Met Office AMV Processing in JEDI
The Next Generation Modelling Systems (NGMS) programme aims to reformulate and redesign our complete weather and climate research and operational/production systems to allow the Met Office and its partners to fully exploit future generations of supercomputer.
A component of NGMS is the Next-Generation Processing and Assimilation of Observations (NG-PAO). This project will replace the current Observation Processing System (OPS) and Variational Assimilation System (VAR) used by our global and regional atmospheric data assimilation (DA) systems. To achieve this, NG-PAO has adopted the Joint Effort for Data assimilation Integration (JEDI) code framework developed by the Joint Center for Satellite Data Assimilation (JCSDA). This adopts a generic, object-oriented programming approach, providing a standard interface between models, observations, and DA (model agnostic).
This paper will give an overview of the NG-PAO project, its expected benefits, and describe the results and challenges of implementing the Met Office Atmospheric Motion Vector (AMV) observation processing code in NG-PAO. This has included porting the Met Office situation-dependent observation error scheme, observation operator, inversion pressure correction, quality control, and observation thinning procedures. Extensive testing and validation have shown that the JEDI-based AMV processing is essentially equivalent to the OPS. This has been confirmed using a 3-month long assimilation experiment which showed a neutral impact when the JEDI processing is used.