Retrievals of atmospheric columnar carbon dioxide and methane from GOSAT observations with photon path-length probability density function (PPDF) method

Abstract:

We retrieved the column-averaged dry air mole fraction of atmospheric carbon dioxide (X_CO2) and methane (X_CH4) from the radiance spectra measured by Greenhouse gases Observing SATellite (GOSAT) for 48 months of the satellite operation from June 2009. Recent version of the Photon path-length Probability Density Function (PPDF)-based algorithm was used to estimate X_CO2 and optical path modifications in terms of PPDF parameters. We also present results of numerical simulations for over-land observations and “sharp edge” tests for sun-glint mode to discuss the algorithm accuracy under conditions of strong optical path modification.

For the methane abundance retrieved from 1.67-µm-absorption band we applied optical path correction based on PPDF parameters from 1.6-µm carbon dioxide (CO₂) absorption band. Similarly to CO₂-proxy technique, this correction assumes identical light path modifications in 1.67-µm and 1.6-µm bands. However, proxy approach needs pre-defined X_CO2 values to compute X_CH4, whilst the PPDF-based approach does not use prior assumptions on CO₂ concentrations.

Post-processing data correction for X_CO2 and X_CH4 over land observations was performed using regression matrix based on multivariate analysis of variance (MANOVA). The MANOVA statistics was applied to the GOSAT retrievals using reference collocated measurements of Total Carbon Column Observing Network (TCCON). The regression matrix was constructed using the parameters that were found to correlate with GOSAT-TCCON discrepancies: PPDF parameters α and ρ, that are mainly responsible for shortening and lengthening of the optical path due to atmospheric light scattering; solar and satellite zenith angles; surface pressure; surface albedo in three GOSAT short wave infrared (SWIR) bands. Application of the post-correction generally improves statistical characteristics of the GOSAT-TCCON correlation diagrams for individual stations as well as for aggregated data.

In addition to the analysis of the observations over 12 TCCON stations we estimated temporal and spatial trends (interannual X_CO2 and X_CH4 variations, seasonal cycles, latitudinal gradients) and compared them with modeled results as well as with similar estimates from other GOSAT retrievals.