The Paraguay River is the main hydrological axis of the Pantanal wetland, the world’s largest tropical floodplain, and plays a key role in the hydrodynamics and ecological functioning of the Upper Paraguay Basin (UPB). This study evaluated the spatio-temporal variability of river levels at six fluviometric stations distributed along the Paraguay River and adjacent sectors of the UPB: Cáceres, Cuiabá, Bela Vista do Norte, Ladário, Forte Coimbra, and Porto Murtinho. Daily water-level data provided by the Brazilian Navy (CHN-6) were used for descriptive, persistence, multivariate, spectral, duration-curve, heatmap, anomaly, and clustering analyses. Long-term trends were evaluated using annual water-level series and the Mann–Kendall test after trend-free pre-whitening (TFPW) to account for serial dependence, with Sen's slope used to quantify the magnitude of change.The results revealed pronounced hydrological seasonality associated with the annual Pantanal flood pulse and strong temporal persistence throughout the series. Bela Vista do Norte exhibited the greatest hydrological stability, whereas Porto Murtinho showed the highest hydrological amplitude and strongest influence of accumulated upstream flows. After accounting for serial dependence using trend-free pre-whitening, significant long-term decreases in annual river stage were detected at Cáceres and Cuiabá, with Sen's slopes of −0.0234 and −0.0191 m year⁻¹, respectively. In contrast, no statistically significant monotonic trends were detected at Bela Vista do Norte, Ladário, Forte Coimbra, or Porto Murtinho. PCA and cluster analysis indicated strong similarity between Cáceres and Cuiabá, whereas Porto Murtinho displayed more independent behavior. Spectral and Singular Spectrum Analysis (SSA) analyses highlighted the dominance of low-frequency seasonal oscillations. These findings improve understanding of Pantanal hydrodynamics and provide valuable information for water resources management, environmental monitoring, and climate adaptation strategies in the Upper Paraguay Basin.
Forecasting total nitrogen (TN) across multiple reservoirs remains challenging because nutrient dynamics are nonlinear, non-stationary, and influenced by interacting hydrological, climatic, and biogeochemical processes. This study proposes a CEEMDAN-enhanced Deep Hybrid Ensemble Framework (CEEMDAN-DHEF) for monthly TN forecasting across eight reservoirs in Central China. Using a 2,304-record monthly reservoir panel spanning 2000–2023, the framework integrates water-quality variables, meteorological drivers, temporally lagged predictors, interaction terms, and CEEMDAN-derived multiscale descriptors within 12-month forecasting sequences. Five complementary neural learners a TFT-inspired temporal-attention network, Long Short-Term Memory, Convolutional Neural Network, Gated Recurrent Unit, and Deep Neural Network were trained on 12-month sequences, with their outputs integrated through constrained convex weighting and Ridge stacking calibrated exclusively on chronologically ordered development data. Under the independent chronological evaluation (November 2019–December 2023; n = 400), CEEMDAN-DHEF achieved R²/NSE = 0.3448, RMSE = 0.658 mg/L, and MAE = 0.481 mg/L, providing a conservative assessment of temporal generalization. A separate retrospective later-period evaluation yielded R²/NSE = 0.8345, indicating stronger predictive agreement under that evaluation setting. Complementary leakage-controlled analyses showed that lagged predictors accounted for 42.1% of the total SHAP importance, followed by observed-history variables (20.2%), rolling predictors (10.4%), and CEEMDAN-derived multiscale features (9.5%). Prediction-interval analysis further quantified forecast uncertainty, with empirical coverages of 92.25% and 98.50% for the nominal 90% and 95% intervals, respectively. These findings demonstrate that decomposition-assisted ensemble learning can capture multi-scale TN dynamics while providing interpretable and uncertainty-aware evidence for reservoir water-quality forecasting.
