Water Resources Department.
Presenter:
Shuwan Jawdat
Deep learning (DL) and artificial intelligence (AI) are revolutionizing dam site selection through enhanced analysis of complex geospatial data, outperforming traditional methods. Key applications include automatic site identification using YOLOv5 with Sentinel-2 imagery, feature analysis of critical factors such as slope and soil type, and suitability mapping via Convolutional Neural Networks (CNN), achieving high sensitivity (93%) in identifying check dam sites. AI models analyze topography, geology, environmental factors, and water availability efficiently, minimizing human bias, integrating diverse datasets, and saving time in feasibility assessments. Notable models employed are CNN, Graph Neural Networks (GNN), and Recurrent Neural Networks (RNN). Challenges include data scarcity, dense vegetation affecting topography visibility, and integrating DL with Multi-Criteria Decision Making (MCDM) for regulatory approval.
12/05/2026
