A12B-01 INVITED
Ground-based Observations of the Total CO2 Column: Method, Validation, and First Results
We describe the development of a ground-based network for measurement of the total CO2 column (Total Carbon Column Observing Network - TCCON). These data are derived from analysis of near-IR solar spectra obtained with high spectral resolution using FTIR spectrometers. Observations are currently made at seven globally-distributed sites, and planned at several others. They have been evaluated against aircraft in situ profiles in Park Falls, Wisconsin, Darwin, Australia, and Los Angeles, California. We show how these observations constrain the strength of exchange of carbon between the surface and the atmosphere. Implications for current and future CO2 observations from space are described. http://www.tccon.caltech.edu
A12B-02
The impact of large-scale dynamics on the variability of total column CO2
Ground-based measurements of total column CO2 have been made at Park Falls, Wisconsin beginning in 2004. Solar spectra are obtained under clear sky conditions using a high-resolution Fourier Transform Spectrometer (FTS). These spectra are analyzed using a line-by-line retrieval to obtain precise and accurate measurements of the CO2 and other greenhouse gas columns. During summertime, significant variability in the total column CO2 is observed. This variability reflects the large footprint of column observations and may facilitate estimation of continental-scale CO2 exchange between the atmosphere and the terrestrial biosphere from a single ground site. We use three approaches to understand the CO2 column variability at Park Falls. We use STILT to determine influence functions for the CO2 at Park Falls. We analyze these regions of influence to determine the correlation between the origin of CO2 at the surface and daily anomalies observed in the total column. Secondly, we use our measurements in conjunction with CarbonTracker reanalyzed CO2 fields to explore the relationship between large-scale dynamic activity and total column CO2 both at Park Falls and in neighboring regions. Finally, we use the GFDL AM2 model to explore the sensitivity of simulated CO2 synoptic variability to various convective parameterizations methods.
A12B-03
Large-Scale Variability of Middle and Upper Tropospheric CO2
It is generally assumed that upper tropospheric CO2 exhibits little spatial variability. We present clear evidence to the contrary. Our analysis centers on CO2 mixing ratios for selected months of 2003 from AIRS retrievals using the Vanishing Partial Derivative (VPD) method. The AIRS middle to upper tropospheric (~500-300 hPa) data CO2 show unequivocally large spatial and temporal variability in the northern midlatitudes in summer. For example, there is a ~3 ppmv longitudinal gradient in the concentrations across North America in July. Aircraft in situ observations from CO2 the COBRA and INTEX-NA campaigns support such large variability. Compared with the AIRS data and aircraft observations, results from CO2 simulations using current global three-dimensional chemistry and transport models (CTMs) including the GEOS- Chem and MOZART-2 models tend to show much dampened variability but with consistent distributions. In addition to the surface sources/uptakes, the distributions of AIRS CO2 are modulated by large-scale circulation such as the midlatitude jet streams in both hemispheres and by synoptic weather systems including warm conveyor belts (WCBs).
A12B-04
AIRS CO2 in the Upper Troposphere
Atmospheric Infrared Sounder (AIRS) retrieved CO2 and O3 by the Vanishing Partial Derivative (VPD) method [Chahine et al., 2005] have been validated with aircraft and ozonesonde data in April 2003. The latitudinal distributions of AIRS CO2 and O3 are consistent with in situ observations as well as results from two- and three- dimensional (2-D and 3-D) chemistry and transport models (CTMs). Following a stratospheric major final warming event, AIRS CO2 increases while O3 decreases in the polar upper troposphere. The seasonal cycle of AIRS CO2 matches that from aircraft data. Model results have smaller seasonal cycle amplitudes in the southern hemisphere compared with those in the northern hemisphere, which are consistent with the aircraft data. Some discrepancies are evident between the model and aircraft data in the mid-latitudes, where models tend to underestimate the amplitude of CO2 seasonal cycle. Sensitivity studies reveal that the convection mass flux is very crucial for the accurate simulation of upper tropospheric CO2. We are also exploring the use of AIRS CO2 to constrain the vertical transport in 3-D CTMs.
A12B-05
4D-Var Atmospheric Data Assimilation and Flux Inversion Using Current and Future CO2 In- Situ and Satellite Observations
Under the umbrella of European Global Monitoring for Environment and Security (GMES) the Global and regional Earth-system (Atmosphere) Monitoring using Satellite and in-situ data (GEMS) project has been running since March 2005. An important part of the GEMS project is the building of a monitoring system for atmospheric greenhouse gases and their surface fluxes. The system consists of a 4-dimensional variational (4D-Var) atmospheric data assimilation system embedded in the ECMWF operational weather data assimilation system, a 4D-Var flux inversion system that uses the output of the atmosphere system, and relevant validation efforts. In this presentation we will present first results for CO2 using data from the Atmospheric Infrared Sounder (AIRS) for 2003. Atmospheric fields will be shown with a particular emphasis on validation with indepedent data. Strengths and weaknesses of the system will be discussed. We will also present early flux inversion results. Finally, expectations for the near future using data from IASI, OCO, and GOSAT will be discussed to illustrate the full potential of the system.
A12B-06
The Constraint on Surface CO2 Fluxes Provided by OCO Column CO2 Measurements: a Realistic Assessment
The current network of in situ CO2 monitoring sites lacks the spatial coverage to provide regional estimates of the sources and sinks of CO2 over most of the globe. NASA's soon-to-be-launched Orbital Carbon Observatory (OCO), a sun synchronous satellite specially designed to be sensitive to near-surface CO2 concentrations, should soon furnish such coverage. Here we use a variational data assimilation method in a simulation framework to estimate how well OCO's column-CO2 measurements can constrain the surface sources and sinks. We start with new estimates of the measurement uncertainties for a single column measurement, given by an analysis of the instrument's inversion procedure as a function of surface type, aerosol optical depth, and solar zenith angle . Next, we examine cloud and aerosol statistics derived from MODIS and other satellites to determine the likely correlations in measurement error (including modelling errors) along the OCO trajectory. These are used to obtained the effective number of independent measurements inside each 2°x5° atmospheric model grid box that the satellite passes over, from which effective multi-shot measurement errors are calculated. These are then used in the simulation study to determine how much of the error in the prior flux estimate can be removed by assimilating the OCO data. The merits of OCO's nadir- and glint-viewing modes are assessed. Spatial discretization errors due to the atmospheric model are quantified and considered. Our analysis highlights the importance of aerosols in the problem: not only do thick aerosol optical depths preclude useful retrievals, but aerosols at lower levels will bias the column measurements; we simulate the impact of such aerosol levels globally to quantify their importance. This study is the first to realistically consider the impact of clouds and aerosols in the problem, and hopefully will help the mission operations personnel decide when to operate in nadir versus glint mode, and which retrievals to process first.
A12B-07
Assessing sampling and representation errors in assimilation of satellite CO2 retrievals
Retrieval of the mean dry mole faction of atmospheric CO2 (XCO2) from spectra measured by dedicated spaceborne instruments will be made a few times each month over a given location and will only represent clear atmosphere columns. Variations in atmospheric CO2 on synoptic time scales may lead to temporal sampling errors in transport inversions, especially if they covary strongly with cloud cover. Note that these are modeling errors, and that they arise if the model used for the inversion is sampled differently than the real atmosphere is sampled by the satellite. To assess the potential magnitude of these errors, we investigated sources of synoptic variability and its relationship to clouds using continuous tower observations, a coupled cloud-resolving model, and a global chemical transport model. An observational assessment of systematic differences between mid-day CO2 on clearsky vs all days used multiyear timeseries of continuous data from the WLEF-TV tower and Harvard Forest in the USA, and Flona Tapajos, Brazil. The WLEF site is a temperate forest in a remote rural area, and Harvard Forest is located in a region with strong anthropogenic emissions. Flona Tapajos is a very remote equatorial forest reserve. We found systematic differences of 1 to 3 ppm in measured mid-day CO2 , with lower values on sunny days than average. At the temperate sites, the differences are greatest in winter and are not attributable to anomalous surface fluxes. The tropical site showed the biggest differences in the rainy season, with very little sampling error in the dry season. We performed cloud-resolving simulations of two cases using the SiB-RAMS coupled ecosystem-atmosphere model, one during summer at the temperate WLEF site and one during the dry season at the tropical Tapajos site. Simulated XCO2 was sampled on 1-km-wide swaths in clear columns only as if nadir sampling from space, and compared to a (100-km) mean which represents the grid column of a global CTM such as might be used in a transport inversion. In both cases, this "local" clearsky sampling error was found to be much smaller than the likely instrument retrieval error. At the temperate site, temporal sampling errors in representing time means from individual swaths were comparable to likely retrieval errors because of systematic XCO2 anomalies associated with fronts that were masked by clouds. In the tropics, these temporal sampling errors were much smaller. The magnitude and seasonal variation of clearsky sampling errors at the three towers were reproduced well using the global PCTM. Global simulations of monthly mean XCO2 were compared to the means of samples taken only under clearsky conditions to assess spatial and seasonal patterns of these errors. Clearsky temporal sampling errors were found to be greatest over land, and were dominated by positive errors over the midlatitudes (especially Asia) in summer and negative errors over the tropics (especially South America) during the rainy season. These errors often exceeded 1 ppm, and must be addressed in a data assimilation system by correct simulation of synoptic variations in XCO2 associated with cloud systems. http://biocycle.atmos.colostate.edu
A12B-08
Ground-based FTIR Measurements of Carbon Dioxide Isotopic Concentrations and Fluxes and Ecosystem-Atmosphere Exchange
Analysis of the isotopic content of atmospheric carbon dioxide provides a wealth of information about the complex interaction between the biosphere and the atmosphere. When combined with micrometeorological flux measurements, a much improved picture emerges of ecosystem-scale carbon cycling dynamics. Since photosynthesis and respiration impart a unique isotopic signature to the atmosphere, knowledge of the concentrations and fluxes of the individual CO2 isotopologues can be used to constrain and partition net ecosystem exchange into its gross photosynthetic and respiration components. In this work, concentrations and fluxes of individual isotopologues of carbon dioxide in a forest ecosystem were determined using Fourier Transform Infrared (FTIR) spectroscopy and the Disjunct Eddy Covariance (DEC) flux measurement technique. We separately measure the concentrations of the 12CO2 and 13CO2 isotopomers of carbon dioxide at approximately one minute intervals with very high signal-to-noise ratio using molecular absorption in a 1-meter cell in the 2100 to 2600 cm-1 region of the isotopic vibration-rotation bands. The FTIR – DEC system was deployed during four field campaigns in the summers of 2005 and 2006 in north central Oregon. Fixed height or vertical profile concentration measurements of 12CO2 and 13CO2 were carried out in the canopy using a 25-meter tower fitted with air inlets at 0.3, 4.1, 7.5, 10.8, 14.0, and 20.6 meters. Isotopic flux measurements above the canopy were taken at constant height of 20.6 meters. Temporal and vertical profiles of CO2 concentrations and carbon isotope ratios will be shown together with the total net ecosystem exchange and isotopic CO2 fluxes.