NG002
Advancing Data Assimilation for Earth System Prediction
Advancing Data Assimilation for Earth System Prediction
Session ID#: 250350
Session Description:
Data Assimilation (DA) is essential for Earth System Prediction (ESP), constraining models with observations for accurate state analysis by optimally blending forecasts and data. Significant challenges persist, particularly in using real-world data and applying DA to complex scenarios or extreme events. Key difficulties include characterizing model error, representing realistic background error covariances (B), observation operators (H), and observation errors (R). These are amplified in coupled DA due to cross-component interactions, complicating error modeling and limiting Earth System Prediction (ESP) skill. This session seeks diverse approaches, including ML/AI, to address these issues by learning error representations, emulating components, improving quality control, and tackling coupled DA problems. We invite submissions showcasing theoretical, methodological, and applied DA advancements across all Earth system domains and timescales that demonstrably improve DA performance and prediction skill. The goal is fostering discussion on progress, challenges (like physical consistency, uncertainty), and innovative solutions, including ML/AI integration, for ESP.
Co-Sponsor(s):
- A - Atmospheric Sciences
- GC - Global Environmental Change
- H - Hydrology
- P - Planetary Sciences
Index Terms:
1816 Estimation and forecasting [HYDROLOGY]
3315 Data assimilation [ATMOSPHERIC PROCESSES]
4430 Complex systems [NONLINEAR GEOPHYSICS]
4499 General or miscellaneous [NONLINEAR GEOPHYSICS]
Primary Convener: Soyoung Ha, National Center for Atmospheric Research, Boulder, CO, United States
Convener: Moha Gharamti, NSF National Center for Atmospheric Research, Computational and Information Systems Lab, Boulder, United States
See more of: Nonlinear Geophysics