Tag: NASA Earthdata Cloud

An improved multi- year record of atmospheric CO₂ from NASA’s Orbiting Carbon Observatory instruments (OCO- 2 and OCO- 3), based on updated data-processing algorithms.

The Eos Research Spotlight (published 30 October 2026 by science writer Sarah Stanley) summarizes a paper by Taylor et al. in Earth and Space Science. Every day, OCO- 2 and OCO- 3 collect more than a million observations of sunlight reflecting off Earth’s surface.

An algorithm converts these into estimates of atmospheric carbon dioxide concentrations (specifically, the column- averaged dry- air mole fraction, XCO₂).

OCO-2 has operated as a free- flying satellite since its 2014 launch; OCO- 3 was installed on the International Space Station in 2019. Together they have provided extensive global coverage and supported studies of the carbon cycle’s responses to events such as wildfires, droughts, and El Niño.

Researchers applied version 11 (v11) of the Atmospheric Carbon Observations from Space (ACOS) Level 2 full- physics retrieval algorithm.

Improvements include:

• Better instrument calibration
• Updated handling of factors such as clouds, aerosols, and how sunlight reflects off the ocean
• Refined quality filtering and bias correction

The resulting record runs from September 2014 through (at least) August 2026-more than a decade. Compared with earlier algorithm versions, the new data generally agree more closely with highly accurate ground-based measurements from the Total Carbon Column Observing Network (TCCON).

OCO- 2 and OCO- 3 also show close agreement with each other (mean difference <0.2 ppm in collocated observations), despite different viewing angles, times, and orbital paths.

The upgraded dataset supports month- by- month (rather than only year- by-year) tracking of atmospheric CO₂ buildup. This enables nearer- real- time monitoring of how extreme events affect the carbon cycle.

The publicly available record is positioned as a benchmark for advancing understanding of the global carbon cycle and for assessing the effectiveness of climate policies.

Thomas E. Taylor et al., “The OCO-2 and OCO-3 ACOS Version 11 XCO₂ Data Products: A Benchmark Record for Space- Based Global Carbon Cycle Research,” Earth and Space Science (2026), https://doi.org/10.1029/2026EA005360.[](https://phys.org/news/2026-10-decade-carbon-dioxide-space.html)

The Eos article itself is available at: https://eos.org/research-spotlights/a-decade-of-carbon-dioxide-data-from-space-gets-an-upgrade.

It primarily improves observational constraints and validation for carbon- cycle and atmospheric inversion models, which in turn support and refine the carbon- cycle components of broader climate- Earth- system models.

NASA’s OCO- 2 (since 2014) and OCO- 3 (since 2019) measure the column-averaged dry- air mole fraction of CO₂ (XCO₂). The v11 algorithm updates (better calibration, handling of clouds- aerosols- and ocean- reflectance, quality filtering, and bias correction) produce a more consistent record from September 2014 onward that agrees more closely with ground- based TCCON measurements and shows reduced variance and less temporal divergence relative to atmospheric inversion models used as “truth proxies”. OCO- 2 and OCO- 3 also agree with each other to within a mean difference of <0.2 ppm.

These improvements affect modeling in several concrete ways:

• Stronger constraints on surface flux inversions
  Atmospheric inversion models (top- down approaches) assimilate XCO₂ data (alongside in- situ measurements) to estimate net CO₂ fluxes between the surface and atmosphere, i.e., the balance of anthropogenic emissions, land sinks, and ocean uptake. The cleaner, more consistent v11 data reduce systematic errors and provide better- characterized uncertainties. This leads to higher-fidelity estimates of regional- to- global fluxes, seasonal- to-interannual variability, and responses to events such as wildfires, droughts, and El Niño. An OCO- 2 v11 Model Intercomparison Project (MIP) is already underway to quantify the impact of the version change on flux estimates.
• Better evaluation and diagnosis of carbon- cycle processes in Earth- system models
  Climate and Earth- system models include interactive land and ocean carbon components that simulate photosynthesis, respiration, fires, ocean uptake, etc. The multi- year satellite record serves as an independent check on simulated atmospheric CO₂ growth rates, seasonal amplitudes, and spatial patterns. Prior OCO data have already highlighted discrepancies (e.g., many dynamic vegetation models underestimating the seasonal amplitude of the terrestrial carbon cycle). The upgraded record, with lower errors and month- by- month tracking capability, allows finer diagnosis of model strengths/weaknesses, especially for extreme- event responses and carbon- climate feedback.⁠
• Support for higher- resolution and near- real- time applications
  The ability to monitor atmospheric CO₂ accumulation on monthly rather than only annual timescales helps evaluate how models represent short-term variability and extreme events. Combined OCO- 2 and OCO- 3 sampling also improves coverage in some regions (though OCO- 3’s orbit introduces its own sampling challenges that inversions must handle).
• Indirect benefits for climate projections and policy assessment
  More accurate flux estimates and carbon- cycle understanding feed into assessments of the remaining carbon budget, the strength of natural sinks, and the effectiveness of mitigation policies. These, in turn, inform the scenarios and feedback parameterizations used in climate projections. The paper positions the v11 record as a “benchmark” for space- based global carbon-cycle research precisely because of these uses.

The data improve constraints and validation, not the underlying model physics directly. Differences between v10 and v11 XCO₂ will propagate into revised flux estimates, so earlier inversion results may need updating. OCO-3’s irregular sampling and limited high-latitude coverage still require careful treatment in inversions. Overall agreement with models and TCCON is improved but not perfect; residual biases remain at the level of tenths of a ppm in some cases.

In short, the upgrade delivers a more reliable observational backbone for carbon -cycle science. This tightens the link between satellite measurements, flux estimates, process understanding, and the carbon- cycle modules embedded in climate models, ultimately supporting more robust assessments of how the Earth system is responding to rising CO₂.

It affects climate projections mainly indirectly, by improving constraints on the carbon cycle, the processes that determine how much of emitted CO₂ stays in the atmosphere versus being absorbed by land and oceans.

This reduces uncertainty in carbon- climate feedback, which are a significant source of spread in future warming estimates for any given emissions pathway.

Climate projections (e.g., those from CMIP Earth – system models used in IPCC assessments) depend on:

• Emissions scenarios (SSPs or similar).
• Climate sensitivity.
• Carbon- cycle feedback: how warming, droughts, fires, and other changes alter natural sinks, potentially leaving more CO₂ in the air and amplifying warming.

The upgraded OCO- 2 and OCO- 3 ACOS v11 XCO₂ record (2014– present, with better calibration, lower systematic errors, closer agreement with ground measurements, and consistency between the two instruments) strengthens the observational foundation for the carbon- cycle side in these ways:

• Better flux estimates via atmospheric inversions
  The cleaner, longer, and more consistent satellite data are assimilated into inversion models to estimate surface CO₂ fluxes (net land and ocean sinks). Reduced variance relative to prior versions and truth proxies (including inversion models themselves) and the ongoing OCO-2 v11 Model Intercomparison Project should yield more reliable regional- to- global flux estimates and better characterization of interannual variability. These observationally constrained fluxes serve as benchmarks for the carbon- cycle modules in Earth-system models.⁠
• Improved evaluation of carbon- cycle processes and feedback
  Prior OCO data already showed that many dynamic vegetation models underestimate the seasonal amplitude of the terrestrial carbon cycle compared with satellite- and in- situ- constrained fluxes; models with larger (more realistic) amplitudes also tend to show greater sensitivity to climate change over the historical period. A higher- quality multi- year record allows tighter evaluation of how models represent responses to extremes (El Niño, droughts, wildfires). Recent analyses using OCO- related data have highlighted unusually weak land sinks in some years, testing whether models adequately include rapid carbon-loss processes (e.g., extreme fires, climate- driven mortality). Deficiencies here can make some projections of remaining carbon budgets overly optimistic.⁠
• Month- by- month monitoring capability
  The improved record supports tracking atmospheric CO₂ accumulation on monthly rather than only annual timescales. This enables nearer- real- time detection of carbon- cycle anomalies linked to extreme events, providing an early-warning capability that can inform updates to near- term projections, carbon budgets, and model benchmarking for decadal predictions.⁠
• Support for remaining carbon budgets and policy assessment
  More accurate knowledge of current sink strengths and their variability helps refine estimates of how much CO₂ can still be emitted while staying below temperature targets (the remaining carbon budget). It also aids evaluation of whether observed atmospheric growth rates are consistent with reported emissions and policy actions, feeding into updated scenario assessments.

The upgrade does not directly alter emissions scenarios, equilibrium climate sensitivity, or the physical climate components of models. Quantitative impacts on specific projected temperature ranges or carbon budgets from the v11 dataset itself are not yet available (the paper is very recent). Effects will appear gradually as the data are incorporated into inversions, model evaluations, and subsequent assessment cycles.

In summary, the upgraded satellite record provides a stronger observational benchmark for the carbon cycle. This helps reduce (or at least better quantify) a key uncertainty in climate projections, the strength of carbon- climate feedback, and supports more reliable estimates of remaining carbon budgets and the effectiveness of mitigation efforts.

The OCO-2 and OCO-3 ACOS Version 11 XCO₂ Data Products: A Benchmark Record for Space-Based Global Carbon Cycle Research

This is the primary research paper behind the Eos spotlight.

NASA’s Orbiting Carbon Observatory- 2 (OCO-2, since September 2014) and OCO-3 (on the ISS since August 2019) measure the column- averaged dry- air mole fraction of CO₂ (XCO₂) from reflected near- infrared sunlight. This paper describes the version 11 (v11) Atmospheric Carbon Observations from Space (ACOS) Level 2 full- physics retrieval algorithm, including updates to quality filtering and bias correction.

Key improvements and results:

• Updates include better instrument calibration, spectroscopy (ABSCO tables), meteorology and prior profiles, ocean surface reflectance parameterization, digital elevation model, and handling of clouds and aerosols.
• The resulting XCO₂ products show reduced systematic errors and better agreement between actual and theoretical uncertainties (especially for OCO- 3) compared with the previous v10 versions.
• Validation against the Total Carbon Column Observing Network (TCCON) and atmospheric inversion models (used as truth proxies) shows reduced variance and less temporal divergence.
• Collocated OCO- 2 and OCO- 3 observations agree to a mean difference of <0.2 ppm across their overlapping period, better than either dataset agrees with the validation sources.
• The unified, inter- calibrated record spans September 2014– August 2026 (OCO-2) and August 2019 onward (OCO- 3). Data are publicly available from the NASA Earthdata Cloud (including both X2007 and X2019 WMO scales, plus low-latency forward-stream files).

Conclusions from the paper

The lower and better- characterized errors “should enable higher- fidelity carbon cycle science studies, at local, regional, and global scales.” Together with related GOSAT v11 data, these products form a long- term global benchmark record for space- based carbon- cycle research.

The paper itself focuses on the retrieval algorithm, data quality, and validation rather than directly quantifying impacts on climate models or projections. Those applications (flux inversions, carbon- cycle evaluation, remaining carbon budgets, etc.) are enabled by the improved dataset, as discussed in our earlier exchanges.

Full text is available via the Wiley link (institutional access or purchase may be required) or through repositories such as CaltechAUTHORS. Data products can be downloaded from NASA Earthdata.

NASA Earthdata is the primary public portal for accessing the upgraded OCO- 2 and OCO- 3 data products discussed in the Taylor et al. (2026) paper.

Main access point

• Website: https://www.earthdata.nasa.gov

All Level 2 Lite File (LtCO2) products, forward- stream files, and related documentation for the ACOS Version 11 XCO₂ data are freely available through the NASA Earthdata Cloud / GES DISC.

Key datasets from the paper

Instrument	Version	Product	DOI / Notes
OCO- 2	v11.2	OCO2_L2_Lite_FP_11.2r	https://doi.org/10.5067/70K2B2W8MNGY
OCO- 2	v11.3	OCO2_L2_Lite_FP (recommended after 1 Jan 2024)	Current production version (includes icing correction)
OCO- 3	v11	OCO3_L2_Lite_FP_11r	https://doi.org/10.5067/8U0VGVQC7HZG

• Data are provided on both the older X2007 and current X2019 WMO CO₂ scales.
• Low- latency (forward- stream) daily files are also available (typically within ~2 days of observation).
• Supporting files include L1b, meteorology, preprocessor, and diagnostic products.

How to get the data

1. Go to earthdata.nasa.gov and create a free Earthdata Login account (required for downloads).
2. Use Earthdata Search and search for “OCO- 2” or “OCO- 3” and “Lite FP” or the specific version (e.g., “OCO2_L2_Lite_FP_11”).
3. Or go directly via the GES DISC portal linked from Earthdata.

Useful documentation

• Data User’s Guide for OCO- 2 (v11.3) and OCO- 3 (v11)
• Data Release Statements for each version
• Available on the GES DISC documentation pages or via the links in the Taylor et al. paper.

Journal: Earth and Space Science, Open Access, Volume13, Issue10

DOI: 10.1029/2026ea005360

First published: 22 October 2026

Provided: Eos

Authors: Thomas E. Taylor, Christopher W. O’Dell, Albert Chang, Abhishek Chatterjee, Cecilia Cheng, Lan Dang, Saswati Das, Brendan Fisher, Dejian Fu, Jonathan Hobbs, Nicole Jacobs …

Abstract

NASA’s Orbiting Carbon Observatory-2 (OCO-2) and OCO-3 have been observing Earth’s atmosphere from space since September 2014 and August 2019, respectively. The version 11 (v11) Atmospheric Carbon Observations from Space (ACOS) Level 2 full physics (L2FP) retrieval algorithm has been used to provide estimates of the column-averaged dry-air mole fraction of carbon dioxide (XCO2$$) from OCO measurements of surface reflected near-infrared solar radiation. In this work, updates to the ACOS v11 retrieval algorithm are discussed, and the quality filtering and bias correction of the XCO2$$ products are described. XCO2$$ estimates from OCO-2 and OCO-3 are compared to Total Carbon Column Observing Network (TCCON) observations and to atmospheric inversion models, which serve as truth-proxies for XCO2$$. In addition, a direct comparison of collocated soundings between OCO-2 and OCO-3 is performed. Variations in v11 XCO2$$ over small areas show improvement in agreement between actual and theoretical uncertainties, suggesting a reduction in systematic errors compared to v10, especially for OCO-3. There is reduced variance in the new v11 XCO2$$ product relative to truth-proxies, including less divergence in the XCO2$$ values with time against models. The collocations of XCO2$$ observations between OCO-2 and OCO-3 indicate an overall mean difference of <0.2 ppm across the overlapping time range. Furthermore, OCO-2 and OCO-3 agree better with each other than they do with the validation sources, suggesting a well conditioned and unified set of XCO2$$ concentrations that can be used concurrently for carbon cycle studies to provide a benchmark record spanning September 2014 through August 2026.