How Well Can We Validate and Characterise GOSAT XCH4 and XCO2 Retrievals?
Byckling, Kristiina; Boesch, Hartmut; Parker, Robert; Webb, Alex
University of Leicester, UNITED KINGDOM

The first dedicated greenhouse gas sensor GOSAT was launched in 2009 and 4 years of global observations, which allow the retrieval of CO2 and CH4 columns, are now available. Such global CO2 and CH4 column data can provide constraints on regional surface fluxes but the requirements on precision and accuracy are very high (e.g. relative accuracy of < 0.2-0.5 ppm for XCO2 and < 1-10 ppb for XCH4; GHG-CCI User Requirement Document).

The primary validation dataset are observations from the ground-based Total Carbon Column Observing Network (TCCON), which itself is calibrated against in-situ aircraft observations and thus provides a vital link to the WMO standard. We have assessed the systematic biases and random errors of the GOSAT CO2 and CH4 ('full-physics' and 'proxy' retrieval method) retrievals by comparisons to the TCCON data. Due to the sampling pattern of GOSAT and the frequent occurrence of clouds, the number of available GOSAT soundings for some sites is small and the statistical significance of observed biases needs to be taken into account. Potential temperature can be used to extend the spatial and temporal coincidence criteria and to increase the statistical significance of the comparisons. Furthermore, using regression analysis, we determine correlations of systematic biases with geophysical parameters, which can help to inform bias-correction schemes.