Figure 6 is here.
Fig. 6 Information-Complexity diagram for the streamflow data
according to different aggregation length.
To give a conceptual understanding about the high frequency findings,
metric entropy and fluctuation complexity according to different
aggregation lengths are displayed in Fig. 7 (a, b). Since the
investigated period is relatively short (i.e. the hourly data from
2016-01 to 2016-06), the results, will not be sufficiently informative,
especially, for the long ranges. However, the comparison manifests some
interesting outcomes. First, the computed metric entropy (Fig. 7a),
shows that there is a notable variation between the streamflow data
records obtained by Ozekiyama, and FAT (since both of these stations are
located at the same site). More importantly, the estimated information
content by means of FAT has higher values compared to RC, particularly,
for AL \(\leq\) 4 hours suggesting that there is an additional scaling
regime occurs during sub-daily scales (i.e. few hours), and hence the
FAT is capable to capture the streamflow fluctuations that occur during
hourly scales. Apparently, both high and low frequencies confirm that
the information contents (i.e. metric entropy) at small aggregation
lengths have consistent slope as streamflow computed by means of RC
approach (Fig. 7a, and 7c). Alternatively, the fluctuation complexity
estimates (Fig. 7b), demonstrates that there is remarkable difference
between FAT and Ozekiyama estimates, therefore, it is advised to
consider high resolution streamflow records to accurately investigate
the hidden phenomena that cannot be observed by conventional discharge
calculation methods.