B53A-01
CERES, the Carboeurope Regional Experiment Strategy in les Landes, South West France, May-June 2005
Quantification of sources and sinks at global and regional scales requires not only a good description of the land sources and sinks of carbon, but also of the synoptic and mesoscale meteorology. An experiment was performed in les Landes, South West France, to determine the variability in concentration gradients and fluxes of CO2 . Several aircraft sampled the CO2 concentration and fluxes over the area, while fixed stations observed the fluxes and concentrations at high precision. Several (mesoscale) meteorological modeling tools were used to plan the experiment and flight patterns. Results show that at regional scale the relation between profiles and fluxes is not obvious, and strongly influenced by airmass history and mesoscale flow patterns. This calls for an experimental and modeling strategy that takes into account the large spatial gradients in concentrations and the variability in sources and sinks that arise from different land use types. We briefly describe such an analysis.
B53A-02
Including Complexity, Heterogeneity and Vegetation Response Characteristics into Carbon Balance Assessments at Continental Scale: Stepwise Development of a Simulation Framework with the Bottom-Up Core Model PIXGRO
The cultural landscapes of the European continent are characterized by highly fragmented land cover, distributed along the gradient from extremely sparse, dry Mediterranean shrublands to sub-arctic wetlands and forests. In addition, several mountain systems, in particular the Alps, modify vegetation and land surface exchange characteristics. In the recently begun CARBOEUROPE project, energy, water and CO2 exchanges of selected ecosystem types are monitored via eddy covariance methods. The model PIXGRO is being developed to help relate measurements at flux tower sites to measurements carried out at larger scale and to CO2 exchange estimates obtained with atmospheric inversion techniques. The main question addressed in this modelling effort is how to efficiently include a spectrum of vegetation characteristics into carbon balance assessments at large scale and in such a way that the influence of local response on regional fluxes may be visualized and analyzed. Furthermore, the modelling attempts to demonstrate both strengths and weaknesses in existing ecosystem process information required to understand carbon balances. The model PIXGRO attempts to define continental distribution and controls on net ecosystem CO2 exchange for ca. 12 ecosystem types, including coniferous forest, deciduous forest, grassland, cropland (assumed dominated by grain crops), northern boreal mixed forest, tundra, wetlands, alpine forest, evergreen forest, evergreen shrubland, and Mediterranean oak woodland. The model uses a single layer canopy with two leaf classes (sun and shade) and a three layered rooting zone to estimate GPP, ecosystem respiration and overall ecosystem gas exchange. Maximum LAI is determined during each year from the MODIS LAI-product. In the case of grassland, crops, tundra and wetlands, classical growth routines estimate dynamic changes in the vegetation canopy. Soil/canopy coupling in response to drought reflects a root system hormonal signalling and is calibrated with observations at flux tower sites during the extremely dry year 2003. PIXGRO is applied over a European grid with 10 km resolution in MODIS sinusoidal projection, where 1 km2 land cover and soils information is used to define average LAI, most frequent soil type and weighting of flux rates in the spatial summary. Phenological influences are based on temperature climate (onset of growth and senescence) and soil water availability (herbaceous dieback and stomatal restrictions). Initial results demonstrate good agreement of the simulations with observations at flux tower sites and year to year variation in the spatial patterns of NEE. These spatial patterns are examined with respect to plausibility, the identification of key factors determining contributions to the overall European flux rates, and future research needs.
B53A-03
Upscaling carbon Budgets From Field to Region by Combining Flux Towers, Airborne Measurements and Remote Sensing - a Case Study From Zealand Denmark
A study on upscaling eddy correlation measurements of CO2 form field to region has been conducted over the island of Zealand, Denmark within an area of 8000 km2. The ground reference measurements are conducted within the framework of the Nordic Centre for Studies of Ecosystem Carbon Exchange and CarboEurope and comprises continuous eddy correlation measurements over four distinct surface type, beech forest, wheat fields, grassland and urban areas. The collection of CO2 flux data is supplemented by an airborne flux campaign conducted in August 2003. MODIS data on leaf area index and surface temperature are used together with standard meteorological data for running a simulation model. The airborn flux measurements which are conducted from an altitude of 150 m are compared to the ground level measurements by use of footprints model. The comparisons show that while the time integrated flux levels are in agreement the small scale variability in the CO2 flux is captured less accurately compared to the water vapour - and the sensible heat flux.. The soil respiration and the CO2 assimilation are then examined separately. For the soil respiration, it is found that soil temperature and soil moisture are of nearly equal significance for predicting the soil respiration. The spatial distribution of soil temperatures are derived from MODIS surface temperature data whereas the soil moisture in the upper soil column is estimated by use of a hydrological model operating in Geographical information system. With respect to carbon assimilation the spatial distribution is calculated on the basis of MODIS vegetation index product (EVI) with a spatial resolution of 250 m in combination with a mechanistic photosynthesis model using meteorological input data. The ground level flux distribution along the flight track is calculated as the sum of modelled carbon sequestration plus the soil respiration. Comparison between the airborn and the simulated CO2 fluxes is found to be in encouraging agreement especially when taking into account the development of the internal boundary layers compared to the flight level. Supplemented by national statistics and land use maps it is finally found that the island as a whole functions as carbon source with a large range in annual carbon budgets between the metropolitan areas of Copenhagen (+7900 gCm-2 yr-1) and the deciduous forests (-550 gCm-2 yr-1) and agriultural land (-300gCm-2 yr-1 for winter wheat).
B53A-04
Factors Affecting the Pattern of Vegetation Biomass and Canopy Height With Elevation at Hubbard Brook Experimental Forest
Understanding patterns of carbon stocks and fluxes on the land surface is important for studies of terrestrial ecology, the carbon cycle, and climate change and is an increasingly high priority for environmental policymakers. This need is especially relevant in areas of mountainous terrain, where methodological challenges limit the usefulness of atmospheric methods such as eddy covariance. At the Hubbard Brook Experimental Forest (White Mountains, New Hampshire), both field data and remote sensing data demonstrate that forests exhibit decreased height and biomass with elevation. In particular, aboveground biomass (AGB) values decline from an average of 280 mg/ha at 250 meters elevation to 145 mg/ha at 910 meters elevation. Correspondingly, average canopy height declines from 28 meters to 15 meters within the same elevational range. Although this trend is well documented by field and LiDar data, the relative influence of various causal factors has not been well established. Potential mechanisms include increased rates of disturbance and mortality, decreased rates of growth and changes in tree allometry. These factors may in turn be influenced by changes in water and nutrient availability, edaphic factors, and climate. This study examines the relative importance of these mechanisms through 2 objectives; statistical analysis of existing Hubbard Brook data and collection and analysis of additional field data. Our analysis of 1999 LiDar data indicates that differences in slope and aspect do not explain the AGB and height trend. Analysis of ground based measurements of tree diameters (DBH) and remote sensing measurements of tree height suggest that allometric changes are not responsible for the observed trends. To evaluate the remaining hypothesis of growth, mortality, and disturbance, we obtained and analyzed 371 previously collected tree cores. Using a stratified random sampling design based on LiDar data, 108 additional tree cores have been collected to better establish rates of growth and mortality along an elevational transect at Hubbard Brook. Analysis of these cores establishes the relative importance of growth rates, mortality, and disturbance on tree height and AGB at elevation. Results from this study will add more detail to the pattern of AGB and height decline with elevation at Hubbard Brook and improve understanding of carbon stocks and fluxes.
B53A-05
An Integrated Approach to a Complete Carbon Budget for the Delaware River Basin
We combined integrated measurement and monitoring of vegetation, soil, and water with process and empirical models to estimate carbon stocks and dynamics of the Delaware River Basin. Agencies operating in the basin accepted the challenge of designing an integrated monitoring strategy by augmenting existing monitoring systems. Watersheds are logical conceptual units for integrating environmental information on a regional scale, because aquatic systems integrate the biogeochemistry of large areas with well-defined boundaries. Scaling is also possible within watersheds because there is a hierarchical physiography, from small catchments to whole river basins. At the largest scale of the whole river basin, we used remote sensing and data from existing sample plot networks maintained by the USDA Forest Service and Natural Resources Conservation Service. We used estimators from the FORCARB-2 carbon accounting model to estimate basin-wide change in forest carbon stocks. At the smallest scale, we added land and water measurements to existing intensive monitoring sites maintained by the U.S. Geological Survey and the National Park Service, and estimated complete land and water carbon budgets. At an intermediate scale, we used remote sensing and intensive sampling of selected small watersheds that had contrasting conditions defined primarily by degree of residential development. At all scales, we used the PnET-CN ecosystem process model to estimate and map biomass and productivity, and the SPARROW empirical model to estimate carbon transport by water. Initial results show that the Basin is a small net carbon sink, although within the basin, southern areas are losing carbon while northern areas are gaining carbon primarily because of land-Use change. To extend this work for decision support, we are developing methods to query data and models from these different scales so that estimates can be made for any watershed within the river basin. We are also evaluating the watershed approach to see how well it may complement an airshed approach based on flux towers and ecosystem process models.
B53A-06
Carbon Dynamics in Mid-Atlantic Temperate Forests: Responses to Changes in Atmospheric Chemistry and Climate
This study examines how multiple stresses -- changing climate and atmospheric composition of CO2, O3 and N deposition -- affect productivity, carbon storage, and sequestration of Mid-Atlantic temperate forests. We use a process-based ecosystem model, PnET-CN, that has a strong foundation of ecosystem process knowledge from experimental studies. Our results suggest that the chronic changes in atmospheric chemistry in the past decades markedly affect carbon dynamics and sequestration in Mid-Atlantic temperate forests. At the regional scale, net primary production has increased by 28% in response to the three major atmospheric chemical changes. In the last 70 years, carbon sequestered in live forest biomass increased by 20%, while the carbon increase in soil organic matter was 18%. More fundamentally, changes in atmospheric chemistry components exert impacts on ecophysiological processes and ecosystem functioning. The modeling results suggest that N deposition is a stronger force than elevated CO2 for increasing primary production. Ozone pollution offsets about 22% of enhanced biochemical capacity for photosynthesis. Changes in interannual variability of climate during past decades add complexity to forest responses to changing atmospheric chemistry. As annual NPP increases along with increased precipitation and temperature in the region, annual NEP appears to decline, likely caused by accelerated decomposition processes. Because of the complexity of interactions among multiple stresses, and the limitations of the experimental approach, the process-based mechanistic model is shown as a powerful tool in this study to predict ecosystem-level responses and attribute causation to various climatic drivers and different environmental factors.
B53A-07
Dissolved Organic Carbon Export From the Penobscot River Basin to the Gulf of Maine
Recent and historical water quality data collected from the Penobscot River in central Maine were analyzed to estimate the flux of dissolved organic carbon (DOC). Historical data collected by the U.S. Geologic Survey during 1973 to 1981 as part of the National Stream Quality Accounting Network (NASQAN) and recently collected data during 2004 and 2005 supported by NASA provide a framework for preliminary estimates of seasonal and interannual variability in DOC export. Monthly average concentrations were computed from samples collected during all seasons. Flux estimates were calculated using monthly average concentrations and average daily discharge. For snowmelt and storms measured during 2004 to 2005 DOC increased with increasing discharge, but the increase was distinctly muted during the snowmelt-dominated high spring flow compared with summer and fall storms. DOC export was highest during the spring snowmelt season in spite of relatively low concentrations, compared with late summer and fall because of the substantially higher spring discharge. DOC export tends to be lowest during July through August and increases progressively to moderately high levels during the fall. Export of DOC during the winter months is intermediate between summer and fall levels. Estimated DOC export varied substantially among years ranging from 4.0 t/sq km/yr during the driest year to 7.8 t/sq km/yr during the wettest year. Average annual DOC export per unit area for all years was 5.7 t/sq km/yr. Under wetter conditions flowpaths are thought to include more organic matter-rich surface soils. DOC export from the Penobscot River is substantially higher than estimated for most larger rivers in North America and smaller rivers draining to the Atlantic Ocean along the east coast of the United States. Preliminary evidence from analysis of DOC concentrations in tributaries to the Penobscot River indicates that wetland area and reservoir volume exert substantial control over DOC export in this basin
B53A-08
Atmosphere-Biosphere Carbon Exchange in New England and Quebec During Summer 2004 From a Receptor Oriented Modeling Framework
We quantify atmosphere-biosphere carbon exchange for spring and summer 2004 in greater New England and Quebec. Our approach is a receptor-oriented modeling framework, consisting of a time-reversed Lagrangian adjoint model (STILT) [Gerbig et al. 2003a,b; Lin et al. 2003] coupled to a vegetation CO2 flux model that calculates gross primary production and respiration by partitioning photosynthetic and non-photosynthetic respiration, the Vegetation Photosynthesis and Respiration Model (VPRM) [Pathmathevan et al., 2005; Xiao et al., 2004]. To drive atmospheric transport, the adjoint transport model utilizes wind fields from the Colorado State Regional Atmospheric Modeling System (RAMS), the Eta Data Assimilation System 40-km product (EDAS-40), or the Weather Research and Forecasting (WRF) model. The biosphere model incorporates MODIS-derived enhanced vegetation index (EVI) and land surface water index (LSWI) together with GOES-derived shortwave radiation [Diak et al. 2004] to capture surface spatial heterogeneity and variations in soil moisture, canopy nutrition, solar input and phenology. To integrate functional dependence of CO2 flux, we optimized parameters within VPRM using Ameriflux eddy covariance data. These observations were separated into eight vegetation types, based on GLCC 2.0 land cover classes. These initial biosphere parameters serve as a priori values in a Bayesian optimization method, where the adjoint atmospheric model links the atmospheric measurement at a receptor point to the appropriate footprint at high spatiotemporal resolution within the domain of the biosphere. We focused a priori model runs on receptor points at Harvard Forest and the NOAA-CMDL Argyle tall tower in Maine. These a priori results demonstrate remarkably close agreement with in-situ CO2 measurements, suggesting the VPRM can be scaled up to determine a priori regional carbon exchange for New England and Quebec in summer 2004. We scale up a posteriori results to produce hourly and monthly average regional carbon exchange estimates optimized in New England and Quebec, with corresponding estimates of uncertainty.