usamimeri/geoInterp

This repo implements both traditional statistical interpolation methods and modern graph neural network(including train and inference) approaches to address the challenge of two-dimensional geospatial plane interpolation.

---

Introduction

This repo implements both traditional statistical interpolation methods and modern graph neural network(including train and inference) approaches to address the challenge of two-dimensional geospatial plane interpolation.

The methodologies used in this project are detailed below:

Traditional Statistical Interpolation Methods

• ADW (Angular Distance Weighting Interpolation) https://dl.acm.org/doi/10.1145/800186.810616
• IDW (Inverse Distance Weighting Interpolation)
• Kriging

Modern Graph Neural Network Approaches

• GAT (Graph Attention Networks)
• GCN (Graph Convolutional Networks)
• KCN (Kriging Convolutional Networks) https://ojs.aaai.org/index.php/AAAI/article/view/5716

todo list

• Add some usage examples
• visualize the interpolated graph