Improving solar wind forecasting using Data Assimilation

In terrestrial weather prediction, Data Assimilation (DA) has enabled huge improvements in operational forecasting capabilities. It does this by producing more accurate initial conditions and/or model parameters for forecasting; reducing the impacts of the “butterfly effect”. However, data assimilation is still in its infancy in space weather applications and it is not quantitatively understood how DA can improve space weather forecasts.

To this effect, we have used a solar wind DA scheme to assimilate observations from STEREO A, STEREO B and ACE over the operational lifetime of STEREO-B (2007-2014). This scheme allows observational information at 1AU to update and improve the inner boundary of the solar wind model (at 30 solar radii). These improved inner boundary conditions are then input into the efficient solar wind model, HUXt, to produce forecasts of the solar wind over the next solar rotation.

In this talk, I will be showing that data assimilation is capable of improving solar wind predictions not only in near-Earth space, but in the whole model domain, and compare these forecasts to corotation of observations from STEREO-B at Earth. I will also show that the DA forecasts are capable of reducing systematic errors that occur to latitudinal offset in STEREO-B’s corotation forecast.