Multi-criteria decision analysis and linear programming for optimal resource allocation in water quality monitoring networks: a case study of California’s EPA monitoring stations

Abstract Water quality monitoring networks require strategic resource allocation to maximize effectiveness while managing budget constraints. This study presents an integrated multi-criteria decision analysis (MCDA) framework combining CRITIC (CRiteria Importance Through Intercriteria Correlation), TOPSIS (Technique for Order Preference by Similarity to Ideal Solution), VIKOR (VIseKriterijumska Optimizacija I Kompromisno Resenje), and linear programming (LP) optimization for prioritizing monitoring stations and allocating resources. Using real data from 347 monitoring stations in California extracted from the United States Environmental Protection Agency (EPA) Water Quality Portal (WQP) spanning January 2023 through December 2024 (16,750 measurements across seven water quality parameters), we demonstrate a reproducible methodology for evidence-based monitoring network optimization. Prior to analysis, physical plausibility filters removed 1106 impossible values (2.4% of raw data), ensuring analytical integrity. Dissolved oxygen was treated as an optimal-range criterion (|DO − 8 mg/L|) rather than a monotonic benefit, correcting a common misclassification in MCDA water quality studies. Spearman’s rank correlation replaced Pearson’s throughout the CRITIC computation to account for skewed parameter distributions. The CRITIC method determined objective criteria weights, with dissolved oxygen deviation (0.1813) and phosphorus (0.1804) as the most informative parameters. TOPSIS analysis identified top-performing stations with closeness coefficients ranging from 0.9813 to 0.3799, with a mean of 0.712 (MAD = 0.063). VIKOR analysis confirmed ranking consistency, yielding Spearman $$\varvec{\rho = 0.912}$$ ρ = 0.912 ( $$\varvec{p}$$ p < 0.001 ) between TOPSIS and VIKOR rankings. LP optimization concentrated resources efficiently, achieving 36.9% improvement over uniform allocation ( $$\varvec{Z}^{\varvec{*}} \varvec{= 0.9749}$$ Z ∗ = 0.9749 vs. $$\varvec{\bar{C}_i = 0.712}$$ C ¯ i = 0.712 under uniform distribution). This integrated MCDA-optimization framework provides water resource managers with a transparent, data-driven tool for strategic planning, enabling efficient allocation of limited monitoring resources while maintaining comprehensive environmental surveillance.