Evidence-concordance audit before multicriteria ranking: reassessing Kabul’s historical water-supply options

Introduction Historical infrastructure decisions are often revisited using evidence developed under different definitions, assumptions, and system boundaries. This study reassesses seven water-supply alternatives from Kabul’s 2009 planning archive to determine whether the archived evidence remains sufficiently comparable for decision modeling. Methods Each source-table cell or narrative quantity was transcribed with an exact locator, unit, construct label, and system boundary, then screened through five ordered binary gates: provenance, semantic typing, comparability, dependence, and uncertainty. The audit was qualitative and rule-based and did not assign a composite evidence-quality score. Costs were compared at 30 million m3/year, a common plotted service level rather than a demand target, with sensitivity analysis from 15 to 100 MCM/year. Reported supply quantities were excluded because the archive combines non-equivalent constructs, whereas availability statements were retained only as source-defined timing indicators. Results Under the baseline cost-timing specification, CRITIC weights were driven mainly by criterion dispersion, and Shatoot ranked first. Across normalized cost weights, the Panjshir fan aquifer ranked first below 0.288, Shatoot from approximately 0.288 to 0.941, and Gat above 0.941. These three alternatives form the non-dominated set. Across tested service levels, the Pareto set and sequence of first-ranked alternatives remained unchanged; only the final positions of Gulbahar and Salang exchanged at 100 MCM/year. Alternative timing assumptions and bounded cost scenarios were descriptive stress tests, not probabilistic predictions. Discussion The findings support conditional historical comparison, not a definitive project winner. A four-dam CRITIC formulation was not retained because estimating correlations across eight criteria from only four alternatives provides insufficient information for reliable weighting. Later JICA reports provide updated context but no comparable contemporary seven-option matrix. Any project recommendation requires renewed, comparable evidence.