WeatherBench is a data set compiled to serve as a standard for evaluating new approaches to artificial intelligence–driven weather forecasting.
New methods that help researchers understand the decision-making processes of neural networks could make the machine learning tool more applicable for the geosciences.
Sea ice area in CMIP6 is similar to previous versions while its sensitivity to external forcing is subtly different and closer to observations, but still not in step with global surface temperature.
The #GreatAntarcticClimateHack; La Jolla, California, 9–12 October 2017
Space weather forecasting is benefiting, and can benefit much further, from concepts and tools already developed by the global meteorological community.
A recent paper in Reviews of Geophysics explored the challenges of validating data collected from Earth observation satellites.
The software could help strengthen ice sheet models to provide a better basis for policy decisions.
Researchers create a new model of the electric currents circulating throughout the atmosphere that will improve the accuracy of global climate models.
A new study uses identical tests to evaluate the accuracy and performance of current models of Earth's magnetic field, then extrapolates the results to anticipated "petascale" supercomputers.
A new study shows that aerosols have a strong impact on the reflectivity of the Earth's atmosphere, absorbing more solar radiation within the atmospheric layers than previously thought.