Fractals are self-similar objects (which look the same at all scales), whose dimensions can be noninteger. They are mathematical concepts, useful to describe various physical systems, as the fractal dimension is a measure of their complexity. In this paper we study how these concepts can be applied to some problems in space plasmas, such as the activity of the Earth's magnetosphere, simulations of plasma turbulence, or identification of magnetic structures ejected from the Sun.

An algorithm is proposed to estimate the spectral index of the turbulence energy spectrum directly in the wavenumber domain using multiple-sensor-array data. In contrast to the conventional method using time series data and Fourier transform of the fluctuation energy onto the frequency domain, the proposed algorithm does not require the assumption of Taylor's frozen inflow hypothesis, enabling direct comparison of the spectra in the wavenumber domain with various theoretical predictions.