International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 4 (2016) pp 2417-2423
© Research India Publications. http://www.ripublication.com
Statistical Analysis of Precipitation over Seonath River Basin,
Chhattisgarh, India
Mani Kant Verma
Assistant Professor, Department of Civil Engineering,
NIT Raipur, Chhattisgarh, India.
Dr. M. K. Verma
Professor, Department of Civil Engineering,
NIT Raipur, Chhattisgarh, India.
Sabyasachi Swain
Research Scholar, IDP in Climate Studies,
IIT Bombay, Maharashtra, India.
first into picture. The variation in rainfall and temperature has
been arising as a challenging issue for the present generation
and it will also remarkably affect the future (Ventura et al.,
2002).From the point of view of India, this may lead to severe
detrimental conditions due to poor adaptation strategies and a
very high population. Intense flooding and severe drought
conditions may prevail in various parts of the country
simultaneously (Gosain et al., 2006, 2011; Swain, 2014). This
will be further accelerated by the rampant human
interventions. But the matter to look into is that, be it a
drought or a flood, the variation in amount of precipitation
will certainly govern these aspects to a significant extent
(Katz et al., 1992). In India, it matters for rainfall due to
South-West monsoon i.e. rainfall during June-September.
Abstract
The impact of climate change in terms of anomalies in
precipitation has risen up as a major challenge in the world, as
it may lead to havoc by intense floods or severe droughts.
This study presents trends in annual and monthly precipitation
data collected from 39 stations for 32 years (1981-2012) for
Seonath river basin, Chhattisgarh, India, using Mann-Kendall
(MK) test and Sen’s slope estimator test. Based on the
analysis, at 5% significance level, only few stations show a
significant change and the analysis for overall Seonath river
basin shows increasing trend of rainfall in monsoon season
and decreasing trend of rainfall in post-monsoon season. It
was also observed that the annual rainfall shows increasing
trend for the Seonath river basin. In addition, a comparison
has been performed between the observed and its prewhitened rainfall data and the result reflects similar pattern for
both the datasets.
Study area and Data used
Seonath (also called Shivnath) river basin is situated in the
fertile plains of Chhattisgarh Region. This basin is situated
between 20° 16’ N to 22° 41’ N Latitude and 80° 25’ E to 82°
35’ E Longitude. The basin occupies a large portion of the
upper Mahanadi valley. Seonath is the longest tributary of
Mahanadi river and it traverses a length of 380 km. It
originates near Panabaras village in Rajnandgaon district,
Chhattisgarh, which is at 624 m above the sea level. Tandula,
Arpa, Kharun, Agar, Hamp and Maniyari are its major
tributaries (Chakraborty et al., 2013). The area of the basin is
30560 km2. The topography of the watershed is almost flat.
The slope ranges from 1% to 2% and the weighted average
slope of the watershed is 1.6%. Seonath basin has a tropical
wet and dry climate; temperatures can be extremely hot from
March to June, although it remains moderate throughout the
year. The basin receives about 1150 mm of mean annual
rainfall and a vast majority of it is contributed by monsoon
season i.e. from June to early October, followed by postmonsoon season (October to December).
Keywords Precipitation, Trend, Pre-whitening, Mann-Kendall
test, Sen’s slope estimator test.
Introduction
Water resource is the prime concern for any project planning,
development and management. Indian agriculture primarily
depends on rainfall and its distribution. Distribution of rainfall
is the major factor in the planning and management of
projects related to water resources like agricultural
production, water requirement changes, irrigation and
reservoir operation (Corte-Real et al., 1998; Chakraborty et
al., 2013). Climate change indicates a different behavior of the
hydro-meteorological parameters comparing between two
different periods. The climatic variability is not a very short
span process. It takes years or decades to have a noticeable
change in climate. Whenever the word ‘Climate change’ is
coined, the changes in temperature and erratic rainfall come
2417
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 4 (2016) pp 2417-2423
© Research India Publications. http://www.ripublication.com
The daily rainfall data of 39 Meteorological Stations over
Seonath river basin for a period of 32 years(from 1981 to
2012) were collected from State Data Centre, Water
Resources Department, Raipur (Chhattisgarh) and Central
Water Commission (CWC), Bhubaneswar, to test the
variability with respect to space and time. Out of these 39
stations, only 5 stations are from Central Water Commission
(CWC) and the rest are the Chhattisgarh Water Resources
Department (WRD) stations. All the stations along with their
latitude (degree North), longitude (degree East) and area
(obtained from Thiessen polygon) are presented in Table 1.
The 5 stations namely Andhiyakore, Jondhra, Kotni,
Patharidih and Simga with suffix CWC represent that these
are the CWC stations. The location of the stations in the basin
and their respective area is presented through Thiessen
polygon map (Figure 1). Thiessen polygon map was prepared
in order to determine the rainfall over the basin as a whole, so
that the trend can be obtained for the overall study area.
Shahspur
Simga CWC
Simga WRD
Sond
Surhi (Palemeta)
20.970
21.900
22.183
21.620
21.220
81.870
81.117
82.167
81.705
81.690
667.415
291.545
590.477
931.621
472.045
Table 1: Location and Area (Thiessen Polygon) for Stations
Station Name
Ambagarh Chowki
Andhiyakore CWC
Balod
Bemetera
Bilaspur
Bodla
Chilhaki
Chirapani
Chuikhadan
Dhamtari
Dongargaon
Dongargarh
Doundi
Durg
Gandai
Ghonga
Gondly
Jondhra CWC
Kawardha
Kendiri
Kharkhara
Khuria
Khutaghat
Kota
Kotni CWC
Madiyan
Mungeli
Nawagarh
Newara
Pandariya
Patharidih CWC
Pindrawan
Raipur
Semartal
Latitude
20.778
21.780
20.733
21.729
22.083
22.182
21.792
22.208
21.533
20.822
20.975
21.183
20.485
21.217
21.667
22.300
20.750
21.720
22.017
21.100
20.967
22.388
22.300
22.267
22.130
21.990
21.133
22.067
21.906
21.550
22.217
21.340
21.400
21.250
Longitude
80.749
81.610
81.233
81.549
82.150
81.223
82.308
81.196
81.017
81.552
80.863
80.767
81.096
81.283
81.117
81.967
81.133
82.340
81.233
81.733
81.033
81.599
82.208
82.033
81.240
83.200
80.617
81.683
81.606
81.833
81.417
81.600
81.850
81.633
Area (km2)
1298.78
364.887
934.391
573.711
1185.89
114.639
1030.25
547.109
1167.67
1158.84
622.855
976.891
789.881
1574.88
684.035
1562.44
641.397
490.286
630.165
563.022
924.559
1519.24
1523.81
382.005
229.803
386.266
1114.24
689.732
1050.78
881.239
552.498
610.785
394.213
432.258
Figure 1: Location of rainfall stations in Seonath river basin
(Thiessen Polygon map)
Methodology
In the present study, daily rainfall data is collected and
arranged in 4 different parts of a year (winter, pre-monsoon,
monsoon and post-monsoon seasons), and thereafter, nonparametric Mann-Kendall test and Sen’s slope estimator test
has been used for trend analysis. For individual stations, only
monsoon and post-monsoon season data are considered as the
randomness of the data will be very high for winter and premonsoon seasons as these phases of a year receive almost no
rainfall over the study area. Then, rainfall over the basin as a
whole is estimated by Thiessen polygon method and is
statistically analyzed.
Pre-whitening of Data:
The detection of trend is significantly affected for
autocorrelation among dataset (Hamed and Rao, 1998;
Serrano et al., 1999; Yue et al., 2002; Blain, 2013). For a
discrete time series, the coefficient of autocorrelation ρk of a
discrete time series for lag k is given by,
(1)
A positive autocorrelation among data may lead to a clear
presence of trend by the non-parametric tests while it may not
2418
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 4 (2016) pp 2417-2423
© Research India Publications. http://www.ripublication.com
Sen’s Slope Estimator Test
Sen’s slope estimator is the most commonly used test to detect
a linear trend (Yue and Hashino, 2003; Karpouzos et al.,
2010; Jain and Kumar, 2012; Swain et al., 2015; Adarsh and
Reddy, 2015). The slope (Ti) of all data pairs is given as as
(Sen, 1968),
actually exist. Similarly, presence of a negative
autocorrelation may ignore the presence of an actual trend
(Hameed and Rao, 1998). Hence, the removal of remarkable
autocorrelation is necessary before carrying on statistical tests.
This method of removal of autocorrelation is referred to as
pre-whitening (Blain, 2013). In this case, pre-whitening has
been done using lag-1 autocorrelation.
For i = 1, 2, 3…..N
(2)
Where and are considered as data values at time j and k
(j>k) respectively.
Here,
represents the pre-whitened dataset i.e. after
removing the positive autocorrelation among the data points.
Then the results of the statistical methods applied to actual
data and pre-whitened data are analyzed and a comparison is
presented.
The Sen’s estimator of slope is given by the median of these
N values of Ti, which is projected as
(8)
Mann-Kendall Test
Mann Kendall test is used for identifying the monotonicity in
a time series (Yue et al., 2002; Kumar et al., 2010; Caloiero et
al., 2011; Jain and Kumar, 2012; Jain et al. 2013; Adarsh and
Reddy, 2015). Being a non-parametric test, the problem due to
data skew can be evaded easily (Swain et al., 2015).
The Mann-Kendall statistic S is given as
Very similar to the Mann-Kendall test, the positive and
negative values of Qi represents a positive and negative trend
respectively.
Result and Discussion
Mann-Kendall and Sen’s slope estimator test is applied on
both the datasets (actual and pre-whitened). In Figure 2, the
stations are marked with different colors according to the
results of the Mann-Kendall on annual rainfall recorded data.
It can be noticed that 20 stations are showing a positive value
of Zmk whereas 19 stations are showing negative values. The
Mann-Kendall co-efficient Zmk value for 5% significance level
is 1.96. Thus, the stations marked blue indicate for significant
increase in annual rainfall and that of red shows significant
decrease. All other stations show no significant (at 5%
significance level) trend for the study period. So, out of 39
stations, only 5 stations namely Dhamtari, Newara, Kota,
Pandariya and Simga CWC shows significant rise in rainfall,
whereas only 3 stations Bodla, Chirapani and Chuikhadan are
showing a decreasing trend for annual rainfall.
From Figure 3, it is clear that most of the stations are showing
a positive trend for monsoon season although only a few of
them can be regarded as significant. Out of 39 stations, a
decreasing trend of rainfall can be observed for 10 stations of
pre-whitened and 12 stations of actual observed data. The 3
stations showing decreasing trend in annual rainfall are also
showing a negative trend for monsoon season whereas, 7
stations are showing an increasing trend in case of prewhitened data and 8 stations for actual data. Similarly from
Figure 4, it can be observed that, most of the stations are
showing insignificant trend for post-monsoon seasons. Only 4
stations are having a significant decreasing trend for prewhitened data and that of 3 stations for actual observed data
whereas, no station is showing a significant positive trend for
rainfall during post-monsoon seasons.
The trend test is applied to a time series xi that is ranked from
i = 1, 2 …n-1, and xj, which is ranked from j = i+1, 2 ….n.
Every data point xi is taken as a reference point for comparing
with the all other data points, xj so that,
=
(7)
(4)
It is observed that when n ≥ 8, the statistic S is approximately
normally distributed i.e. with zero mean and variance given
by,
(5)
Where, ti represents the number of ties up to sample i (Zhu et
al., 2010).
(6)
Zmk is assumed to follow a standard normal distribution.
Hence, its value being positive indicates a rising trend and that
of negative indicates a decreasing trend. A significance level α
is also utilized for testing monotonicity of trend (a two-tailed
test). If Zmk is greater than Zα/2 where α depicts the
significance level, then the trend is considered to be
significant (Babar and Ramesh, 2014).Generally, Zmk values
are 1.645, 1.960 and 2.576 for significance level of 10%, 5%
and 1% respectively (Swain et al., 2015).
2419
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 4 (2016) pp 2417-2423
© Research India Publications. http://www.ripublication.com
(a)
(a)
(b)
(b)
Figure 2: Mann-Kendall Test results for annual rainfall for (a)
pre-whitened data; (b) observed data
Figure 3: Mann-Kendall Test results for monsoon season for
(a) pre-whitened and (b) actual data
2420
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 4 (2016) pp 2417-2423
© Research India Publications. http://www.ripublication.com
From the results of Mann-Kendall test, it is evident that the
significant trend of rainfall in monsoon season and over whole
year is almost same, the reason being, rainfall during monsoon
phase contributes more than 85% of the annual rainfall over
Seonath river basin. One more thing is to be noticed is that
pre-whitening doesn’t have much effect on trend for this data
from Mann-Kendall test. The results obtained for actual
observed data and pre-whitened data are hardly different from
each other.
Very similar to the Mann-Kendall test, the results are almost
same for annual rainfall and that of monsoon season by
Sen’sslope estimator test. The regions marked with blue color
indicates increasing trend and yellow color indicates
decreasing trend. For other regions, Sen’s slope value is zero.
From Figure 5 (a), it can be observed that, out of 39 stations,
20 stations are showing increasing trend in annual rainfall for
pre-whitened data and that of 18 stations for actual data. Rests
are showing a decreasing trend in both cases. From Figure 5
(b), in monsoon season, 11 stations are having a decreasing
trend for both pre-whitened and actual data. Only one station
is showing absolutely no trend for actual data. For most of the
stations, a positive Sen’s slope is obtained i.e. an increasing
trend can be observed for the entire study period. It can be
seen from Figure 5 (c) that most of the stations are showing a
decreasing trend for both pre-whitened and actual rainfall data
in post-monsoon season.
(a)
(a) Sen’s Slope values for different stations for Annual rainfall
(b)
Figure 4: Mann-Kendall Test results for post-monsoon season
for (a) pre-whitened and (b) actual data
(b) Sen’s Slope values for different stations for rainfall in
Monsoon season
2421
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 4 (2016) pp 2417-2423
© Research India Publications. http://www.ripublication.com
(b) Seasonal variation of rainfall over Seonath river basin
(C) Sen’s Slope values for different stations for rainfall in
Post-Monsoon season
Figure 6: Mann-Kendall test results for pre-whitened data for
monthly and seasonal rainfall
Figure 5: Sen’sslope estimator test results for pre-whitened
data and actual data for (a) Annual rainfall; (b) Rainfall in
monsoon season; (c) Rainfall in post-monsoon season
From Table 2, it can be observed that the Sen’s slope is
positive for April, July and September, zero for November
and negative for all other months.
The trend of rainfall over different stations is determined
using the non-parametric tests. But it is essential to determine
the behavior of rainfall over the whole basin. The Thiessen
polygon method was used to determine the rainfall over the
whole Seonath river basin considering the area covered by
each station. The monthly and seasonal variation over the
whole basin was analyzed by Mann-Kendall and Sen’s slope
estimator test.
Table 2: Monthly Sen’s slope results for rainfall
Median Standard Sen’s
(%
Deviation slope(β) change α)
314.34833 13.30411 3.67972 17.72987 -0.03946 -0.094904329
117.38473 9.446052 5.39712 10.83442 -0.22121 -0.749369926
88.979965 8.214486 5.336704 9.432919 -0.02805 -0.109279824
29.140871 6.213002 4.745383 5.398229 0.024628 0.126847691
131.97939 10.84519 7.09259 11.48823 -0.17392 -0.51316768
3481.5742 150.3759 140.6504 59.00487 -0.33211 -0.070672857
9137.973 303.7481 294.1403 95.59275 3.528383 0.371716677
4333.2955 298.6511 303.0616 65.82777 -0.14358 -0.015384195
3056.0606 165.8706 163.2684 55.28165 1.059016 0.204307029
751.28343 41.5091 38.57057 27.40955 -0.72155 -0.556257041
373.96891 9.93016 1.533419 19.33828 0
0
164.88507 5.32481 1.092111 12.84076 -0.00489 -0.029405063
Month Variance Mean
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
From Figure 6, it can be observed that, the monthly rainfall is
showing an insignificant trend for overall Seonath river basin,
taking into consideration the pre-whitened data. But, for the
month of July and September, an increasing trend can be
observed whereas decreasing trend for January, February,
May and October months. The months of July and October are
showing almost significant (5% significance level) increasing
and decreasing trend respectively. Talking about the seasonal
variation, an increasing trend can be marked for monsoon
season whereas it is decreasing for winter, pre-monsoon and
post-monsoon season, although they are not significant at 5%
significance level. As the rainfall in monsoon season is
showing an increasing trend, the Zmk value for annual rainfall
is also positive i.e. 0.7622. Hence, over the study period
(1980-2012), the rainfall over Seonath basin has a rising trend.
Figure 7 shows seasonal rainfall over Seonath basin where, a
high positive slope is observed i.e. Sen’s slope value is 3.983
for monsoon season and negative for post-monsoon
season.The Sen’s slope value for annual rainfall over entire
Seonath river basin is also 2.832, showing a clear increase in
annual rainfall for the study area in these 32 years.
(a) Monthly variation of rainfall over Seonath river basin for
pre-whitened data
Figure 7: Sen’sslope results for Seasonal rainfall
2422
International Journal of Applied Engineering Research ISSN 0973-4562 Volume 11, Number 4 (2016) pp 2417-2423
© Research India Publications. http://www.ripublication.com
[8]
Conclusion
Trend analysis of monthly and annual rainfall data for Seonath
river basin, Chhattisgarh, for 32 years (1981-2012) using
Mann-Kendall and Sen’s slope estimator test, has been
presented in this article.For monsoon season, the Zmk value of
Mann-Kendall Test was positive for 7 stations and negative
for only 3 stations at 5% significance level for pre-whitened
data, although most of the stations showed positive value of
Mann-Kendall co-efficient. The Sen’s slope values for most of
the stations were also found to be positive. For overall
Seonath river basin, an observable rising trend was there for
the months of July and September and decreasing trend for
January, February, May and October. For seasonal variation,
the trend is clearly positive for monsoon season, both from
Mann-Kendalltest and Sen’s slope estimator test. Since the
study area represents more of agricultural lands and
agriculture is primarily dependent on monsoon season, such
an increasing trend is desirable for existing conditions. Further
study in this regard may reveal other aspects which will be
helpful to understand the changes in hydrological process
along with agricultural development.
[9]
[10]
[11]
[12]
[13]
[14]
Acknowledgment
The authors are thankful to ‘State Data Centre, Chhattisgarh’,
‘Central Water Commission Office, Bhubaneswar’ and ‘Water
Resources Department, Chhattisgarh’ for providing data for
this study.We are also grateful to all those individuals whose
suggestions have helped to improve the quality of this article.
[15]
[16]
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
Adarsh, S., & Janga Reddy, M. “Trend analysis of
rainfall in four meteorological subdivisions of southern
India using nonparametric methods and discrete
wavelet transforms”, International Journal of
Climatology, 35(6), pp. 1107-1124, 2015.
Babar, S., & Ramesh, H. “Analysis of extreme rainfall
events over Nethravathi basin”, ISH Journal of
Hydraulic Engineering, 20(2), pp. 212-221, 2014.
Blain, G. C. “The Mann-Kendall test: the need to
consider the interaction between serial correlation and
trend”, Acta Scientiarum. Agronomy, 35(4), pp. 393402, 2013.
Caloiero, T., “Coscarelli, R., Ferrari, E., & Mancini, M.
“Trend detection of annual and seasonal rainfall in
Calabria (Southern Italy)”, International Journal of
Climatology, 31(1), pp. 44-56, 2011.
Chakraborty, S., Pandey, R. P., Chaube, U. C., &
Mishra, S. K, “Trend and variability analysis of rainfall
series at Seonath River Basin, Chhattisgarh (India)”,
International Journal of Applied Sciences and
Engineering Research, 2(4), pp. 425-434, 2013.
Corte-Real, J., Qian, B., & Xu, H. “Regional climate
change in Portugal: precipitation variability associated
with large-scale atmospheric circulation”, International
Journal of Climatology, 18(6), pp. 619-635, 1998.
Gosain, A. K., Rao, S., & Basuray, D. “Climate change
impact assessment on hydrology of Indian river
basins”, Current science, 90(3), pp. 346-353, 2006.
[17]
[18]
[19]
[20]
[21]
[22]
2423
Gosain, A. K., Rao, S., & Arora, A. “Climate change
impact assessment of water resources of India”, Current
Science, 101(3), pp. 356-371, 2011.
Hamed, K. H., & Rao, A. R. “A modified MannKendall trend test for autocorrelated data”, Journal of
Hydrology, 204(1), pp. 182-196, 1998.
Jain, S. K., & Kumar, V. “Trend analysis of rainfall and
temperature data for India”, Current Science, 102(1),
pp. 37-49, 2012.
Jain, S. K., Kumar, V., & Saharia, M. “Analysis of
rainfall and temperature trends in northeast India”,
International Journal of Climatology, 33(4), pp. 968978, 2013.
Karpouzos, D. K., Kavalieratou, S., & Babajimopoulos,
C. “Trend analysis of precipitation data in Pieria
Region (Greece)”, European Water, 30, pp. 31-40,
2010.
Katz, R. W., & Brown, B. G. “Extreme events in a
changing climate: variability is more important than
averages”, Climatic change, 21(3), pp. 289-302, 1992.
Kumar, V., Jain, S. K., & Singh, Y. “Analysis of longterm rainfall trends in India”, Hydrological Sciences
Journal–Journal des Sciences Hydrologiques, 55(4), pp.
484-496, 2010.
Sen, P. K. “Estimates of the regression coefficient
based on Kendall's tau”, Journal of the American
Statistical Association, 63(324), pp. 1379-1389, 1968.
Serrano, A., Mateos, V. L., & Garcia, J. A. “Trend
analysis of monthly precipitation over the Iberian
Peninsula for the period 1921–1995”, Physics and
Chemistry of the Earth, Part B: Hydrology, Oceans and
Atmosphere, 24(1), pp. 85-90, 1999.
Swain, S. “Impact of climate variability over Mahanadi
river basin”, International Journal of Engineering
Research and Technology, 3(7), pp. 938-943, 2014.
Swain, S., Verma, M., Verma, M. K. “Statistical trend
analysis of monthly rainfall for Raipur district,
Chhattisgarh”, International Journal of Advanced
Engineering Research and Studies /IV/II/Jan-March,
pp. 87-89, 2015.
Ventura, F., Pisa, P. R., & Ardizzoni, E. “Temperature
and precipitation trends in Bologna (Italy) from 1952 to
1999”, Atmospheric Research, 61(3), pp. 203-214,
2002.
Yue, S., Pilon, P., Phinney, B., & Cavadias, G. “The
influence of autocorrelation on the ability to detect
trend in hydrological series”, Hydrological Processes,
16(9), pp. 1807-1829, 2002.
Yue, S., Hashino, M. “Long term trends of annual and
monthly precipitation in Japan”, Journal of the
American Water Resources Association, 39(3), pp.
587-596, 2003.
Zhu, Q., Jiang, H., Liu, J., Wei, X., Peng, C., Fang, X.,
Liu, S., Zhou, G., Yu, S., Ju, W. “Evaluating the
spatiotemporal variations of water budget across China
over 1951–2006 using IBIS model”, Hydrological
processes, 24(4), pp. 429-445, 2010.