Probabilistic Suitability Assessment and Stability-Sensitivity Zoning of Urban Underground Space under Weighting-Method-Induced Uncertainty

GIS-based multi-criteria decision-making assessments of urban underground space suitability commonly rely on deterministic weights， yet the reliability of such maps remains sensitive to differences among weighting mechanisms. Using the 0-15 m shallow underground space of central Ningbo， China， this study develops a probabilistic suitability assessment framework focused on weighting-method-induced uncertainty. AHP， EWM and CRITIC are used as complementary expert-judgement， dispersion-based and contrast-and conflict-based weighting anchors to define a study-specific empirical feasible weight domain. Accepted valid weight vectors are then generated through constrained weight perturbation scenario sampling and propagated through the WLC model to obtain a Monte Carlo suitability ensemble. The convergence assessment shows that N = 5，000 accepted valid weight vectors provide stable estimates relative to the N = 10，000 reference ensemble， with a global mean suitability difference of 0.0194% and an MAE of 0.0160. The results further identify stability-sensitivity zones， quantify classification confidence， and determine dominant constraints. High-confidence class assignments account for 67.25% of the complete-case probabilistic classification area， while the RCI results show that C3， C9， C7 and C5 jointly account for 68.17% of the normalized contribution to suitability-loss. The framework supports reliability-oriented interpretation of shallow underground-space suitability while remaining conditional on the selected indicators， preprocessing scheme and weighting anchors.