Reads detailed CO2 soil emissions out of a MAgPIE gdx file using Shapley-style decomposition with interaction terms
Details
This function uses a structural decomposition approach (Shapley decomposition) to attribute soil carbon emissions to different drivers and their interactions. The framework is based on counterfactual analysis where all possible combinations of driver states are evaluated following the inclusion-exclusion principle.
For four drivers, the total emission is decomposed as:
totEmissions = main_effects + first_order_interactions + second_order_interactions + third_order_interactions + residual
This creates 16 counterfactual scenarios (2^4) to isolate individual and combined effects. Main effects capture changes when one driver varies while others stay constant. Interaction terms capture synergies where driver combinations amplify or dampen effects beyond simple addition.
Drivers included: - Climate (C): Changes in reference soil carbon density - Land-use (LU): Transitions between land types - Management (M): Changes in crop type shares, fallow, and treecover * Can be split into treecover vs other management in attribution - Soil Carbon Management (SCM): Enhanced soil carbon practices
Legacy effects are incorporated using exponential decay (85 over 100 years. Interaction terms are proportionally attributed back to main drivers in the final output.