See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323297328

Change Point Detection and Trend Analysis of Rainfall and Temperature Series
over the Vellar River Basin

Article in Polish Journal of Environmental Studies · February 2018

DOI: 10.15244/pjoes/77080

CITATIONS READS

26 1,105

2 authors:

Supriya P. Krishnaveni Muthiah

Anna University, Chennai Anna University, Chennai
7 PUBLICATIONS 48 CITATIONS 58 PUBLICATIONS 271 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

hydrological modelling of climate change impacts of a semi arid basin View project

Climatological homogenity and trend analysis of Anantapur precipitation View project

All content following this page was uploaded by Supriya P. on 27 February 2020.

The user has requested enhancement of the downloaded file.

Pol. J. Environ. Stud. Vol. 27, No. 4 (2018), 1673-1681
DOI: 10.15244/pjoes/77080 ONLINE PUBLICATION DATE: 2018-02-20

Original Research
Change Point Detection and Trend Analysis
of Rainfall and Temperature Series over the Vellar
River Basin

Supriya Palaniswami*, Krishnaveni Muthiah

Centre for Water Resources, College of Engineering, Anna University,
Guindy, Chennai-600025, India.

Received: 31 May 2017

Accepted: 18 September 2017

Abstract

Rainfall and temperature have been extensively considered as an initial point toward the apprehension
of climate change progressions, establishing one of the important constituents of the hydrological cycle.
The purpose of the present study is to examine the variability of rainfall and temperature for better
understanding of the hydrological environment of the river basin located in northern Tamil Nadu.
Mann-Kendall and Sen’s slope tests were employed for monthly, seasonal, and annual temperatures, and
also for annual maximum daily rainfall, seasonal, and annual rainfall statistics. A change point detection test
was applied for annual maximum and minimum, mean temperature, and annual precipitation series. The
results revealed that all the monthly, seasonal, and annual maximum, minimum, and mean temperatures
have a significant greater rising trend. The magnitude of increasing trends in NEM (northeast monsoon)
and SWM (southwest monsoon) are greater than that of summer and winter seasons for almost all the rain
gauge stations. The maximum temperature and minimum temperature change points are identified in the
years 1985 and 2001 and 1987 and 2013, respectively. From the mean annual temperature, it is seen that
the change point is present in 1983-88 and 2000-04 at 100% confidence interval.

Keywords: trend, change point, climate change, hydrological environment, river basin

Introduction in particular extreme events can intensely increase the

vulnerable population and affect the adaptation capability
The Fifth Assessment Report (AR5) released by the under the predicted future climate change scenarios by
Intergovernmental Panel on Climate Change documented the middle and at the end of the 21st century [1]. The
that the rise in global temperature since the last century work [2] reviewed numerous studies on trend analysis of
is definite both over land and ocean and that the last three rainfall, temperature, and rainy days all over India and
decades have been warmer compared to any other previous perceived that the trend and its magnitude is assessed
decades in instrumental records. Changes in precipitation recurrently by Mann-Kendall and Sen’s slope estimation
tests. Furthermore, they documented that the river basin
in the northern part of Tamil Nadu is experiencing an
increase in rainfall and decrease in rainy days, which
*e-mail: [email protected] implies that more intense floods will occur. A significant
1674 Palaniswami S., Muthiah K.

change in discharge in the middle reach of the Yellow points (BCP), wild binary segmentation (WBS),
river analyzed by the non- parametric Mann-Kendall and E-agglomerative algorithm (E-Agglo.), and iterative
Pettitt tests revealed that the methods act as a powerful robust detection (IR). Their results implied that BCP
tool for evaluating the previously implemented soil and gives incorrect change points, and that E-Agglo and WBS
water conservation measures [3]. A Mann-Kendall trend provide correct location and the number of change points,
test is utilized to analyze evapotranspiration sensitivity but E-Agglo is computationally costlier and WBS takes
to changes in meteorological variables, revealing that the less time for estimation. If a dataset has both trend and
method can be used as a water management tool with change point, analyzing with only one method can give
proper measures to evade climate change impacts [4]. A erroneous results. A new approach is developed using
highly increasing annual temperature trend is identified locally weighted polynomial regression and segmented
with the long-term data 1901-2012 using Mann-Kendall regression for trend and change point detection for water
test at 95% confidence level for Tbilisi, Georgia, and quality parameters. The method helps in environmental
also reported that monthly temperature data provided and pollution control decision making especially suited
insignificant trends [5]. for irregular datasets [10]. Several studies have been
A study [6] investigated trend and change point carried out by researchers all over the world in trend
detection of annual and seasonal precipitation and analysis and change point detection [11-14].
maximum, minimum, and mean temperature for The present work attempts to understand the
southwest Iran. The results revealed insignificant variability of rainfall and temperature in order to
trends in precipitation and significant increasing trends improve the hydrological environment of the river basin.
in temperature using Mann-Kendall and Pettit’s test. The study area is located in northern Tamil Nadu and
The change points are observed in temperature time covers an area of 7,520 km2. The boundaries of the basin
series that began in the 1990s. The author [7] analyzed are delineated by the Bay of Bengal in the eastern side,
the time series of trend and change point detection for the Kalvarayan Hills in the western side, by Ponnaiyar
temperature, relative humidity, wind speed, sunshine Basin in the northern side, by Paravanar Basin on the
hour, and pan evaporation using non-parametric tests. northeastern side, and by Cauvery Basin on the southern
Their results showed that a significant change point was side (Fig. 1).
present during 1900-2000, and a significant rising trend
observed in temperature and falling trend is reported for
other climate variables. The agricultural and livestock Material and Methods
sectors are affected not only by rainfall and temperature,
but also by other climatic variables such as wind velocity Mann-Kendall Test for Trend Detection
and relative humidity [8].
A work [9] compared four different change point The Mann-Kendall test is a widely used non-parametric
detection methods such as Bayesian analysis of change test utilized for trend detection in environmental sciences

Fig. 1. Vellar River Basin of Tamil Nadu, India.

Change Point Detection and Trend Analysis... 1675

[15-16]. The standardized test statistics Z and Mann- …where S0 = 0; initial cumulative sum S0; i = 1, 2,
–
Kendall statistic S is given by, 3,……………, 25; and X is the average of the sample
series.
3.1 The difference between current value and the average
to the previous sum are added to compute the CUSUM
values. The upward and downward slope indicates that
the period of time where the values are above or below the
overall average. At the change point, the series having two
3.2 means are split into segments and through the iterative
process mean square of estimates is computed till the
minimum mean square of error (MSE) is reached. This
3.3 shows the change of time and ultimately endorses the
change in the climate system. The boot strapping-based
confidence level is utilized to determine the significance
of the change point as given in Equation 3.7:

3.7
3.4
…where S ~ is the difference between the maximum
The statistical parameters are: Xi and Xj consecutive
and minimum CUSUM values, X is the number of
data values in the year i and j, n is the number of recorded
bootstrapped S ~ values greater than the original S ~
data, tp is the number of ties for the pth values, and q is
values, and N is the number of bootstraps.
the number of tied values.
The change point analyzer is a Microsoft Excel-based
The rising trend values are identified with the
add-on software utilized for change point detection in the
positive Z values, whereas the negative Z values indicate
dataset. This analyzer is an efficient tool that detects the
a decreasing trend of the corresponding time series.
exact year when the change or shift that has been taken
The null hypothesis is represented as H0 and alternate
place by using MSE and CUSUM charts. The level of
hypothesis is denoted as H1 in the two-tailed test. If there
change and confidence levels indicates the major change
is no trend in the series, H0 is rejected. If there is a trend in
point in the entire dataset. The likelihood of climate shift
the series, Z value is acquired from the standard normal
from the observation year is shown by the steady rise or
distribution with the pre-stated significance level.
decline lines in the CUSUM plots, which are computed
by equation 3.7. If there is no change present in the data
Sen’s Slope Estimator
series, then the values normally oscillate between the
The magnitude of the trend is quantified by a median- horizontal axis.
based non-parametric slope estimator known as Theil-
Sen method [17-18]. The computation of slope is given by
Equation 3.5: Results and Discussion

Based on the Mann-Kendall test and Change point

3.5 analyzer, the trends and change points were identified for
the rainfall and temperature data. Twenty-two rain gauge
…where xj and xk are the consecutive data values of series stations and one temperature station is involved in the
in the years j and k, and β is the magnitude of the trend present study as depicted in Fig. 2.
slope of data values.
Analyzing Monthly, Seasonal, and Annual
Change Point Detection Method Temperatures

The change point detection method is an effective We utilized the past 38 years of data for trend
tool to recognize the Climate erraticism in the historical analysis and change point detection studies. The monthly,
climate data series. The recognition of change point seasonal, and annual means of maximum, minimum, and
location in the climate series is vital since it modifies mean temperatures indicated that there is an increasing
processes in the hydrological cycle. A combination of trend. For all cases, the trends are considered statistically
cumulative sum charts (CUSUM) and boot strapping is a significant when α value is equal to 0.05. The maximum
method used to detect changes by several iterations [19]. temperatures in March and September showed a
The cumulative sum Si is estimated by equation 3.6 as significantly higher rising trend, with highest Z values of
follows: 2.96 and 2.73, respectively. The minimum temperatures
in May and June specified that significant higher rising
3.6 trend with highest Z values as 4.46 and 3.69, respectively.
1676 Palaniswami S., Muthiah K.

Fig. 2. Rain gauge and climate station of Vellar River Basin.

However, the minimum increasing trend is observed that June and November see a significant rising trend with
in the months of October and December for maximum greater Z values as shown in Table 1. All the monthly,
temperature, and January and September for minimum seasonal, and annual maximum, minimum, and mean
temperature (Table 1). The mean temperature indicated temperature results obtained from the Mann-Kendall test

Table 1. Mann-Kendall statistics for monthly, seasonal, and annual temperatures.

Time Scale Maximum Temperature Minimum Temperature Mean Temperature
Sen’s Sen’s
Monthly Statistics Z Sen’s slope P Z P Z P
slope slope
January 2.13 0.087 0.033 1.96 0.068 0.050 2.28 0.081 0.022
February 2.69 0.093 0.007 3.00 0.081 0.028 2.59 0.076 0.010
March 2.96 0.099 0.003 2.64 0.032 0.008 2.11 0.048 0.035
April 2.59 0.059 0.010 3.52 0.025 0.005 2.00 0.025 0.046
May 2.68 0.051 0.048 4.46 0.056 <0.0001 2.43 0.028 0.014
June 2.02 0.05 0.044 3.69 0.045 <0.0001 3.35 0.06 <0.0001
July 2.07 0.061 0.037 2.96 0.03 0.003 2.27 0.044 0.023
August 2.45 0.068 0.014 2.86 0.036 0.004 3.28 0.051 <0.0001
September 2.73 0.078 0.006 2.00 0.033 0.046 3.09 0.054 0.002
October 1.48 0.053 0.011 2.18 0.037 0.029 2.95 0.057 0.003
November 2.53 0.039 0.028 2.03 0.036 0.041 3.48 0.077 <0.0001
December 1.52 0.065 0.034 2.37 0.085 0.018 2.28 0.083 0.022
Seasonal statistics
Winter 2.27 0.085 0.022 2.57 0.067 0.009 2.69 0.092 0.007
Summer 1.68 0.048 0.047 4.96 0.068 <0.0001 2.36 0.023 0.018
Swm 2.86 0.087 0.004 3.07 0.04 0.002 2.96 0.043 0.003
Nem 2.35 0.052 0.016 2.03 0.046 0.041 1.75 0.029 0.081
Annual Statistics 2.23 0.037 0.025 3.10 0.043 0.002 2.11 0.04 0.035
Change Point Detection and Trend Analysis... 1677

Table 2. Mann-Kendall statistics for annual maximum daily, seasonal, and annual rainfall.

Seasonal Statistics
Stations Amdr Statistics Annual Statistics
SWM NEM Winter Summer
S Z S Z S Z S Z S Z S Z
Ariyalur 1.00 1.12 -5.491 -2.06 6.175 2.03 0.00 -0.13 -0.755 -0.26 -0.217 1.12
Athur 0.794 1.61 1.672 1.25 6.133 2.42 0.00 -1.03 1.311 1.28 9.066 1.61
Chettikulam 0.25 0.50 -2.221 -0.32 1.682 0.79 0.00 -0.72 0.067 0.23 -0.877 0.50
Chidambaram 0.726 0.62 -2.625 -1.62 8.245 1.88 0.00 0.44 0.11 0.39 6.39 0.62
Kallakurichi -3.132 -0.92 -10.82 -4.29 -10.46 -3.58 0.00 -2.60 -1.729 -2.88 -23.17 -0.92
Kattumylore 1.192 1.45 -6.182 -2.45 9.639 3.39 0.00 -1.00 -0.99 -0.29 3.98 1.45
Keelacheruvai 2.393 2.40 2.961 1.31 13.57 3.10 -0.109 -1.24 1.674 1.74 18.836 2.40
Memathur 2.129 3.05 -12.46 -3.16 4.943 1.99 -0.024 -1.24 -0.987 -0.45 -8.84 3.05
Parangipettai 0.684 0.96 -1.618 -0.28 10.15 2.25 0.041 0.57 2.078 1.83 11.85 0.96
Pelandurai 0.854 1.07 -0.346 0.31 13.36 3.65 0.00 0.74 2.613 1.87 14.284 1.07
Perambalur 0.836 1.78 2.01 1.31 8.133 2.16 0.00 -0.24 2.748 2.42 16.299 1.78
Rasipuram -0.365 -0.36 6.04 -2.42 1.443 0.79 0.00 -1.88 -0.351 0.11 -6.418 -0.36
Sendamangalam -1.129 -1.93 -4.50 -1.69 1.661 0.60 0.00 -0.26 -0.972 -0.24 -1.944 -1.93
Sethiothope 2.218 2.34 -0.764 0.05 14.68 3.00 0.074 0.56 1.438 1.65 13.814 2.34
Sethiothope Anicut 2.342 2.66 -3.616 -1.59 9.832 2.19 0.00 0.43 1.683 1.46 4.967 2.66
Srimushnam 2.776 2.87 -2.502 -0.54 8.976 2.16 0.00 0.10 -0.081 0.28 3.519 2.87
Tholudur 2.867 2.39 4.121 1.77 16.42 4.52 0.00 -0.25 3.063 2.21 29.00 2.39
Thuraiyur -0.271 -0.45 -1.587 -0.31 1.092 0.79 0.00 0.03 2.829 1.64 3.046 -0.45
Ulundurpet 1.913 2.64 -4.323 -1.25 10.13 3.32 0.00 0.00 1.227 1.41 7.145 2.64
Vembanur 0.643 0.49 0.00 0.45 0.00 0.63 0.00 -0.26 2.206 1.95 1.189 0.49
Virudhachalam 1.512 1.35 -1.764 -0.32 12.60 3.36 0.00 0.05 2.088 1.01 13.092 1.35
Virudhachalam Anicut 0.429 0.58 -10.46 -3.09 -8.445 -1.34 0.00 -1.75 -2.057 -2.21 -24.07 0.58

revealed that a significant greater rising trend is present three decades. The greatest monthly temperatures of
in the series. The magnitude of Sen’s slope of all the 41.6ºC and 40.4ºC are noticed in April and March, and
monthly, seasonal, and annual maximum, minimum, and the lowest monthly temperatures of 17ºC and 18ºC are
mean temperatures specifies that there is a significant observed in January and December.
rising trend in the study area. The maximum temperature
represented the greatest values of 41.6ºC, 40.4ºC, and Analying Annual Maximum, Seasonal,
38.9ºC in 2002 during April, March, and summer, and Annual Precipitation
respectively. The greatest values of 26.9ºC, 25.7ºC, and
25.1ºC in 2003 during May, April, and June is identified The Mann-Kendall trend test is carried out for the
for minimum temperature. The magnitude of annual rainfall series of the 22 rain gauge stations located in
mean maximum temperature shows an increasing trend the river basin. The estimated Mann-Kendall Z and
of 0.2ºC and 0.6ºC per decade. p values of each station for annual maximum, seasonal,
A warming trend is observed from both maximum and annual time scales are shown in Tables 2-3. The
and minimum temperature characteristics. A significant seasonal rainfall time series are more substantial than
rising trend of mean annual temperature shows that there annual time series because they are subjected to greater
is an alarming global warming signal expressed over inter-annual variability. The trend results were abbreviated
the entire river basin. An increase of 0.5ºC in the annual as IT (increasing trend), DT (decreasing trend), and NT
mean temperature is noted in the recent decade within (no trend) in Table 3. A statistically significant increasing
the study period of 38 years. Thus the results indicate that trend is seen in the AMDR statistics of six stations:
there is a consistent warming trend during the previous Keelacheruvai, Memathur, Sethiothope, Sethiothope
1678 Palaniswami S., Muthiah K.

Table 3. Trend analysis of annual maximum daily, seasonal, and annual rainfall.
Seasonal Statistics Annual
Stations Amdr Statistics
SWM NEM Winter Summer Statistics

P Trend P Trend P Trend P Trend P Trend P Trend

Ariyalur 0.263 NT 0.024 DT 0.111 NT 0.644 NT 0.535 NT 0.988 NT
Athur 0.108 NT 0.433 NT 0.019 IT 0.196 NT 0.433 NT 0.045 IT
Chettikulam 0.614 NT 0.438 NT 0.701 NT 0.342 NT 0.951 NT 0.842 NT
Chidambaram 0.538 NT 0.063 NT 0.173 NT 0.815 NT 0.926 NT 0.317 NT
Kallakurichi <0.0001 DT 0.0001 DT 0.001 DT 0.006 NT 0.002 DT <0.0001 DT
Kattumylore 0.148 NT 0.0009 DT 0.002 IT 0.209 NT 0.457 NT 0.415 NT
Keelacheruvai 0.016 IT 0.274 NT 0.003 IT 0.132 NT 0.239 NT 0.009 IT
Memathur 0.002 IT 0.001 DT 0.092 NT 0.133 NT 0.403 NT 0.223 NT
Parangipettai 0.338 NT 0.570 NT 0.065 NT 0.651 NT 0.092 NT 0.190 NT
Pelandurai 0.284 NT 0.939 NT 0.001 IT 0.729 NT 0.085 NT 0.002 IT
Perambalur 0.074 NT 0.348 NT 0.065 NT 0.749 NT 0.024 IT 0.005 IT
Rasipuram 0.721 NT 0.007 DT 0.634 NT 0.038 NT 0.724 NT 0.223 NT
Sendamangalam 0.050 DT 0.059 NT 0.509 NT 0.634 NT 0.466 NT 0.770 NT
Sethiothope 0.019 IT 0.747 NT 0.011 IT 0.639 NT 0.204 NT 0.065 NT
Sethiothope Anicut 0.008 IT 0.051 NT 0.075 NT 0.923 NT 0.239 NT 0.415 NT
Srimushnam 0.004 IT 0.348 NT 0.086 NT 0.923 NT 0.877 NT 0.701 NT
Tholudur 0.017 IT 0.179 NT <0.0001 IT 0.575 NT 0.107 NT 0.0001 IT
Thuraiyur 0.650 NT 0.451 NT 0.678 NT 0.924 NT 0.086 NT 0.364 NT
Ulundurpet 0.008 IT 0.104 NT 0.003 IT 0.759 NT 0.386 NT 0.179 NT
Vembanur 0.626 NT 0.582 NT 0.507 NT 0.634 NT 0.029 IT 0.521 NT
Virudhachalam 0.180 NT 0.476 NT 0.003 IT 0.799 NT 0.227 NT 0.08 NT
Virudhachalam Anicut 0.559 NT 0.002 DT 0.17 NT 0.059 NT 0.014 DT 0.013 DT

anaicut, Srimushnam, and Tholudur and Ulundurpet. years. Sendamangalam station has a decreasing rainfall
Seasonal trend statistics indicated an increasing trend trend magnitude of 1.9 mm/10 years. The magnitude of
observed for Perambalur and Vembanur stations and a increasing trends in NEM and SWM are greater than that
decreasing trend in Kallakurichi and Virudhachalam of summer and winter seasons for almost all the stations.
anaicut stations in summer. A statistically insignificant The 2005 year is the surplus year in the period of study,
trend is perceived in all the stations for winter, and a with rainfall of 1,470 mm, which is above the average
decreasing trend was observed in six stations, and the rainfall.
other stations are having a statistically insignificant
trend for the SWM season shown in Table 3. For the Analysis of Change Points
NEM season, a statistically significant increasing trend
is specified in eight stations except Kallakurichi, which The recognition of change points is a statistical
shows a decreasing trend and other stations are having an technique that plays a vital role in spotting climate jumps
insignificant trend as depicted in Table 3. A statistically in the whole climatological data period. The change
significant increasing trend resulted in five stations except points were identified for maximum, minimum, and mean
Kallakurichi and Vridhachalam Anaicut stations, which temperatures. It is clearly evident that the maximum,
shows a decreasing trend, and other stations are having an minimum, and mean temperature values appear to be
insignificant trend as depicted in Table 3. outside the control limits, which reveals that change took
The Sen’s slope is utilized to compute trend magnitude. place during the examined data period (Fig. 3).
In AMDR statistics, the greatest increase in trend The level represented in Fig. 4 implies change and
magnitude is seen in Memathur station by 3.05 mm/10 is strongly linked with climate change. The maximum
Change Point Detection and Trend Analysis... 1679

Fig. 5. CUSUM chart for annual temperature.

Fig. 3. Identifying change points for annual temperature.

2000-2004 at 100% confidence interval at levels 4 and 3,

temperature change point occurred in 1985 and 2001 at
respectively. The change point years are clearly shown in
a confidence level of 100% and 94% at levels 2 and 1,
Fig. 4. The change point magnitude is computed using the
respectively. Before the existence of the change the mean
ratio of average of climate data after the change point to
maximum temperature was 32.94°C during 1985, and
the before the change point occurrence is found to be 0.99,
after the change the value was 34.46°C. In the case of
1.00, and 1.00, correspondingly.
minimum temperature, the change point appeared in 1987
The present study shows a statistical importance of
and 2013 at 100% and 91% confidence interval at levels 1
change point that can be seen from the late 1980s at a
and 3, respectively. Earlier to the change, the minimum
confidence interval setting of 95% with 1,000 boot straps
temperature is 23.11°C and after the change, it had a
and a method of mean square of estimates. The process
value of 23.97°C. From the mean annual temperature, it
shifts estimation and generation of trends are well
is seen that the change point is present in 1983-1988 and

Fig. 4. Estimated change point years for annual temperature.

1680 Palaniswami S., Muthiah K.

Acknowledgements

The authors acknowledge sincere thanks for data

support from the Vellar River Basin Project, Centre for
Water resources, Anna University.

References
Fig. 6. Estimated change point years for annual rainfall. 1. HARTMANN D.L., TANK A.M.K., RUSTICUCCI M.,
ALEXANDER L.V., BRONNIMANN S., CHARABI
Y.A.R., DENTENER F.J., DLUGOKENCKY E.J., EAST-
indicated using CUSUM charts. The CUSUM chart of ERLING D.R., KAPLAN A., SODEN B.J., THORNE P.W.,
WILD M., ZHAI P.M. Observations: atmosphere and sur-
maximum, minimum, and mean temperatures obtained
face. In: Climate Change 2013: The Physical Science Ba-
by the boot-strapping technique has a noticeable change sis. Working Group I Contribution to the Fifth Assessment
at α = 0.05, which is depicted by modified background Report of the Intergovernmental Panel on Climate Change,
color in Fig. 5. A clear upward trend is represented in Cambridge University Press, Cambridge, United Kingdom
CUSUM charts and a sudden temperature change is and New York, 159, 2013.
found in the late 1980s and 2000s. The increasing trend 2. JAIN S.K., KUMAR V. Trend analysis of rainfall and tem-
continues until 2014. The estimated change point years of perature data for India. Current Science. 102 (1), 37, 2012.
annual rainfall is shown in Fig. 6. The results of annual 3. GAO P., DENG J., CHAI X., MU X., ZHOU G., SHAO H.,
rainfall change point years shows no major change, which SUN W. Dynamic sediment discharge in the Hekou-Long-
man region of Yellow river and soil and water conservation
is due to the characteristic of highly erratic and inter
implications. Science of The Total Environment. 578, 56,
annual variability. 2017.
4. NOURI M., HOMAEE M., BANNAYAN M. Quantitative
trend, sensitivity and contribution analyses of reference
Conclusions evapotranspiration in some arid environments under cli-
mate change. Water Resources Management. 31 (7), 2207,
The variability of rainfall and temperature due to 2017.
probable human involvement or change in climate is 5. OZGUR K.I.S.I., OKAN E.R.A.Y., CIHAN M.E.R.T. Trend
essential for planning and managing water resources at analysis of long term temperatures in Tbilisi, Georgia. Jour-
nal of Technical Science & Technologies. 5 (2), 59, 2017.
a regional level. Different non-parametric tests were used
6. ZARENISTANAK M., DHORDE A.G., KRIPALANI R.H.
for trend analysis and change point detection. The results Trend analysis and Change point detection of annual and
show that all monthly, seasonal, and annual maximum, seasonal precipitation and temperature series over south-
minimum, and mean temperatures have a significant west Iran. Journal of earth system science. 123 (2), 281,
greater rising trend. The magnitude of increasing trends 2014.
in NEM and SWM are greater than those of summer and 7. JAISWAL R.K., LOHANI A.K., TIWARI H.L. Statistical
winter seasons for almost all the rain gauge stations. The analysis for change detection and trend assessment in cli-
previous studies also validated that the river basin in the matological parameters. Environmental Processes. 2 (4),
Northern part of Tamil Nadu is experiencing an increase 729, 2015.
8. SIVAKUMAR A.T., THENNARASU A., RAJKUMAR
in temperature and rainfall, and a decrease in rainy days.
J.S.I. Effect of season on the incidence of infectious dis-
The maximum temperature and minimum temperature eases of bovine in Tamilnadu. Elixir Meteorology. 47, 8874,
change points are identified in 1985 and 2001, and 1987 and 2012.
2013, respectively. From the mean annual temperature, it 9. SHARMA S., SWAYNE D.A., OBIMBO C. Trend analysis
is seen that the change point is present in 1983-88 and and change point techniques: a survey. Energy, Ecology
2000-04 at 100% confidence interval. The important and Environment, 1 (3), 123, 2016.
change points observed in maximum, minimum, and 10. HUANG H., WANG Z., XIA F., SHANG X., LIU Y.,
mean temperatures implies that there is an influence of ZHANG M., DAHLGREN R.A., MEI K. Water quality
fast-growing industrial activities in the basin and also trend and change-point analyses using integration of locally
weighted polynomial regression and segmented regression.
the increase in temperature occurred by change in land
Environmental Science and Pollution Research. 24, 1, 2017.
use pattern, which is caused by the effect of excessive 11. BATISANI N., YARNAL B. Rainfall variability and trends
stream flow during monsoon season and thereafter a long in semi-arid Botswana: implications for climate change ad-
dry period prevailed in the basin. The present study is aptation policy. Applied Geography. 30 (4), 483, 2010.
very helpful in framing climate change scenarios and it is 12. KARPOUZOS D.K., KAVALIERATOU S., BABAJI-
based on the revealed characteristics of climate variables MOPOULOS C. Trend analysis of precipitation data in Pie-
at the regional level. ria region (Greece). European Water. 30, 31, 2010.
Change Point Detection and Trend Analysis... 1681

13. OBASI R.A., IKUBUWAJE C.O. Analytical study of rain- 17. SEN P.K. Estimates of the regression coefficient based on
fall and temperature trend in catchment states and stations Kendall’s tau. Journal of the American Statistical Associa-
of the Benin-Owena river basin, Nigeria. Journal of Envi- tion. 63 (324), 1379, 1968.
ronment and Earth Science. 2 (3), 9, 2012. 18. THEIL H. A rank-invariant method of linear and polyno-
14. SALARIJAZI M., AKHOND-ALI A.M., ADIB A., mial regression analysis, Part I, II, III, In Proceedings of
DANESHKHAH A. Trend and change point detection for Koninalijke Nederlandse Akademie van Weinenschatpen
the annual stream-flow series of the Karun River at the Ah- A. 53, 386, 512, 1397, 1950.
vaz hydrometric station. African Journal of Agricultural 19. TAYLOR W.A. Change-Point Analyzer 2.3 software pack-
Research. 7 (32), 4540, 2012. age, Taylor Enterprises, Libertyville, Illinois, 2000, online
15. KENDALL M.G. Rank correlation methods, Charles Grif- may be assessed at http://www.variation.com/cpa/tech/
fin, London, 1975. change point (3 January 2017).
16. MANN H.B. Non parametric tests against trend. Econo-
metrica: Journal of the econometric society. 13 (3), 245,
1945.
View publication stats