Distinguishing non- stationarity from finite interval effects in the intermittent solar wind.

The solar wind flow has a magnetic Reynolds number estimated ~ 105 and fluctuations in solar wind bulk plasma parameters typically show a clear region of power law scaling in the power spectrum with an exponent close to -5/3. Quantitative analysis of solar wind fluctuations are thus often performed in the context of intermittent turbulence and center around methods to quantify statistical scaling, such as generalized structure functions which assume a weakly stationary process. The solar wind exhibits large scale secular changes and so the question arises as to whether the timeseries of the fluctuations is non- stationary. One approach is to seek a local stationarity by restricting the time interval over which statistical analysis is performed. However multifractality implies that the scaling exponents are local in time. Computing scaling exponents over different intervals of a stationary multifractal process can thus potentially yield anomalously time varying values for the scaling exponents, suggestive of non stationarity. We investigate this using synthetic datasets generated from a (self affine) Levy flight and from a (multifractal) p- model. We are able to estimate the minimum interval (number of datapoints) needed to quantify the scaling exponents in a stationary multifractal process. With fewer datapoints the stationary timeseries becomes indistinguishable from a nonstationary process and we illustrate this with nonstationary synthetic datasets. Finally we apply these ideas to in- situ solar wind observations.