Land Deformation Prediction via Slope-Aware Graph Neural Networks

We introduce a slope-aware graph neural network (SA-GNN) to leverage continuously monitored data and predict the land displacement. Unlike general GNNs tackling tasks in plain graphs, our method is capable of generalizing 3D spatial knowledge from InSAR point clouds. Specifically, we structure surface, while preserving correlations among adjacent points. The cloud can then be efficiently converted near-neighbor where GNN methods applied displacement slope surface. conducted experiments on real-world datasets results demonstrate that SA-GNN outperforms existing CNN methods.