Detection of a long-range time dependence in the radial cross-correlation f
unction is normally difficult because of the oscillatory behavior of the cr
oss-correlation tail, its low level of coherence, and noise contamination.
This problem persists, even with large statistical samples. In this paper,
a method for investigating long-range dependence in a single time series is
extended to the calculation of the cross-correlation function. With this m
ethod and for time series with long-range time correlations, the accuracy o
f the determination of the cross-correlation function for long time lags is
improved. The method is tested by applying it to fractional Gaussian noise
and to the fluxes in a running sandpile model. This analysis technique can
be applied to the detection of avalanche-type transport in magnetic confin
ement devices. (C) 1999 American Institute of Physics. [S1070-664X(99)02402
-7].