Researchers extend classical CNOP method for deep learning forecasting models with multi-time-slice-input structure.  It reveals when—not just where—input errors matter most in targeted observations. This improves forecasts for ocean-atmospheric variables, especially high-impact environmental events.

A new online platform, SMRT-Flames, can identify areas in need of land management in order to reduce future smoke exposures from uncontrolled fires. While most wildfire tools predict fire risk, SMRT-Flames explicitly considers smoke exposure across populations

A Dartmouth study uses machine learning to reexamine whether climate change is causing large waves in the polar jet stream that have brought Arctic-like temperatures and storms to temperate regions of the United States in recent years. The researchers constructed a timeline of the jet stream's wintertime variability since 1901 and found it's in the latest of several “wavy” periods from the past 125 years, most of which predate significant effects of climate change. The authors report that climate change is likely not amplifying extreme winter weather by making the jet stream wavier, but through more direct links such as a warmer atmosphere that retains more moisture.