1. Sol-Terra: A Roadmap to Operational Sun-to- Earth Space Weather Forecasting Mike Marsh 1, David Jackson 1, Alastair Pidgeon 2, Gareth Lawrence 2, Simon Reid 2, Mario Bisi 3, Mike Hapgood 3, Yulia Bogdanova 3, Jason Byrne 3 1 Met Office 2 RHEA TECH 3 STFC RAL Space

2. © Crown copyright Context Sol-Terra Motivation “Develop a roadmap for the realisation of an operational end-to-end space weather forecast modelling system, and identify the main areas where further work is needed.” Objective Timeliness Robust (24/7/365) Verified Evolution Quality software engineering System/architecture High quality service

3. © Crown copyright Scientific Domain Overview Current space weather modelling capabilities Scientific Background & Model Scope Operational/Computational Aspects Inputs & Coupling Operational Sustainability Model Review Operational Constraints Time Constraints – Computational/Data Latencies Source Code Quality Assurance Factors Documentation Standards Version Control (e.g. Git, Subversion) Error Handling Languages, Dependencies Verification & Ensembles Operational Forecast Potential Owens et al., 2014

4. © Crown copyright Coupling “Understand the highly coupled chain of processes that link solar activity, the solar wind and the Earth’s magnetosphere to its upper atmosphere.” Long Term Scientific Goal Replicate this process chain with a suite of coupled models that is operationally viable Forecasting Goal Coupling Gap Identification

5. © Crown copyright Architecture Platforms Distributed/cloud computing Operational suite definition Workflow/Scheduling Operational System Architecture

6. © Crown copyright Operational Requirements Research Operations 2Model Transition

7. © Crown copyright Operational Requirements Development Practices R2O Model Uptake Research Impact

8. © Crown copyright Robustness: models should run successively for a range of space weather conditions and handle errors appropriately and informatively, allowing operational service and IT support teams to understand and resolve problems. Forecast Cycle: The model should run fast enough to be used within a forecasting cycle excluding data latencies (varies by domain and conditions). Quality Assurance: High standard of code structure, documentation, error handling and version control allowing systematic model management i.e. perform acceptance tests, reviewing procedure for code changes and allow operational service and IT support teams to resolve problems. Environment: Developed using appropriate operating system of operational centre (for Met Office this is Linux operating system). Language: Source code available and model written in appropriate language (for Met Office this is Fortran, Python or Java. C and IDL may also be acceptable for critical models, however with lower operational support). Licensing: License, IPR and terms of use for model and input data should be appropriate and obtainable. Efficiency: Unless model is computationally very cheap, code should be parallelised to ensure HPC operation (supported OpenMP and MPI protocols). Resilience: Fall back option of using a simpler configuration, other initialisation, repeat forecast, or alternative input data source to maintain continuous forecasting capability (e.g. solar or geomagnetic drivers as input) in case of technical issues (e.g. data availability). Dependencies: Models should not have dependencies on non- standard libraries not under the operational centre control (e.g. SolarSoft). Coupling: Model should be suitable for coupling to other appropriate models with compatible boundary conditions and input/output parameters. System Requirements Met Office User & System Requirements Timeliness: Data assimilated (NRT) and forecasts produced in a timely manner. Data Model: Models should be data driven rather than climatology. User Documentation: High standard of user documentation, describing model overview, input/output and limitations on validity. Evaluation: Model statistical verification skill scores defined to inform forecaster interpretation. Skill scores available to benchmark impact of replacing standalone model with coupled model, or to benchmark model upgrades. Ensembles: Ensemble operation possible (unless deterministic models are shown to perform better). Autonomy: Models should have the potential to run automatically, without human intervention, producing output for forecaster assessment. User Requirements Space Weather Operational Meteorologist User MOSWOC Infrastructure System Space Weather Operations Centre (MOSWOC)

9. © Crown copyright Sol-Terra Outputs/Impacts Identify gaps in capability/Areas of further development Future research needs Roadmap for coupled proto-operational forecast system Synthesis of Domain Model Reviews Capability of current space weather models to contribute to coupled operational suite Requirements for R2O Knowledge gaps  academic community Awareness of operational issues  enabling R2O  Impact Impacts

10. © Crown copyright Research Operations Research Impact Research Funding Improved ScienceR2O