Christian Sampson, Joint Center for Satellite Data Assimilation

Numerical Weather Prediction, JEDI, and a Hybrid Tangent Linear Model

Accurate weather prediction matters to everyone and has become even more important in our changing climate. We often refer to the combination of weather models and Data Assimilation as Numerical Weather Prediction (NWP), which aims to combine a model of the Earth's atmosphere with observations of it to produce as accurate a forecast as possible. This work is carried out in large operational centers across the globe which require massive compute power, many people, tons of code, and lots of science. Too often though, it can be difficult to incorporate new innovations into these systems, and when one innovation is proven, porting it to other agencies or groups requires a repetition of work slowing research to operations. The Joint Effort for Data assimilation Integration (JEDI) project aims to help alleviate this problem. JEDI is a generic Data Assimilation software suite developed for use by NOAA, NASA, The Navy, The Air Force and the UK Met Office at the Joint Center For Satellite Data Assimilation (JCSDA). This model agnostic system allows anyone with a model interface to JEDI to use its DA algorithms, observation operators, covariance matrices, and facilitate timely ingest of the newest observations. Further, when an innovation is added to JEDI it is available immediately for testing or deployment by multiple agencies and groups. In this talk I will briefly describe NWP, the JEDI project, and highlight one new innovation in the JEDI system, a hybrid tangent linear model for use in 4d-var schemes. I will also briefly describe how you can use JEDI or contribute to the project to accelerate your research to operations. Time permitting I may also highlight some other recent innovations in the JEDI system.

Passcode for this talk is math-geo