Simultaneous retrieval of trace gases and aerosol using RemoTAP – the global orbit ensemble study for the CO2M mission

In the support of the Copernicus Anthropogenic Carbon Dioxide Monitoring (CO2M) mission, SRON Netherlands Institute for Space Research developed the Remote sensing of Trace gas and Aerosol Product (RemoTAP) algorithm. RemoTAP is able to achieve simultaneous retrieval of trace gases and aerosol using measurements from the Multi-Angle Polarimeter (MAP) and CO2I Imager aboard the CO2M mission. At the same time, it has the capability to perform the retrieval of trace gas from only CO2I measurements.

This study evaluates the performance of RemoTAP for combined MAP-CO2I and CO2I-only retrievals, respectively. We base our evaluation on synthetic CO2M measurements simulated for realistic atmospheric (aerosol, cirrus), surface, geometry conditions. MAP-CO2I retrieval method can reduce the regional bias in column-averaged dry-air mole fraction of CO2 (XCO2) by a factor of 3. It shows that only by the inclusion of MAP measurements, the large aerosol-induced biases can be mitigated, resulting in the retrievals that meet the mission requirement (precision <0.7 ppm and bias <0.5 ppm).

In addition, RemoTAP has other functions to further improve the accuracy of trace-gas retrievals. It is able to retrieve cirrus at pixels with an optically thin layer of cirrus cloud. Instead of retrieving cirrus, we also provide the option to filter cirrus-contained pixels by non-scattering retrieval, which results in more accurate XCO2 retrievals but with less yield. To account for the uncertainties in surface pressure, which is related to the calculation of dry air column, RemoTAP has the option to retrieve O2 column which is proportional to dry air column from measurements of NIR band. Besides, RemoTAP retrieves Solar-Induced chlorophyll Fluorescence (SIF) over vegetation surface. The retrieval of cirrus, surface pressure/O2 column and SIF is performed simultaneously with the retrieval of other physical parameters, and the error induced by these factors in the XCO2 retrieval can be mitigated. Finally, we developed a method to perform bias correction and quality filtering using a neural network approach.