Rainfall Trends of Historical Climate in Semarang
38 Kendal statistical test as performed in Table 3.2. However, this result will slightly
different with spatial distribution of rainfall trends in the city as performed in Figure 3.4. The absence of trend in the averaged rainfall is likely caused by uneven spatial
distribution of rainfall trends within Semarang city Figure 3.4. This contributes to the average value of rainfall that seems to show no significant trend in all seasons.
Figure 3.3: Observed seasonal rainfall over Semarang averaged from five rain gauge
Table 3.2: Mann-Kandal Test for Detecting Trend Season
Slope Lower
Upper Z
DJF -0.58574
-4.73124 2.799827 -0.31889 MAM
-1.2 -3.72318 1.572046 -0.74932
JJA -0.20784
-1.71996 1.347403 -0.20847 SON
0.859804 -2.92664 4.105129 0.625407
Figure 3.4: Spatial patterns of seasonal rainfall trends over Semarang. It should be noted that the selection of different ranges of data will affect the trend
and the result of statistical test conducted in this study. Therefore, despite the use of ~30 year observed data, a longer period of rainfall is also needed in order to confirm
100 200
300 400
500 600
1970 1975
1980 1985
1990 1995
2000
Year S
e a
s o
n a
l R
a in
fa ll
A v
e ra
g e
DJF MAM
JJA SON
the consistency of rainfa grid data from Climate
Jones 2005. The data s area for a period of 1901
spatially averaged data extracted from the datase
By using the CRU data t of rainfall, especially in
indicate that the wet sea supported by the declini
found inconsistency betw and from ~100 year grid
differences are more lik the analysis. A longer da
linked to climate change that is more affected by
frequency climate events
Figure 3.5: Trends of se 6.9
In addition to the above frequency Figure 3.6.
Figure 3.6 demonstrates performed in Figure 3.5
during wet seasons SON frequency at the same se
century over Semarang c probability of floods in
season, indicating a dec chance of dry season tha
JJA, the trend seems to b nfall trend over Semarang. Therefore, we use obse
te Research Unit CRU, namely CRU TS2.0 a set has 0.5x0.5 degrees grid resolution covering
01-2002. In order to analysis the rainfall data over ta within Semarang city 110.25E-110.51E, 7.1
asets. ta that has a longer period, Figure 3.5 shows an inc
in SON and DJF. The upward rainfall trends in easons tend to come earlier and end slower than u
ining trend of rainfall in dry season MAM and JJ etween the trends resulted from using ~30 year
ridded observations of CRU TS2.0 in Semarang ci likely caused by the dissimilar length of time pe
data will give better description of the actual tren ge impact, while a shorter data will tend to repres
by climate variability, especially by the oscillat nts.
f seasonal rainfall in Semarang city 110.25E-110 6.95S extracted from CRU TS2.0 dataset
ve analysis, we also investigate the trends in seaso 6. The data is also collected from the CRU T
tes similar trends of wet days frequency with the r 5 for all seasons except in JJA. The upward tren
ON and DJF are associated with increasing trend seasons. This indicates that the increasing rainfa
g city is caused by the rains that came more ofte in the region. In contrast, a downward trend appe
ecrease of wet days frequency that is associated that comes earlier. Especially for the wet days freq
o be relatively flat with a very slow increase.
39 bserved rainfall
0 Mitchell and ring global land
ver Semarang, a 7.12S-6.95S is
increasing trend in both seasons
n usual. This is JJA. Here we
ar observations city. The trend
periods used in rend that can be
resent the trend llations of low-
10.51E, 7.12S- asonal wet days
TS2.0 dataset. e rainfall trends
rends of rainfall nds of wet days
nfall during 20
th
ften, raising the ppears in MAM
ted with greater requency during
Figure 3.6: Trends of 110.51E, 7
Given the result shown i in rainfall data, it is evid
In addition, the low freq the low-frequency of I
Oscillation IPO Folla Mantua Hare 2002
relationships between th ENSO intensity and freq
al. 2008; Barnett et al., the PDO IPO, an incr
compared to El Nino, a during the positive phas
events considerably incr long-term rainfall variab
also by the low-frequenc show changes from curr
the conditions of future component of rainfall de
s of seasonal wet days frequency in Semarang city , 7.12S-6.95S extracted from CRU TS2.0 dataset
n in Figure 3.10 and the fact of significant upward vident that global warming has important role to th
requency oscillations found in rainfall data could f Indo-Pacific climate drivers such as Interdec
lland et al. 1999 or Pacific inter-Decadal Oscil 2 ; Mantua et al. 1997. Several studies have s
these interdecadal climate phenomena with the requency Saji Yamagata 2003; Salinger et al. 2
l., 1999, White Cayan, 2000. During the nega crease in the number of La Nina events are qui
, as happened in the period between 1948-1976. ase, e.g. in the period of 1972-1990s, the numbe
ncreased than in the negative phase. This study sh iability in Semarang is not only affected by climat
ency climate drivers. If both of these component urrent conditions, this may result in uncertainty o
re rainfall over the city. Figure 3.7 shows the lo defined by a simple moving average.
40 ity 110.25E-
set ard trend found
o these changes. ld be related to
decadal Pacific cillation PDO
e shown strong the changes of
. 2001; Wang et gative phase of
uite significant 76. Conversely,
ber of El Nino shows that the
ate change, but ents continue to
y of changes in low-frequency
Figure 3.7: Low-frequen simple 13-year moving a