A river flow regime describes an average seasonal behaviour of river flow,
usually representing a set of long-term monthly mean values. Seasonal patte
rns of flow can be regular, repeating in principle the same pattern from ye
ar to year, or irregular, i.e. alternating between a couple of different re
gime types during individual years. By tradition, a river flow regime has b
een considered as a static characteristic that does not change in large tem
poral scale, yet this may be an oversimplification with regard to constantl
y changing environmental conditions. Such a "static" description of a flow
regime, based on long-term mean values is in discrepancy with the dynamic c
haracter of the system described. The dimension of this system in terms of
fractal and intrinsic dimensions has been investigated on an example of Sca
ndinavian runoff series by different methods: from a simple graphic method
to determine the fractal dimension to empirical orthogonal function (EOF) a
nd entropy-based aggregation to outline the number of patterns necessary fo
r representing the regimes of different intrinsic dimension of a set of poi
nt data. The series studied demonstrated a variety of fractal and intrinsic
dimensions that were well in agreement with the stability character of the
investigated regime types. The less stable the regime, the higher were its
fractal and intrinsic dimensions and the number of variables required for
its description.