Maximising Weather Forecasting Accuracy through the Utilisation of Graph Neural Networks and Dynamic GNNs

TL;DR

This paper explores the use of Graph Neural Networks and Dynamic GNNs to improve weather forecasting accuracy by analyzing heterogeneous multivariate meteorological data from various sensors.

Contribution

It introduces the application of GNNs and Dynamic GNNs specifically for weather forecasting, comparing their performance with traditional machine learning models.

Findings

GNNs outperform traditional models in weather prediction accuracy.

Dynamic GNNs adapt better to changing meteorological data.

The approach demonstrates potential for more accurate climate modeling.

Abstract

Weather forecasting is an essential task to tackle global climate change. Weather forecasting requires the analysis of multivariate data generated by heterogeneous meteorological sensors. These sensors comprise of ground-based sensors, radiosonde, and sensors mounted on satellites, etc., To analyze the data generated by these sensors we use Graph Neural Networks (GNNs) based weather forecasting model. GNNs are graph learning-based models which show strong empirical performance in many machine learning approaches. In this research, we investigate the performance of weather forecasting using GNNs and traditional Machine learning-based models.