The prediction of solar wind values at L1 is of limited value without confidence limits on these predictions. These can only easily be derived from ensemble runs of solar wind codes driven from observations. These also need to be real predictions for expected values 1-2 days into the future.

The approach of this WP is to use boundary driving at 21.5 solar radii from the University of Michigan AWSoM-R code, or from the DEF code to be developed in WP2, to predict solar wind values at L1 based on ensemble runs of the SWIFT code. These ensemble runs will be guided by a sensitivity analysis of SWIFT from boundary variations and the output of the ensemble at L1, along with in situ measurements used to complete a Bayesian inference improvement of each ensemble elements weight. The final output will therefore be a prediction at L1 with quantified and reliable estimates of confidence levels from an optimised suite of simulation codes. A library of AWSoM-EEGGL runs with various CME initial conditions will be generated throughout the project timeline, including testing of the method of incorporation into the SWIFT input dataset.

Lead: The University of Warwick (Prof Tony Arber, Dr Keith Bennett)