with repercussions on air quality. With the introduction of atmospheric chemistry and transport in the CCSM, we will be able to better understand the full range of effects of BVOC emissions and their variability. In this first attempt at simulating BVOC emissions in the CCSM we recognize some shortcomings. The plant functional types used in CLM are the same types that can be sensed by satellite and simulated by dynamic vegetation models. BVOC emission capacities (Table 1) are not easily categorized using the same list of plant functional types. For example, large amounts of isoprene are emitted from mosses and ferns [Sharkey and Yeh, 2001], which are not represented in CLM. Future studies may include region-specific emission capacities and plant types grouped according to alternate criteria. For the present study, the list of plant functional types is kept unchanged. The BVOC emissions model in CLM can be updated with more process-based algorithms. CLM calculates photosynthesis and stomatal conductance at every time step. 12 Emissions of certain BVOC are linked to such plant physiological processes [Guenther et al. 2000], but these links have not, yet, been parameterized in global models. Acknowledgments The National Center for Atmospheric Research is sponsored by the National Science Foundation. References Andr
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks