Precision road extraction via remote imagery plays a vital role in smart transportation and urban digitalization. Current deep learning methods, however, confront three key limitations: high computational complexity from intricate architectures, road fragmentation caused by weak topological continuity, and environmental interference-induced errors in thin road detection. To address the aforementioned challenges, this paper proposes FAMNet, which establishes a collaborative edge surface perception mechanism to achieve high-precision road extraction in complex scenarios while maintaining topological continuity. Firstly, a lightweight MobileNetv3 is redesigned as the backbone network to reduce computational overhead and model complexity. Then, we develop an MCASPP combined with a Convolutional Block Attention Module to expand receptive fields while enhancing extraction capability and connectivity for slender road features. Thirdly, a composite loss function combining Focal loss and Dice loss is implemented to mitigate the impact of class imbalance between roads and background samples, suppress environmental interference, and emphasize road-specific information. Experimental results for DeepGlobe and CHN6-CUG datasets demonstrate that FAMNet significantly outperforms state-of-the-art methods in both the accuracy and structural completeness of road extraction, while maintaining advantages in parameter count and computational efficiency. Furthermore, the proposed method shows promising potential for advancing natural resource surveying and monitoring through enhanced geospatial environment perception modeling capabilities.

Flood-season lake spatiotemporal dynamics are vital for ecological security and socioeconomic development, requiring consistent high-resolution monitoring. However, precipitation fluctuations and sediment turbidity significantly alter water quality, while blurred boundaries between water and floodplain wetlands challenge precise monitoring. To address these issues, this study proposes a water body extraction method leveraging polarimetric Synthetic Aperture Radar data. utilizes the maximum between-class variance algorithm for initial segmentation, optimizes the threshold via a genetic algorithm, and employs dynamic morphological operations to refine boundary details. The method was validated using 2015–2025 Sentinel-1 flood-season time series of Dongting Lake on Google Earth Engine. The results demonstrate that the proposed method achieves stable and accurate water extraction across various years and seasons, with an overall accuracy surpassing 0.93, confirming its robustness and broad applicability. Furthermore, the spatiotemporal hydrodynamics and driving mechanisms of Dongting Lake were analyzed by integrating the extracted water areas with multi-source data, including water level, precipitation, discharge, temperature, and sunshine duration. Findings indicate that the flood-season water area exhibited a fluctuating trend, initially increasing and subsequently decreasing, peaking at 2202.26 km² in 2020 and dropping to 614.04 km² in 2025, a pattern primarily driven by extreme meteorological events such as heavy rainfall and prolonged droughts. Spatially, inundation patterns were characterized by deeper water in the north and shallower depths in the south, separated by a topographically higher central region. Regression analysis revealed a robust correlation between water area and water level with an R² of 0.931, providing a quantitative reference for water level estimation in ungauged regions. Additionally, discharge and precipitation were positively correlated with water area, whereas temperature and sunshine duration exerted a negligible influence. This study supports flood regulation in the Dongting Lake basin and provides a robust framework for analyzing lake dynamics using long-term SAR data.