Designated neural network modules are combined to mimic numerically-discretized diffusion-sorption equations, which allows learning “missing pieces” in system understanding and their uncertainties.
machine learning & AI
Interactive Learning for Better AI-Based Subgrid-Scale Modeling
A study shows that interactive learning can significantly enhance the performance of artificial intelligence-based parameterization of small-scale processes, a critical component of climate models.
Scientists EEAGER-ly Track Beavers Across Western United States
Efficiently tracking nature’s engineers—beavers—at the scale of entire watersheds over time is now possible, thanks to a new artificial intelligence–trained model called EEAGER.
Using Big Data for Monitoring Network Design and Beyond
Large data sets can be generated using deep learning to improve the design of observation networks for monitoring subsurface flow and transport.
Enhancing Earthquake Detection from Orbit
A new application of machine learning boosts scientists’ ability to use data from satellite navigation systems to detect and warn of earthquakes.
Are We Entering The Golden Age Of Climate Modeling?
Thanks to the advent of exascale computing, local climate forecasts may soon be a reality. And they’re not just for scientists anymore.
Machine Learning Helps to Solve Problems in Heliophysics
A new special collection invites papers pertaining to the use of machine learning techniques in all sub-fields of heliophysics.
Machine Learning Looks Anew at Isotope Ratios in Oceanic Basalts
While past attempts to define isotopic endmembers and assign them a geodynamic significance ended in controversy, a machine-learning clustering algorithm offers a solution to this classical problem.
Deep Learning for Hydrologic Projections Under Climate Change
Extrapolation or not? Big data may help deep learning to go places where it has not been before by transferring learned hydrologic relationships.
How Quantum Computing Can Tackle Climate and Energy Challenges
The day is coming when quantum computers, once the stuff of science fiction, will help scientists solve complex, real-world problems that are proving intractable to classical computing.