Accurately forecasting extreme temperature events remains a critical challenge for climate adaptation and disaster preparedness. Now, researchers have developed Hankelformer, a novel deep learning architecture that integrates structured Hankel-based augmentation with contrastive learning to significantly improve extreme weather forecasting. Tested on real-world extreme events including the 2021 Texas winter storm and the Pacific Northwest heat dome, Hankelformer achieved up to 34% reduction in prediction error compared to leading baseline models, offering a promising tool for reliable extreme temperature forecasting.

Water is a crucial catalyst for proton transport, but what is the minimal structural threshold for the onset of efficient catalysis? For the first time, researchers from East China Normal University tracked the birth of size-resolved hydrated protons in real time. They discovered a dramatic “kinetic collapse” when three water molecules form a ring. This cyclic trimer acts as a structural switch, eliminating reaction barriers and triggering highly efficient catalytic dynamics in 70 femtoseconds.

Ovarian ageing is a primary cause of declining female fertility, characterised by a reduction in both oocyte number and quality. A new study from Professor Lin Liu and Professor Zhengmao Zhu’s team at the College of Life Sciences, Nankai University, published in Science China Life Sciences, reveals that induced pluripotent stem cells (iPSCs) reprogrammed from granulosa cells of reproductively aged mice exhibit a markedly reduced capacity to differentiate into primordial germ cell‑like cells (PGCLCs). This defect cannot be rescued by simply improving mitochondrial function, but inhibition of the ERK/MAPK signalling pathway significantly enhances differentiation efficiency. The findings provide new insights into the decline of germ‑cell regenerative potential during ovarian ageing and identify a potential intervention target for future fertility restoration using autologous cells.