← Back to all news
arXiv AI By Etienne Lehembre (CA, LIFO), Pascal Audigane (BRGM), Vincent Nguyen (LIFO), Christel Vrain (LIFO, CA), Thi-Bich-Hanh Dao (LIFO, CA)

Physical knowledge on historical data matters more than enforcing physical constraints on the forecast

Read the original on arXiv AI →

The paper introduces a Physics Informed Recurrent Neural Network (PIRNN) that simultaneously predicts target time series and unobservable intermediate physical variables, enhancing robustness and interpretability. It adapts to any physical model with multiple equations and variables, demonstrated on groundwater level predictions using the Gardenia model. Experiments on twelve real‑world datasets show PIRNN outperforming several neural network baselines and the Gardenia model, with an ablation study confirming the value of physical knowledge.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

Related stories

arXiv Machine Learning
Aug 14

History-informed Lagrangian Neural Networks

arXiv:2608. 13215v1 Announce Type: new Abstract: Forecasting the long-horizon evolution of mechanical systems from position-only observations is a pivotal yet difficult task, as hidden velocities and trajectory-specific physical properties must be inferred simultaneously.

By Tianshuo Zhang, Xianglei Xing, Wenzhe Zhai, Jia Gao, He Cao
benchmarks
More like this →
arXiv Machine Learning
Jul 14

Parameter estimation for land-surface models using Neural Physics

arXiv:2505. 02979v4 Announce Type: replace-cross Abstract: We propose a novel inverse-modelling approach that estimates the parameters of a simple land-surface model (LSM) by assimilating data into a differentiable, physics-based forward model formulated using convolutional operations.

By Ruiyue Huang, Claire E. Heaney, Maarten van Reeuwijk
More like this →
arXiv Machine Learning
Sep 2

GenONet: A Generative operator Network for High-Resolution Precipitation Nowcasting

GenONet introduces a Spatio-Temporal U-DeepONet architecture that serves as a generator in a GAN framework for high‑resolution precipitation nowcasting up to three hours ahead. By learning continuous‑time precipitation dynamics with a Deep Operator Network and enforcing physics through a moisture‑conservation loss, the model produces sharp, physically consistent forecasts that outperform baselines, especially for high‑intensity events and longer lead times. Ablation studies confirm the added value of the physics‑informed regularizer and the synergy of operator learning with adversarial training.

By Mohammad Kian Golkar, Luciano Alves de Oliveira, Mohammad Khanjani
diffusionbenchmarkssafety
More like this →