Is there evidence for the existence of nonlinear behavior within the interplanetary solar sector structure?

Using data from the Sweden and Britain Radar Experiment (SABRE) VHF coheren t radar, Yeoman ef nl. [1990] found evidence for two and four sector struct ures during the declining phase of solar cycle (SC) 21. No such obvious har monic features were present during the ascending phase of SC 22. It was sug gested that the structure of the heliospheric current sheet might exhibit n onlinear behavior during the latter period. A direct test of this suggestio n, using established nonlinear methods, would require the computation of th e fractal dimension of the data, for example. However, the quality of the S ABRE data is insufficient for this purpose. Therefore we have tried to answ er a simpler question: Is there any evidence that the SABRE data was genera ted by a (low-dimensional) nonlinear process? If this were the case, it wou ld be a powerful indicator of nonlinear behavior in the solar current sheet . Our approach has been to use a system of orthogonal linear filters to sep arate the data into linearly uncorrelated time series. We then look for non linear dynamical relationships between these time series, using radial basi s function models (which can be thought of as a class of neural networks). The presence of such a relationship, indicated by the ability to model one filter output given another, would equate to the presence of nonlinear prop erties within the data. Using this technique, evidence is found for the pre sence of low-level nonlinear behavior during both phases of the solar cycle investigated in this study. The evidence for nonlinear behavior is stronge r during the descending phase of SC 21. However, it is not possible to dist inguish between nonlinear dynamics and a nonlinearly transformed colored Ga ussian noise process in either instance, using the available data. Therefor e, in conclusion we find insufficient evidence within the SABRE data set to support the suggestion of increased nonlinear dynamical behavior during th e ascending phase of SC 22. In fact, nonlinear dynamics would seem to exert very little influence within the measurement time series at all, given the observed data. Therefore it is likely that stochastic or unresolved high-d imensional nonlinear mechanisms are responsible for the observed spectrum c omplexity during the ascending phase of SC 22.