Trend and Homogeneity in Subsurface Hydrologic Variables: Case Study in a Hard-Rock Aquifer of Western India - Hydrologic Time Series Analysis: Theory and Practice - page 170

not have significant variance at D = 0.05. It is apparent from Table 8.1 that the
calculated test-statistic values of the Levene's median test do not vary
significantly for the ten rainfall stations. Thus, the results of the Levene's
median test are in agreement with those of the Levene's ANOVA test. This
finding suggests that the annual rainfalls over the area are spatially
homogeneous.
8.5 Trend and Homogeneity in Seasonal Groundwater
Levels
8.5.1 Results of Trend Tests
Results of the three trend tests indicating number of sites with presence/
absence of the trends in pre- and post-monsoon groundwater levels are shown
in Figs 8.4 and 8.5, respectively. It is apparent that the number of sites with
significant trends of increasing groundwater level at 5% significance level
(D = 0.05) is approximately same for Mann-Kendall test (49% and 7% of the
sites in pre- and post-monsoon seasons, respectively) and Kendall Rank
Correlation test (51% and 8% of the sites in pre- and post-monsoon seasons,
respectively). However, Spearman Rank Order Correlation test does not indicate
nature of trend (i.e., increasing/decreasing) and results in relatively large
number of sites with significant trends (i.e., 61% in pre-monsoon and 20% in
post-monsoon) at D = 0.05 compared to two earlier trend tests.
Fig. 8.4. Trends in pre-monsoon groundwater levels based on
the three trend tests.
It is apparent from Figs 8.4 and 8.5 that the number of sites with significant
trends of increasing groundwater levels (at D = 0.05) is relatively more in the
pre-monsoon season than in the post-monsoon season. Presence of the
increasing trends of post-monsoon groundwater level at relatively small number
of sites compared to that in the pre-monsoon groundwater level is due to the