AI-based Ensemble Troposphere Conditions Forecasting using GNSS tomography
Weather phenomena forecasting is undergoing a significant transformation with the latest research, introducing a AI-based to atmospheric prediction. The new study, published in IEEE Transactions on Geoscience and Remote Sensing, marks the pioneering use of GNSS tomography for ensemble forecasting, setting a new pathway in the field. Traditional troposphere models often struggle with low spatial resolution and irregular data distribution. By leveraging GNSS troposphere tomography, the study generates high-resolution, three-dimensional wet refractivity fields, effectively addressing these critical challenges. Unlike previous deterministic models, the research employs ensemble forecasting using Generative Adversarial Networks (GAN), enhancing predictive accuracy by generating realistic time series data and providing probability-based forecasts. At the core of the model are Long Short Time Memory (LSTM) networks, optimized with Genetic Algorithms (GA), ensuring robust predictive performance.

The model was rigorously tested in Poland and California, two regions with distinct climatic conditions. It accurately predicted various weather phenomena, including rain bands and storms, demonstrating versatility and reliability. The results highlight the model’s low false positive rates (FPR) of 0.027 and 0.011 for Poland and California, respectively, underscoring its high sensitivity and dependability.

For more detailed information and results, please refer to the full article:

Saeid Haji-Aghajany, Witold Rohm, Maciej Kryza and Kamil Smolak (2024). Machine Learning-Based Wet Refractivity Prediction through GNSS Troposphere Tomography for Ensemble Troposphere Conditions Forecasting. IEEE Transactions on Geoscience and Remote Sensing, doi:

Poczta / Logowanie do systemu
GISLab - Laboratorium GIS
Stacja permanentna GNSS 'WROC'
Stacja permanentna GNSS 'WROC'
Wroclaw University of Environmental and Life Sciences
Grunwaldzka 53
50-357 Wroclaw

NIP: 896-000-53-54, REGON: 00000 18 67

Phone +48 71 3205617
Fax +48 71 3205617