JADA-Regional is designed to:

  • Produce very short-range ensemble forecasts by drawing on the latest local observations.  


  • Operate as a rapidly cycling ensemble system (hourly or less).  


  • Focus on analyzing small-scale phenomena like convective scales.  


  • Integrate with Nowcasting (up to T+2h) and Mogreps UK (from T+24h).  


  • Enhance forecast performance up to T+12/24h.  


  • Capture extreme weather events.  


The development plan is divided into two horizons:

  • Horizon A focuses on establishing an Ensemble Kalman Filter (EnKF) framework, including large JEDI development and deciding on the EnKF flavor.

  • Horizon B aims to improve the use of convective-scale observations through correlated observation error and to balance large-scale and small-scale information via large-scale blending and localization.

The development timeline shows that JADA-Regional is targeted for implementation at PS51, with key development activities planned between 2024 and 2029.  


The presentation also details the code development aspects, including:

  • The current code structure using local volume solvers like LETKF and GETKF.  


  • The proposed code design, which involves refactoring and enhancements like cross-validation, correlated observation error, multiscale localization, and filters for handling bounded variables.  


  • The methods for computing and applying weights in deterministic and stochastic LETKF/GETKF, with and without cross-validation.

  • A delivery timeline with specific deliverables planned for each quarter from Q1 2025 to Q3 2025.  


  • No labels