Amazonian forests play an important and complex role in the global carbon cycle, contributing substantially to increases (via land use change emissions) and possibly to net sequestration (in intact forests) of atmospheric CO₂. Predicting these processes of net carbon uptake and release depends crucially on understanding ecosystem response to both seasonal and interannual variations. However, prominent ecosystem modeling studies of the Amazonian carbon cycle [Tian et al., 1998; Botta 2002] appear to make seasonal predictions (wet-season carbon uptake and dry-season loss) at odds with both some site-specific observations (which show the opposite pattern, Saleska et al., [2003]) and basin-wide satellite observations (which imply large-scale increases in the activity of photosynthetic vegetation during the dry season, Huete et al., [2005]).

Three techniques for separation of total CO₂ efflux from soil into root and microbial respiration were compared: component integration, root exclusion and pulse labelling of shoots in ¹⁴CO₂ atmosphere. The contribution of rhizosphere to total CO₂ efflux from soil varied from 19 to 49% (including root respiration amounted to 9-32%). The share of non-rhizosphere respiration was 51-80%. The results obtained by component integration and root exclusion techniques were similar. Rhizosphere respiration estimated by pulse labelling were less as estimated by two non-isotopic methods.

The Ags parameterization of canopy conductance from ISBA-Ags is implemented in TESSEL, the ECMWF land surface scheme. We present first results of the investigation of the model behavior in view of an operational use in a data assimilation system. It is shown that the performance of the Ags module is sensitive to the land surface model in which it is embedded.

The global rate of fossil fuel combustion continues to rise, but the amount of CO₂ accumulating in the atmosphere has not increased accordingly (Tans et al., 1990; Conway et al., 1994; Wofsy, 2001).  The causes for this discrepancy are widely debated (Houghton, 2003).  In particular, the location and drivers for the interannual variability of atmospheric CO₂ are highly uncertain.  Here we examine links between global atmospheric CO₂ growth rate (CGR) and the climate anomalies of biomes based on ten years (1986-1995) of global climate data and accompanying satellite data sets.  Our results show that four biomes, the tropical rainforest, tropical savanna, C₄ grassland and boreal forest, and their responses to climate anomalies, are the major climate-sensitive CO₂ sinks/sources that control the CGR.  The nature and magnitude by which these biomes respond to climate anomalies are generally not the same.  However, one common influence did emerge from our analysis; the extremely high CGR that was observed for the one extreme El Niño year was caused by the response of the tropical biomes (rainforest, savanna and C₄ grassland) to temperature.

The overarching goal of a long-term, multi-investigator, regional study of ecosystem-atmosphere carbon cycling in a mixed forest ecosystem in the upper Midwest of the USA is to observe ecosystem-atmosphere exchange of carbon dioxide at scales of relevance to the global carbon balance, while simultaneously understanding the mechanisms governing this exchange. This study, the Chequamegon Ecosystem-Atmosphere Study (ChEAS), brings together multiple approaches to observing carbon fluxes, including chamber flux, sap flux and biometric measurements at the plot scale (~1 m²), multiple stand-level (~1 km²) eddy-covariance flux towers, landscape-scale (~10-100 km²) eddy-covariance flux measurements from the WLEF tall tower, multiple regional (10³-10⁵ km²) atmospheric boundary layer (ABL) budget approaches using tall tower mixing ratio measurements, and a regional (~10⁵ km²) ABL budget using a network of CO₂ mixing ratio measurements on communications towers. Flux measurements have been up-scaled to the region using a variety of approaches, and compared to the regional ABL budget methods. Top-down and bottom-up methods fall within a range of values for growing-season flux estimates that suggests a level of precision for regional flux estimates of approximately 0.5 gC m^-2 d^-1. A multi-tower inverse study should increase the level of precision of the ABL budget flux estimates. Interpreting the mechanisms governing these fluxes requires plot- and stand-level data. These data show that variability in seasonal and annual fluxes among flux towers is large, refuting hypotheses that ecosystem-atmosphere exchange can be explained simply by climate, or that a sparse flux tower network can be used to map carbon fluxes over continental domains. Stand age and stand type (e.g. aspen, wetland, northern hardwood forest) explain a large fraction but not all of the observed variability among stands. More sophisticated land classification schemes may be needed to improve the precision of bottom-up methods. Multi-year records are used to examine interannual variability in the carbon balance of the region and show that interannual variability at WLEF is clearly correlated with climate variability. Limited multi-year records at the plot- and stand-level partly support the hypothesis that year-to-year variability in carbon fluxes are coherent across the region, and begin to describe the causes of the observed interannual variability. Further study is needed to evaluate the network design required to describe both the magnitude and mechanisms of interannual variability in the regional carbon balance.

Evaluation of the difference between annual net CO₂ ecosystem exchanges (NEE) from the open- and the closed-path data is important for site intercomparison studies. However, long-term measurements of NEE using both systems have been limited. We report the comparison of eddy CO₂ fluxes measured with open- and closed path systems for three years from 2001 through 2003. The annual GPP estimated from closed-path data was 8–10% less negative than that from the open-path data, whereas the annual RE was 11–16% more positive for closed-path data. Consequently, the annual NEE from the closed-path data was less negative by 301–333 gC m^-2 y^-1. The bias of NEE between two systems is large and an extremely important issue. Ecophysiological approaches are needed to validate of the eddy covariance technique.

Progress in determining CO₂ sources, sinks, and their response to environmental forcing will rely on utilization of more extensive and intensive CO₂ and related observations including those from satellite remote sensing.  Full exploitation of new observations will require new modeling and analysis techniques, especially those that can use information at finer spatial and temporal scales than has traditionally been employed in “top-down” carbon flux studies.  We report on a modeling effort to reduce uncertainty in carbon cycle processes that create the so-called missing terrestrial sink of atmospheric CO₂ using transport fields derived from NASA’s GEOS-4 meteorological assimilation analyses.  Our overall objective is to improve characterization of CO₂ source/sink processes globally with improved formulations for atmospheric transport, terrestrial uptake and release, biomass and fossil fuel burning, and observational data analysis.  We show results from an advanced biosphere model (SiB3) constrained by remote sensing data and coupled to the global transport model to produce distributions of CO₂ fluxes and concentrations that are consistent with actual meteorological variability.  Use of analyzed meteorological data allows comparison to observations on a wide range of temporal and spatial scales.  Here we compare with local-to-global data for hourly to annual CO₂ simulation.  The results will help to prepare for the use of satellite CO₂ and other data in a multi-disciplinary carbon data assimilation system for analysis and prediction of carbon cycle changes and carbon/climate interactions.

The CO₂ flux from coarse woody debris (R_CWD) in a deciduous broad-leaved forest was measured using chamber measurements. The relationships between R_CWD and environmental factors, such as temperature (T) and the water content (θ) of the coarse woody debris (CWD), were determined from long-term continuous measurements. Measurements of the R_CWD of many CWD samples revealed relationships between R_CWD and CWD characteristics, such as wood density (ρ) and diameter (D). A field survey conducted in 2003 estimated the mass of CWD as 9.30tC·ha^-1, with snags amounting to 60% of the total CWD mass. Scaling R_CWD to the ecosystem while considering environmental factors according to the type (snag or log) of CWD and CWD characteristics, we estimated that the annual R_CWD in the forest was 0.50tC·ha^-1·y^-1 in 2003. This came to 13-19% of the total heterotrophic respiration in the forest. The mean annual CWD input mass from 2000 to 2004 was 0.61tC·ha^-1·y^-1. Therefore, 0.11tC·ha^-1·y^-1 were sequestered by CWD, which amounted to 7% of the net ecosystem production (NEP) in the forest. In a younger forest, it is difficult to assume that the CWD input and decomposition are balanced, so the R_CWD and CWD input mass should be quantified to evaluate the forest carbon cycle and NEP.

Vegetation strongly influences the spatial distribution of sensible and latent heat fluxes, and also controls ecosystem-atmosphere CO₂ exchange. We describe here a methodology to estimate surface energy fluxes and Net Ecosystem Exchange (NEE) of CO₂ continuously over the Southern Great Plains, using (1) data from the U.S. Department of Energy Atmospheric Radiation Measurement (ARM) program in Oklahoma and Kansas; (2) meteorological forcing data from Mesonet facilities; (3) U.S. Geological Survey (USGS) soil database; (4) MODIS NDVI at 250 meters resolution; and (5) a tested carbon and isotope land-surface model (ISOLSM, based on LSM1.0 [Bonan 1996]). The need for distributed ecosystem modeling was demonstrated by the large spatial variability in CO₂ fluxes across the region, which is typically modeled as homogeneous cropland. This work addresses U.S. national goals of estimating regional CO2 sources and sinks, and provides inputs to forward and inverse models.

Precise CO₂ concentration measurements at marine stations and tall towers are crucial for quantifying global trends in atmospheric CO₂ concentrations. We propose that measurements in the continental planetary boundary layer—the poor cousin of the clean background stations—can be used to understand trends in, and controls, of atmospheric CO₂ concentrations at local and regional scales as well as global scales. The key is choosing appropriate time scales of integration for the data. In the US Southern Great Plains, we are measuring precise CO₂ concentrations continuously at 2–60 m and weekly at 300 and 3300 m above ground level (agl). CO₂ flux is measured in individual crop fields and pastures (4 m towers) and at 60 m. The precise CO₂ concentrations show strong continental influence in both diurnal and seasonal cycles. In continental regions, atmospheric CO₂ profiles are strongly influenced by atmospheric dynamics as well as ecosystem and anthropogenic fluxes. Relating site level measurements or atmospheric profiles to regional CO₂ budgets requires methods to represent or evaluate these influences. We observe inter-annual differences in the climatology of diurnal cycles (seasonal average diurnal cycles). Using the several years’ data for boundary layer concentrations, the annual trend in CO₂ growth nearly matches the value estimated by National Oceanic and Atmospheric Administration (NOAA) Climate Monitoring Diagnostic Laboratory for our latitude band.