AMTAtmospheric Measurement TechniquesAMTAtmos. Meas. Tech.1867-8548Copernicus PublicationsGöttingen, Germany10.5194/amt-9-683-2016Consistent evaluation of ACOS-GOSAT, BESD-SCIAMACHY, CarbonTracker, and MACC
through comparisons to TCCONKulawikSusansusan.s.kulawik@nasa.govWunchDebrahttps://orcid.org/0000-0002-4924-0377O'DellChristopherFrankenbergChristianhttps://orcid.org/0000-0002-0546-5857ReuterMaximilianhttps://orcid.org/0000-0001-9141-3895OdaTomohiroChevallierFrederichttps://orcid.org/0000-0002-4327-3813SherlockVanessaBuchwitzMichaelhttps://orcid.org/0000-0001-7616-1837OstermanGregMillerCharles E.WennbergPaul O.https://orcid.org/0000-0002-6126-3854GriffithDavidhttps://orcid.org/0000-0002-7986-1924MorinoIsamuhttps://orcid.org/0000-0003-2720-1569DubeyManvendra K.https://orcid.org/0000-0002-3492-790XDeutscherNicholas M.NotholtJustusHaseFrankWarnekeThorstenSussmannRalfRobinsonJohnStrongKimberlyhttps://orcid.org/0000-0001-9947-1053SchneiderMatthiashttps://orcid.org/0000-0001-8452-0035De MazièreMartineShiomiKeiFeistDietrich G.https://orcid.org/0000-0002-5890-6687IraciLaura T.WolfJoyceBay Area Environm. Res. Inst., Sonoma, CA 95476, USACalifornia Institute of Technology, Pasadena, CA, USAColorado State University, Fort Collins, CO, USAJet Propulsion Laboratory, California Institute of Technology, Pasadena,
CA, USAUniversity of Bremen, Institute of Environmental Physics, Bremen,
GermanyGoddard Earth Sciences Technology and Research, Universities Space
Research Association, Columbia, MD, USAGlobal Modeling and Assimilation Office, NASA Goddard Space Flight
Center, Greenbelt, MD, USALaboratoire de Meteorologie Dynamique, Palaiseau, FranceThe National Institute of Water and Atmospheric Research, Wellington and
Lauder, New ZealandUniversity of Wollongong, Wollongong, New South Wales, AustraliaCenter for Global Environmental Research, National Institute for
Environmental Studies (NIES), Tsukuba, Ibaraki, JapanEarth and Environmental Science, Los Alamos National Laboratory, Los
Alamos, NM 87545, USAKarlsruhe Institute of Technology, Institute for Meteorology and
Climate Research (IMK-ASF), Karlsruhe, GermanyDepartment of Physics, University of Toronto, Toronto, Ontario, CanadaRoyal Belgian Institute for Space Aeronomy, Brussels, BelgiumEarth Observation Research Center, Japan Aerospace Exploration Agency,
Tsukuba, Ibaraki, JapanMax Planck Institute for Biogeochemistry, Jena, GermanyNASA Ames Research Center, Atmospheric Science Branch, Moffett Field,
CA, USAretiredSusan Kulawik (susan.s.kulawik@nasa.gov)29February20169268370919May201522June201524January201629January2016This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/This article is available from https://amt.copernicus.org/articles/9/683/2016/amt-9-683-2016.htmlThe full text article is available as a PDF file from https://amt.copernicus.org/articles/9/683/2016/amt-9-683-2016.pdf
Consistent validation of satellite CO2 estimates is a prerequisite for
using multiple satellite CO2 measurements for joint flux inversion, and
for establishing an accurate long-term atmospheric CO2 data record.
Harmonizing satellite CO2 measurements is particularly important since
the differences in instruments, observing geometries, sampling strategies,
etc. imbue different measurement characteristics in the various satellite
CO2 data products. We focus on validating model and satellite
observation attributes that impact flux estimates and CO2 assimilation,
including accurate error estimates, correlated and random errors, overall
biases, biases by season and latitude, the impact of coincidence criteria,
validation of seasonal cycle phase and amplitude, yearly growth, and daily
variability. We evaluate dry-air mole fraction (XCO2) for Greenhouse gases Observing SATellite (GOSAT) (Atmospheric
CO2 Observations from Space, ACOS b3.5) and SCanning Imaging Absorption spectroMeter
for Atmospheric CHartographY (SCIAMACHY) (Bremen Optimal Estimation DOAS, BESD v2.00.08) as well as the CarbonTracker (CT2013b)
simulated CO2 mole fraction fields and the Monitoring Atmospheric Composition and Climate (MACC) CO2 inversion
system (v13.1) and compare these to Total Carbon Column Observing Network (TCCON) observations (GGG2012/2014). We
find standard deviations of 0.9, 0.9, 1.7, and 2.1 ppm vs. TCCON for
CT2013b, MACC, GOSAT, and SCIAMACHY, respectively, with the single
observation errors 1.9 and 0.9 times the predicted errors for GOSAT and
SCIAMACHY, respectively. We quantify how satellite error drops with data
averaging by interpreting according to error2=a2+b2/n
(with n being the number of observations averaged, a the systematic (correlated)
errors, and b the random (uncorrelated) errors). a and b are estimated by
satellites, coincidence criteria, and hemisphere. Biases at individual
stations have year-to-year variability of ∼ 0.3 ppm, with
biases larger than the TCCON-predicted bias uncertainty of 0.4 ppm at many
stations. We find that GOSAT and CT2013b underpredict the seasonal cycle
amplitude in the Northern Hemisphere (NH) between 46 and 53∘ N, MACC overpredicts between
26 and 37∘ N, and CT2013b underpredicts the seasonal cycle amplitude in the
Southern Hemisphere (SH). The seasonal cycle phase indicates whether a data set or
model lags another data set in time. We find that the GOSAT measurements
improve the seasonal cycle phase substantially over the prior while
SCIAMACHY measurements improve the phase significantly for just two of seven
sites. The models reproduce the measured seasonal cycle phase well except
for at Lauder_125HR (CT2013b) and Darwin (MACC). We compare the variability
within 1 day between TCCON and models in JJA; there is correlation between
0.2 and 0.8 in the NH, with models showing 10–50 % the variability of
TCCON at different stations and CT2013b showing more variability than MACC.
This paper highlights findings that provide inputs to estimate flux errors
in model assimilations, and places where models and satellites need further
investigation, e.g., the SH for models and 45–67∘ N for GOSAT and CT2013b.
Introduction
Carbon–climate feedbacks are a major uncertainty in predicting the climate
response to anthropogenic forcing (Friedlingstein et al., 2006). Currently,
about 10 Gigatons (Gt) of carbon are emitted per year from human activity
(e.g., fossil fuel burning, deforestation), of which about 5 Gt stays in the
atmosphere, causing an annual CO2 increase of approximately 2 ppm yr-1.
The yearly increase is quite variable, estimated at 1.99 ± 0.43 ppm yr-1
(http://www.esrl.noaa.gov/gmd/ccgg/trends/global.html), however always
positive (Houghton et al., 2007). The remaining 5 Gt of carbon is taken up
by the ocean and the terrestrial biosphere; however, there are uncertainties
in the location and mechanism of these sinks, e.g., the distribution of land
sinks between the Northern Hemisphere and the tropics (e.g., Stephens et al.,
2007), and the localization of sources and sinks on regional scales
(Canadell et al., 2011; Baker et al., 2006). The uncertainties in top-down
source and sink estimates are a consequence of uncertainties in model
transport and dynamics (e.g., Prather et al., 2008; Stephens et al., 2007)
and sparseness of available surface-based CO2 observations
(Hungershoefer et al., 2010; Chevallier et al., 2010). Satellites offer a
much denser and spatially contiguous data set for top-down estimates, but are
much more susceptible to biases as compared to ground-based measurements
(e.g., see summary in Sect. 3.3.2 of Ciais et al., 2014).
This paper tests different characteristics of model and satellite CO2
(e.g., seasonal cycle amplitude and phase, regional and seasonal biases,
effects of averaging, and diurnal variations) through a series of
specialized comparisons to the Total Carbon Column Observing Network
(TCCON). The findings from this work can be propagated into assimilation
systems to determine the influence of various findings on top-down flux
estimates (e.g., see Miller et al., 2007; Deng et al., 2014; Chevallier and
O'Dell, 2013; Chevallier et al., 2014). For example, this paper
characterizes biases by latitude and season; these biases can be assimilated
to determine their effects on flux estimates (e.g., Kulawik et al., 2013).
This paper also shows a set of comparisons and tests that may be useful for
evaluating bottom-up flux estimates or transport schemes in models.
The remainder of the paper is organized as follows: Sect. 2 describes
the satellite XCO2 data and models used. The various satellite and model
XCO2 data are compared against TCCON observations in Sect. 3.
Temporal characteristics of the different XCO2 data and models,
including the seasonal cycle amplitude and phase, are compared in Sect. 4.
Diurnal behavior is evaluated in Sect. 5. Section 6 summarizes our
findings and discusses their impact on evaluating terrestrial carbon fluxes.
Data and models used
The characteristics of the sets of carbon dioxide that will be compared to
TCCON are summarized in Table 1. The following sections contain detailed
descriptions of the data set versions and characteristics.
The TCCON
The TCCON consists of ground-based Fourier transform spectrometers (FTSs)
that measure high spectral (0.02 cm-1) and temporal (∼ 90 s) resolution spectra of the direct
sun in the near infrared spectrum (Wunch et al., 2011a). Column abundances of CO2, O2, and other atmospheric gases
are determined from their absorption signatures in the solar spectra using
the GGG software package, which employs a nonlinear least squares spectral
fitting algorithm to scale an a priori volume mixing ratio profile.
Absorption of CO2 is measured in the weak CO2 band centered on
6220 and 6339 cm-1, and of O2 in the band centered on
7885 cm-1.
Summary of the CO2 data sets and models we used, showing
the coverage for several different CO2 products. The obs/day are the
approximate number of CO2 observations which passed quality screening.
TCCON site locations used for this work. The color indicates the
year when each station started collecting data.
The total column dry-air mole fractions of CO2 (XCO2) are computed
by ratioing the column abundances of CO2 and O2. The resulting
dry-air mole fractions have been calibrated against profiles of CO2
measured by WMO-scale instrumentation aboard aircraft (Wunch et al., 2010;
Messerschmidt et al., 2011). The precision and accuracy of the TCCON
XCO2 product is ∼ 0.8 ppm (2σ) after calibration
(Wunch et al., 2010). The TCCON data used in this paper are from the GGG2012
and GGG2014 releases, available from http://tccon.ornl.gov/. In this
paper we use 90 min average TCCON values which have 1σ precision of
0.4 ppm (see Appendix B).
We use 20 TCCON stations, distributed globally (see Fig. 1), and these data
have been used extensively for satellite validation and bias correction
(e.g., Butz et al., 2011; Morino et al., 2011; Wunch et al., 2011b; Reuter et
al., 2011; Schneising et al., 2012; Oshchepkov et al.; 2012), in flux
inversions (Chevallier et al., 2011), and in model comparisons (Basu et al., 2011; Saito et al., 2012). We use the GGG2014 data when available, and the
GGG2012 data from sites Four Corners and Tsukuba_120HR. Note
that the overall bias between the GGG2012 and GGG2014 XCO2 is 0.3 ppm
(with the GGG2014 results smaller) (Wunch et al., 2015). The GGG2012 sites
have corrections based on the instructions from the TCCON partners, listed
on the TCCON website (https://tccon-wiki.caltech.edu/Network_Policy/Data_Use_Policy/Data_Description_GGG2012#Laser_Sampling_Errors). We also apply a 0.9972 factor to Four
Corners, as indicated here: https://tccon-wiki.caltech.edu/Network_Policy/Data_Use_Policy/Data_Description_GGG2012. Two instruments have been operated at
the Lauder site. We identify them using 120HR (for the period of 20 June 2004 through
28 February 2011) and 125HR (for 2 February 2010 through to the present) when
results are instrument-specific. We find biases in the
Tsukuba_125HR for the time ranges in this paper (possibly
from high humidity combined with solar tracker issues during this time
period that were later corrected, and/or bias updates that are needed) and
do not use Tsukuba_125HR in this paper.
Stations which have special circumstances regarding validation and are
considered locally influenced are Garmisch which is in the midst of
complicated terrain, for which local atmospheric transport is difficult to model
and to measure from space; Four Corners, which is located in the vicinity of
two power plants with large CO2 emissions (Lindenmaier et al., 2014).
The meteorology is such that Four Corners regularly samples large localized
plumes with column CO2 increases of several ppm that last hours in the
late morning. Therefore, the low bias in models and satellite data relative
to the Four Corners TCCON is attributed to the smaller scale enhancements
from the power plants measured in TCCON which are significantly diluted in
the model and satellite results; Bremen is also affected by local urban
sources, and satellites and models would be expected to be biased low, though it is similar to adjacent stations; and JPL, Pasadena,
and Edwards are in or adjacent to a megacity with complex adjacent terrain.
The Izaña TCCON station is on Tenerife, a small island (about 50 × 90 km) with complex topography located about 300 km west of southern
Morocco. The TCCON station is located at 2.37 km above sea level (about 770 hPa), whereas models and satellite observations are of the surrounding ocean
at about 1013 hPa. These stations are not included in averages (e.g., average
bias, average seasonal cycle amplitude differences). In the future,
either targeted observations or more spatially resolved models could make
better use of these TCCON stations. Although we do not use these stations in
averaging, we show results from all TCCON stations in this paper.
GOSAT CO2
The Greenhouse gases Observing SATellite (GOSAT) takes measurements of
reflected sunlight in three shortwave bands with a circular footprint of
approximately 10.5 km diameter at nadir (Kuze et al., 2009; Yokota et al.,
2009; Crisp et al., 2012). The first useable science measurements were made
in April 2009, but due to changing observational modes in the early months,
we use data beginning in July 2009. In this work, we use column-averaged dry-air mole fraction (XCO2) retrievals produced by NASA's Atmospheric
CO2 Observations from Space (ACOS) project, version 3.5 (see O'Dell et
al. (2012) for retrieval details). For each sounding, the retrieval produces
an estimate of XCO2, the vertical sensitivity of the measurement (i.e.,
the averaging kernel), and the posterior uncertainty in XCO2. It also
produces a number of other retrieval variables, such as surface pressure and
aerosols, which are used in both filtering and bias correction.
Post-retrieval filter is employed based on a number of variables associated
with the retrieval. In addition to filtering, a revised bias correction
scheme has been developed for the v3.5 retrievals. This scheme is similar to
the approach described in Wunch et al. (2011b), which characterized the
errors in earlier versions of the ACOS retrieval using a simple spatial
uniformity assumption of XCO2 in the Southern Hemisphere (sometimes
referred to as the Southern Hemisphere Approximation) to assess errors
and biases in the retrievals. V3.5 has corrections of GOSAT high (H) and
medium (M) gain data over land, as well as glint-mode data over the ocean,
by using not only the Southern Hemisphere Approximation, but also TCCON
observations, and comparisons to an ensemble mean of multiple transport
model output. Details of the post-retrieval filter and the bias correction
scheme can be found in the ACOS v3.5 user's guide which will soon be at
https://co2.jpl.nasa.gov/.
SCIAMACHY CO2
The following description of the SCanning Imaging Absorption spectroMeter
for Atmospheric CHartographY (SCIAMACHY) CO2 retrieval algorithm
summarizes important aspects of Reuter et al. (2010 and 2011) and is
adopted in parts from the algorithm theoretical basis document (Reuter et
al., 2012b).
The Bremen Optimal Estimation DOAS (BESD) algorithm is designed to analyze
SCIAMACHY sun normalized radiance measurements to retrieve the column-averaged dry-air
mole fraction of atmospheric carbon dioxide (XCO2). BESD is a so-called
full physics algorithm, which uses a two-band retrieval, with the O2 A
absorption band used to retrieve scattering information of clouds and
aerosols, while the 1580 nm band additionally contains CO2 information.
Similar to the ACOS three-band retrieval for GOSAT, the explicit
consideration of scattering by this approach reduces potential systematic
biases due to clouds or aerosols.
The retrieved 26-elements state vector consists of a second-order polynomial
of the surface spectral albedo in both fit windows, two instrument
parameters (spectral shift and slit functions full width at half maximum in both fit windows, described in Reuter et al., 2010), a temperature
profile shift, a scaling of the H2O profile and a default aerosol
profile, cloud water/ice path, cloud top height, surface pressure, and a 10-layer CO2 mixing ratio profile. Even though the number of state vector
elements (26) is smaller than the number of measurement vector elements
(134), the inversion problem is generally underdetermined, especially for
the CO2 profile. For this reason BESD uses a priori knowledge as a
side constraint. However, for most of the state vector elements the a priori
knowledge gives only a weak constraint and therefore does not dominate the
retrieval results. The degree of freedom for XCO2 typically lies within
an interval between 0.9 and 1.1.
A post-processor adjusts the retrieved XCO2 to a priori CO2
profiles generated with the simple empirical CO2 model described
by Reuter et al. (2012a). Additionally the post-processor performs quality
filtering and bias correction. The bias correction is based on, e.g.,
convergence, fit residuals, and error reduction. The bias correction
follows the idea of Wunch et al. (2011b) using TCCON as a reference to derive
an empirical bias model depending on solar zenith angle, retrieved albedo,
etc. The theoretical predicted errors have been scaled to agree with the
errors vs. TCCON (Reuter et al., 2011). More details can be found in
BESD's algorithm theoretical basis document (Reuter et al., 2012b).
CarbonTracker
CarbonTracker (CT) is an annually updated analysis of atmospheric carbon
dioxide distributions and the surface fluxes that create them (Peters et al.,
2007). CarbonTracker uses the Transport Model 5 (TM5) offline atmospheric
tracer transport model (Krol et al., 2005) driven by meteorology from the
European Centre for Medium-Range Weather Forecasts (ECMWF) operational
forecast model and from the ERA-Interim reanalysis (Dee et al., 2011) to
propagate surface emissions. TM5 runs at a global 3∘× 2∘ resolution and at a 1∘× 1∘ resolution over
North America. CarbonTracker separately propagates signals from fossil fuel
emissions, air–sea CO2 exchange, and terrestrial fluxes from wildfire
emissions and non-fire net ecosystem exchange. Similar to other existing
CO2 inverse models, oceanic and terrestrial biosphere surface fluxes
are optimized to agree with atmospheric CO2 observations, while fossil
fuel and wildfire emissions are specified. First-guess fluxes from
terrestrial biosphere models and surface ocean carbon analyses are modified
by applying weekly multiplicative scaling factors estimated for 126 land and
30 ocean regions using an ensemble Kalman filter optimization method. The
CT2013b release of CarbonTracker assimilates in situ observations between 2000 and
2012 from 103 data sets around the world, including time series from NOAA
observatories, tall towers, and flasks sampled by the NOAA Cooperative Air
Sampling Network, and flask and continuous measurements from partners
including Environment Canada, the Australian Commonwealth Scientific and
Industrial Research Organization, the National Center for
Atmospheric Research, the Lawrence Berkeley National Laboratory, and
the Brazilian Instituto de Pesquisas Energéticas e Nucleares.
In order to explicitly quantify the impact of transport uncertainty and
prior flux model bias on inverse flux estimates from CarbonTracker, the
CT2013b release is composed of a suite of inversions, each using a different
combination of prior flux models and parent meteorological model. Sixteen
independent inversions were conducted, using two terrestrial biosphere flux
priors, two air–sea CO2 exchange flux priors, two estimates of imposed
fossil fuel emissions, and two transport estimates in a factorial design.
CT2013b results are presented as the performance-weighted mean of the
inversion suite, with uncertainties including a component of across-model
differences. All CarbonTracker results and complete documentation can be
accessed online at http://carbontracker.noaa.gov.
MACC
Monitoring Atmospheric Composition and Climate (MACC, http://www.copernicus-atmosphere.eu/) was the European Union-funded
project responsible for the development of the pre-operational Copernicus
atmosphere monitoring service. MACC monitors the global distributions of
greenhouse gases, aerosols, and reactive gases, and estimates some of their
sources and sinks. Since 2010, it has been delivering an analysis
of the carbon dioxide in the atmosphere and of its surface fluxes every year, based on
the assimilation of air sample mole fraction measurements (Chevallier et al., 2010). It relies on a variational inversion formulation, developed by LSCE,
that estimates 8-day grid point daytime/nighttime CO2 fluxes and the
grid point total columns of CO2 at the initial time step of the
inversion window. The Bayesian error statistics of the estimate are computed
by a robust randomization approach. The MACC inversion scheme relies on the
global tracer transport model Laboratoire de Météorologie Dynamique (LMDZ; Hourdin et al., 2006), driven by the
wind analyses from the ECMWF. For release v13.1 of the MACC inversion, used
here, LMDZ was run at a horizontal resolution 3.75∘ longitude × 1.9∘ latitude with 39 vertical layers. The other elements of
the inversion configuration follow Chevallier et al. (2011), with
climatological (i.e., not interannually varying), terrestrial, and ocean prior
fluxes and interannually varying fossil fuel and biomass burning emissions.
The variational formulation of the inversion allowed the 1979–2013 period to
be processed in a single assimilation window, therefore ensuring the
physical and statistical consistency of the inversion over the full 35-year
period. Mole fraction records from 131 measurement sites have been used from
the NOAA Earth System Research Laboratory archive, the World Data Centre for
Greenhouse Gases archive, and the Réseau Atmosphérique de
Mesure des Composés à Effet de Serre (RAMCES) database (see the list
in the Supplement of Peylin et al., 2013).
Direct comparisons to TCCON
We show comparisons between satellite XCO2, model-simulated mole
fraction fields, and TCCON XCO2 at 25 different TCCON sites, shown in
Fig. 1. These sites span the northern and southern hemispheres and cover a
wide range of latitudes and longitudes. In our analysis we mainly use
stations with start dates in or before 2012 which have coverage in all
seasons so that seasonal cycle fits, which require 2 years of data, can be
made. Newer sites are not used for SCIAMACHY comparisons, as this data set
ends in mid-2012 and our version of CarbonTracker (CT2013b) ends at the end
of 2013. We show representative time series for CT2013b, MACC, GOSAT, and
SCIAMACHY for a Northern Hemisphere site (Lamont, OK, US at 37∘ N)
and two Southern Hemisphere sites (Lauder, NZ at 45∘ S or
Wollongong, New South Wales, Australia at 35∘ S) in Fig. 2. Lauder
is shown for the models to show the phase lag for CT2013b seen at this site
in Table 5. Since there are not enough coincidences for the satellites at
Lauder, we show Wollongong for satellites. These plots show matches using
the geometric coincidence criteria described in Table 2 below (for
satellites) and give an idea of the number of coincidences for each data set
using these criterion. As only TCCON/satellite-matched pairs are shown,
different subsets of TCCON are included for models and the two satellites.
These sites were chosen as they have the most coincidences in the northern
and southern hemispheres, respectively, for satellites. All sets compare
well; the 30-day moving averages show differences most easily, such as a
repeating blip in CT2013b comparisons at the summer drawdown at Lamont and a
seasonal mismatch in CT2013b comparisons to Lauder, which will be
discussed later in the paper.
Time series for matches of CT2013b, MACC, SCIAMACHY, and GOSAT
vs. TCCON at Lamont (a–d) and Lauder (for models) or Wollongong
(for satellites) (e–h). The top plot of each set (a1, b1, c1, …) shows a time series of all geometric matching pairs. The
middle panel of each set (a2, b2, c2, …) shows the
difference vs. TCCON, with the blue line denoting the 30-day average difference.
The bottom panel of each set (a3, b3, c3, …) shows a
histogram of the differences, indicating an approximate error and bias.
Fit of a and b in Eq. (2), with stations having data out to at least
n= 50 for dynamic coincidence criteria, n= 10 (GOSAT) or n= 40 (SCIAMACHY)
for geometric coincidence criteria. The CT-CT column describes the CT2013b
coincidence error, providing a lower bound on the satellite/TCCON
differences. The subtr. co-location error row estimates the correlated
error for satellites only, subtracting the quadrature TCCON error (0.35 ppm)
and co-location error from the CT-CT column.
a Bremen, Garmisch, Four Corners, JPL, and Izaña are influenced by local
effects or complex terrain and are not included in averages. b Ascension data
start in 2013 after CT2013 stops.
Bias (left panel) and standard deviation (right panel) for CT2013b, MACC,
BESD-SCIAMACHY, and ACOS-GOSAT vs. TCCON stations, arranged from high to
low latitude. Comparisons which have a particularly low number of matches
are TSUKUBA and Lauder for SCIAMACHY and Lauder for GOSAT.
Coincidence criteria and other matching details
The SCIAMACHY and GOSAT comparisons in this paper are based on two different
definitions of coincidence criteria between TCCON and satellite data.
Satellite measurements, which satisfy the so-called geometric criteria, are
within ±1 h, ±5∘ latitude and longitude of an
unaveraged TCCON observation. Following the match-ups, all TCCON
observations matching one satellite observation which are within 90 min
are averaged, reducing the TCCON random error. The dynamical criteria (Wunch
et al., 2011b; Keppel-Aleks et al., 2011, 2012) are
designed to exploit information about the dynamical origin of an air parcel
through a constraint on the free-tropospheric temperature. This allows us to
relax the geometric constraints and find more coincident satellite soundings
per TCCON measurement. Briefly, a match is found when the measurements are
within 5 days and the following is satisfied:
ΔLatitude102+ΔLongitude302+ΔTemperature22<1,
where ΔTemperature is the co-located NCEP temperature
difference at 700 hPa (Kalnay et al., 1996). Matches are found with
unaveraged TCCON data; TCCON observations matching a single satellite
observation are averaged within 90 min intervals. Table A2 summarizes the
coincidence criteria and data versions that are used. Other matching schemes
not included in this paper include a method implemented by S. Basu described
in Guerlet et al. (2013), which utilizes model CO2 fields to determine
coincidences, and Nguyen et al. (2014)
which uses a weighted average of distance, time, and mid-tropospheric
temperature. Dynamic and geometric coincidence criteria are compared in
Sect. 3.3 and geometric coincidence criteria are used to spot-check dynamic
coincidence criteria results.
The choices used in this paper regarding model/TCCON match-ups are linearly
interpolating to the TCCON latitude, longitude, and time for the models, and
using the TCCON surface pressure for calculating XCO2. In the cases
where the TCCON surface pressure is greater than the model surface pressure,
the model surface CO2 value is replicated to the missing pressure
values. When comparing models vs. TCCON, the TCCON averaging kernel is
applied to the model. When using the model to assess satellite coincidence
error, the satellite averaging kernel is applied to the model at the
satellite and TCCON coincidences. Note that the TCCON averaging kernel
cannot be applied to satellite data and vice versa because a profile of
higher resolution than the comparison observation is needed to apply an
averaging kernel. In the satellite/TCCON comparison, both products have
∼ 1 ∘ of freedom. (The satellites do initially retrieve a
profile with ∼ 1.6∘ of freedom for GOSAT, but the
profile is not what we are validating.)
Because of the earth's curvature, high-latitude sites could have relaxed
coincidence in longitude, particularly for geometric coincidence. However
stations north of 60∘ N have gaps in the winter months in the satellite record
such that relaxed criteria do not add additional stations to the analysis.
Overall bias (left panel, with error bars showing the standard deviation
of the bias) and standard deviation (right panel, with stars showing the predicted
error for satellites) for most stations (some stations removed, see text).
Bias and standard deviation for individual matches
Figure 3 shows a summary of the comparisons for geometric criteria where
satellite matches are not averaged. Averaging and the effects of coincidence
criteria and satellite averaging are addressed in Sect. 3.4. The black box
shows five European stations which are very close, geographically, yet have
different biases. The gray bars labeled TCCON bias uncertainty in Fig. 3
signify the overall calibration uncertainty in TCCON which is estimated to be
0.4 ppm (Wunch et al., 2010, 2011a). The significance of the bias vs. TCCON is
estimated by the 5 % t test (looking up bias/standard deviation
times the square root of the number of comparisons in a t test table). The gray box is the
uncertainty in the TCCON station-to-station bias and all comparisons with
|bias| > 0.4 were found by the t test to be
significant. Therefore, from the above two pieces of information, all biases
differing from TCCON more than 0.4 ppm are significantly different than
TCCON. For GOSAT, biases larger than the TCCON bias uncertainty occur at
Sodankylä(+), Karlsruhe(+), Lamont(-), Tsukuba(+), and
Wollongong(+). SCIAMACHY has the same outliers as GOSAT with additional
biases at Bialystok(+), Orléans(+), Park Falls(+), and
Lauder(+). The sites with identified local influences and the general bias
with respect to these sites are Bremen(+), Four Corners(-), JPL2007(-),
and JPL2010(-). The stations which have complex conditions are Garmisch and
Izaña, which show similar biases to surrounding stations. For unaveraged
results, model standard deviations are lower than either satellite as
satellite differences result from both systematic and random measurement
error, the latter which does not occur for models. The standard deviations
show some variability from station to station which are investigated below.
The effects of averaging and coincident criteria are investigated in Sect. 3.3.
Figure 4 shows the biases and standard deviations grouped globally and over
the northern and southern hemispheres. To estimate the overall bias and
standard deviations for single observations, we take out the outliers as
follows. As described in Sect. 2.1, we take out JPL, Four Corners, Bremen,
Garmisch, and Izaña for averaging. For satellites, we remove the above
plus Tsukuba and Lauder due to limited numbers of comparisons for SCIAMACHY.
For the mean NH bias, we take out stations poleward of 60∘ N, which show a large
positive bias for GOSAT and SCIAMACHY. There is an overall bias vs. TCCON on the order of 0.7 ppm for CT2013b,
and 0.2–0.3 ppm for the other three sets. The overall bias is less of a
concern than the bias variability in satellite data which indicates regional
errors that will translate to regional errors in flux estimates. The bias
variability is 0.4, 0.4, 0.5, and 0.3 ppm for CT2013b, MACC, SCIAMACHY
non-polar, and GOSAT non-polar respectively. Note SCIAMACHY data are
corrected to have an average zero bias with respect to TCCON GGG2012 which
is 0.3 ppm higher than GGG2014 (https://tccon-wiki.caltech.edu/Network_Policy/Data_Use_Policy/Data_Description#CORRECTIONS_AND_CALIBRATIONS).
The overall standard deviations are 0.9 ppm for CT2013b, 0.9 ppm for MACC,
2.1 ppm for SCIAMACHY (all latitudes), and 1.7 ppm for GOSAT (all
latitudes). These values represent the overall performance of CT2013b, MACC,
and single soundings from SCIAMACHY and GOSAT. The standard deviations are
somewhat lower for the SH stations used, with values of 0.8, 0.8, 2.0, and
1.6 for CT2013b, MACC, SCIAMACHY, and GOSAT, respectively. Reuter et al. (2013) validated earlier retrieval versions of BESD-SCIAMACHY and ACOS-GOSAT
with TCCON and found 2.1 ppm (BESD) and 2.3 ppm (ACOS) for the single sounding
precision and 0.9 ppm for the station-to-station biases. Their findings for
BESD are consistent with the findings of Dils et al. (2014). The
station-to-station biases are lower in our analysis due to corrections in
TCCON, improvements in satellite estimates, and removal of several stations
from the estimates. The polar station Sondankyla north of 67∘ N is compared to
GOSAT and SCIAMACHY. A similar standard deviation is found vs. other
TCCON stations, but a higher bias, 1.6 ppm for GOSAT and SCIAMACHY. Most of
the remaining analysis in the paper does not have enough coincidences at
high latitudes so it is important to note this result.
We test whether the biases seen in Figs. 3 and 4 are persistent from year to
year. When at least two full-year averages exist for a station, the standard
deviation of the yearly bias is calculated. The average over all stations of
the yearly bias standard deviation is 0.3 ppm for all sets (CT2013b, MACC,
SCIAMACHY, GOSAT).
Another important comparison is of the predicted and actual errors. The
predicted error (also referred to as the a posteriori error) is reported for each
satellite product and the actual error we take to be the standard deviation
of the satellite observation vs. TCCON. These two quantities should agree
if the TCCON error is much smaller than the a posteriori error and the coincidence
criteria do not degrade the agreement. The predicted and actual errors
vary from site to site, e.g., from variations in albedo, aerosol composition, and
solar zenith angle. We calculate the correlation between the standard
deviation vs. TCCON and the predicted error for each site as follows: the
standard deviation of the satellite vs. TCCON is calculated at each TCCON
station. The correlation of the standard deviation and predicted errors by
station are calculated. ACOS-GOSAT has a 0.6 correlation and BESD-SCIAMACHY
has a 0.5 correlation. This indicates that the predicted error should be
utilized, e.g., when assimilating ACOS-GOSAT, as the variability in the
predicted error represents variability in the actual error, though not
perfectly. A scale factor should also be applied to the predicted errors.
For ACOS-GOSAT the predicted error averaged over all TCCON sites is 0.9 ppm,
as compared to the actual error of 1.7 ppm and can be corrected by applying
a factor 1.9 to the reported GOSAT errors. For BESD-SCIAMACHY, the
prediction error of 2.3 ppm multiplied by 0.9 agrees with the 2.1 ppm actual
error.
Errors as a function of coincidence criteria and averaging
We now directly compare performance of geometric and dynamic coincidence
criteria and averaging in terms of error. Figure 5 shows SCIAMACHY and GOSAT
standard deviations vs. TCCON for geometric and dynamical coincidence
criteria in the Northern Hemisphere. The stations used were those that had
entries for all comparisons, listed in the Fig. 5 caption. For n= 1 no
averaging is done and the dynamic coincidence criteria perform similarly to
the geometric criteria, though the dynamic error is ∼ 0.1 ppm
higher for both satellites. For n= 2, exactly two satellite observations
were averaged for each coincidence. The error drops substantially, but not
as 1/2, which would be expected if the error were uncorrelated. For
n= 4, the error again drops but it is not half the n= 1 error, which is
shown by the dotted line. At n= 4, the dynamic coincidence error is the same
as the geometric error for SCIAMACHY, likely because dynamic coincidence
involves averaging observations farther apart in location and time, which
are less likely to have correlated errors. The last bar is the maximum n,
which has results for all stations included. The dynamic criteria allow far
more coincidences, resulting in significantly lower average errors. Note
that all averaged satellite observations match one particular TCCON
observation.
Standard deviation of SCIAMACHY and GOSAT minus TCCON for
different coincidence criteria and number of satellite observations
averaged, n, in the Northern Hemisphere. The following consistent set of
stations was used for all comparisons: Bialystok, Karlsruhe, Orléans,
Garmisch, Park Falls, and Lamont. The dotted line shows the error if it
scaled as the inverse square root of the number of averaged observations.
The far right case for each of the categories contains the maximum n that has
results for all stations.
Averaging matches of satellite data vs. TCCON at Lamont. As the
number averaged increases, the standard deviation vs. TCCON decreases.
CT2013 at the satellite vs. at CT2013 at TCCON (purple) is used to
quantitate spatiotemporal mismatch error. The points are fit to Eq. (2)
(black). For GOSAT the uncorrected data are also fit (black dashed). The
initial guess minus TCCON standard deviation is shown as a green dashed
line. We see that in this case, for GOSAT at Lamont, averaging more than
about four observations improves over the initial guess.
Errors vs. averaging: random and correlated error
To test the effects of spatial averaging, we calculate station by station
standard deviations of satellite–TCCON matched pairs as a function of n,
where n is the number of satellite observations that are averaged, which are
chosen randomly from available matches (so there should be no difference in
the characteristics of chosen points for larger vs. smaller n). Figure 6
shows plots from Lamont for SCIAMACHY and GOSAT for standard deviation
difference to TCCON vs. n. Initially the error drops down rapidly with
n, however the decrease slows with larger n. The curve fits well to the
theoretically expected form:
error2=a2+b2/n,
where a represents correlated errors which do not decrease with averaging for
similar cases (including smoothing errors, errors from interferents such as
aerosols, TCCON error, and co-location error), b represents uncorrelated
errors which decrease with averaging, and n represents the number of
satellite observations that are averaged. The purple dashed line represents
the standard deviation of CT2013b at the satellite time and location vs.
CT2013b at the TCCON time and location and represents a lower bound of the
spatiotemporal mismatch error (co-location error). As expected, this value
is much smaller for geometric than for our dynamic coincidence criteria
(other dynamic coincidence criteria may do better; in our case the dynamic
criteria consider points ±30∘ longitude, ±10∘
latitude, ±5 days, and ±2 K temperature vs. 1 h, 5∘ for geometric criteria). The co-location error for large n is shown
in the CT-CT columns of Table 2, with Northern Hemisphere averages of 0.3, 0.6, 0.4, and 0.7
ppm, for SCIAMACHY geometric, SCIAMACHY dynamic, GOSAT geometric, and GOSAT
dynamic coincidences, respectively. There is not much difference between the
estimated co-location error at North American sites of Lamont and Park Falls
where the CT2013b model is at 1 × 1∘ vs. other Northern Hemisphere sites where the
CT2013b model is at 3 × 2∘. The co-location error is subtracted from
the a value in quadrature to estimate results without co-location error.
Bias for 3-month groups for each station, where each station is
normalized to have 0 yearly bias. For satellites, stations are included when
at least 20 matches are found in each season. Dynamic coincidence criteria
are used. The station colors are coded by location: far NH gray, European and
Park Falls red/yellow, midlatitude green, SH blue.
Bias for 3-month groups for Southern Hemisphere (left panel), 0–45∘ N
(middle panel), and poleward of 45∘ N (right panel). Each station composing each group is
normalized to have zero average over the year. The Southern Hemisphere
(left panel, Lauder (except SCIAMACHY), Wollongong, and Darwin) has relatively
small biases. 0–45∘ N includes Tsukuba and Lamont. > 45∘ N includes
Karlsruhe, Park Falls, and Orléans for satellites, and those plus Sodankylä
and Bialystok for models. The results that are significant according to the
t test are shown with a * in the bar plots.
We calculate a and b by station in Table 2, splitting out geometric and
dynamic coincidence criteria. The a term, which does not reduce with
averaging, is the correlated error, and the b term, which reduces with
averaging, is the uncorrelated error. There is more correlated error, a, for
SCIAMACHY geometric vs. dynamic matches in 5/6 stations in the Northern Hemisphere, indicating that averaging is more effective when it is over a
larger spatial/temporal area, probably due to variability in the source of
the correlated errors. GOSAT geometric vs. dynamic averages show larger
correlated error for 4/8 stations. The GOSAT dynamic correlated error of 0.9 ppm
is significantly reduced by subtraction of the 0.7 ppm co-location
error, whereas the 0.9 ppm geometric error is not strongly changed by the
subtraction of 0.4 ppm co-location error. Subtracting (in quadrature) the
co-location error (CT-CT column in Table 2) and TCCON error of 0.35 ppm
(Appendix B) results in the corrected correlated error, a, shown in the
mean NH: subtr. co-location error row of Table 2.
The northern hemispheric average values, corrected by co-location and TCCON
error, are a= 1.4 ± 0.3 ppm, b= 1.7 ± 0.2 ppm for SCIAMACHY
geometric, a= 1.0 ± 0.3 ppm, b= 2.1 ± 0.1 ppm for SCIAMACHY
dynamic, a= 0.8 ± 0.2 ppm, b= 1.6 ± 0.1 ppm for GOSAT geometric,
a= 0.5 ± 0.2 ppm, b= 1.9 ± 0.1 ppm for GOSAT dynamic. These
values indicate the expected error when averaging GOSAT or SCIAMACHY
observations within 5 ∘ and 90 min for geometric, and for 5 day
regional averages for dynamic coincidences.
The green dashed line in Fig. 6 shows the standard deviation of the
satellite prior vs. TCCON. Although using an optimal constraint will
result in an error lower than the prior error in the absence of systematic
errors, these satellite retrievals of CO2 have been set up to value
average results over single observations; while the error increases from the
prior for a single observation, average results have both less error and
minimal prior influence.
Seasonally dependent biases
It is important to determine whether there are seasonally dependent biases,
as these will impact flux distributions. We look at 3-month periods (DJF,
MAM, JJA, SON), with the overall yearly bias at each site subtracted out to
isolate the seasonal biases. To get enough comparisons, we use the dynamical
criteria for satellite coincidences, as using the geometric criteria cuts
down the comparisons with sufficient seasonal coverage to three stations (Park
Falls, Lamont, and Wollongong). This is a simple averaging method which will
later be compared to seasonal cycle amplitude fit results.
Figure 7 shows the biases for stations that have at least 20 matches in each
season, and Fig. 8 shows the results averaged by SH, 0–45∘ N (which includes
Tsukuba and Lamont), > 45∘ N. The error bars shown in Fig. 8 are
the standard deviation of the results in each bin, once the
station-dependent average biases are subtracted. Significance was tested
using the t test with biases larger at least 0.2 ppm. For the north
midlatitudes, the seasonal cycle peak is in MAM and the minimum is in JJA,
so the seasonal cycle error should be approximately bias (MAM) minus bias (JJA).
For Lamont and Tsukuba in JJA, SCIAMACHY is biased low vs. TCCON by about 1.4 ppm (the low bias is 0.8 ppm
for Lamont). In 45–53∘ N, SCIAMACHY is biased low vs. TCCON in MAM and high in JJA, with about a
0.9 ppm spread. GOSAT has the same pattern with a 0.5 ppm spread. SCIAMACHY
is biased low vs. TCCON in DJF by 0.7 ppm, but this does not affect the seasonal cycle
amplitude. GOSAT is similarly low but has more station-dependent variability
so is not considered significant. CT2013b is biased high vs. TCCON in JJA by about 0.2 ppm and
biased low vs. TCCON in DJF by about 0.3 ppm. MACC is biased low vs. TCCON in MAM by 0.2 ppm in 28–37∘ N. In the
Southern Hemisphere, the seasonal cycle maximum is in the October to January
time frame, and the minimum is in the April time frame. The significant findings
in the SH are that CT2013b is biased about 0.3 ppm low vs. TCCON in DJF and biased 0.2 ppm high in
MAM, which should lead to a 0.5 ppm underestimate of the SH seasonal cycle.
Geometric results, which including fewer stations due to the cutoff of at
least 20 observations per season, corroborate the low bias for in MAM, high
bias in JJA, and low bias in DJF in 45–53∘ N for SCIAMACHY, and the low bias
in MAM and high bias in JJA in 45–53∘ N for GOSAT.
Seasonal cycle amplitude for different latitudes. Stations included
for satellites are Karlsruhe, Orléans, Park Falls, Lamont, Darwin,
Reunion (GOSAT), and Wollongong. Stations included for models are
Sodankylä, Bialystok, Karlsruhe, Orléans, Park Falls, Lamont,
Tsukuba, Saga, Darwin, Reunion, Wollongong, Lauder_120HR, and
Lauder_125HR. Bold shows entries with statistically
significant differences.
ComparisonRegionSeasonalSeasonal amp.Seasonal amp.End date Co-loc.BootstrapAveraging choicesBin SDamp. (ppm)TCCON (ppm)difference (ppm)(ppm)(ppm)(ppm)(ppm)(ppm)CT2013b67–79∘ N (n=2)9.69.7-0.1 ± 0.10.000.14––46–53∘ N(n=4)7.98.1-0.2± 0.10.040.13–0.128–37∘ N (n=1)6.37.4-0.6 ± 0.50.030.14–0.5SH(n= 4)0.91.6-0.5± 0.20.050.09–0.2MACC67–79∘ N (n=2)10.910.20.7 ± 0.50.30.33–0.3346–53∘ N (n=4)8.48.20.2 ± 0.10.060.16–0.1328–37∘ N(n=3)5.34.70.6± 0.20.050.16–0.16SH (n=5)1.11.3-0.2 ± 0.20.050.08–0.19GOSAT46–53∘ N(n=3)6.77.1-0.4± 0.10.05+0.10.060.070.1(v3.5)28–37∘ N (n=1)5.15.50.4 ± 0.10.03-0.10.010.02–SH (n=3)2.02.1-0.1 ± 0.20.09+0.10.020.030.2SCIAMACHY46–53∘ N (n=3)7.57.80.4 ± 0.80.06+0.40.090.030.8(BESD-v02.00.08)28–37∘ N (n=1)7.26.40.8 ± 0.10.11-0.00.030.04–SH (n=2)2.22.00.2 ± 0.30.06-0.20.030.010.3Comparisons of seasonal cycle amplitude, phase, and yearly increase
We compare model and satellite XCO2 to TCCON using the NOAA fitting
software CCGCRV (Thoning et al., 1989) to calculate seasonal cycle amplitudes
and yearly increases. At least 2 years are needed to distinguish the seasonal
cycle from the yearly increase. The errors are calculated using the bootstrap
method (Rubin, 1981; Efron, 1979), the standard deviation of differences within one bin
divided by the square root of the number of stations minus 1
(n > 1), the variability of results when choosing different
averaging (for satellites) where one, two or four satellite observations are
averaged for GOSAT and SCIAMACHY, coincidence errors estimated by the mean
difference between CT2013b matched to the satellite and CT2013b matched to
TCCON, and the variability resulting from small changes in the time range.
Two data sets at a time are matched, using the dynamic criteria, with the
satellite averaging four observations for SCIAMACHY and two for GOSAT, which
reduces the fit errors. Since different data sets will have different data
gaps and time ranges, the TCCON results will be somewhat different for each
comparison. Plots are individually examined to ensure that there are adequate
data. Stations are removed when there are large errors for any of the above
errors, and locally influenced stations are also removed (see Sect. 2.1). The
stations that were excluded based on variability of results for changes in
the time range on the order of a few months are Eureka for the models and
Sodankylä, Bialystok, Saga, and Lauder_125HR for the satellites. At Bialystok, for example, the
GOSAT-TCCON seasonal cycle was 0.3 or -0.9 ppm depending on whether the
time series ended in November 2013 or on 1 January 2014, respectively. The
time series shows large gaps in the data in the winter and a gap around the
seasonal cycle minimum in 2012. The findings show that the GOSAT seasonal
cycle amplitude at Bialystok may be influenced by a small subset of the GOSAT
points.
Seasonal cycle amplitude. The TCCON values are shown by the
circles. The averaging is done over 10 × 10∘ bins every 5∘.
Comparisons vs. TCCON may be different than Table 4, since Table 4 has very
close criteria for models and dynamic coincidence criteria for satellite
data. The models are sampled at GOSAT observations.
Yearly increases. Each comparison uses matched pairs with TCCON
using locations which have at least 2 years of data for comparisons. See
Table 3 for stations included. The start date and end date are averaged for
the stations in each bin and are shown in the second to last column. Bold text
shows one difference larger than predicted errors. The last column shows the
average global yearly increase for the time period using Table 6.
The seasonal cycle amplitude is important for estimating source and sink
estimates and global distributions. Table 3 shows the seasonal cycle
amplitudes grouped by latitude. As described in Sect. 4, the error is
calculated by several different methods. The error in Table 3 is root mean
square of the end date choice, bootstrap error, averaging choice, and bin
standard deviation. The co-location error (only relevant for satellites) is
calculated as an average bias in the bin; this bias should be subtracted
from the satellite–TCCON differences to remove the effects of co-location
error. When n > 1, results are considered significant using the
t test of all the results in the bin (comparison of the mean difference/standard
error difference to a standard look up table). The calculated
seasonal cycle amplitudes are shown in Table 3, and results are compared for
consistency to the simpler seasonal cycle biases calculated in Sect. 3.5
to understand what season the discrepancy occurs.
The significant findings from Table 3 are as follows. (1) In northern latitudes
(46–53∘ N), GOSAT underestimates the seasonal cycle by 0.4 ppm. This
latitudinal range is composed of two European sites and Park Falls, all of
which follows this pattern, so it is not just in Europe but also North
America at this latitude. This finding is consistent with Lindqvist et al. (2015). The simple bias calculation from Sect. 3.5 estimates that GOSAT
should underestimate the seasonal cycle by 0.5 ppm, with half the issue being a
low bias in MAM and half the issue being a high bias in JJA. The co-location error
is estimated as +0.1 ppm, so removing the co-location error should
increase the GOSAT-TCCON discrepancy. The geometric criteria bias for this
same bin is -0.7 ± 0.4 ppm, which is consistent with the dynamic
results, though with larger error bars and significantly sparser time series
plots. (2) There is an underestimate in the CT2013b seasonal cycle for 45–53∘ N
compared to TCCON by 0.2 ppm. This is a small discrepancy but the error bars
are also small. Similarly in Sect. 3.5, a 0.2 ppm low bias is predicted.
(3) There is an overestimate in the MACC seasonal cycle in the 28–37∘ N range by 0.6
ppm. A smaller 0.2 ppm low bias is seen in MAM for MACC. (4) CT2013b
underestimates the SH seasonal cycle by 0.5 ppm. This is partially from a
low bias in DJF and partly from a high bias in MAM.
Findings from Sect. 3.5 which did not reach significance in Table 3 are as follows.
(1) SCIAMACHY should underestimate the seasonal cycle in 45–53∘ N by 2.5 ppm,
which it does not. There is a 0.4 ppm high bias with large error bars. (2) SCIAMACHY
should overestimate the seasonal cycle for Lamont by 1.6 ppm. In
Table 3, SCIAMACHY overestimates by 0.8 ppm but statistical significance is
not calculated for one station. (3) MACC should underestimate the seasonal
cycle by 0.2 ppm in 28–37∘ N. In Table 3 MACC overestimates the seasonal cycle
by 0.6 ppm. The seasonal biases are generally consistent with the seasonal
cycle amplitude differences vs. TCCON and can be used to pinpoint which
months are the cause of the seasonal cycle amplitude differences vs. TCCON; however, the mean seasonal biases find more significant differences
than the seasonal cycle fits.
Figure 9 shows a global map of fits of the seasonal cycle amplitude of
SCIAMACHY, GOSAT, CT2013b, and MACC, with TCCON having at least 2 years of
matches shown by circles. This map shows how the results of Table 4 fit
into the global pattern (with the model fields matched to GOSAT locations
and times). Interestingly, the seasonal cycle amplitude varies
longitudinally; this pattern is seen in both satellite data sets and both
models. Since the amplitude is taken from the sampled harmonic there is no
extrapolation, although the seasonal cycle will be underpredicted at high
latitudes where there are significant data gaps. The model data in Fig. 9 is
co-located with GOSAT, so the same gaps will occur in GOSAT and the two
models, other than fit errors larger than 10 % of the amplitude, were
screened out. This map is consistent with Lindqvist et al. (2015), Fig. 8,
which also finds high values in the 45–50∘ N, 120–180∘ E range.
CO2 yearly growth rate
The same fitting program in the above section, CCGCRV, also calculates a
yearly increase. In Table 4 we compare the fitted yearly increase for TCCON,
which ranges from 1.92 to 2.55 ppm yr-1 for the different stations and
time ranges, to each of the data sets. None of the TCCON/satellite or
TCCON/model differences reach significance for the t test, which is done
when there is more than one result per bin. However, the GOSAT Lamont station
is low at -0.3 ppm yr-1, with -0.1 of this estimated to result from co-location
error. To see how much of the observed variability in the growth rate is
temporal vs. spatial variability in the growth rate, we compare the satellite and model yearly XCO2 increase to the
global annual increase (growth rate) from surface measurements (http://www.esrl.noaa.gov/gmd/ccgg/trends/global.html) which are 1.74,
2.12, 1.77, 1.67, 2.39, 1.70, 2.40, 2.51, 1.89 ppm yr-1 for global yearly
increases 2006–2014, and 1.76, 2.22, 1.60, 1.89, 2.42, 1.86, 2.63, 2.06,
2.17 ppm yr-1 for Mauna Loa yearly increases 2006–2014, with error bars
0.05–0.09 ppm yr-1 for global and 0.11 ppm yr-1 for Mauna Loa. The average
global yearly increase predicted from the above using the time periods in
Table 4 is shown in the last column of Table 4. The correlation r value
between the yearly increase in TCCON and the above ESRL global yearly increase is
0.61, and the correlation with the Mauna Loa yearly increase is 0.67, whereas
the correlation r value between the yearly increase in TCCON and the yearly increase in
columns is 0.82. Therefore, the variability seen in Table 4 is partly from
the time range used but also partly from variations due to geographic
location (and sampling).
Top: cross correlation between TCCON and SCIAMACHY (top left)
and GOSAT (top right) with matches using dynamic criteria at Park Falls.
The x axis shows results when satellite data are offset in days vs. TCCON.
The dashed line shows the expected maximum correlation based on the error
(see Eqs. 2 and 3). The gray line is the correlation for the satellite priors,
which are each out of phase by at least 10+ days. The * is the peak of a
polynomial fit of the correlation between -25 and +25 days. Bottom:
standard deviation between TCCON and SCIAMACHY (bottom left) and GOSAT
(bottom right). The dashed line shows the Eq. (2) predicted error.
The left two numerical columns show standard deviation drop within
±2 days of zero offset. Higher values with either model indicate sites
where temporal co-location is more important. The next two numerical columns
show calculated phase difference between CT2013b, MACC, and TCCON in days,
with entries larger than 10 in bold. A phase difference of -10 days means
that the model seasonal cycle is 10 days behind TCCON. The next four columns
show the same calculations for SCIAMACHY and GOSAT prior and retrieved
values, with entries larger than 10 in bold. Blank values are those for
which a good fit was not found.
a indicates TCCON stations which are locally influenced (see Sect. 2.1).
Reuter et al. (2011, JGR, Table 2) found agreement within the calculated
errors at Park Falls and Darwin for BESD-SCIAMACHY and CT2009 vs. TCCON.
However, older data sets were used for this result. Looking specifically at
Park Falls, we see 1.80 ± 0.14 and 2.10 ± 0.22 for SCIAMACHY and
TCCON, respectively and at Darwin 1.67 ± 0.08 and 2.16 ± 0.05 for
SCIAMACHY and TCCON, respectively, where the errors represent the standard
deviation of SCIAMACHY fits for similar latitudes.
Top: cross correlation examples between TCCON and CT2013 (left) or
MACC (right). Each panel shows the correlation and second-order polynomial fits
(top) and standard deviation (bottom) vs. offset in days of TCCON vs. satellite data. The correlation should be at a maximum and standard
deviation at a minimum at days offset = 0. The top panels show examples
of stations with a phase difference of less than 10 days. The bottom set shows example
stations which have phase differences of at least 10 days.
Seasonal cycle phase
This section looks at the time offset correlation and standard deviation
between the test data sets and TCCON. This checks whether, for example, a
seasonal cycle is delayed or ahead of the TCCON seasonal cycle, which has
important implications for flux estimates (Keppel-Aleks et al., 2012),
whether there are seasonally dependent biases that are affecting the
seasonal cycle, and whether the data sets are seeing the same seasonal
cycle.
Diurnal variability of CT2013b and MACC13.1 vs. TCCON in JJA
arranged by latitude. TCCON variability and maximum theoretical correlation
are shown, as well as actual correlation and slope for both models. The
slope is the mode vs. TCCON fit to a straight line. a indicates sites
expected to be strongly influenced by local sources. Max correlation is calculated
using Eq. (3), TCCON SD, and the lower of the model–TCCON standard
deviations from Appendix B. TCCON range, SD, shows the range of the values
seen for the daily variability for TCCON, as well as the standard deviation
of the TCCON daily differences. The average row gives averages for all entries above it.
The average rows, e.g., DJF, are for stations Bialystok, Orléans, Park
Falls, and Lamont. Eureka and Karlsruhe are not included because fewer than five
matches were found.
TCCONMax corr.CT2013b MACC 13.1 StationLat.Range, SD(Eq. 3)CorrelationSlopeCorrelationSlope(deg)(ppm, ppm)Sodankylä67-1.5 to 1.5, 0.50.60.350.230.400.19Bialystok53-2 to +1, 0.50.60.650.550.660.36Bremena53-2 to +1, 0.70.50.240.06-0.20-0.03Orléans48-1 to +0.5, 0.70.70.720.510.700.31Garmischa47-3 to 0, 0.80.60.550.500.690.45Park Falls46-3.5 to +2, 1.00.80.770.320.540.23Four Cornersa37-4.5 to +0.5, 1.80.90.390.030.220.02Lamont37-3.0 to +1.0, 0.80.70.280.190.320.14Tsukuba36-2 to +1.5, 0.80.60.790.470.640.24JPL2011a34-1 to 1, 0.50.50.180.06-0.05-0.01Saga33-0.5 to 0, 0.30.4-0.14-0.07-0.11-0.06Izañaa28-0.5 to +0.5, 0.10.20.730.110.450.04Darwin-12-0.5 to +0.5, 0.20.30.040.030.100.03Reunion-21-0.3 to 0,0.20.3-0.33-0.040.720.10Wollongong-34-1.5 to +1.0, 0.40.50.330.470.270.15Lauder_120HR-45-0.5 to +0.5, 0.30.30.050.010.200.02Lauder_125HR-45-0.5 to +0.5, 0.20.40.410.230.300.06Average JJA for all stations 0.80.300.210.400.15Average DJF (see caption for station list) 0.80.520.160.540.17Average MAM (see caption for station list) 0.80.590.290.430.15Average JJA (see caption for station list) 0.80.510.390.560.26Average SON (see caption for station list) 0.80.180.090.190.06
To compare seasonal cycle amplitudes, all data sets have 2 ppm yr-1
subtracted to approximately remove secular increases (over the ±60
days' offset this has a very small effect). For a 0 day offset, the data sets
are matched as usual. For a 1 day offset, TCCON is moved forward by 1 day
and compared to the data set. This is repeated for all offset times.
Correlations are fit to a second-order polynomial to determine the phase
minimum difference. As TCCON is moved forward or backward in time, different
points will match up, particularly when there are data gaps in either
data set. This can cause difficulties in interpretation. The maximum
correlation is limited by the ratio of the error to the variability. It
follows from the definition of correlation that
corr_max=corro11+εxσx21+εyσy2,
where corro is the noise-free correlation, εx is the
error on x and σx is the true variability for x, εy is the error on y and σy is the true variability for y.
Because we are estimating σ and ε, there is
uncertainty on the correlation maximum. In our case σy is taken
to be the TCCON variability and εy is estimated using Table 2 with 10 SCIAMACHY and 4 GOSAT averages. The error bars on the correlations
are calculated from Fisher's z test. (Fisher, 1915, 1921). Figure 10 shows
SCIAMACHY and GOSAT results at Park Falls. Although the prior performs well
in regards to the standard deviation vs. TCCON, Fig. 10 shows the prior has
a clear seasonal cycle phase error which is corrected by the satellite
retrievals for both SCIAMACHY and GOSAT at Park Falls.
Results of the seasonal cycle phase error are tabulated in Table 5. Stations
not shown have either too few match-ups (e.g., Sodankylä) or too little
variability compared to the noise (e.g., Wollongong) to have useful
comparisons. The GOSAT retrieval markedly improves over the prior seasonal
cycle phase vs. TCCON at 12 out of 13 stations. For the six stations that
are not locally influenced (with no a by the station name), GOSAT improves
over the prior at all six stations, and additionally has a smaller phase
difference than one or more of the models at four of the six stations, is the
same at one station, and is worse at one station. SCIAMACHY improves over
its prior for Park Falls and Four Corners, mildly improves at Karlsruhe, and
stays the same or gets slightly worse in four cases. Mismatches in SCIAMACHY
phase could be from mismatches in vertical sensitivity (as higher altitudes
have lagged seasonal cycles), effects of coincidence criteria, or
seasonal-dependent biases. To check the coincidence criteria,
cross-correlations were calculated for the geometric coincidence criteria which
had significantly fewer match-ups. Similar results for geometric coincidence
criteria were found for GOSAT and SCIAMACHY for Lamont and Park Falls; the
other stations are too noisy to draw conclusions.
Table 5 also shows the phase differences for the models, which have closer
spatial/temporal matches and lower single-match-up errors. Model–TCCON phase
differences could result from errors in model flux distributions, seasonal
timing, or transport errors. Table 5 shows the phase differences, which vary
from -20 to +10 days. Phase differences more
than 10 days are noticeable by eye when looking at time series data and
occur in the NH at Bremen, Four Corners (negative), Orléans, and
Izaña (positive, CT2013b only), although Bremen, Four Corners, and
Izaña are locally influenced. Large phase differences also occur at some
stations in the Southern Hemisphere. Although the seasonal cycle is weaker
in the Southern Hemisphere, the phase offset can be seen in the CT2013b
Lauder_125HR plots in Fig. 2. The correlations vs. offset
days show a phase difference of -20 days for CT2013b and +0 days for MACC
at Lauder_125HR, as seen in Fig. 11. Note that the fits of
the seasonal cycle in the SH display more complexity, such as multiple local
maximum, than fits in the NH and “phase lag” could easily be an indication
of an issue with the fit shape. Figure 11 shows correlations and standard
deviations vs. day offset for three stations that have the seasonal cycle
peak within ±10 days for CT2013b and MACC (top panels), and for stations
which have a larger phase lag compared to TCCON (bottom panels). There is
often a small peak within ±3 days, which indicates the models'
capability of picking up variations that occur day to day (i.e., synoptic-scale variability), which indicates both the strength of synoptic activity
and matching between models and TCCON. This peak is not seen in satellite
data for dynamic coincidence criteria likely due to matching, or geometric
coincidence criteria likely due to the noise – not that this synoptic peak
occurs at 0, even when the seasonal cycle has a phase lag (e.g., MACC model at
Bremen, in the lower right panel, or Lauder_125HR
comparisons). The synoptic-scale correlation varies between 0 and 0.17, as
seen in Table 5.
Izaña will be briefly discussed. As noted in Sect. 2.1, the TCCON station
is located on a small island at 2.37 km above sea level (about 770 hPa),
whereas the MACC and CT2013b models at ∼ 2∘× 3∘ resolution do not resolve topography and consequently have mean
surface pressure at sea level, about 1013 hPa at this location. Our standard
treatment is to interpolate the model to the TCCON pressure grid, then
calculate XCO2 using the TCCON pressure weighting function. At
Izaña this has the effect of chopping off the lower atmosphere. The
CT2013b result for this treatment has a +10 day seasonal cycle phase
difference at Izaña; whereas MACC has no phase difference at Izaña.
If, however, the model surface pressure is used to calculate XCO2, MACC
goes from a 0 to a -10 day phase lag, and CT2013b has 0 phase difference. An
argument for using the model surface pressure would be if the upslope winds
at Izaña (Sancho et al., 1991; Bergamaschi et al., 2000) shifted the
profile upwards rather than chopping it off, which would occur if the air
deviated around the island instead. This finding has important implications
for the choice of the comparison methodology and the ideal location for
validation sites. Validation sites within complex geographical terrain have
to be treated as special cases as (a) the atmospheric models usually do not
resolve these variations and (b) satellite measurements rarely have a perfect
co-location with the ground-based site, meaning that they could sample a
substantially different altitude level. This holds for both mountains (e.g., Izaña) and valleys (e.g., Garmisch). This highlights one of the many
choices that are made when comparing two products (e.g., whether to apply the
averaging kernel, whether to use interpolation, how to treat the surface
pressure, or what coincidence criteria to use).
Another finding worth noting is the comparisons at Lauder. In 2010 the
Lauder_125HR instrument began routine operation, while the Lauder_120HR
instrument continued to take TCCON data through to the end of 2010. Both
MACC and CT2013b show no seasonal cycle correlation with the 120HR time
series at Lauder, but do show correlation with Lauder_125HR time series. We
attribute this to the improved precision of the 125HR data, and an increase
of the seasonal cycle amplitude in 2011 and 2012 as compared to other years
(e.g., compare 2011 vs. 2007). The phasing error found in the CT2013b
comparison with the Lauder_125HR may be due to CT2013b not modeling the
drivers of the seasonal cycle amplification in 2011 and 2012.
At Bremen and Four Corners, local effects that are not resolved at 3 × 2∘ likely dominate, particularly since Bremen is clustered with
Orléans, Garmisch, and Karlsruhe, which all compare fine, and because
the correlation of daily variability, as seen in the next section, is also
very low at these two stations.
(a) Time series for models and TCCON from 1 to 8 August 2011 at
Bialystok. The end points of the solid lines show the time points used for
comparing daily variability; both diurnal and synoptic variations are seen.
(b) Change throughout the day at Bialystok for 2 August 2011. The large
diamonds at 06:00 and 18:00 show the two times with largest difference that
have the same solar zenith angle that are used in the analysis. The
difference between these times are -2, -1.7, and -0.7 ppm for TCCON, CT2013,
and MACC, respectively. Right: CarbonTracker (c) and MACC (d) daily trends
vs. TCCON daily trends for BIALYSTOK in JJA, from compiling differences
as shown in (b). Correlation is seen in the daily trends as compared to
TCCON with the daily amplitude for the models smaller than TCCON.
Daily variability (models vs TCCON)
At the surface, CO2 shows a strong diurnal cycle in areas with active
vegetation, e.g., Park Falls, during summer, and synoptic trends based on
regional dynamics. Even though the diurnal cycle is markedly smaller in the
total column (Olsen and Randerson, 2004), it can be observed both by TCCON
and also in models, in our case CT2013b and MACC, as seen in Fig. 12. Both
diurnal variations and synoptic trends can be seen in Fig. 12. Validating
the amplitude of the diurnal variability in the column is important as the
column diurnal variability better represents the amount of CO2 emitted
or absorbed by surface processes as compared to surface measurements, which
are more impacted by boundary layer height. To our knowledge this is the
first comparison of model fields to TCCON to compare the diurnal cycle. As
TCCON itself has not been validated at multiple times in 1 day, this is
considered a comparison, not a validation. We compare the difference between
morning and afternoon in models and TCCON. To minimize potential TCCON
biases that depend on the solar zenith angle (through the air mass factor),
we compare at two points in each day separated by the largest time with the
same solar zenith angle (SZA). The methodology is to (1) identify two
points, t1 and t2, from the same day with the largest time difference but
with the same SZA. As the TCCON data used in this paper have been averaged
over 90 min, t1 or t2 may be interpolated between two time points. (2) We compare TCCON at t2 minus TCCON at t1 and the same times for each model.
We look at the variability within 1 day for one season (JJA). Looking at
different seasons for the Northern Hemisphere at the bottom of Table 6, both
models showed clearly higher correlations and slopes in MAM and JJA vs. DJF
and SON, with SON showing the lowest correlation and slope. In the SH,
correlations are highest in DJF, second highest in SON, and lowest in MAM.
However in the SH, the slopes are on the order of 15 % of the slope of TCCON. For
CT2013b, at Darwin, correlations are highest in DJF, less in SON, with
slopes on the order of 15 % of the slope of TCCON. Table 6 shows correlations
between CT2013b or MACC vs. TCCON in the daily variability. In the NH, on
average, the correlations are about two thirds as large as could be expected, given
the relative sizes of the variability and errors (see text around Eq. 3).
Additionally, the two models have about one third to one half the daily variability of
TCCON (as seen from the smaller slopes). In the far north (Eureka,
Sodankylä), the correlations indicate agreement but the model daily
variability is less than 1/4 TCCON. In the midlatitudes (excluding locally
influenced stations) there is the highest correlation (∼ 0.3–∼ 0.7) with
model daily variability 20 to 60 % of the variability of TCCON. Sites influenced by local sources, Bremen, Four Corners,
and JPL2011, the models do not show the diurnal variability that is seen by
TCCON, as evidenced by very small or zero slopes. Sites which have
complicated terrain, Garmisch and Izaña, do show correlations in the
daily variability similar to other nearby stations.
The CT2013b model in general shows more daily variability and higher
correlations, which are in better agreement with TCCON. Since the satellite
observations are coincident ∼ once per day, the diurnal
pattern will not be well constrained by satellite observations, except as
preserved in transported air coincident with satellite measurements
downwind. Model Observing System Simulation Experiments (OSSEs) can determine
the impact of the diurnal cycle strength on flux estimates to determine the
importance of independently verifying the diurnal cycle in
models.
Discussion and conclusions
We find standard deviations of 0.9, 0.9, 1.7, and 2.1 ppm vs. TCCON for
CT2013b, MACC, GOSAT, and SCIAMACHY, respectively, with the single target
errors 1.9 and 0.9 times the predicted errors for GOSAT and SCIAMACHY,
respectively. There is a correlation r value of 0.5 for SCIAMACHY and 0.6
for GOSAT for the actual and predicted errors grouped by station.
Equation (2) and Table 2 show how errors decrease when satellite results are averaged and
estimate the magnitude of the correlated and random errors components for
averaged satellite results, where random error components decrease with an
increasing number of averaged observations. When satellite data are averaged
and interpreted according to the model error2=a2+b2/n
(where n is the number of observations averaged, a is the systematic
(correlated) errors, and b is the random (uncorrelated) errors). The
northern hemispheric average values from Table 2 are a= 1.4 ± 0.3 ppm,
b= 1.7 ± 0.2 ppm for SCIAMACHY geometric; a= 1.0 ± 0.3 ppm,
b= 2.1 ± 0.1 ppm for SCIAMACHY dynamic; a= 0.8 ± 0.2 ppm,
b= 1.6 ± 0.1 ppm for GOSAT geometric; a= 0.5 ± 0.2 ppm,
b= 1.9 ± 0.1 ppm for GOSAT dynamic. The lowest correlated errors are
found when using dynamic coincidence criteria where values are averaged from
a larger spatiotemporal region, but the lower value for GOSAT is only after
the estimated co-location error is subtracted. The Southern Hemisphere
errors are uniformly smaller, both in correlated and uncorrelated errors.
These data represent averaging of satellite data which match one averaged
TCCON value. The above error model should help in assigning realistic retrieval
error correlations in assimilation systems in place of current ad hoc
hypotheses. For example, in Basu et al. (2013) observations within 500 km
and 1 h are assumed to have 100 % correlated errors, and are inflated
by a factor such that when observations are later treated as if the errors
were random, the final error of the average is the same as the error of one
observation. This can be improved by setting the inflation factor so that
the average observation error is a2+b2/n, with a and b set by the
geometric values from Table 2, which should result in a lower error than
assuming 100 % correlation.
Regarding the quality of the dynamic relative to the geometric coincidence
criteria, the coincidence error estimated by models is larger for dynamic
coincidences by about 0.3 ppm, as seen in Table 2. However, the coincidence
error (0.3 to 0.4 ppm for geometric criteria and 0.6 to 0.7 ppm for dynamic
criteria) is not the dominant error. As seen in Fig. 5, the dynamic
coincidence criteria average 0.1 ppm higher error for unaveraged satellite
comparisons. This is small compared to a total error of 2.0 and 2.2 ppm,
respectively, for stations in the Northern Hemisphere. With maximum
averaging, as seen in Fig. 5, the errors are lower for dynamic vs. geometric because
the dynamic criteria finds more observations to average. Figure 6 shows that at Lamont the average difference between geometric and
dynamic observations is 0.4 ppm for unaveraged satellite observations,
which is higher than average. This error difference reduces to less than 0.2 ppm
when all available observations are averaged, also seen in Fig. 6.
While coincidence error is an important error source, it is not the dominant
error source. Although dynamic coincidence criteria allow the inclusion of
more stations in analyses because of the larger number of coincidences,
comparisons to geometric coincidence results are done when possible. Biases
at individual stations have a year-to-year variability of ∼ 0.3 ppm, with biases larger than the TCCON predicted bias uncertainty of 0.4 ppm
at many stations. Using fitting software, we find that GOSAT and CT2013b
underpredict the seasonal cycle amplitude in the Northern Hemisphere between
46 and 53∘ N, MACC overpredicts in 26–37∘ N, and CT2013b underpredicts the seasonal
cycle amplitude in the Southern Hemisphere. The seasonal cycle phase
indicates whether a data set or model lags another data set in time. We find
that the GOSAT phase improves substantially over the prior and the SCIAMACHY
retrieved phase improves substantially for two of seven sites. The models
reproduce the measured seasonal cycle phase well except for at Lauder_125HR
(CT2013b) and Darwin (MACC). We compare the variability within 1 day
between TCCON and models in JJA; there is correlation between 0.2 and 0.8 in
the NH, with models showing 10 to 50 % the variability of TCCON at
different stations and CT2013b showing more variability than MACC. This
paper highlights findings that provide inputs to estimate flux errors in
model assimilations, and places where models and satellites need further
investigation, e.g., the SH for models and 45–67∘ N for GOSAT and CT2013b.
We focus on validating aspects of model and satellite data which may be
important for accurate flux estimates and CO2 assimilation, including
accurate error estimates, overall biases, biases by season and latitude,
impact of coincidence criteria, validation of seasonal cycle phase and
amplitude, yearly growth, and daily variability. Our methodologies can be
used to correct known biases and data deficiencies (e.g., Basu et al. (2013)
accounted for global land/sea biases; Nassar et al. (2011) corrected for
hemispheric gradients). Alternatively, biases can be mitigated through data
assimilation, such as the inversion method of Reuter et al. (2014) which is
insensitive to seasonal and regional biases outside a targeted region. The
bias evaluation performed in the present study has been restricted to
various satellite and model CO2 data products. To quantify the
importance of each bias (seasonal biases, location-dependent biases,
seasonal cycle differences, seasonal cycle phase differences, and diurnal
cycle differences) on carbon flux estimates requires detailed Observing
System Simulation Experiments (OSSEs). For example, Kulawik et al. (2013)
tested the effect of a NH bias of 0.3–0.5 ppm in JJA, finding flux biases
were not insignificant, and were comparable to flux updates from GOSAT in
some regions.
Biases vary by station (See Fig. 3); these station-dependent biases have a
standard deviation of ∼ 0.3 ppm from year to year. Biases also
vary by season as seen in Figs. 7–8. Seasonal biases affect the seasonal
cycle amplitudes, which are important for biospheric flux attribution. All
sets show the same general patterns for the different latitude bands (SH,
28–37, 46–53, 67–79∘ N). The statistically significant differences are (1) a
low bias in CT2013b seasonal cycle amplitude of 0.2 ± 0.1 ppm for
45–53∘ N caused by high values in JJA, (2) a high bias in the MACC seasonal
cycle amplitude for 46–53∘ N of 0.6 ppm, (3) a low bias in the CT2013b
Southern Hemisphere seasonal cycle amplitude, and (4) a low bias in the
GOSAT seasonal cycle amplitude of 0.4 ± 0.1 ppm for 46–53∘ N, also seen
in Lindqvist et al. (2015), caused by a combination of higher values in JJA
and lower values in MAM vs. TCCON. A preliminary study of how a seasonal
bias in JJA in GOSAT of 0.5 ppm in the NH affect fluxes using a global
assimilation showed that the effect was not minor (Kulawik et al., 2013).
The seasonal cycle phase is a sensitive indicator of seasonally dependent
biases in satellite data as well as issues with model fluxes or transport
errors. The GOSAT root mean square (RMS) phase difference vs. TCCON across all sites is 16.1 days for the prior; this improves to 6.8 days for the GOSAT retrieved
XCO2. The SCIAMACHY RMS phase difference vs. TCCON across all sites is
16.4 days for the prior; this improves to 13.2 days for the SCIAMACHY
retrieved XCO2, reflecting the fact that SCIAMACHY data significantly
improved the seasonal cycle phase at just two of the seven TCCON
sites.
Model comparisons to TCCON are much less noisy as there are many more
matches. Most NH stations show the expected seasonal drop-off (e.g., see Fig. 11), with the peak correlation near 0 days, and an additional spike within
±3 days indicating the capture of synoptic variability. Stations that
showed phase differences larger than 10 days are Darwin (Macc only), and
Lauder_125HR (CT2013b only), as well as all of the locally influenced
stations.
In comparing CO2 diurnal variability between TCCON and models, both
models show up average 0.5 correlation to the TCCON CO2 change between
afternoon and morning for select stations in the NH. This is on the order of
about 2/3 of the maximum possible correlation given the error vs. variability (except the locally influenced JPL2011, Bremen, and Four Corners
which had little correlation and no slope). The amplitude of the variability
is higher in TCCON vs. the models, with CT2013b closer to TCCON than
MACC. However, note that TCCON daily variability has not been validated.
Diurnal pattern will not be constrained by satellite observations, except as
preserved in transported air coincident with satellite measurements
downwind, and therefore may be important to independently verify the diurnal
cycle in models to ensure accurate flux attribution. The importance of the
diurnal cycle on flux estimates needs to be tested.
Spatial and seasonal-dependent biases are obstacles to accurate and better
resolved CO2 flux estimates. This paper highlights findings that
provide inputs to estimate flux errors in model assimilations, and places
where models and satellites need additional validation or improvement. Some
of the issues which need further investigation are the GOSAT and CT2013b
seasonal cycles in the 46–53∘ N latitude range (which are 0.4 and 0.2 ppm smaller
than TCCON, respectively), the MACC seasonal cycle in 28–37∘ N latitude range
(which is 0.6 ppm larger than TCCON), seasonal cycle amplitude and phase
differences at SH stations, differences in the diurnal cycle amplitude
between models and TCCON, and high biases for GOSAT and SCIAMACHY north of
67∘ N.
Coincidence criteria, data versions, and terminology used in our
analysis.
Coincidence criteriaGeometric5∘ in lat./long., ± 1 hDynamicalfrom Wunch et al. (2011b); considers free-tropospheric temperature, ± 10∘ lat, ± 30∘ long. and 5 days (see Sect. 3.1)Averaging: all averaging is done by station first and then averaging over station results Model choices: MACC is interpolated to TCCON latitude, longitude, and time; CT2013b special output is interpolated to TCCON time; model XCO2 uses TCCON surface pressure Data setsGOSATACOS-GOSAT version 3.5 with corrections and quality flags from the user's guideSCIAMACHYBESD-SCIAMACHY v02.00.08TCCONGGG2014 when available, GGG2012 data from Four Corners, Tsukuba, and Bremen with GGG2012 bias corrections applied as described in Sect. 2.3CarbonTrackerCT2013bMACCMACC v13.1TCCON and model errors by station
We estimate the 90 min average TCCON standard deviation error by
calculating the standard deviation of adjacent time points and model
standard deviation of CT2013b and MACC13.1 vs. TCCON by station. These
values are used to estimate theoretical maximum correlations for seasonal
cycle and diurnal correlations using Eq. (3). a denotes stations strongly
influenced by local effects and b Tsukuba has higher TCCON error.
These stations have been removed in the NH average (subset) row.
90 min average TCCON errors and model error vs. TCCON by station StationTCCON adjacentCT2013b standardMACC standardstandard deviation (ppm)deviation (ppm)deviation (ppm)Eureka0.20.80.9Ny Alesund0.80.80.8Sodankylä0.30.70.8Bialystok0.30.70.7Bremena0.51.31.4Karlsruhe0.40.90.9Orléans0.30.70.7Garmisch0.30.90.9Park Falls0.40.80.8Four Cornersa0.71.11.1Lamont0.30.80.8Tsukubab0.91.01.1JPL2007a0.61.31.1Izaña0.20.80.6NH average0.440.910.92NH average (subset)0.350.790.79Darwin0.40.91.0Wollongong0.40.80.7Lauder_120HR0.40.80.8Lauder_125HR0.30.60.5SH average0.380.780.75
Susan Kulawik set the direction of the research and did much of the
analysis. The following authors were involved with discussions of results
with specific knowledge in the listed areas: Debra Wunch, TCCON, Christopher
O'Dell, ACOS-GOSAT, Christian Frankenberg, ACOS-GOSAT, Maximilian Reuter,
BESD-SCIAMACHY, Tomohiro Oda, CarbonTracker, Frederic Chevallier, MACC,
Vanessa Sherlock, TCCON, Michael Buchwitz, BESD-SCIAMACHY, Greg Osterman,
ACOS-GOSAT, Charles Miller, CO2 data records. The TCCON data providers,
who also provide expertise regarding TCCON sites are Paul Wennberg, David
Griffith, Isamu Morino, Manvendra Dubey, Nicholas M. Deutscher, Justus
Notholt, Frank Hase, Thorsten Warneke, Ralf Sussmann, John Robinson,
Kimberly Strong, and Matthias Schneider. Joyce Wolf is a science programmer
and provided technical expertise.
Acknowledgements
Funded by NASA Roses ESDR-ERR 10/10-ESDRERR10-0031, “Estimation of biases
and errors of CO2 satellite observations from AIRS, GOSAT, SCIAMACHY, TES,
and OCO-2”.
Maximilian Reuter and Michael Buchwitz received funding from ESA (GHG-CCI
project of ESA's Climate Change Initiative) and from the University and
state of Bremen.
Information about all TCCON sites and their sources of funding can be found
on the TCCON website (https://tccon-wiki.caltech.edu/).
Manvendra K. Dubey is grateful for the funding for monitoring at Four Corners by LANL-LDRD, 20110081DR.
Frédéric Chevallier received funding from the EU H2020 Programme
(grant agreement no. 630080, MACC III). NCEP Reanalysis data used in dynamic
coincidence criteria were provided by the NOAA/OAR/ESRL PSD, Boulder, Colorado,
USA, from their website at http://www.esrl.noaa.gov/psd/.
Thanks to Andrew R. Jacobson for help with CarbonTracker.
Edited by: H. Worden
ReferencesBaker, D. F., Law, R. M., Gurney, K. R., Rayner, P., Peylin, P., Denning, A.
S., Bousquet, P., Bruhwiler, L., Chen, Y.-H., Ciais, P., Fung, I. Y.,
Heimann, M., John, J., Maki, T., Maksyutov, S., Masarie, K., Prather, M.,
Pak, B., Taguchi, S., and Zhu, Z. : TransCom 3 inversion intercomparison: impact
of transport model errors on the interannual variability of regional CO2
fluxes, 1988–2003, Global Biogeochem. Cy.,20, GB1002, 10.1029/2004GB002439, 2006.Basu, S., Houweling, S., Peters, W., Sweeney, C., Machida, T., Maksyutov,
S., Patra, P. K., Saito, R., Chevallier, F., Niwa, Y., Matsueda, H., and Sawa,
Y.: J. Geophys. Res., 116, 1–14, 10.1029/2011JD016124, 2011.Basu, S., Guerlet, S., Butz, A., Houweling, S., Hasekamp, O., Aben, I.,
Krummel, P., Steele, P., Langenfelds, R., Torn, M., Biraud, S., Stephens, B.,
Andrews, A., and Worthy, D.: Global CO2 fluxes estimated from GOSAT
retrievals of total column CO2 , Atmos. Chem. Phys., 13, 8695–8717,
10.5194/acp-13-8695-2013, 2013.Bergamaschi, P., Bräunlich, M., Marik, T., and Brenninkmeijer, C. A. M.:
Measurements of the carbon and hydrogen isotopes of atmospheric methane at
Izaña, Tenerife: Seasonal cycles and synoptic-scale variations, J. Geophys.
Res., 105, 14531–14546, 10.1029/1999JD901176, 2000.Butz, A., Guerlet, S., Jacob, D. J., Schepers, D., Galli, A., Aben, I.,
Frankenberg, C., Hartmann, J.-M., Tran, H., Kuze, A., Keppel-Aleks, G.,
Toon, G. C., Wunch, D., Wennberg, P. O., Deutscher, N. M., Griffith, D. W.
T., Macatangay, R., Messerschmidt, J., Notholt, J., and Warneke, T.:
Geophys. Res. Lett., 38, 2–7, 10.1029/2011GL047888, 2011.
Canadell, J. G., Ciais, P., Gurney, K., Le Quéré, C., Piao, S.,
Raupach, M. R., and Sabine, C. L.: An international effort to quantify regional
carbon fluxes, EOS Trans. AGU, 92, 81–82, 2011.Chevallier, F. and O'Dell, C. W.: Error statistics of Bayesian CO2 flux inversion schemes as seen from GOSAT, Geophys. Res. Lett., 40, 1252–1256, 10.1002/grl.50228,
2013.Chevallier, F., Ciais, P., Conway, T. J., Aalto, T., Anderson, B. E.,
Bousquet, P., Brunke, E. G., Ciattaglia, L., Esaki, Y., Fröhlich, M.,
Gomez, A. J., Gomez-Pelaez, A. J., Haszpra, L., Krummel, P., Langenfelds,
R., Leuenberger, M., Machida, T., Maignan, F., Matsueda, H., Morguí, J.
A., Mukai, H., Nakazawa, T., Peylin, P., Ramonet, M., Rivier, L., Sawa, Y.,
Schmidt, M., Steele, P., Vay, S. A., Vermeulen, A. T., Wofsy, S., and Worthy, D.: CO2
surface fluxes at grid point scale estimated from a global 21-year
reanalysis of atmospheric measurements, J. Geophys.
Res., 115, D21307, 10.1029/2010JD013887, 2010.Chevallier, F., Deutscher, N. M., Conway, T. J., Ciais, P., Ciattaglia, L.,
Dohe, S., Fröhlich, M., Gomez-Pelaez, A. J., Griffith, D. W. T., Hase,
F., Haszpra, L., Krummel, P., Kyrö, E., Labuschagne, C., Langenfelds,
R., Machida, T., Maignan, F., Matsueda, H., Morino, I., Notholt, J.,
Ramonet, M., Sawa, Y., Schmidt, M., Sherlock, V., Steele, P., Strong, K.,
Sussmann, R., Wennberg, P. O., Wofsy, S. C., Worthy, D., Wunch, D., and Zimnoch,
M.: Geophys. Res. Lett., 38, 1–5, 10.1029/2011GL049899,
2011.Chevallier, F., Palmer, P. I., Feng, L., Boesch, H., O'Dell, C. W.,
and Bousquet, P.: Toward robust and consistent regional CO2 flux estimates from in situ and space-borne measurements of atmospheric CO2, Geophys. Res.
Lett., 41, 1065–1070, 10.1002/2013GL058772, 2014.Ciais, P., Dolman, A. J., Bombelli, A., Duren, R., Peregon, A., Rayner, P.
J., Miller, C., Gobron, N., Kinderman, G., Marland, G., Gruber, N.,
Chevallier, F., Andres, R. J., Balsamo, G., Bopp, L., Bréon, F.-M.,
Broquet, G., Dargaville, R., Battin, T. J., Borges, A., Bovensmann, H.,
Buchwitz, M., Butler, J., Canadell, J. G., Cook, R. B., DeFries, R., Engelen,
R., Gurney, K. R., Heinze, C., Heimann, M., Held, A., Henry, M., Law, B.,
Luyssaert, S., Miller, J., Moriyama, T., Moulin, C., Myneni, R. B., Nussli,
C., Obersteiner, M., Ojima, D., Pan, Y., Paris, J.-D., Piao, S. L., Poulter,
B., Plummer, S., Quegan, S., Raymond, P., Reichstein, M., Rivier, L., Sabine,
C., Schimel, D., Tarasova, O., Valentini, R., Wang, R., van der Werf, G.,
Wickland, D., Williams, M., and Zehner, C.: Current systematic carbon-cycle
observations and the need for implementing a policy-relevant carbon observing
system, Biogeosciences, 11, 3547–3602, 10.5194/bg-11-3547-2014, 2014.Crisp, D., Fisher, B. M., O'Dell, C., Frankenberg, C., Basilio, R., Bösch,
H., Brown, L. R., Castano, R., Connor, B., Deutscher, N. M., Eldering, A.,
Griffith, D., Gunson, M., Kuze, A., Mandrake, L., McDuffie, J.,
Messerschmidt, J., Miller, C. E., Morino, I., Natraj, V., Notholt, J.,
O'Brien, D. M., Oyafuso, F., Polonsky, I., Robinson, J., Salawitch, R.,
Sherlock, V., Smyth, M., Suto, H., Taylor, T. E., Thompson, D. R., Wennberg,
P. O., Wunch, D., and Yung, Y. L.: The ACOS CO2 retrieval algorithm – Part
II: Global XCO2 data characterization, Atmos. Meas. Tech., 5, 687–707,
10.5194/amt-5-687-2012, 2012.Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, P.,
Kobayashi, S., Andrae, U., Balmaseda, M. A., Balsamo, G., Bauer, P.,
Bechtold, P., Beljaars, A. C. M., van de Berg, L., Bidlot, J., Bormann, N.,
Delsol, C., Dragani, R., Fuentes, M., Geer, A. J., Haimberger, L., Healy, S.
B., Hersbach, H., Hólm, E. V., Isaksen, L., Kållberg, P.,
Köhler, M., Matricardi, M., McNally, A. P., Monge-Sanz, B. M.,
Morcrette, J.-J., Park, B.-K., Peubey, C., de Rosnay, P., Tavolato, C.,
Thépaut, J.-N., and Vitart, F.: The ERA-Interim reanalysis:
configuration and performance of the data assimilation system, Q. J. Roy.
Meteor. Soc., 137, 553–597, 10.1002/qj.828, 2011.Deng, F., Jones, D. B. A., Henze, D. K., Bousserez, N., Bowman, K. W.,
Fisher, J. B., Nassar, R., O'Dell, C., Wunch, D., Wennberg, P. O., Kort, E.
A., Wofsy, S. C., Blumenstock, T., Deutscher, N. M., Griffith, D. W. T.,
Hase, F., Heikkinen, P., Sherlock, V., Strong, K., Sussmann, R., and Warneke,
T.: Inferring regional sources and sinks of atmospheric CO2 from GOSAT XCO2
data, Atmos. Chem. Phys., 14, 3703–3727, 10.5194/acp-14-3703-2014,
2014.Dils, B., Buchwitz, M., Reuter, M., Schneising, O., Boesch, H., Parker, R.,
Guerlet, S., Aben, I., Blumenstock, T., Burrows, J. P., Butz, A., Deutscher,
N. M., Frankenberg, C., Hase, F., Hasekamp, O. P., Heymann, J., De Mazière,
M., Notholt, J., Sussmann, R., Warneke, T., Griffith, D., Sherlock, V., and
Wunch, D.: The Greenhouse Gas Climate Change Initiative (GHG-CCI):
comparative validation of GHG-CCI SCIAMACHY/ENVISAT and TANSO-FTS/GOSAT
CO2
and CH4 retrieval algorithm products with measurements from the TCCON, Atmos.
Meas. Tech., 7, 1723–1744, 10.5194/amt-7-1723-2014, 2014.
Efron, B.: Bootstrap Methods – Another Look at the Jackknife, Ann. Stat., 7, 1–26,
1979.
Fisher, R. A.: Frequency distribution of the values of the
correlation coefficient in samples of an indefinitely large population,
Biometrika, 10, 507–521, 1915.
Fisher, R. A.: On the probable error of a coefficient of correlation
deduced from a small sample, Metron, 1, 3–32, 1921.Friedlingstein, P., Cox, P., Betts, R., Bopp, L., von Bloh, W., Brovkin, V.,
Cadule, P., Doney, S., Eby, M., Fung, I., Bala, G., John, J., Jones, C.,
Joos, F., Kato, T., Kawamiya, M., Knorr, W., Lindsay, K., Matthews, H. D.,
Raddatz, T., Rayner, P., Reick, C., Roeckner, E., Schnitzler, K.-G., Schnur,
R., Strassmann, K., Weaver, A. J., Yoshikawa, C., and Zeng, N.:
Climate-Carbon Cycle Feedback Analysis: Results from the C4MIP Model
Intercomparison, J. Climate, 19, 3337–3353, 10.1175/JCLI3800.1, 2006.Guerlet, S., Butz, A., Schepers, D., Basu, S., Hasekamp, O. P., Kuze, A.,
Yokota, T., Blavier, J.-F., Deutscher, N. M., Griffith, D. W. T., Hase, F.,
Kyro, E., Morino, I., Sherlock, V., Sussmann, R., Galli, A., and Aben, I.: Impact of aerosols and thin cirrus on retrieving and
validating XCO2 from GOSAT shortwave infrared measurements, J. Geophys. Res.-Atmos., 118, 4887–4905, 10.1002/jgrd.50332,
2013.Houghton, R. A.: Balancing the Global Carbon Budget, Annu. Rev. Earth Pl. Sc., 35, 313–347, 10.1146/annurev.earth.35.031306.140057, 2007.Hourdin, F., Musat, I., Bony, S., Braconnot, P., Codron, F., Dufresne, J.-L.,
Fairhead, L., Filiberti, M.-A., Friedlingstein, P., Grandpeix, J., Krinner,
G., LeVan, P., Li, Z.-X., and Lott, F.: The LMDZ4 general circulation model:
climate performance and sensitivity to parametrized physics with emphasis on
tropical convection, Clim. Dynam., 27, 787–813, 10.1007/s00382-006-0158-0, 2006.Hungershoefer, K., Breon, F.-M., Peylin, P., Chevallier, F., Rayner, P.,
Klonecki, A., Houweling, S., and Marshall, J.: Evaluation of various
observing systems for the global monitoring of CO2 surface fluxes, Atmos.
Chem. Phys., 10, 10503–10520, 10.5194/acp-10-10503-2010, 2010.
Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L.,
Iredell, M., Saha, S., White, G., Woollen, J., Zhu, Y., Chelliah, M.,
Ebisuzaki, W., Higgins, W., Janowiak, J., Mo, KC., Ropelewski, C., Wang, J.,
Leetmaa, A., Reynolds, R., Jenne, R., and Joseph, D.: The NCEP/NCAR 40-year
reanalysis project, B. Am. Meteorol. Soc., 77, 437–470, 1996.Keppel-Aleks, G., Wennberg, P. O., and Schneider, T.: Sources of variations
in total column carbon dioxide, Atmos. Chem. Phys., 11, 3581–3593,
10.5194/acp-11-3581-2011, 2011.Keppel-Aleks, G., Wennberg, P. O., Washenfelder, R. A., Wunch, D., Schneider,
T., Toon, G. C., Andres, R. J., Blavier, J.-F., Connor, B., Davis, K. J.,
Desai, A. R., Messerschmidt, J., Notholt, J., Roehl, C. M., Sherlock, V.,
Stephens, B. B., Vay, S. A., and Wofsy, S. C.: The imprint of surface fluxes
and transport on variations in total column carbon dioxide, Biogeosciences,
9, 875–891, 10.5194/bg-9-875-2012, 2012.Krol, M., Houweling, S., Bregman, B., van den Broek, M., Segers, A., van
Velthoven, P., Peters, W., Dentener, F., and Bergamaschi, P.: The two-way
nested global chemistry-transport zoom model TM5: algorithm and applications,
Atmos. Chem. Phys., 5, 417–432, 10.5194/acp-5-417-2005, 2005.
Kulawik, S. S., Baker, D., Wunch, D., Jacobson, A. R., Oda, T., Frankenberg,
C., Reuter, M., Buchwitz, M., Osterman, G., O'Dell, C., Wennberg, P.,
Olsen, E., Griffith, D., Sherlock, V., Deutscher, N. M., Notholt, J.,
Warneke, T., Morino, I., Sussmann, R., and Wolf, J.: Spatial and Temporal Biases
in ACOS-GOSAT and BESD-SCIAMACHY Carbon Dioxide and Effects on Flux
Estimates, American Geophysical Union Fall Meeting, 9–13 December 2013, San Francisco, CA, USA, 2013.
Kuze, A., Suto, H., Nakajima, M., and Hamazaki, T.: Thermal and near
infrared sensor for carbon observation Fourier-transform spectrometer on the
Greenhouse Gases Observing Satellite for greenhouse gases monitoring, Appl.
Opt., 48, 6716–6733, 2009.Lindenmaier, R., Dubeya, M. K., Hendersonb, B. G., Butterfielda, Z. T., Hermanc,
J. R., Rahna, T., and Leea, S.-H.: Multiscale observations of CO2,
13CO2, and pollutants at Four Corners for emission verification and
attribution, P. Natl. Acad. Sci. USA, 111, 8386–8391, 10.1073/pnas.1321883111,
2014.Lindqvist, H., O'Dell, C. W., Basu, S., Boesch, H., Chevallier, F.,
Deutscher, N., Feng, L., Fisher, B., Hase, F., Inoue, M., Kivi, R., Morino,
I., Palmer, P. I., Parker, R., Schneider, M., Sussmann, R., and Yoshida, Y.:
Does GOSAT capture the true seasonal cycle of carbon dioxide?, Atmos. Chem.
Phys., 15, 13023–13040, 10.5194/acp-15-13023-2015, 2015.Messerschmidt, J., Geibel, M. C., Blumenstock, T., Chen, H., Deutscher, N.
M., Engel, A., Feist, D. G., Gerbig, C., Gisi, M., Hase, F., Katrynski, K.,
Kolle, O., Lavric, J. V., Notholt, J., Palm, M., Ramonet, M., Rettinger, M.,
Schmidt, M., Sussmann, R., Toon, G. C., Truong, F., Warneke, T., Wennberg, P.
O., Wunch, D., and Xueref-Remy, I.: Calibration of TCCON column-averaged CO2:
the first aircraft campaign over European TCCON sites, Atmos. Chem. Phys.,
11, 10765–10777, doi:10.5194/acp-11-10765-2011, 2011.Miller, C. E., Crisp, D., DeCola, P. L., Olsen, S. C., Randerson, J. T.,
Michalak, A. M., Alkhaled, A., Rayner, P., Jacob, D. J., Suntharalingam, P.,
Jones, D. B. A., Denning, A. S., Nicholls, M. E., Doney, S. C., Pawson, S.,
Boesch, H., Connor, B. J., Fung, I. Y., O'Brien, D., Salawitch, R. J.,
Sander, S. P., Sen, B., Tans, P., Toon, G. C., Wennberg, P. O., Wofsy, S.
C., Yung, Y. L., and Law, R. M.: Precision requirements for space-based
XCO2 data, J. Geophys. Res., 112, D10314 ,10.1029/2006JD007659, 2007.Morino, I., Uchino, O., Inoue, M., Yoshida, Y., Yokota, T., Wennberg, P. O.,
Toon, G. C., Wunch, D., Roehl, C. M., Notholt, J., Warneke, T.,
Messerschmidt, J., Griffith, D. W. T., Deutscher, N. M., Sherlock, V.,
Connor, B., Robinson, J., Sussmann, R., and Rettinger, M.: Preliminary
validation of column-averaged volume mixing ratios of carbon dioxide and
methane retrieved from GOSAT short-wavelength infrared spectra, Atmos. Meas.
Tech., 4, 1061–1076, 10.5194/amt-4-1061-2011, 2011.Nassar, R., Jones, D. B. A., Kulawik, S. S., Worden, J. R., Bowman, K. W.,
Andres, R. J., Suntharalingam, P., Chen, J. M., Brenninkmeijer, C. A. M.,
Schuck, T. J., Conway, T. J., and Worthy, D. E.: Inverse modeling of CO2
sources and sinks using satellite observations of CO2 from TES and surface
flask measurements, Atmos. Chem. Phys., 11, 6029–6047,
10.5194/acp-11-6029-2011, 2011.Nguyen, H., Osterman, G., Wunch, D., O'Dell, C., Mandrake, L., Wennberg, P.,
Fisher, B., and Castano, R.: A method for colocating satellite XCO2
data to ground-based data and its application to ACOS-GOSAT and TCCON, Atmos.
Meas. Tech., 7, 2631–2644, 10.5194/amt-7-2631-2014, 2014.O'Dell, C. W., Connor, B., Bösch, H., O'Brien, D., Frankenberg, C., Castano,
R., Christi, M., Crisp, D., Eldering, A., Fisher, B., Gunson, M., McDuffie,
J., Miller, C. E., Natraj, V., Oyafuso, F., Polonsky, I., Smyth, M., Taylor,
T., Toon, G. C., Wennberg, P. O., and Wunch, D.: Corrigendum to “The ACOS
CO2
retrieval algorithm – Part 1: Description and validation against synthetic
observations” published in Atmos. Meas. Tech., 5, 99–121, 2012, Atmos. Meas.
Tech., 5, 193–193, 10.5194/amt-5-193-2012, 2012.Olsen, S. C. and Randerson, J. T.: Differences between surface and column
atmospheric CO2 and implications for carbon cycle research, J. Geophys. Res.-Atmos., 109,
D02301, 10.1029/2003JD003968, 2004.Oshchepkov, S., Bril, A., Yokota, T., Morino, I., Yoshida, Y., Matsunaga,
T., Belikov, D., Wunch, D., Wennberg, P., Toon, G., O'Dell, C., Butz, A.,
Guerlet, S., Cogan, A., Boesch, H., Eguchi, N., Deutscher, N., Griffith, D.,
Macatangay, R., Notholt, J., Sussmann, R., Rettinger, M., Sherlock, V.,
Robinson, J., Kyrö, E., Heikkinen, P., Feist, D. G., Nagahama, T.,
Kadygrov, N., Maksyutov, S., Uchino, O., and Watanabe, H.: J. Geophys. Res., 117, 1–18, 10.1029/2012JD017505,
2012.
Peters, W., Jacobson, A. R., Sweeney, C., Andrews, A. E., Conway, T. J.,
Masarie, K., Miller, J. B., Bruhwiler, L. M. P., Petron, G., Hirsch, A. I.,
Worthy, D. E. J., van der Werf, G. R., Randerson, J. T., Wennberg, P. O.,
Krol, M. C., and Tans, P. P.: An atmospheric perspective on North American
carbon dioxide exchange: CarbonTracker, P. Natl. Acad. Sci. USA, 104,
18925–18930, 2007.Peylin, P., Law, R. M., Gurney, K. R., Chevallier, F., Jacobson, A. R., Maki,
T., Niwa, Y., Patra, P. K., Peters, W., Rayner, P. J., Rödenbeck, C., van
der Laan-Luijkx, I. T., and Zhang, X.: Global atmospheric carbon budget:
results from an ensemble of atmospheric CO2 inversions, Biogeosciences,
10, 6699–6720, 10.5194/bg-10-6699-2013, 2013.Prather, M. J., Zhu, X., Strahan, S. E., Steenrod, S. D., and Rodriguez, J.
M.: Quantifying errors in trace species transport modeling, P. Natl.
Acad. Sci. USA, 105, 19617–19621, 10.1073/pnas.0806541106, 2008.Reuter, M., Buchwitz, M., Schneising, O., Heymann, J., Bovensmann, H., and
Burrows, J. P.: A method for improved SCIAMACHY CO2 retrieval in the presence
of optically thin clouds, Atmos. Meas. Tech., 3, 209–232,
10.5194/amt-3-209-2010, 2010.Reuter, M., Bovensmann, H., Buchwitz, M., Burrows, J. P., Connor, B. J.,
Deutscher, N. M., Griffith, D. W. T., Heymann, J., Keppel-Aleks, G.,
Messerschmidt, J., Notholt, J., Petri, C., Robinson, J., Schneising, O.,
Sherlock, V., Velazco, V., Warneke, T., Wennberg, P. O., Wunch, D.:
Retrieval of atmospheric CO2 with enhanced accuracy and precision from
SCIAMACHY: Validation with FTS measurements and comparison with model
results, J. Geophys. Res.-Atmos., 116, D04301, 10.1029/2010JD015047,
2011.Reuter, M., Buchwitz, M., Schneising, O., Hase, F., Heymann, J., Guerlet, S.,
Cogan, A. J., Bovensmann, H., and Burrows, J. P.: A simple empirical model
estimating atmospheric CO2 background concentrations, Atmos. Meas. Tech., 5,
1349–1357, 10.5194/amt-5-1349-2012, 2012a.Reuter, M., Bovensmann, H., Buchwitz, M., Burrows, J. P., Heymann, J.,
Schneising, O., Schnisa, H., Velazco, V. : Algorithm Theoretical Basis
Document (ATBD) Bremen Optimal Estimation DOAS (BESD) Version 1, available
at: http://www.iup.uni-bremen.de/~mreuter/atbd_besd_v1.pdf (last access: 23 February 2016), 2012b.Reuter, M., Bösch, H., Bovensmann, H., Bril, A., Buchwitz, M., Butz, A.,
Burrows, J. P., O'Dell, C. W., Guerlet, S., Hasekamp, O., Heymann, J.,
Kikuchi, N., Oshchepkov, S., Parker, R., Pfeifer, S., Schneising, O., Yokota,
T., and Yoshida, Y.: A joint effort to deliver satellite retrieved
atmospheric CO2 concentrations for surface flux inversions: the ensemble
median algorithm EMMA, Atmos. Chem. Phys., 13, 1771–1780,
10.5194/acp-13-1771-2013, 2013.Reuter, M., Buchwitz, M., Hilker, M., Heymann, J., Schneising, O., Pillai,
D., Bovensmann, H., Burrows, J. P., Bösch, H., Parker, R., Butz, A.,
Hasekamp, O., O'Dell, C. W., Yoshida, Y., Gerbig, C., Nehrkorn, T.,
Deutscher, N. M., Warneke, T., Notholt, J., Hase, F., Kivi, R., Sussmann, R.,
Machida, T., Matsueda, H., and Sawa, Y.: Satellite-inferred European carbon
sink larger than expected, Atmos. Chem. Phys., 14, 13739–13753,
10.5194/acp-14-13739-2014, 2014.Rubin, D. B.: The Bayesian Bootstrap, Ann. Stat. 9, 130–134,
10.1214/aos/1176345338, 1981.Saito, R., Patra, P. K., Deutscher, N., Wunch, D., Ishijima, K., Sherlock,
V., Blumenstock, T., Dohe, S., Griffith, D., Hase, F., Heikkinen, P., Kyrö,
E., Macatangay, R., Mendonca, J., Messerschmidt, J., Morino, I., Notholt, J.,
Rettinger, M., Strong, K., Sussmann, R., and Warneke, T.: Technical Note:
Latitude-time variations of atmospheric column-average dry air mole fractions
of CO2, CH4 and N2O, Atmos. Chem. Phys., 12, 7767–7777,
10.5194/acp-12-7767-2012, 2012.
Sancho, P., De La Cruz, J., Diaz, A., Martin, F., Hernandez, E., Valero, F.,
and Albarran, B.: A five-year climatology of back-trajectories from the Izaña
baseline station, Tenerife, Canary Islands, Atmos. Environ., 26, 1081–1096,
1992.Schneising, O., Bergamaschi, P., Bovensmann, H., Buchwitz, M., Burrows, J.
P., Deutscher, N. M., Griffith, D. W. T., Heymann, J., Macatangay, R.,
Messerschmidt, J., Notholt, J., Rettinger, M., Reuter, M., Sussmann, R.,
Velazco, V. A., Warneke, T., Wennberg, P. O., and Wunch, D.: Atmospheric
greenhouse gases retrieved from SCIAMACHY: comparison to ground-based FTS
measurements and model results, Atmos. Chem. Phys., 12, 1527–1540,
10.5194/acp-12-1527-2012, 2012.Stephens, B. B., Gurney, K. R., Tans, P. P., Sweeney, C., Peters, W., Bruhwiler, L., Ciais, P.,
Ramonet, M., Bousquet, P., Nakazawa, T., Aoki, S., Machida, T., Inoue, G.,
Vinnichenko, N., Lloyd, J., Jordan, A., Heimann, M., Shibistova, O., Langenfelds, R. L., Steele, L. P.,
Francey, R. J., Denning, A. S.: Weak northern and strong tropical land carbon
uptake from vertical profiles of atmospheric CO2, Science, 22, 1732–1735, 2007.Thoning, K. W., Tans, P. P., and Komhyr, W. D.: Atmospheric Carbon
Dioxide at Mauna Loa Observatory 2. Analysis of the NOAA GMCC Data,
1974–1985, J. Geophys. Res., 94, 8549–8565,
10.1029/JD094iD06p08549, 1989.Wunch, D., Toon, G. C., Wennberg, P. O., Wofsy, S. C., Stephens, B. B.,
Fischer, M. L., Uchino, O., Abshire, J. B., Bernath, P., Biraud, S. C.,
Blavier, J.-F. L., Boone, C., Bowman, K. P., Browell, E. V., Campos, T.,
Connor, B. J., Daube, B. C., Deutscher, N. M., Diao, M., Elkins, J. W.,
Gerbig, C., Gottlieb, E., Griffith, D. W. T., Hurst, D. F., Jiménez, R.,
Keppel-Aleks, G., Kort, E. A., Macatangay, R., Machida, T., Matsueda, H.,
Moore, F., Morino, I., Park, S., Robinson, J., Roehl, C. M., Sawa, Y.,
Sherlock, V., Sweeney, C., Tanaka, T., and Zondlo, M. A.: Calibration of the
Total Carbon Column Observing Network using aircraft profile data, Atmos.
Meas. Tech., 3, 1351–1362, 10.5194/amt-3-1351-2010, 2010.Wunch, D., Toon, G. C., Blavier, J.-F. L., Washenfelder, R. A., Notholt, J.,
Connor, B. J., Griffith, D. W. T., Sherlock, V., and Wennberg, P. O.: The Total
Carbon Column Observing Network (TCCON), Philos. T. R. Soc.
A, 369, 2087–2112, 10.1098/rsta.2010.0240, 2011a.Wunch, D., Wennberg, P. O., Toon, G. C., Connor, B. J., Fisher, B., Osterman,
G. B., Frankenberg, C., Mandrake, L., O'Dell, C., Ahonen, P., Biraud, S. C.,
Castano, R., Cressie, N., Crisp, D., Deutscher, N. M., Eldering, A., Fisher,
M. L., Griffith, D. W. T., Gunson, M., Heikkinen, P., Keppel-Aleks, G., Kyrö,
E., Lindenmaier, R., Macatangay, R., Mendonca, J., Messerschmidt, J., Miller,
C. E., Morino, I., Notholt, J., Oyafuso, F. A., Rettinger, M., Robinson, J.,
Roehl, C. M., Salawitch, R. J., Sherlock, V., Strong, K., Sussmann, R.,
Tanaka, T., Thompson, D. R., Uchino, O., Warneke, T., and Wofsy, S. C.: A
method for evaluating bias in global measurements of CO2 total columns from
space, Atmos. Chem. Phys., 11, 12317–12337, 10.5194/acp-11-12317-2011,
2011b.Wunch, D., Toon, G. C., Sherlock, V., Deutscher, N. M, Liu, X., Feist, D. G.,
and Wennberg, P. O.: The Total Carbon Column Observing Network's GGG2014
Data Version. Carbon Dioxide Information Analysis Center, Oak Ridge National
Laboratory, Oak Ridge, Tennessee, USA, available at: 10.14291/tccon.ggg2014.documentation.R0/1221662, 2015.Yokota, T., Yoshida, Y., Eguchi, N., Ota, Y., Tanaka, T., Watanabe, H., and
Maksyutov, S.: Global Concentrations of CO2 and CH4 Retrieved from GOSAT: First Preliminary Results, Sola, 5, 160–163, 10.2151/sola.2009-041, 2009.