Currently the state of the art in the operational prediction of the background solar wind speed and the interplanetary magnetic field (IMF) polarity at earth via the so called WSA model (http://www.swpc.noaa.gov/ws/) is based on using an expertly and careful processing of the input data given by solar observations, e.g. the GONG magnetogram synoptic map <http://gong.nso.edu/data/magmap/>s. We investigate here the use of statistical techniques for data assimilation currently widely in use in oceanographic and atmospheric modeling. We base our approach on previous work (Barrero Mendoza, Oscar , Data assimilation in magnetohydrodynamics systems using Kalman filtering , KU Leuven thesis, 2005; Rigler, E.; Arge, C.; Mayer, L., Optimizing Coronal and Solar Wind Model Inputs with Data Assimilation, AGU Fall Meeting 2008) that has already proven the capability. The approach uses multiple runs of the modeling approach on statistically guided modifications of the input (such as the GONG magnetograms). Past observed outputs (such as solar wind speed and IMF obtained from the ACE satellite) can be used to better predict future values. Different statistical approaches will be compared.