Submitted abstract to the 24th General Essembly in Den Hague, Netherlands, April 19-23, 1999.
We have performed nonlinear analysis of high resolution time series of the solar wind density and velocity obtained from proton monitor onboard the SOHO spacecraft. Using several different nonlinear predictability measures we demonstrate that the data are significantly different from so-called surrogate data sets. Those share identical linear properties with the original time series, i.e. the power spectrum as well as the probability distribution, but are realizations of a linear, stochastic process. The existence of nonlinearities in the density time series on minute time scales is shown.
We further discuss the possibility and eligibility of applying nonlinear prediction algorithms to the time series to obtain good short term predictions.