Value-added diagnostics for the assessment and validation of integrated assessment models

We evaluated three Integrated Assessment Models (IAMs: IGSM, MERGE, MiniCAM) by: (i) comparing their global Primary Energy year-2000 initializations and projections for 2010 and 2015 to historical data; (ii) mapping their CO₂ emissions projections against observations; and (iii) examining model-output diagnostics. The IAMs underestimated historical primary energy consumption and initial/projected CO₂ emissions in both reference and stabilization scenarios (particularly for combustion fuels) but overestimated usage of non-biomass renewables, causing underestimates of future CO₂ emissions that, for the stabilization scenarios, are wildly optimistic. Mitigation technology breakdowns in the policy scenarios vary enormously across IAMs, suggesting that confidence in their projections might be misplaced, or that options for mitigation have greater scope than is supposed. Most increases in carbon-free technologies in the stabilization scenarios are already captured in the reference cases. Energy-conversion efficiencies in electricity generation improve over time, but, (except for gas-powered generation in IGSM), efficiencies in the policy scenarios are less than in the reference. Electrification results diverge widely: IGSM has little change over the 21st century, while MiniCAM and MERGE have major electrification increases in their policy scenarios. We suggest: 1) comprehensive model output suitable for secondary analysis should be more readily available; 2) directly comparable reference and policy-driven mitigation scenarios are essential for assessing mitigation measures; 3) model validation using historical, source-specific energy data is crucial for assessing model credibility; 4) separation of mitigation contributions into no-policy and policy-driven amounts is needed to assess the effectiveness of mitigation policies; and 5) detailed inter-model comparisons can provide important insights into model credibility.