Description
Diagnostics are used throughout weather forecasting. They can identify problems, aid model development, and highlight successes. Fundamental goals for ECMWF and others include the improvement of ensemble systems and their underlying Earth system models (ESM). This workshop discussed how novel diagnostics can help us achieve these goals in physics-driven and data-driven systems. Themes include:
- Diagnostics of predictability – key initial condition and model uncertainties, pathways of error growth
- Process-oriented diagnostics – identifying issues, digging deeper, ESM interactions, learning from km-scale forecasts
- Data assimilation and ML as diagnostic tools – flow-dependent bias and uncertainty growth, explainable AI
- Community diagnostic packages and data archives – the benefits of comparison and communication
The use of examples of applications to weather phenomena (tropical waves and cyclones, extratropical storm-tracks and blocking, for example) helped make the workshop accessible to a wide-range of participants. Highlighting diagnostic success stories (where diagnostics have made a difference) was valuable.
Format
The workshop took place at ECMWF’s Headquarters in Reading on 9-12 September 2024, and consisted of in-person talks and poster presentations. Talks were live streamed.