Atmospheric Greenhouse Gas inverse modelling

SNIC 2022/5-575


SNIC Medium Compute

Principal Investigator:

Guillaume Monteil


Lunds universitet

Start Date:


End Date:


Primary Classification:

10501: Climate Research



The project is a continuation of the medium SNIC 2021/5-549 project. The aims are: - to support developments of atmospheric inverse modeling system in the inversion group at the department of Physical Geography, at Lund University. - to enable the annual updating of CO2 and CH4 atmospheric inverse modeling products, produced for the EU H2020 CoCO2 project. The tasks include the development of the LUMIA regional atmospheric inversion system, which aims to compute estimates of regional greenhouse gas sources and sinks using a combination of prior information from process models, such as LPJ-GUESS developed in Lund, and of atmospheric observations of greenhouse gas concentrations, such as observations from the ICOS network. In short, the principle is to use an atmospheric transport model to compute the concentrations corresponding to a “prior” estimate of the fluxes, which are then compared to observed concentrations. The aim of the inversion is then to find the set of fluxes that minimizes the mismatch with observations, while minimizing departures from the prior estimate. That optimal solution is searched for iteratively, which requires many (>100) applications of the transport model. LUMIA simulation typically rely on pre-computed footprints from Lagrangian particle dispersion models, such as FLEXPART, both of which (LUMIA and FLEXPART) are efficiently parallelized on tetralith nodes. The tasks expected to be started/continued in 2023 are: - annual release of European CO2 and CH4 inversions: such simulations are not demanding in term of developments, but they are computationally intensive as they cover a long period (> 15 years). - Further developments towards the extension of the atmospheric inversion approach to other tracers (14CO2, Black Carbon aerosols) - Assimilation of satellite retrievals of CO2 and CH4 in LUMIA. This will require the implementation of an additional transport model (CMAQ), more adapted than FLEXPART for that task.