The ecological environment of urban lakes affected by human activities is deteriorating rapidly. As a source and sink of pollutants in the lake environment, sediments have become the focus of environmental assessments. At present, most of the studies only conduct pollution assessments on surface sediments. In this study, taking the typical urban lakes GanTang Lake and NanMen Lake (G&N Lake) as the background, not only is the planar spatial distribution of their nutrient elements, seven kinds of heavy metals, and As analyzed in detail, but risk assessments are also carried out on the pollution conditions at different depths. The causes of pollution at different depths are analyzed. It is found that in this lake, with the increase in depth, the pollution situation decreases slightly, but the pollution of nutrient elements is severe. There is severe pollution of nutrient elements at a depth of up to 1 m in the whole lake sediment. In the sediments with a depth of up to 1 m, more than 90% of the areas in the whole lake are at or above the moderate pollution level of Hg, and more than 70% of the areas are under slight pollution of Cd, resulting in the ecological risk level of the whole lake being at or above the high-risk level. Urban lake sediment management is inherently complex, driven by multifaceted factors where intensive anthropogenic activities constitute the primary pollution source. This research provides insights to guide restoration strategies and sustainable development policies for lacustrine ecosystems.

Flash-flood and floodplain inundation simulations are highly sensitive to the spatiotemporal variability of convective rainfall, particularly during the initial runoff generation stage. However, coarse-resolution numerical weather prediction (NWP) forcing tends to smooth localized rainfall extremes, limiting its ability to accurately represent hydrological responses in low-relief floodplains. In this study, we couple a diffusion-enhanced radar nowcasting model, Diff_ConvLSTM, with a spatial resolution of 1 km and a temporal resolution of 6 min, to assess the hydrological value of high-resolution rainfall forcing over the middle Yangtze River floodplain. We introduce a monotone piecewise cubic Hermite interpolation scheme to ensure a stable transition from discrete high-frequency rainfall inputs to continuous hydrodynamic integration. Evaluation using a radar dataset from 2023 to 2024 shows that Diff_ConvLSTM better preserves intense convective echoes and rainband structures compared to the baseline ConvLSTM, increasing the Probability of Detection at the 40 dBZ threshold by 65.8%. A forcing-replacement experiment for the flood event on 30 June 2023 demonstrates that AI-based nowcasting rainfall forcing reduces peak-discharge underestimation, improves volumetric consistency, and produces inundation patterns that are closer to the observation-driven reference than those generated by low-resolution forecast forcing, although positive biases in inundation area and water depth persist. An additional event in 2024 confirms that the improvements are primarily reflected in discharge magnitude and flood volume representation, while enhancements in peak timing remain limited. Overall, the results illustrate both the added value and the remaining limitations of AI-enhanced nowcasting for hydrologically informed flood forecasting.

Accurate calibration of urban drainage models is critical for reliable stormwater management. This study applies the OSTRICH-SWMM framework, which integrates the Storm Water Management Model (SWMM) with multiple parallel optimization algorithms, to systematically evaluate calibration performance in a large-scale urban drainage system with a total area of 25.6 km ² located in the southern part of Jiujiang, China. Three algorithms were tested: Asynchronous Parallel Dynamically Dimensioned Search (ParaDDS), Real-coded Genetic Algorithm (RGA), and Simulated Annealing (SA). Results show that ParaDDS delivered the most robust performance, achieving Nash–Sutcliffe efficiency (NSE) values of 0.864 for calibration rainfall event and 0.757 for validation event, with minimal variability among top 25-percentile solutions. RGA also performed well with a NSE value of 0.861 for calibration and 0.601 for validation whereas SA had a NSE value of 0.778 for calibration and 0.482 for validation. These findings demonstrate the effectiveness of the OSTRICH-SWMM framework for automatic calibration of complex urban drainage systems and underscore the capability of parallel optimization algorithms, particularly ParaDDS, in achieving stable and reliable parameter sets.