AI for Climate Risk

AI Foundation Model for Weather and Climate

• Prithvi WxC(open-source) In collaboration with NASA, IBM built a general-purpose AI foundation model that could be customized for a range of practical weather and climate applications, at varying spatial scales. Potential applications include creating targeted forecasts from local weather data, predicting extreme weather events, improving the spatial resolution of global climate simulations, and improving the representation of physical processes in conventional weather and climate models.
• Aurora A foundation model developed by Microsoft that can do high-resolution (11-km) weather forecast and even air pollution forecast.

AI for Weather forecast

• GraphCast Global weather forecast model developed by Google DeepMind.
• Panggu(open-source) Global weather forecast model developed by Huawei, China
• Fuxi(open-source) Global weather forecast model developed by Fudan University, China
• FourCastNet
• Spherical Fourier Neural Operators
• Aardvark weather: end-to-end data-driven weather forecasting An end-to-end weather forecasting system proposed to replace the entire operational numerical weather forecast pipeline. Aardvark directly ingests raw observations and is capable of outputting global gridded forecasts, as well as local station forecasts.

Climate Downscaling

• Localized weather GNN by Earth Intelligence Lab It downscales gridded weather forecasts, such as ERA5, to provide accurate off-grid predictions.

Climate Risk Forecasting

• A nonstationary stochastic simulator for clustered regional hydroclimatic extremes to characterize compound flood risk
• A stochastic ML model of wildfire activity in western US
• A statistical model to create gridded (4-km spatial resolution), monthly predictions of burn area as a function of climatic variables.