Beyond Outcome Variance: Addressing Ranking Instability via an RSSI-SA-ADF Framework for Adaptive Ecological Restoration Decision-Making
Description
Research Hypothesis This dataset supports the study "Beyond Outcome Variance: Addressing Ranking Instability via an RSSI-SA-ADF Framework for Adaptive Ecological Restoration Decision-Making," which challenges the reliance on variance-based sensitivity analysis (e.g., Sobol indices) in environmental decision-making. We hypothesize that parameters driving numerical outcome variance are distinct from those triggering Ranking Structural Instability (RSI). Minor fluctuations in "structural" parameters (e.g., stakeholder preferences) often cause strategy ranking reversals despite contributing little to total variance. Data Content and Methodology The files provide documentation and empirical data from three ecological restoration projects in China (Nanyang, Shangqiu, and Jiaozuo). Key components include: Frameworks (S2, S3, S6): Mathematical derivations for the MD-LEA evaluation model, the RSSI algorithm, and NSGA-II/TOPSIS optimization procedures. Parameter Distributions (S7): Probability distributions and sources for inputs like Planting/Maintenance costs and ESV preference weights. Stakeholder Analysis (S4, S12): Aggregated AHP matrices for risk threshold calibration (τ = 0.15) and post-implementation feedback surveys. Validation Metrics (S8, S9, S15): Efficiency benchmarks, convergence plots, and ROI calculations comparing SA-ADF against traditional frameworks. Notable Findings Risk-Driver Reversal (S10): Technical parameters dominate outcome variance, but social preference parameters are the primary drivers of ranking reversals. Performance (S3, S4): The RSSI early warning reduced false-positive alerts by 7.0% and plant replanting rates by 68.6%. Economic Impact (S15): The framework achieved a 109.0% ROI by triggering interventions only when structural instability was detected. Interpretation and Usage Replication: Use (S7) distributions to test the MD-LEA model; code is available in the linked GitHub repository. Calibration: Refer to (S4) to set the risk threshold (τ) based on specific cost ratios of false positives to negatives. Contextual Analysis: (S10) illustrates how risk structures vary by project type (e.g., preference-driven vs. duration-constrained) to guide sensitivity analysis for new projects.
Files
Institutions
- North China University of Water Resources and Electric PowerHenan, Zhengzhou