Constrain CMIP6 projections using observationally anchored emergent constraints.
npx skills add https://github.com/xiechy/climate-ai --skill xai-cons
Please help me install this Agent Skill. Skill: xai-cons Source: https://github.com/xiechy/climate-ai/tree/main/climate-methods/xai-cons Command: npx skills add https://github.com/xiechy/climate-ai --skill xai-cons
This Skill specializes in Observational Constraint (Emergent Constraint, EC) analysis - a method that uses historical observations to constrain future climate projections across CMIP multi-model ensembles, reducing prediction uncertainty. It includes inter-model regression analysis, EC relationship establishment, physical mechanism diagnostics (residual analysis, teleconnection pathways, Walker circulation, lead-lag correlation, and SVD), uncertainty quantification (variance reduction, and confidence intervals), and reliability assessment (binning analysis, random EC comparison).
Use the EC workflow to constrain a CMIP6 variable pair (e.g., South Atlantic SST vs East Asia TAS) by applying historical observations to the emergent relationship and obtaining a constrained future projection.
High-intent search queries and answers about installing and using this skill.
Emergent constraint (EC) analysis uses historical observations to constrain future CMIP6 projections by establishing linear relationships between model variables. Load multi-model historical and future data, compute inter-model regression to derive constraint coefficients (β), then apply observational values to produce constrained predictions with reduced uncertainty.
An emergent constraint identifies a strong relationship between a historical observable variable and a future climate target across CMIP models. By anchoring this relationship to observations, you shift the model ensemble distribution toward more realistic projections, quantified through variance reduction and posterior confidence intervals.
Yes. Beyond regression, EC analysis includes residual analysis, mediation analysis, and teleconnection diagnostics (Walker circulation, lead-lag patterns, SVD) to validate the physical pathways behind constraints and confirm the relationship is mechanistically sound, not spurious.
Apply binning analysis to test constraint stability across model subsets and run random EC comparisons against shuffled data to measure statistical significance. These diagnostics produce prior/posterior distributions and reliability metrics, distinguishing robust constraints from artifacts.
EC analysis works across diverse pairs—SST with temperature (TAS), precipitation with circulation indices—where historical covariance is strong and mechanistically linked. The Skill outputs correlation (r), p-values, and R² to evaluate relationship strength before constraining projections.
Load xarray-formatted CMIP6 data (multi-model ensembles with historical and future scenarios). The Skill requires numpy, scipy, pandas, matplotlib, and xarray; it outputs regression coefficients, constrained predictions, variance metrics, and diagnostic visualizations as standard arrays and plots.
xjtulycCompute climate trends and ETCCDI extreme indices with Mann-Kendall and Sen's slope.
SPIRAL-EDWINRun large-scale Monte Carlo simulations on stochastic models to quantify uncertainty.
jimmc414Build, fit, and validate Bayesian models with PyMC.
reaatechCalculate calibrated confidence scores and uncertainty estimates for classification predictions.
