Data assimilation for AU index

The auroral electrojet indices (AU, AL, AE) are a proxy of substorm as well as auroral activity, so that the forecast of these indices is important for the space weather forecast. In this study, we develop a data assimilation code for estimating the AU index based on Goertz's [1993] model. As the assimilation scheme, we use the particle filter and smoother. The state vector includes the AU index and coupling parameters between the solar-wind, the magnetosphere and the ionosphere. We try to improve these coupling parameters for better forecast of the AU index. Using the data assimilation, we estimate the AU index and find that forecast performance is significantly improved using the data assimilation. We also estimate long-term trend of the coupling parameters. The estimated coupling parameters have semi-annual modulations as well as the long-term modulations. The semi-annual modulations seem to be consistent with the seasonal variations of the ionospheric conductivity, so that our data assimilation detects the ionospheric effects on evolution of the AU index.