The new graph convolutional recurrent neural network (GCRNN) will enable water utilities to forecast water use, even if some sensors fail.
More accurate aftershock zones reveal that the rupture areas of megathrust Aleutian–Alaska earthquakes are larger than we thought and partly overlap, in contradiction with the seismic gap hypothesis.
Researchers present a new approach to modeling the stochastic mixing process of convection using a machine learning technique.
Hurricane winds can lead to coast downwelling, which brings warmer surface water near the coast and can contribute to the intensification of the landfalling hurricane.
As cities face health threats from heat and air pollution—both expected to worsen from climate change—researchers pilot a community scientist effort to map air quality and improve urban health.
Adjoint tomography employing 3D wavefield simulations for 72 well recorded regional earthquakes in the western U.S. yields spectacular improvements to waveform fits.
The Lut Desert in Iran is an exceptional natural laboratory to study how wind moves sediment across the landscape. A new study quantifies erosional and depositional sediment fluxes of the desert.
A new study reveals how small cracks turn into gigantic submarine slides.
Cloud-to-ground lightning is found to be the most important controller of wildfire occurrence in the Artic tundra of Alaska from 2001 to 2019.
Models show that several puzzling features about Ceres’ topography, gravity anomalies, and crater size distribution may be explained by asymmetric hemispherical convection due to radiogenic heating.