Analysis of Trends in Low-Flow Time Series of Canadian Rivers - Hydrologic Time Series Analysis: Theory and Practice - page 214

the LTP case compared to those shown in Figs 10.10(a-c) for the STP case
(note the scale difference) suggests that it is very less likely to find trends with
the LTP based tests. In other words, it means that many of the significant
trends with the independence assumption can be explained merely on the
basis of the serial structure of the time series.
Overall, without reference to positive and negative type of trends, the
trends in annual, winter and summer low flows identified with the MK test
and those with the STP based tests were found field significant at the 5%
significance level using the FDR test; for the case of LTP based tests, trends
in summer low flows were found field significant at the 5% significance level
and those in winter low flows were found field significant only at 10%
significance level. A similar analysis performed separately for positive and
negative trends led to the following conclusions. Negative trends in annual
and summer low flows were found field significant at the 5% significance
level for the case of MK and both STP based tests and those in winter flows
were not found field significant even at the 10% significance level. Positive
trends in annual and summer low flows were found field significant at 5%
significance level only for the case of MK and MMK1 tests and not for the
case of MK-BBS test, while those in winter low flows were found field
significant for the case of MK and both STP based tests. For the LTP based
tests, only negative (positive) trends in summer (winter) low flows were
found field significant at 5% (10%) significance level. The Canadian RHBN
spans over many different climatic regions and hence it would be more sensible
to perform field significance analysis on the basis of suitably defined

Fig. 10.10. The effect of assuming (a-c) only STP and (d-f) LTP and/or STP, where
applicable, on the significance of trends shown as differences in

p -values (e.g.

p -value for the MK test). The time series for

which the first (or the first few) autocorrelation(s) was (were) found significant
are circled. Station indices are the same as shown in Fig. 10.5. The six figures
are adopted from Khaliq et al. (2009b) and modified to improve their clarity.

-value for the MK-BBS test minus p