Climate Data Imputation (GNN)
Python
PyTorch
Graph Neural Networks
Climate Data
Project Overview
This project built a Graph Neural Network (GNN) to predict missing precipitation data across Nepal’s weather stations. The dataset included ERA5 weather data and local station measurements, addressing data gaps in rural regions.
Key Features
- Converted NetCDF data to time-series for PyTorch DataLoader.
- Modeled weather stations as graph nodes for spatial dependencies.
- Optimized GNN to predict missing precipitation values with high accuracy.