Dynamic Lag Encoding and Multi-Head Attention Improve Deep Learning Flood Forecasting
A deep learning model combining dynamic lag encoding and temporal multi-head attention significantly reduces peak flow forecasting errors across multi-hour ...
A deep learning model combining dynamic lag encoding and temporal multi-head attention significantly reduces peak flow forecasting errors across multi-hour ...
A long-term satellite and modeling investigation reveals that human activities outweighed climate shifts in driving terrestrial water storage depletion across ...
A satellite-based evaluation across the Loess Plateau revealed that while two primary Chinese soil erosion models captured an overall decline ...
Combining Lutan-1 and Sentinel-1 satellite radar with a random forest model, researchers identified 65 potential landslides in a Chinese oilfield ...
A coupled biogeochemical and hydraulic modeling study projects that future warming will increase water consumption and hydraulic failure risk in ...
A synthesis of 1,221 soil measurements from 86 studies finds that carbon gains on the Loess Plateau depend on both ...
Thirty-five years of outlet records from the Luoyugou catchment show sediment yield and sediment efficiency falling sharply after 2010 while ...
A new study combines two physically based models to estimate slope failure probabilities across China under 15 rainfall scenarios, 10 ...
A 2000–2024 study finds transpiration-driven evapotranspiration rising 2.58 mm per year in the middle reaches of the Yellow River, with ...
Researchers in Linfen, China, built an erosion-sensitive ecological security pattern whose modeled corridors stretched 814.22 kilometers and shifted priority restoration ...
© 2026 Scienmag - Science Magazine
© 2026 Scienmag - Science Magazine