Professional Documents
Culture Documents
ISSN 1818-4952
IDOSI Publications, 2013
DOI: 10.5829/idosi.wasj.2013.28.07.946
Abstract: This study aims to analyze the long-term precipitation series using nonparametric methods (i.e. the
MannKendall and Time Series techniques) spanning from 1956 to 2005 in Gorgan station. The monthly total
of precipitation data collected from the Bureau of Meteorology and then the time series plots produced using
AnClim software.The outputs indicated that the direction of wintertime precipitation trend in polynomial
regression is, in general, downward in Gorgan station as well as the confidence limit bands are close to the
polynomial fit line in the middle periods.The normal distribution indicated a right skewness in precipitation time
series and the curve is completely unsymmetric and slightly bell-shaped. The spectrum of precipitation time
series (red-noise) in Blackman-Tukey method showed a continuous decrease of spectral amplitude with slightly
decreasing in precipitation frequency. As well as, the power spectra for analysis of cycles illustrated thatthere
are significant cycles of 5-years in the series. The results of u and u' time series diagrams in Mann-Kendall
method also indicated that there is not a meaningful trend or a sudden change (jump) in the time series.
A natural negative/descending trend is seen without jump, though.
Key words: Winter Precipitation Series
Frequency Mann-Kendall
INTRODUCTION
Mohsen Soltani, Department of Climatology, Faculty of Geography, University of Tehran, Tehran, Iran
Tel: +98 9119772934.
902
Fig. 2: (a) winter total of precipitation time series together with 95% confidence limits (blue dashed lines) for the
polynomial regression trend line (red bold line) and (b) the residual plot for same data as a. The winter total of
precipitation ranged from 1956 to 2005
903
50 years
Arithmetic Mean
173.8735 mm
Standard Deviation
66.6584`
Variance
4443.3487
Coefficient of Variance
38.34%
Coefficient of Skewness
2.1551
Coefficient of Kurtosis
8.4883
Maximal Precipitation
451.0 mm (1959)
Minimal Precipitation
75.0 mm (1963)
118.6
Median
172.9
197.9
Fig. 3: (a) time series frequency (blue bar-chart) of the winter total of precipitation with a normal distribution overlay (red
curve-line); (b) plot of cumulative deviation; (c) Blackman-Tukey method for variance spectrum analysis in
frequency time series along with white noise (pink dashed line-WN) and red noise with 95% confidence limits
(red dashed line-RN);and (d) plot of autocorrelation coefficient with confidence limits 95% (red dashed lines). The
wintertotal of precipitation ranged from 1956 to 2005
The sum of the deviations across the entire set of all
observations from the overall sample mean is always zero
and the average deviation is zero [28] Xue and
Titterington as follows:
s=
n t ( x i2 )t-( x i )2t
n t (n t - 1)
where, nt( xi)t2 is the total of the sums of all the squared
individual values and ( xi)t2 is the square of the total of
the sums of all the individual values and nt is the total
number of measurements in the time period of interest.
Thus, the magnitude of the total precipitation indicates
that the size of the difference in the years ranging from
1978 to 1995 is big. While it represents a small deviation
differences at the beginning of the examined period.
Frequency, by definition, measures the number of
times something occurs in a specific amount of time. It is
also referred to as temporal frequency, which emphasizes
the contrast to spatial frequency [29] Zwillinger and
Kokoska (2010).Frequencyof the total of precipitation time
series witha Blackman-Tukey method in variance
spectrum analysis overlaytogether with white noise (pink
dashed line-WN) and red noise with 95% confidence
limits (red dashed line-RN) shown in Fig. 3c. This method
Px ( f ) =|
M 1
wkrk e 2 ifk |
k =0
Fig. 4: (a) The power spectra(PS) analysis of winter total of precipitations frequency; and (b) same as a but for
dynamicalPS of frequency in 3D plot. The wintertotal of precipitation ranged from 1956 to 2005
In statistics, the autocorrelation of a random process
describes the correlation between values of the process
at different times, as a function of the two times or of the
time difference. It is a mathematical tool for finding
repeating patterns, such as the presence of a periodic
signal obscured by noise, or identifying the missing
fundamental frequency in a signal implied by its harmonic
frequencies. The definition of the autocorrelation between
times s and t is [30] Toros as follow:
R (s , t ) =
E[X t
t ](X s
t s
s)
t=
ni
i =1
t E (t )
Var (t )
Fig. 5: A 5-year moving average with a polynomial trendoverlay (left-side) and u and u'time series diagrams (right-side)
for winter total of precipitation series in Gorgan weather station
The u and u' components were successively tested
for the winter total of precipitation series. Using graphical
presentation of the standardized test statistic, the
development of trend in the series was easily traced.
A 5-year moving average time series with a polynomial
trend overlay (left-side) and u and u' time series diagrams
(right-side) are shown in Fig. 5. The moving average time
series illustrates a deep descending trend over a 50-year
period. As of 1998 forward, the trend rose up till the end
of the period, though. The u and u' time series diagrams
also indicate thatthere is not a meaningful trend and/or a
sudden change (jump) in the time series, as the u & u'
lines are almost moving parallel to each other except for a
few contacts, which it has not resulted in achange in
direction
of
the
trend
(jump).Anatural
negative/descending trend is seen without jump, though.
CONCLUSIONS
Long-term analysis of winter total of precipitation
records at Gorgan weather stationin north of Iran has
been carried out through two procedures viz time series
technique and Mann-Kendall non-parametric test using
monthly total series. The trend in polynomial regression
indicated a slightly descending tendency toward the end
of the period. The confidence limit bands were close to
the polynomial fit line in the middle terms - ranging from
1970 to 1990, when the fit was good. The direction of
precipitation trend was, in general, downward in Gorgan
station. The residual plot showed some odd data or
curved trends in the plot, the average of the residuals was
zero and the bars were not equally represented about the
x-axis. In addition, the winter total of precipitation in
Gorgan station with a normal distribution had a right
skewness and the curve was completely unsymmetric and
slightly bell-shaped due to having small values in both
standard deviation and kurtosis.Analysis of the
ACKNOWLEDGMENTS
The authors wish to express their sincere thanks to
Petr t pnek- authorof AnClim software - for helpful
support during some series plots analysis used in this
research. The monthly total of precipitation data provided
by Iranian Meteorological Organization (IRIMO).
REFERENCES
1.
2.
3.
907
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
908