Predicting Atmospheric Rivers with AI: The Power of Convolutional Autoencoders
Автор: Water, Climate and Sustainability Lab, IIT Indore
Загружено: 2025-12-18
Просмотров: 10
An innovative approach to predict atmospheric rivers: Exploring convolutional autoencoder
Authors: Shivam Singh, Manish Kumar Goyal
Atmospheric rivers are narrow corridors of intense moisture transport that play a critical role in shaping floods and water availability in mid-latitude regions. This video explains a novel data-driven approach to model and predict atmospheric rivers using deep learning. By applying an autoencoder architecture, the study captures the complex, nonlinear dynamics of integrated water vapor transport (IVT) associated with atmospheric rivers.
The model is trained using the Adam optimizer with ReLU and Sigmoid activation functions, and its predictive skill is evaluated using standard performance metrics including SSIM, MSE, RMSE, and PSNR. High similarity and accuracy scores demonstrate the model’s effectiveness in forecasting atmospheric river occurrence at future time steps. The findings highlight the potential of deep learning–based forecasting to improve early warnings, reduce flood risk, and enhance resilience in regions vulnerable to atmospheric river impacts.
DOI: https://doi.org/10.1016/j.atmosres.2023.10...
Доступные форматы для скачивания:
Скачать видео mp4
-
Информация по загрузке:
