The present project makes use of recent Monte Carlo advances to design a numerical tool addressing a quite extreme climate-science calculation: evaluating the radiative forcing of an ensemble of greenhouse gases, at the global scale, integrated over climatic durations, using a reference radiative transfer model with high-resolution spectral data and 4D atmospheric fields (space and time). Starting from the null-collision concept, the feasibility was recently established and the present challenge consists in defining an efficient strategy for sampling the space of molecular transitions combined with all the geometrical and temporal complexity of atmospheric variables. For this purpose, we gather climatologists, spectroscopists, specialists in the statistical-engineering of complex systems and specialists in geometry-modelling for computer graphics applications.
MCG-Rad benefits of a prototype-algorithm available and already tested for clear-sky atmospheres. This algorithm will be implemented at the launching of the project, allowing first climatic evaluations, in particular within the Radiative Forcing Model Intercomparison Project. But outside this first practicability level, a major issue is convergence-enhancement: the size of the community interested by such reference radiative-transfer computations at the climatic scale will be highly dependent on the required computational-resource. We therefore concentrate an essential part of the present research on the theoretical exploration of all means to reduce the variance of the Monte Carlo estimate, i.e. using the most advanced propositions of both the transport-physics and computer-graphics communities to reduce the computational-resources.
These theoretical considerations extend widely the objective of the original algorithm (that was restricted to clear-sky single atmospheric columns). The question is scalability when thinking of:
- the 4D fields produced by climate General Circulation Models (GCM), including multiple-scattering in 3D clouds below the column-scale (clouds reconstructed from the GCM outputs);
- the set of all radiative-observables of interest, as elaborated by the climate-change community, i.e. with an accuracy-requirement formulated in terms of integrated radiative-forcing (a quantity that is small compared to the involved infrared and shortwave fluxes which makes it difficult to evaluate) and including sensitivities to the concentration of absorption gases.
Three neighbor communities will be connected to the project as they face similar challenges: heat-transfer/combustion engineering, planet atmospheric-studies, cloud dynamics. A second circle of scientific expertise will therefore join the project via annual meetings and workshops. They will contribute to our software-design choices, structuring the libraries so that the present work addresses a broader audience as an example of jointly sampling large spectroscopic databases and complex 3D/4D fields for accurate radiative-transfer objectives.
Monsieur Jean-Louis Dufresne (Laboratoire de météorologie dynamique)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
IRIT Institut de Recherche en Informatique de Toulouse
LMD/CNRS Laboratoire de météorologie dynamique
LAPLACE LABORATOIRE PLASMA ET CONVERSION D'ENERGIE
Help of the ANR 505,112 euros
Beginning and duration of the scientific project: October 2018 - 48 Months