We present a dimension analysis of a set of solar type I storms and type IV
events with different kind of fine structures, recorded at the Trieste Ast
ronomical Observatory. The signature of such types of solar radio events is
highly structured in time. However, periodicities are rather seldom, and l
inear mode theory can provide only limited interpretation of the data. Ther
efore, we performed an analysis based on methods of the nonlinear dynamics
theory.
Additionally to the commonly used correlation dimension, we also calculated
local pointwise dimensions. This alternative approach is motivated by the
fact that astrophysical time series represent real-world systems, which can
not be kept in a controlled state and which are highly interconnected with
their surroundings. In such systems pure determinism is rather unlikely to
be realized, and therefore a characterization by invariants of the dynamics
might probably be inadequate.
In fact, the outcome of the dimension analysis does not give hints for low-
dimensional determinism in the data, but we show that, contrary to the corr
elation dimension method, local dimension estimations can give physical ins
ight into the events even in cases in which pure determinism cannot be esta
blished. In particular, in most of the analyzed radio events nonlinearity i
n the data is detected, and the local dimension analysis provides a basis f
or a quantitative description of the time series, which can be used to char
acterize the complexity of the related physical system in a comparative and
non-invariant manner.
In this frame, the degree of complexity we inferred for type I storms is on
the average lower than that relevant to type IV events. For the type TV ev
ents significant differences occur with regard to the various subtypes, whe
re pulsations and sudden reductions can be described by distinctly lower va
lues than spikes and fast pulsations.