Artificial intelligence applied in disaster management must be reliable, accurate, and, above all, transparent. But what does transparency in AI mean, why do we need it, and how is it achieved?
machine learning & AI
Physics Meets Machine Learning for Better Cyclone Predictions
A new hybrid modeling approach combines physics-based and machine learning models to extend—and improve—path and intensity predictions of tropical cyclones.
Equation Discovery for Subgrid-Scale Closures
Machine learning can discover closure equations for fluid simulations. A new study finds that common algorithms rediscover known, unstable closures, which can be stabilized with higher-order terms.
New Model Can Better Predict Areas Vulnerable to Forest Fires in India
Researchers incorporated local atmospheric parameters and terrain data to more accurately estimate the probability of fire in a specific area.
Introducing the New Editor-in-Chief of the ESS Open Archive
Learn about the person taking the helm of the Earth and Space Science Open Archive and their vision for the coming years.
Physics + Machine Learning Provide a Better Map of Ocean Measurements
A new study offers a compelling example where the merger of dynamical modeling, machine learning, and ocean measurements enhances oceanographic understanding, monitoring, and mapping.
Machine Learning Masters Weather Prediction
Community datasets and evaluation standards are needed to further advance machine learning for weather prediction.
Autocalibration of the E3SM Atmosphere Model Improves Model Fidelity
A surrogate model was trained to predict E3SM atmosphere model spatial fields as a function of uncertain physical parameters and used to optimize the parameters for present-day climate.
Harmonizing Theory and Data with Land Data Assimilation
Land data assimilation advances scientific understanding and serves as an engineering tool for land surface process studies, reflecting the trend of harmonizing theory and data in the big data era.
Learning Data Assimilation Without the Help of the Gaussian Assumption
Major Earth system processes are non-linear and non-Gaussian, and so should be our data assimilation approaches.