EGU Journal Logos (RGB) Open Access Open Access Open Access Advances in Annales Nonlinear Processes Geosciences Geophysicae in Geophysics Open Access Open Access Natural Hazards Natural Hazards and Earth System and Earth System Sciences Sciences Discussions Open Access Open Access Atmos. Chem. Phys., 13, 3205–3225, 2013 Atmospheric Atmospheric www.atmos-chem-phys.net/13/3205/2013/ doi:10.5194/acp-13-3205-2013 Chemistry Chemistry © Author(s) 2013. CC Attribution 3.0 License. and Physics and Physics Discussions Open Access Open Access Atmospheric Atmospheric Measurement Measurement Techniques Techniques Discussions Open Access Comparison of improved Aura Tropospheric Emission Spectrometer Open Access CO2 with HIPPO and SGP aircraft profile measurements Biogeosciences Biogeosciences Discussions S. S. Kulawik1, J. R. Worden1, S. C. Wofsy2, S. C. Biraud3, R. Nassar4, D. B. A. Jones5, E. T. Olsen1, R. Jimenez6, S. Park7, G. W. Santoni2, B. C. Daube2, J. V. Pittman2, B. B. Stephens8, E. A. Kort1, G. B. Osterman1, and TES team Open Access 1Jet Propulsion Laboratory, California Institute of Technology, CA, USA Open Access 2 Abbott Lawrence Rotch Professor of Atmospheric and Environmental Chemistry School of EngineeringClimate and Applied Climate Science and Department of Earth and Planetary Science Harvard University, Cambridge, MA, USA of the Past 3Lawrence Berkeley National Laboratory, Berkeley, CA, USA of the Past 4Environment Canada, Toronto, Ontario, Canada Discussions 5Department of Physics, University of Toronto, Tonronto, Canada Open Access 6Air Quality Research Group, Department of Chemical and Environmental Engineering, Universidad Nacional de Colombia, Open Access Bogota, DC 111321, Colombia Earth System Earth System 7Department of Oceanography, College of Ecology and Environmental Science, Kyungpook National University, Daegu, Dynamics South Korea Dynamics Discussions 8National Center for Atmospheric Research, Boulder, CO, USA Open Access Correspondence to: S. S. Kulawik ([email protected]) Geoscientific Geoscientific Open Access Received: 18 January 2012 – Published in Atmos. Chem. Phys. Discuss.: 29 February 2012Instrumentation Instrumentation Revised: 23 February 2013 – Accepted: 25 February 2013 – Published: 18 March 2013 Methods and Methods and Data Systems Data Systems Abstract. Thermal infrared radiances from the Tropospheric the sub-tropical TES CO estimates are lower than expected Discussions Open Access 2 Open Access Emission Spectrometer (TES) between 10 and 15 µm con- based on the calculated errors. Comparisons to land aircraft Geoscientific tain significant carbon dioxide (CO ) information, however profiles from the United StatesGeoscientific Southern Great Plains (SGP) 2 Model Development the CO2 signal must be separated from radiative interference Atmospheric RadiationModel Measurement Development (ARM) between 2005 from temperature, surface and cloud parameters, water, and and 2011 measured from the surface to 5 km to TES CO2 Discussions other trace gases. Validation requires data sources spanning show good agreement with an overall bias of −0.3 ppm to Open Access the range of TES CO2 sensitivity, which is approximately 2.5 0.1 ppm and standard deviations of 0.8 to 1.0 ppmOpen Access at differ- to 12 km with peak sensitivity at about 5 km and the range ent pressure levels. ExtendingHydrology the SGP aircraft and profiles above Hydrology and ◦ ◦ of TES observations in latitude (40 S to 40 N) and time 5 km using AIRS or CONTRAILEarth System measurements improves Earth System (2005–2011). We therefore characterize Tropospheric Emis- comparisons with TES. Comparisons to CarbonTracker (ver- sion Spectrometer (TES) CO2 version 5 biases and errors sion CT2011) show a persistent spatiallySciences dependent bias pat- Sciences through comparisons to ocean and land-based aircraft pro- tern and comparisons to SGP show a time-dependent bias of Discussions − 1 Open Access files and to the CarbonTracker assimilation system. We com- −0.2 ppm yr . We also find that the predictedsensitivityOpen Access of pare to ocean profiles from the first three Hiaper Pole-to-Pole the TES CO2 estimates is too high, which results from us- ◦ ◦ Ocean Science Observations (HIPPO) campaigns between 40 S and 40 N ing a multi-step retrievalOcean for CO 2Scienceand temperature. We find with measurements between the surface and 14 km and find that the averaging kernel in the TES product corrected by a Discussions that TES CO2 estimates capture the seasonal and latitudinal pressure-dependent factor accurately reflects the sensitivity gradients observed by HIPPO CO2 measurements. Actual er- of the TES CO2 product. Open Access rors range from 0.8–1.8 ppm, depending on the campaign and Open Access pressure level, and are approximately 1.6–2 times larger than Solid Earth the predicted errors. The bias of TES versus HIPPO is within Solid Earth 1 ppm for all pressures and datasets; however, several of Discussions Open Access Open Access Published by Copernicus Publications on behalf of the European Geosciences Union. The Cryosphere The Cryosphere Discussions 3206 S. S. Kulawik et al.: Comparison of improved Aura Tropospheric Emission Spectrometer CO2 1 Introduction 45◦ N. Based on the findings of Kulawik et al. (2010), up- dates were made to the retrieval strategy which significantly Over the past decade, measurements of carbon dioxide improved the accuracy of the TES CO2 retrieval over land (CO2) from space have become increasingly prevalent, with and changed the overall bias of TES CO2 from a 1.8 % to a CO2 measurements from SCIAMACHY, AIRS, TES, IASI, 0.3 % low bias. Results with the new version, processed with ACE, and GOSAT (e.g. Reuter et al., 2011; Chahine et the TES v5 production code, are shown in this paper. The al., 2008; Kulawik et al., 2010; Crevosier et al., 2009; TES CO2 netcdf “lite” products, on 14 pressure levels, were Foucher et al., 2011; Yoshida et al., 2011; Crisp et al., 2012; used for this analysis, available through links from the TES Butz et al., 2011). Robust calculation of errors in the CO2 website, at http://tesweb.jpl.nasa.gov/data/. Special runs, e.g. estimates is critical because errors in interferences can be processed with a constant initial guess and prior, were run larger than the expected variability. There is also a need to using the TES prototype code which has minor differences understand and validate biases and errors with great accu- from the v5 TES production code. These runs are used to racy for the data to be useful for estimating CO2 sources and assess the linearity of the retrieval system and validate the sinks. Consistent validation and intercomparisons for satel- vertical sensitivity of the CO2 estimates. lite data, necessary for combining or utilizing multiple satel- lite results, are challenging since the different products have different coverage, vertical sensitivity, and averaging strate- 2 Measurements gies (as summarized in Table 1). In this paper, we present 2.1 The TES instrument comparisons of TES CO2 to aircraft profile data from the HIPPO campaigns and from the Southern Great Plains ARM TES is on the Earth Observing System Aura (EOS-Aura) site to quantify errors, biases, and correlations between TES satellite and makes high spectral resolution nadir measure- and the validation data. The techniques and methods shown ments of thermal infrared emission (660 cm−1 to 2260 cm−1, in this paper are applicable to validation of other instruments with unapodized resolution of 0.06 cm−1, apodized resolu- with coincident aircraft profiles. tion of 0.1 cm−1). TES was launched in July 2004 in a sun- Multiple studies have estimated the precision and bias re- synchronous orbit at an altitude of 705 km with an equatorial quired to utilize atmospheric CO2 measurements for source crossing time of 13:38 (local mean solar time) and with a and sink estimates. Using simulated observations, Rayner repeat cycle of 16 days. In standard “global survey” mode, and O’Brien (2001) showed that satellite measurements of 2000–3000 observations are taken every other day (Beer, CO2 total column abundances with a precision of 2.5 ppm, 2006). CO is estimated for TES observations between 40◦ S ◦ × ◦ 2 averaged monthly on spatial scales of 8 10 , would offer and 45◦ N. In 2006, TES averaged 1570 “global survey” ob- more information on CO2 fluxes than can be obtained from servations per day. Of these, 743 per day are between 40◦ S the existing surface network. Houweling et al. (2004) also and 45◦ N, and 505 per day have cloud < 0.5 optical depth carried out simulations suggesting that latitude-dependent (OD) and are of good quality. There are additional targeted biases of less than 0.3 ppm are necessary for upper tropo- “special observations”, which are not used in this analysis as spheric CO2 data to be useful for estimating sources and they are less spatially and temporally uniform. TES global ◦ × ◦ sinks. Nassar et al. (2011) showed that 5 5 monthly- survey observations were consistently taken from late 2004 averaged TES observations at 500 hPa (about 5.5 km altitude) through June, 2011. For details on the TES instrument, see over ocean with mean errors of 4.7 ppm between 40◦ S and ◦ Beer (2006), and for information on the retrieval methods 40 N provided information that was complementary to flask see Bowman et al. (2006) and Kulawik et al. (2006, 2010). data and especially helped constrain tropical land regions. Nassar et al. (2011) mitigated latitude and seasonally depen- 2.2 HIPPO aircraft measurements dent biases of 1–2 ppm using 3 different correction meth- ods to estimate sources and sinks from combined TES mid- For validation of observations over oceans, we compare to tropospheric CO2 and surface flask CO2. Although the exact the HIPPO-1, HIPPO-2, and HIPPO-3 campaigns (Wofsy, magnitude of regional fluxes differed based on the bias cor- 2011; Daube et al., 2002; Kort et al., 2011) over the Pa- rection approach used, key results are generally robust within cific from 85◦ N to 67◦ S for January, 2009, November, 2009, the predicted errors.