To address challenges towards building an operational Monitoring and Verification System (MVS) for CO2, we report here on (a) the creation of new sets of input data for data assimilation (DA) systems, and (b) their use in a large number of alternative or new DA configurations. Together, these sets provide new insights on the data processing for XCO2, their assimilation, and the best method to maximize data usage, minimize biases, separate biogenic and anthropogenic fluxes, while maintaining a fast operating chain. This report documents the technical part of our efforts, and the data availability of a first set of input and output products. More and updated results will be created during Task 1.3 when a benchmarking and QA/QC pipeline will be demonstrated for our DA systems.
We find that a new fast and efficient XCO2 production algorithm (FOCAL, UB) can create a suitable alternative to full-physics retrievals for OCO-2, and that propagation of error covariances into an XCO2 retrieval seems feasible (ULEIC & CEA), but that the processing chain is currently not suited for near real-time. Data assimilation of SIF is feasible, but a proper observation operator is missing because of incomplete knowledge of SIF emissions from the canopy, and because the pre-TropOMI SIF products are too coarse to separate ecosystems. COS as a biogenic tracer requires further work on ocean and soil source terms, specifically in the tropical regions. XCO2-based DA systems can now match in-situ based flux estimates on global to regional scales and suggest a smaller tropical sink and seasonal cycle. They are ready for wider application. Atmospheric transport models need to be upgraded in resolution to match the satellite products, which can be achieved by either (a) moving towards online calculation of meteorology and mass-fluxes, and (b) shortening the DA windows towards the daily-to-weekly scale. These system improvements are scheduled for the second part of task 1.2.