Water Resources Management Risks Across Kazakhstan’s Transboundary Basins: A Multi-Criteria Assessment (Zhaiyk-Caspian, Shu-Talas, and Balkhash-Alakol Basins)

Water management in Kazakhstan’s arid basins, which depend on transboundary rivers, faces intense pressure from climate variability, anthropogenic load, and land degradation. This study develops and applies a unified multi-criteria decision analysis (MCDA) framework to compare water-management risk across three structurally different basins: Zhaiyk-Caspian, Shu-Talas, and Balkhash-Alakol. It tests the hypothesis that this risk is determined not only by climatic factors, but equally by structural, institutional, and economic factors. The composite index combines ten indicators in four risk categories (natural-climatic, anthropogenic, economic, and social), weighted by a hybrid of entropy and Analytic Hierarchy Process (AHP) weights from six experts. Among the three basins, Balkhash-Alakol has the lowest relative risk (0.263) under all tested weighting schemes and indicator sets. Within this lower total, the largest contribution comes from industrial and anthropogenic pressure. The Zhaiyk-Caspian (0.484) and Shu-Talas (0.499) basins are comparable in overall risk but differ in their sources: unstable precipitation, water quality, and yield vulnerability in the former, and structural water stress, land degradation, and a rural water access deficit in the latter. Their order reverses at the crossover point α ≈ 0.59: Shu-Talas ranks higher under mostly expert-based weighting, Zhaiyk-Caspian under mostly entropy-based weighting. Thus, the hypothesis is partly confirmed, and differentiated water policy is needed for all basins. By showing which risk category leads in each basin, the framework links water-management risk to the sustainable use of water resources in arid transboundary basins and provides a basin-level basis for monitoring progress toward the Sustainable Development Goals, primarily SDG6. Limitations of the study include the small number of basins, min-max normalization, unweighted regional averaging, and proxy indicators.