Complexity-Entropy Characterization of Storage Dynamics in Semiarid Reservoirs: Linking Ordinal Patterns with Elevation-Storage Curve Shifts in Paraíba, Brazil

Hydrological reservoirs are complex systems in which storage variations integrate climate forcing, catchment response, releases, withdrawals, evaporation, and monitoring procedures. This study presents an information-theoretic characterization of storage dynamics in five semiarid reservoirs in Paraíba, Brazil. The main analytical layer is the complexity–entropy causality plane (CECP), computed from daily storage increments by permutation entropy and Martín-Plastino-Rosso statistical complexity. CECP was estimated for fixed periods and for sliding windows of 120 observations, with sensitivity tests for embedding dimension, delay, and window length. The workflow also benchmarks CECP distance against conventional descriptors, quantifies ties and zero increments, and tests window overlap. As physical context, monotonic elevation-storage curves were reconstructed for 2009–2014, 2015–2019, and 2020–2026, and storage differences at equivalent water levels were quantified by bootstrap confidence intervals. The reservoirs occupied a high-entropy, low-to-moderate-complexity region of the CECP, but their distances from the maximum-entropy/minimum-complexity vertex differed across reservoirs and periods. Sliding windows revealed temporal mobility that was hidden by fixed-period summaries, especially in Engenheiro Arcoverde, Jatobá I, and Mãe d’Água. Rankings remained strongly concordant when window overlap decreased from 94.2% to 0% (Spearman ρ=0.943), although absolute coordinates were sensitive to the treatment of reported plateaus. Elevation-storage shifts provided an independent structural context: negative shifts were compatible with possible useful-capacity reduction, although not uniquely attributable to sedimentation. The results show that CECP descriptors can reveal ordinal organization and regime mobility in reservoir storage increments, while V(H) curves supply the physically interpretable storage-capacity context. The combined evidence prioritizes Engenheiro Arcoverde and Mãe d’Água for bathymetric, curve history, and operational verification. The proposed workflow is therefore an exploratory information-theoretic screening tool for data-limited reservoir monitoring, not a substitute for bathymetric validation.