Coordinator : Jean Raynald De Dreuzy _ Main participants : Alain Rapaport, Philippe Baveye, Jérome Harmand, Térence Bayen, Alejandr Rojas, Catherine Henault, Raia Massad, Patricia Garnier, 25% post-doc, requested Phd2, 1master student

Objectives :

Two different top-down approaches will be adopted to predict the response functions of soils under the different conditions considered in WP1 and WP3.

-       The first approach uses a Hierarchical Bayesian inference method to link the functions statistically to descriptors identified in WP2

-        The second approach constructs progressively more complex models, incorporating various structural elements of soils, until satisfactory predictions are obtained.

In both cases, the objective is to produce functions, e.g., describing CO₂ or N₂O evolution for soils, that can be substituted for the function currently included in models used in WP5.

WP4.1: Correlation analysis of heterogeneity descriptors and core-scale responses

25% post-doc, requested Phd2,, Philippe Baveye, Térence Bayen, Alejandr Rojas

WP4.2: Hierarchical Bayesian inference of macroscopic response functions

25% post-doc, Philippe Baveye, Catherine Henault, Raia Massad

WP4.3: Relation between pore-scale distribution and biological activity

Jean Raynald De Dreuzy, Alain Rapaport, Jérome Harmand , Patricia Garnier, requested Phd2

WP4.4: Derivation of equivalent lumped soil activities

Jean Raynald De Dreuzy, Alain Rapaport, Jérome Harmand , requested Phd2, 1 master student

Deliverables :

-       New macroscopic functions accounting for microscale heterogeneity, and explaining soil CO₂ and N₂O emissions, for the models of WP5

-       New compartmental models using simplified soil structures defined as simple connectivity patterns and of potential influence to microbial fluxes derived from WP1 and WP3.

-       Software with an intuitive user interface for the simulation of compartment models (1-10 compartments) with pre-defined or graphical definition of porosity structures (connectivity graph) and a selection of the most classical biological growth models including those used in WP5

Potential risks and solutions: The second approach, involving the elaboration of conceptual compartment models has never been applied to situations as complex as those considered in this project. However, the risk in this respect is mitigated by the use of a parallel approach, based on Bayesian inference, which has already been used in a similar context, with success.