Abstract
NASA's Orbiting Carbon Observatory 2 (OCO-2) was launched on July 2, 2014, and completed in-orbit checkout on September 6, 2014. To enable precise retrievals of the column-average carbon dioxide dry air mole fraction (XCO2), high-resolution spectra of reflected sunlight are acquired in three infrared channels. Channel 1, centered at 766 nm, is used to measure molecular oxygen and solar-induced fluorescence. Channels 2 and 3, centered at 1608 nm and 2064 nm, measure weak and strong carbon dioxide absorption bands. The three grating spectrometers share a common entrance telescope with a 0.8˚ field of regard, divided into 8 along-slit footprints. The 3 Hz frame rate and 705 km altitude yield ground footprints covering up to 3 km2, depending on slit orientation relative to the ground track.
The OCO-2 project has performed four public data releases to date and is currently releasing and maintaining Version 11. The OCO-2 on-board calibrator rotates between an open position used for collecting science data, a transmissive diffuser for solar calibration, and an opaque reflective diffuser for collecting lamp and dark data. Calibration data collected on over 10 orbits each day are used to estimate gain degradation caused by accumulation of ice on the focal plane arrays. Periodic decontamination cycles remove this ice. Additional sensitivity loss is tracked using clean observations recorded immediately after each decontamination. To separate solar calibrator and instrument throughput degradation, OCO-2 has viewed the 75% illuminated waxing gibbous Moon monthly for a decade. The final operational estimate of degradation, relative to first light, is the product of the solar and lunar ratios. These throughput degradation estimates are then multiplied by the absolute gain determined preflight.
Following Version 11 delivery, additional activities have yielded insight into OCO-2 absolute radiometry, specifically a small bias in the weak CO2 band. First, a series of 29 vicarious calibrations were performed at Railroad Valley, NV, USA between June 29, 2015, and September 3, 2021. In the previous version of the data (Version 10), the mean ratios for the three channels were 102.47 ± 2.31 %, 104.80 ± 2.97 %, and 102.89 ± 4.11 %, respectively. Next, radiance ratios were derived from a time series of simultaneous nadir land overpasses with OCO-3 spanning August 6, 2019, to August 31, 2023. Dividing OCO-2 Version 11 by OCO-3 Version 10.3, the modes of the distributions were 100.5 %, 104.7 %, and 101.5 % for the three channels. Inspired by those studies, water cloud optical property retrievals using standard two-channel lookup table approaches show inconsistencies when using channels 1 & 2 vs. channels 1 & 3. Scaling channel 2 radiance by 0.95 reduces the inconsistency between the two retrievals. While studies using the Level 2 algorithm indicate that scaling weak CO2 radiance causes minimal changes to retrieved XCO2, we do intend to include these updates in future (Version 12) data products.
Multiple Lines of Evidence for Bias in the OCO-2 Weak CO2 Channel (1598-1618 nm)
NASA's Orbiting Carbon Observatory 2 (OCO-2) was launched on July 2, 2014, and completed in-orbit checkout on September 6, 2014. To enable precise retrievals of the column-average carbon dioxide dry air mole fraction (XCO2), high-resolution spectra of reflected sunlight are acquired in three infrared channels. Channel 1, centered at 766 nm, is used to measure molecular oxygen and solar-induced fluorescence. Channels 2 and 3, centered at 1608 nm and 2064 nm, measure weak and strong carbon dioxide absorption bands. The three grating spectrometers share a common entrance telescope with a 0.8˚ field of regard, divided into 8 along-slit footprints. The 3 Hz frame rate and 705 km altitude yield ground footprints covering up to 3 km2, depending on slit orientation relative to the ground track.
The OCO-2 project has performed four public data releases to date and is currently releasing and maintaining Version 11. The OCO-2 on-board calibrator rotates between an open position used for collecting science data, a transmissive diffuser for solar calibration, and an opaque reflective diffuser for collecting lamp and dark data. Calibration data collected on over 10 orbits each day are used to estimate gain degradation caused by accumulation of ice on the focal plane arrays. Periodic decontamination cycles remove this ice. Additional sensitivity loss is tracked using clean observations recorded immediately after each decontamination. To separate solar calibrator and instrument throughput degradation, OCO-2 has viewed the 75% illuminated waxing gibbous Moon monthly for a decade. The final operational estimate of degradation, relative to first light, is the product of the solar and lunar ratios. These throughput degradation estimates are then multiplied by the absolute gain determined preflight.
Following Version 11 delivery, additional activities have yielded insight into OCO-2 absolute radiometry, specifically a small bias in the weak CO2 band. First, a series of 29 vicarious calibrations were performed at Railroad Valley, NV, USA between June 29, 2015, and September 3, 2021. In the previous version of the data (Version 10), the mean ratios for the three channels were 102.47 ± 2.31 %, 104.80 ± 2.97 %, and 102.89 ± 4.11 %, respectively. Next, radiance ratios were derived from a time series of simultaneous nadir land overpasses with OCO-3 spanning August 6, 2019, to August 31, 2023. Dividing OCO-2 Version 11 by OCO-3 Version 10.3, the modes of the distributions were 100.5 %, 104.7 %, and 101.5 % for the three channels. Inspired by those studies, water cloud optical property retrievals using standard two-channel lookup table approaches show inconsistencies when using channels 1 & 2 vs. channels 1 & 3. Scaling channel 2 radiance by 0.95 reduces the inconsistency between the two retrievals. While studies using the Level 2 algorithm indicate that scaling weak CO2 radiance causes minimal changes to retrieved XCO2, we do intend to include these updates in future (Version 12) data products.