Using airborne HIAPER Pole-to-Pole Observations (HIPPO) to evaluate model and remote sensing estimates of atmospheric carbon dioxide

Abstract

Abstract. In recent years, space-borne observations of atmospheric carbon dioxide (CO 2 ) have been increasingly used in global carbon-cycle studies. In order to obtain added value from space-borne measurements, they have to suffice stringent accuracy and precision requirements, with the latter being less crucial as it can be reduced by just enhanced sample size. Validation of CO 2 column-averaged dry air mole fractions (XCO 2 ) heavily relies on measurements of the Total Carbon Column Observing Network (TC-CON). Owing to the sparseness of the network and the requirements imposed on space-based measurements, independent additional validation is highly valuable. Here, we use observations from the High-Performance Instrumented Airborne Platform for Environmental Research (HIAPER) Pole-to-Pole Observations (HIPPO) flights from 01/2009 through 09/2011 to validate CO 2 measurements from satellites (Greenhouse Gases Observing Satellite -GOSAT, Thermal Emission Sounder -TES, Atmospheric Infrared Sounder -AIRS) and atmospheric inversion models (CarbonTracker CT2013B, Monitoring Atmospheric Composition and Climate (MACC) v13r1). We find that the atmospheric models capture the XCO 2 variability observed in HIPPO flights very well, with correlation coefficients (r 2 ) of 0.93 and 0.95 for CT2013B and MACC, respectively. Some larger discrepancies can be observed in profile comparisons at higher latitudes, in particular at 300 hPa during the peaks of either carbon uptake or release. These deviations can be up to 4 ppm and hint at misrepresentation of vertical transport.Comparisons with the GOSAT satellite are of comparable quality, with an r 2 of 0.85, a mean bias µ of −0.06 ppm, and a standard deviation σ of 0.45 ppm. TES exhibits an r 2 of 0.75, µ of 0.34 ppm, and σ of 1.13 ppm. For AIRS, we find an r 2 of 0.37, µ of 1.11 ppm, and σ of 1.46 ppm, with latitude-dependent biases. For these comparisons at least 6, 20, and 50 atmospheric soundings have been averaged for GOSAT, TES, and AIRS, respectively. Overall, we find that GOSAT soundings over the remote Pacific Ocean mostly meet the stringent accuracy requirements of about 0.5 ppm for space-based CO 2 observations.

Gradients of column CO<sub>2</sub> across North America from the NOAA Global Greenhouse Gas Reference Network

“…The NW, SW, SM, and SE inflow regions have significant delays of more than 1 month in the 2-5 km layer compared with the surface layer, which is likely due to the delayed phase of the seasonal cycle in well-mixed air coming from the oceans. Vertical homogeneity of air over ocean was observed during the HIAPER Pole-to-Pole Observations (HIPPO) aircraft campaign (Wofsy et al, 2011;Frankenberg et al, 2016). As air masses are transported further inland, we observe reduced discrepancies of the timing of CO 2 drawdown between surface and upper-layer air (2-5 km), which may be associated with the increased influence of the land surface in the mid-troposphere due to strong convection over land.…”

Lower-tropospheric CO<sub>2</sub> from near-infrared ACOS-GOSAT observations

“…Frankenberg et al (2016) were recently successful in evaluating satellite measurements of column CO 2 over ocean (including GOSAT) using HIPPO. In this paper, we look at comparisons between GOSAT and HIPPO 2-5 (HIPPO 1 occurs prior to GOSAT launch) using the HIPPO-identified profiles and the CO2_X field, based on 1 s data averaged to 10 s, from two (harmonized) sensors: CO2-QCLS and CO2-OMS.…”

“…All three criteria gave similar results overall, with different criteria performing better at different stations, but there was no clear overall best criteria. For HIPPO data, which mainly test latitude gradients over ocean, the dynamic coincidence approach was used following Frankenberg et al (2016). Different variations on the dynamic coincidence criteria were tested, e.g., using temperature comparisons at the surface, averaging from the surface to 2.5 km, or weighting temperature differences by the pressure weighting function.…”

“…CarbonTracker CT2015 (http://carbontracker.noaa.gov; Peters et al, 2007) is used to extend aircraft profiles from the stratosphere to the top of the atmosphere (similarly to in Frankenberg et al, 2016 andInoue et al, 2013) and to quantify colocation error (similarly to Kulawik et al, 2016).…”

Diagnostic methods for atmospheric inversions of long-lived greenhouse gases

“…A key difference, however, is that a number of validation data sets are available to help diagnose the retrieval algorithm (e.g., Osterman et al, 2011). These include, among others, observations from ground-based remote sensing instruments (that look up at the sun, rather than down at the Earth, e.g., Wunch et al, 2011) and targeted campaigns of in situ airborne observations that can capture CO 2 concentration variability within a portion of the atmospheric column (e.g., Tadić et al, 2014;Frankenberg et al, 2016). Unlike in the flux estimation problem, there is no direct conflict between using these additional measurements for validation/diagnosis versus using them to directly inform the solution of the inverse problem itself, as there is no clear mechanism by which these additional observations could be routinely incorporated within the core retrieval algorithm, although they can be used for additional empirical bias correction.…”