Meteorology and Atmospheric Physics (2021) 133:1089–1109

https://doi.org/10.1007/s00703-021-00796-z

ORIGINAL PAPER

Spatio‑temporal analysis of rainfall pattern in the Western Ghats

region of India
B. Venkatesh1 · P. C. Nayak2 · T. Thomas3 · Sharad K. Jain4 · J. V. Tyagi4

Received: 15 May 2020 / Accepted: 25 March 2021 / Published online: 9 April 2021
© The Author(s), under exclusive licence to Springer-Verlag GmbH Austria, part of Springer Nature 2021

Abstract
This paper investigates the rainfall pattern and its distribution, which is one of the key drivers for the water availability in the
Western Ghats region of India. The daily precipitation data from 1901 to 2014 were investigated to understand the rainfall
pattern and its variability. The study area has been divided as southern, coastal, central, and northern regions and rainfall
amount into 0.2–10 mm, 10–20 mm, and > 50 mm classes. The trends in the annual rainfall, number of rainy days, and various
classes within these regions have been investigated. The mean annual rainfall varies between 4000 mm in the coastal areas
and greater than 6000 mm at the mountain crest. The Mann–Kendall test indicates significant decrease of annual rainfall,
at 5% significance level in central and coastal region. On Contrary, the coastal region recorded significant decrease in the
number of rainy days at 5% significance level. An increasing trend in the number of rainy days has been observed in the
northern region. No significant trends have been detected in the southern region. A decreasing trend has been detected in the
number of rainy days in the rainfall class with 10–20 mm in coastal region. Similarly a significant decrease of rainy days was
observed in > 50 mm rainfall class both in the southern and northern regions. The change point probability and homogeneity
test indicates that there is a significant increase in rainfall in few stations under the northern region and similarly decrease
in the rainfall in central region during the decade of 1970 and normally the rainfall has been homogeneous in the study area.
From the precipitation concentration analysis found that Mysore, Kodagu, and adjoining areas are influenced by the bi-modal
rainfall, receiving the rainfall during both south–west and north–east monsoons. This analysis helps to understand the spatial
variability of rainfall in the Western Ghats of India, which depict mixed trends in the quantum of rainfall and decreasing
trend of rainy days across various rainfall classes. These changes detected in the historical rainfall, both in its occurrence
and quantum will have significant bearing on the water availability scenario in the Western Ghats region.

1 Introduction agricultural sectors. During the last few decades, there are
evidences of increase of extreme rainfall events all around
Water scarcity is becoming an increasing problem world- the globe. Global warming is thought to be linked with the
wide due to increasing population, improving living stand- increase in occurrences of heavy rainfall events due to an
ards, and increasing water demands by the industrial and increase in atmospheric vapour and the warmer air (Gos-
wami et al. 2006; Turner and Annamalai 2012). Under-
standing the changes in the monsoon rainfall in a warm
Responsible Editor: Emilia Kyung Jin.

1
* P. C. Nayak Hard Rock Regional Centre, National Institute of Hydrology,
nayakpc@gmail.com Visvesvaraya Nagar, Belgavi, Karnataka 590019, India
2
B. Venkatesh Deltaic Regional Centre, National Institute of Hydrology,
bvenki30@gmail.com Sidhartha Nagar, Kakinada, Andhra Pradesh, India
3
T. Thomas Central India Hydrology Regional Centre, National Institute
thomas_nih@yahoo.com of Hydrology, WALMI Campus, Near Kaliasote Dam,
Bhopal, Madhya Pradesh, India
Sharad K. Jain
4
s_k_jain@yahoo.com National Institute of Hydrology, Jalvigyan Bhavan, Roorkee,
Uttarakhand, India
J. V. Tyagi
jv_tyagi@yahoo.com

13
Vol.:(0123456789)
1090 B. Venkatesh et al.

environment is challenging and of great societal importance. was reported for Hyderabad, Jaipur, Kanpur, and Nagpur
Careful study of past changes in the climate variables dur- showed by − 10.4%, − 10.5%, − 7.1%, and − 4.8%, respec-
ing the historical time horizons may provide valuable evi- tively. Thomas et al. (2015) investigated the trend in extreme
dences for understanding the global warming impacts on the rainfall events using the daily rainfall of 23 grid cells cover-
monsoon precipitation including the occurrences of extreme ing the Narmada basin and reported a significant increasing
event. trend of 1 day maximum rainfall at 5% significance level.
The Indian Summer Monsoon Rainfall (ISMR) primar- Kundu and Modal (2019) analyzed long-term annual and
ily regulates the hydrology of the Indian sub-continent. The seasonal rainfall trends along with change point of annual
country’s dependence on monsoon rainfall is a well-known rainfall in West Bengal, India for 102 years (1901–2002)
fact, as more than 50% of the cultivated area is dependent using monthly rainfall data of 18 rainfall stations using MK
on the vagaries of the monsoon. The agricultural activities and Sen’s slope. Malik et al. (2019) investigated the spatial
of India have thus evolved around the monsoon season, even and temporal patterns of trends on seasonal (pre-monsoon,
though the onset of the monsoon in Kerala has a standard monsoon, post-monsoon, and winter) and the annual rain-
deviation of about 7 days (Subash and Gangwar 2014). An fall time series data (1966–2015) at 13 stations located in
interesting question is, whether the length and/or the spatial the central Himalayan region of the Uttarakhand State of
extent of the monsoon season are changing. The increasing India. Malik and Kumar (2020) investigated the spatial and
occurrence of climatic extremes in recent years has led to temporal patterns of trends and magnitude of rainfall on
the hypothesis that the characteristics of hydro-climatic vari- monthly, seasonal, and annual time scales of 13 districts of
ables are changing. Uttarakhand State located in Central Himalayan region of
Several past studies have investigated the trend and mag- India. The temporal trend was analyzed using Mann–Kendall
nitude of rainfall on the basin scale. Singh et al. (2008) (MK), Modified Mann–Kendall (MMK), and Kendall Rank
detected decreasing trends in the annual rainfall over major Correlation (KRC) tests at 10%, 5%, and 1% significance
Central Indian river basins, viz., Sabarmati, Mahi, Narmada, levels. Tirkey et al. (2000) reported spatial and temporal
Tapi, Godavari, and Mahanadi, since the 1960s, whereas variation of precipitation, long-term precipitation data of
increasing trends were detected over the major North Indian 113 years (1901–2013) using non-parametric Mann–Kendall
river basins, viz., Indus, Ganga, and Brahmaputra, and South (MK) and modified Mann–Kendall (MMK) tests.
Indian river basins, viz., Krishna and Cauvery. Using the The Western Ghats of India acts as a barrier to the
daily gridded rainfall dataset for India during 1951–2003, south–west monsoon clouds and influence the spatio-tem-
Krishnamurthy et al. (2009) observed a significant increase poral distribution of rainfall in the region. The undulating
in frequency of 90th ‰ daily rainfall and a decreasing trend landscape, slope aspect, and direction of exposure of these
in extreme precipitation frequency over many parts of India. mountains to the monsoonal winds have been posing many
Rakhecha and Soman (1994) studied the annual extreme a challenge to the scientific community, in understanding
rainfall for 1–3 days duration over India during 1901–1980 the spatio-temporal distribution of rainfall. The spatio-
and most of the stations did not show any trend and persis- temporal rainfall distribution in the Himalayan region has
tence. However, the Western Ghats and the central peninsu- been studies in greater depth compared to the Western
lar area indicated significant increasing trends at 5% signifi- Ghats (WG) despite the lack of data (Shrestha et al. 2008;
cance level. Goswami et al. (2006) reported significant rising Bharati et al. 2016). Chandrasekhar et al. (2017) analyzed
trends in the frequency and magnitude of extreme rainfall the seasonal spatio-temporal variation in trends of long-term
events and significant decreasing trends in the frequency (1901–2013) observed high resolution (0.25° × 0.25°) grid-
of moderate rainfall during monsoon season in the central ded daily precipitation data of the Indian Meteorological
India during 1951–2000. Kumar and Jain (2010) observed Department over the Western Ghats and the coastal region
an increasing trend in annual rainfall in 15 sub-divisions out of Karnataka, vulnerable to the risks of climate change.
of 30 sub-divisions of India during 1871–2005. Revadekar et al. (2018) examined long-term as well as short-
Taxak et al. (2014) studied long-term spatial and tem- term trends and variability in summer monsoon rainfall over
poral trends of gridded rainfall data of Wainganga river different sub-divisions of WG using monthly rainfall data
basin in central India and reported 8.45% decrease in for the period 1871–1920. Varikoden et al. (2019) reported
annual rainfall during 1901–2012. Kharolet al. (2013) trends of rainfall in the northern and southern regions of
analyzed the trends in monsoon rainfall and the num- WG and examines possible reasons for the phenomenon
ber of rainy days over six major Indian cities, viz., Delhi, and found an average trend of + 0.3 (− 0.39) mm ­day−1
Hyderabad, Kanpur, Jaipur, Nagpur, and Bangalore dur- ­decade−1 is estimated in the northern (southern) WG for the
ing 1951–2007. They reported an increase in the number 1931–2015 period.
of rainy days by 7.4% and 22.9% increase over Delhi and The scarce data in Western Ghats region has been a major
Bangalore, respectively. However, decrease in the rainy days impediment, which is hindering the understanding of the

13
Spatio‑temporal analysis of rainfall pattern in the Western Ghats region of India 1091

regional weather system. The major rivers of peninsular 80–85%) during these 4 months with longer drier month
India originate in this mountain range and the livelihood (Gadgil 1987). These ten districts are grouped into Southern
of people in this region dependent on the water available in Region (Kodagu, Mysore, Chamarajanagar), Central Region
these rivers. There are many major water resources projects (Chickamangalore, Hassan, Shimoga), Coastal Region
constructed on these rivers to cater the needs of the arid and (Mangalore, Udupi and windward parts of Uttara Kannada),
semi-arid areas of these states. Any changes in the rainfall and Northern Region (Belgaum and leeward part of Uttara
pattern would result in the variation of water availability Kannada district). The formation of these groups helps to
and would directly impact on livelihood of the people and understand the latitudinal variation and also the influence
economy of the region. Therefore, there is an urgent need of Western Ghats on rainfall pattern in leeward side (Fig. 1).
for carrying out a comprehensive analysis of rainfall for their This analysis is crucial for the region in two ways; i.e., (i)
distribution, trends and homogeneity, which would help in many major rivers originate from these mountain ranges;
developing water resources management plans for the states and (ii) the leeward side is dominated by the rainfed agricul-
such as Goa, Karnataka, Kerala, and Maharashtra covering ture. So, any change in the pattern in the rainfall will have a
the Western Ghats. greater impact on the water availability of these rivers and
Previously, very few attempts have been made to study on agricultural activities and the economy of the region.
the spatio-temporal behavior of rainfall in the Western There are number of rain gauge stations being maintained
Ghats region. A literature review reveals that, most of the by the Water Resource Development Organisation (WRDO),
studies have been carried out in the Western Ghats located Govt. of Karnataka and the data are available for more than
in Kerala (Bhowmik and Durai 2008; Suji Kumar et al. 100 years. However, at many locations, the data are not con-
2013). Gopalkrishnan et al. (2011) observed that northern tinuous. Therefore, this analysis uses 122 rainfall stations
and central part of the Western Ghats are vulnerable to cli- which have continuous rainfall for a period of 1901–2014.
mate change and reported an increase in temperature may The rain gauge stations considered for the analysis, along
cause high intensity rainfall. But, there are no comprehen- with the two transects which are considered for detailed
sive analyses of rainfall carried out until now in this region. analysis are shown in Fig. 1 and basic statistics are tabu-
Therefore, this study makes an attempt to understand spa- lated in Table 1. Rainfall data have been primarily tested for
tio-temporal variation in the rainfall which is an important homogeneity and identified 122 stations have qualified the
driver for the water availability of these non-perennial rivers. test. These 122 stations are subsequently grouped as coastal,
Keeping this in view, the analysis has been carried out with central, southern and northern regions. Various statistical
the following objectives: (i) analysis for understanding the analyses have been carried out to assess the spatio-temporal
spatio-temporal variation of annual rainfall and number of variation of rainfall. The daily rainfall data have been used
rainy days; (ii) the variation of rainfall in medium, moderate to compute the rainfall amount, number of rainy days for
and high rainfall classes; and (iii) to explore the controls that each year, and for four rainfall classes, viz., 0.2–10 mm,
may be responsible for these variations including the impact 10–20 mm, 20–50 mm and > 50 mm. The derived datasets
of climate change. have been used for trend analysis using the non-parametric
Mann–Kendall test and Sens’s slope method.
Two transects in the northern and southern regions have
2 Materials and methods been selected to investigate spatial variability of rainfall.
These transects are selected primarily on fact that, the south-
Gunnell (1997) outlined the climatic features of the Western ern region is characterized by the bi-modal rainfall and the
Ghats as distinctly monsoonal. The rainfall is dominated northern region is uni-modal with a specific rainfall charac-
by the orographic uplift of moist, southwest monsoon air- teristic as observed by Gunnel (1997).
flow with the topographic barrier by the mountain range
during the months June–September. Geomorphologically,
the study region consists of sharp, contrasting topography 3 Methodology
often with gentle undulating hills which rise steeply from a
narrow coastal strip bordering the Arabian Sea. The altitude Even though various approaches have been used for trend
range from sea level to a height of 1800 m with dissected and step change analysis, the present analysis is intended to
hills rising above 600–900 m before the major peaks of the understand the temporal variation of rainfall and the reasons
mountain ranges. The districts which fall within the Western and consequences for the variations. As many hydrologic
Ghats region in Karnataka state are Belgaum, Chamarajana- time series are not normally distributed, non-parametric test
gar, Chickamangalore, Hassan, Kodagu, Mangalore, Mysore, has been preferred over parametric test, as they are generally
Shimoga, Uttara Kannada, and Udupi (Fig. 1). These dis- distribution free but, however, do not quantify the size of the
tricts receive their major part of the annual rainfall (upto identified trend.

13
1092 B. Venkatesh et al.

Fig. 1  Location of rain gauge stations and the two transects taken for detailed analysis

3.1 The Mann–Kendall test (MK test) ⎧ +1, if �X − X � > 0 ⎫

� � ⎪ � j k� ⎪
This MK test is a non-parametric rank-based procedure used
sign Xj − Xk = ⎨ 0, if �Xj − Xk �= 0 ⎬ (2)
⎪ −1, if Xj − Xk < 0 ⎪
to detect trend of variables in meteorology and hydrology ⎩ ⎭
fields. The MK test is robust enough to the influence of
extremes and suitable for application with skewed variables where n is the length of the sample, Xk and Xj are from
(Mann 1945; Kendall 1975). This rank-based test, which k = 1, 2, ….. n − 1 and j = k + 1, …… n. Sign() is the sign
makes no assumption about the underlying probability dis- of the difference between Xj and Xk as can be visualized in
tribution of data and linearity of trend is robust to the effect Eq. (2). The S statistic, in cases where the sample size ‘n’
of outliers in extreme indices, has been widely used to assess is larger than 10, is assumed to be asymptotically normal,
trends in hydro-climatic time series (De Lima et al. 2007). with mean, E(S) = 0 and the variance of S can be acquired
According to this test, the null hypothesis H0 indicates as follows:
that there is no trend in the time series (X1…Xn), which is n(n − 1)(2n + 5)
a sample of ‘n’ independent and identically distributed ran- var(S) = (3)
18
dom variables.
The standard normal variate, Z is computed as given
∑ ∑
n−1 n
( ) below:
S= sign Xj − Xk (1)
k=1 j=k+1

13
Spatio‑temporal analysis of rainfall pattern in the Western Ghats region of India 1093

Table 1  Annual rainfall statistics of the study area

ID Annual rainfall statistics ID Annual rainfall statistics
Mean Max Min PSD Mean Max Min PSD

Northern 91 5787.29 8383.00 3186.50 17.95 Central 40 558.75 1123.00 129.40 40.93
92 6166.59 7147.90 3950.20 72.72 41 1002.75 2303.20 461.80 31.61
93 5632.72 9786.40 3394.30 30.62 42 689.40 1360.90 129.10 30.48
94 1148.45 2505.80 531.60 38.91 43 532.41 1031.30 247.90 39.07
96 3577.02 6536.30 2160.40 26.73 44 623.94 1540.50 117.80 42.58
97 5067.44 8075.50 2339.20 20.19 45 648.40 1255.70 396.70 38.97
98 4290.85 8529.30 2679.50 31.41 46 612.90 1186.60 324.50 36.88
99 1272.21 1850.40 839.20 18.79 47 2266.45 5045.50 1190.90 26.06
101 2326.55 4181.50 1467.10 29.26 48 699.49 1421.80 318.30 27.80
104 1190.71 1676.20 655.80 20.28 49 1995.70 3241.20 1269.20 21.72
106 1218.45 3402.10 557.60 30.45 50 4933.51 8508.20 2091.20 31.73
107 2455.73 4200.90 1447.80 19.98 70 4143.85 7525.30 2800.40 23.84
108 2487.53 3507.20 1583.00 18.09 71 6407.48 8764.00 4680.70 17.67
109 2465.72 4205.60 1362.60 23.09 72 2628.83 5310.00 1806.50 29.99
111 3064.94 5228.30 1978.00 25.56 73 4334.18 7742.80 2892.20 24.53
113 1430.63 2926.50 767.60 37.52 74 2774.95 4360.00 1122.30 33.78
114 1765.03 2430.40 1305.40 17.64 75 1008.29 2214.90 390.90 37.07
115 2954.21 5645.80 1852.10 26.06 76 2104.99 4826.90 993.00 26.58
116 2146.53 4212.50 1031.50 40.52 77 1614.86 4111.00 531.40 43.49
117 2918.47 6128.10 703.10 54.72 78 1594.64 2963.90 732.60 26.08
118 3748.84 7654.70 1032.50 51.85 79 4459.42 7181.20 2831.00 26.71
119 4471.04 8150.80 1382.40 51.69 80 1223.53 2503.00 690.40 36.66
120 770.66 1209.40 481.90 27.65 Southern 1 681.05 1347.00 376.40 36.46
121 1795.62 3162.90 756.10 23.88 2 837.12 2082.40 110.30 37.28
122 1323.12 2233.90 732.80 22.42 3 764.87 1200.90 379.00 24.68
Central 13 3252.76 5629.80 1221.30 23.21 4 862.79 3336.20 885.10 34.75
14 644.12 998.40 234.70 29.52 5 985.69 3425.70 227.80 19.21
15 2208.88 4599.20 1060.70 30.98 6 825.97 1646.40 277.60 35.86
16 614.61 1160.60 277.70 27.54 7 741.27 1153.40 429.20 23.98
17 2955.59 4625.90 1662.40 18.83 8 703.44 1101.80 320.40 25.22
18 2314.20 4336.50 1269.70 22.55 9 612.25 1294.70 231.70 35.26
19 1745.64 2921.90 960.20 21.15 10 1103.73 1444.80 799.90 16.88
20 2453.94 4050.00 1236.10 21.88 11 850.41 1540.20 477.80 32.38
21 523.38 904.00 133.70 38.20 12 713.32 1400.60 298.40 28.76
22 704.72 1305.20 107.10 29.61 51 2620.77 5175.30 1649.30 22.30
23 704.72 1103.80 326.50 29.61 52 2343.09 4663.40 1153.70 23.03
24 3783.56 6093.50 2362.30 18.82 53 2966.10 5107.80 1663.60 20.28
25 3356.05 5488.80 1413.00 27.27 54 2189.16 3599.90 1290.40 21.82
37 2898.15 5712.90 1050.70 28.53 55 4033.96 2227.00 100.90 24.50
38 1018.01 1714.40 278.70 25.65 56 1081.51 2992.60 1284.20 29.35
39 698.38 1763.40 265.50 31.95 57 2166.08 3156.20 1019.90 21.99

13
1094 B. Venkatesh et al.

Table 1  (continued)
ID Annual rainfall statistics ID Annual rainfall statistics
Mean Max Min PSD Mean Max Min PSD

Southern 58 2824.73 4020.40 1670.00 19.93 Coastal 34 3995.17 5248.40 2738.30 15.96
59 728.55 1341.20 367.30 24.97 35 4142.12 6794.30 2500.60 24.47
60 851.21 1564.50 475.20 24.83 36 4532.06 6943.60 2978.80 15.55
61 727.26 1455.50 375.20 40.91 81 4739.02 6678.60 1770.70 20.41
62 608.88 1070.00 246.50 43.83 82 5516.87 7585.90 3842.10 20.16
63 1033.77 4975.50 539.70 79.45 83 5611.24 7362.70 3221.00 18.00
64 822.58 1224.60 345.80 27.58 84 3421.94 4505.70 2064.20 16.90
65 866.34 1206.00 297.80 23.26 85 5324.33 8900.00 2671.20 34.81
66 866.34 1440.00 470.60 23.26 86 4003.50 5609.90 2627.50 19.29
67 780.46 1416.00 428.60 25.59 87 4101.89 7525.30 2800.40 25.53
68 733.28 1408.00 339.90 31.65 88 4943.33 6893.90 2065.00 28.85
69 701.99 1348.30 332.30 26.16 89 3745.53 5897.20 2635.90 17.28
Coastal 26 4065.29 5464.00 2620.50 18.59 90 4758.76 6582.70 3242.30 15.36
27 3754.20 6939.80 2468.00 25.40 95 3634.52 6666.90 2125.10 25.10
28 3799.51 6056.90 2746.60 16.22 100 4315.24 6574.90 2747.60 23.75
29 3932.89 7205.00 2406.80 17.27 102 3424.66 5620.00 1826.00 18.98
30 3783.84 5197.70 2453.60 16.65 103 3930.54 6788.60 2402.10 19.18
31 4488.00 6437.00 2539.20 19.68 105 3775.67 6418.40 2241.70 57.49
32 4488.00 6339.90 3070.30 19.68 110 4048.73 5188.20 2775.30 16.03
33 3574.83 6791.40 1088.50 23.62 112 3403.46 4980.20 2408.60 18.68

⎧ √S−1 , if S > 0 ⎫ 3.3 Precipitation concentration index

⎪ Var(S) ⎪
Z = ⎨ 0, if S = 0 ⎬ (4) The precipitation concentration index (PCI) that can be used
⎪ √S+1 , if S < 0 ⎪
⎩ Var(S) ⎭ as index in gathering insight into the variability of rainfall
and their influence on hydrological processes. This index
The null hypothesis that there is no trend in the time can also be used for deriving the necessary information on
series is rejected if the test statistic satisfies the condition long-term total variability in the amount of rainfall received
|Z| > Z1−∝ ∕2 or |Z| < Z1−∝ ∕2 Therefore, the alternative at a location and combined for a region.
hypothesis H1 is that there exists a negative or positive trend The precipitation concentration index (PCI) can be esti-
in the time series. mated an annual scale for each measuring point as:
Given a confidence level∝ , the sequential data would
∑12
be supposed to experience statistically significant trend if i=1 P2i
|Z| > Z1−∝ ∕2 , where Z1−∝ ∕2 is the corresponding value of PCIannual = � �2 (5)
∑12
P = ∝∕2 following the standard normal distribution. In this i=1
Pi
study, 0.05 confidence levels have been used. Therefore, if
Z >  + 1.96 it indicates a positive trend at 95% significance The precipitation concentration index was also calculated
level and if Z < 1.96 indicates a negative trend at 95% sig- on a seasonal scale for monsoon season (June to Sept.) using:
nificance level. ∑4
i=1P2i
PCIMonsoon = � �2 (6)
3.2 The Sen’s estimator of slope ∑4
P
i=1 i

Sen’s slope estimation (Sen 1968) is another non-parametric

The PCI values of less than 10 represent a uniform pre-
method for trend analysis of hydro-climatic data set. This
cipitation distribution (i.e., low precipitation concentration);
involves computing slopes for all the pairs of points and the
PCI values from 11 to 15 denote a moderate precipitation
median of these slopes is used as an estimate of the overall
concentration; values from 16 to 20 denote irregular distribu-
slope. Sen’s slope is insensitive to outliers and can be used
tion and values above 20 represent a strong irregularity (i.e.,
to detect the trend in the data.
high precipitation concentration) of precipitation distribution.

13
Spatio‑temporal analysis of rainfall pattern in the Western Ghats region of India 1095

3.4 Homogeneity test ∑
t
∑
T
Ui,T = Dij (7)
A climatic series is said to be homogeneous, if the observed i=1 j=i+1

variation is resulting from fluctuations in weather and cli- The Pettitt’s statistic for the various alternative hypoth-
mate exclusively. It is important to carry out different homo- eses is given by:
geneity tests. For example, standard normal homogeneity
test (SNHT) is sensitive to change point towards the begin- KT = max |U |, for the two - tailed case,
1≤t≤T | t,T |
ning and end of the data series while Buishand’s and Pettitt’s
tests are sensitive to changes in the middle of a series. KT = −max U , for the left - tailed case,
1≤t≤T t,T

3.5 Pettitt’s test
KT = max U , for the right - tailed case.
1≤t≤T t,i,T
Pettitt test is a rank-based test used for detecting signifi-
cant changes in the mean of time series data when the exact
time of change is unknown. The test is considered robust to 3.6 Alexandersson’s SNHT test
changes in the distributional form of time series and rela-
tively powerful compared to Wilcoxon–Mann–Whitney test, A statistic Ty is used to compare the mean of the first y years
Cumulative Sum, and Cumulative derivations. Furthermore, with the last of (n-y) years and can be written as below:
Pettitt test has been widely adopted to detect changes in the Ty = yz1 + (n − y)z2 , y = 1, 2, 3, … , n (8)
climatic and hydrological time series.
The null and alternative hypotheses will be reformulated where
as follows: 1 ∑
n
(y1 −y) 1 ∑
n
(y1 −y)
z1 = and z2 = .
Ho: T variables follow one or more distributions that have y
t=1
s n−y
t=y+1
s

the same location parameter, The year y consisted of break if value of T is maximum.
Two-tailed test—Ha: there is a time ‘t’ when there is To reject null hypothesis, the test statistics:
change of location parameters in the variables, To = max Ty
Left tailed test—Ha: there is a time ‘t’ when the location 0≤y≤n

parameter in the variable is reduced by D.

is greater than the critical value, which depends on the
Right tailed test—Ha: there is a time ‘t’ when the location
sample size.
parameter in the variable is augmented by D.
The statistic used for Pettitt’s test is computed as follows:
( ) ( ) ( )
Let Dij = −1, if xi − xj > 0, Dij = 0, if xi − xj = 0, Dij = 1, if xi − xj > 0,

Fig. 2  Spatial distribution of a mean annual rainfall, b percent standard deviation (PSD) of annual rainfall, and c mean annual number of rainy
days

13
1096 B. Venkatesh et al.

4 Results and discussion to semi-arid); and (iii) the intermediate hill ranges may influ-
ence the variation in the local rainfall with a short distance.
The Western Ghats run parallel to the west coast of India The higher mean annual rainy days are recorded in the
and exert influence on the monsoonal rainfall in this region. coastal region (154 days) and lowest (32 days) in central
An earlier research by Venaktesh and Jose (2007) identi- region (Table 2 and Fig. 2c) with higher PSD of 22% in the
fied three rainfall homogeneous regions within the Western coastal region. The southern region which is bordering the
Ghats of Karnataka. While, Madhavan (2009) identified coastal region has recorded 146 rainy days and the central
the strong spatial variation in precipitation amount falling region recorded 132 rainy days with higher PSD (45%). The
close to the summit and their sharp decrease towards pla- southern region, the rainfall events are of higher intensity
teau. This work is an extension of these two studies which with shorter duration resulting in lower number of rainy days
focuses on assessing the influence of the Western Ghats on (Putty et al. 2000). Such variability in the rainfall distribu-
the Spatio-temporal distribution of the rainfall for the four tion are influenced by the orography and topography of the
distinct regions, viz., northern, central, southern, and coastal region. However, Venkatesh et al. (2006) observed that, the
regions (Fig. 1). As a first step, the autocorrelation was car- rainfall of longer duration with lower intensity may have
ried out to check the randomness of the data (Modarres and contributed for the higher number of rainy days in the north-
Silva 2007). As all lag-1 serial correlation coefficients were ern region. Overall it is observed that, higher variability and
statistically not significant, there was no need to pre-whiten more number of dry days and dry spells in the central region
the data, and all statistical tests described in the previous as compared to the other regions in the study area. The stud-
section has been applied to the original time series. ies carried out by Sumner and Bonell (1986) and Lyons and
Bonell (1992) reported similar pattern of rainfall in coastal
4.1 Statistical characteristics of annual rainfall areas of Australia, wherein the influence of local topography
and rainy days and exposure to moist air flow were noted as important fac-
tors responsible for such spatial variation.
The basic statistics computed for all stations are presented
in Table 1 and Fig. 2a. As reported elsewhere (Sumner 4.2 Spatio‑temporal variation of rainfall quantum
and Bonell 1986; Jackson and Weinand 1994; Lyons and and rainy days
Bonell 1994; Connor and Bonell 1998; Goswami et al. 2006;
Lacombe and McCartney 2014), the rainfall over the tropical The trend analysis of annual rainfall was carried out using
region is characterized by high inter-annual variability. This MK test for all the four regions. The spatial distribution of
is true to the study area as well, as it covers both high and rainfall trend is represented by a symbolic representation of
low rainfall regimes. The mean annual rainfall varies from upward and downward arrows for increasing and decreas-
6400 mm to a low of 486 mm in the central region. How- ing trends, respectively, and is given in Fig. 3a. Tables 3,
ever, stations close to ridge of the mountain often receive 4, 5 and 6 show the magnitude of Sen’s slope (β) of annual
more than 6000 mm rainfall. Whereas as the mean rainfall total rainfall and Z values of MK test. It is interesting to
in the coastal region varies from about 4000 mm in coastal note that, by and large, the decreasing trend is dominating
regions to > 6000 mm at mountain crust. This pronounced the entire study area. The coastal region has witnessed the
change in the rainfall happens within a very short distance extreme rate of decrease (32 mm) and increase (71 mm).
from the coast to the mountain summit on windward as well However, the central and southern regions are dominated by
as on leeward side of the Western Ghats. The topography of the decreasing trend, whereas the northern region is domi-
the mountain may possibly be influencing these changes, as nated by increasing trend.
it acts as a barrier to the moisture laden monsoon clouds. The southern (Table 4) region of the study area, covered
This effect is evident in the spatial variation of rainfall pat- by Mysore, Kodagu, and Chamarajanagar districts, receives
tern along the higher altitudes. The mean annual rainfall rainfall during both South–West (SW) and North–East (NE)
reduces in the plains located adjacent to the Western Ghats, rainfall systems. The SW rainfall is the principal contributor
as it falls in the rain shadow area. The variability of rainfall to the annual rainfall. The Sen’s slope for all the stations in
is presented as Percent of Standard Deviation (PSD), which the southern region is tabulated in Table 4. Figure 3a indi-
varies between 34 and 79% in the study area (Fig. 2b). The cates a decreasing trend of rainfall in most parts of the south-
northern, central, and southern regions recorded very high ern region. This decreasing trend may be due to the fact that
values of PSD (> 70%) indicating very high variability. This these stations are on the leeward side of the Western Ghats
could be attributed to: (i) it geographical setting (stations covering plateau. Krishnakumar et al. (2009), observed a
falls on leeward side of the Western Ghats); (ii) many of significant decreasing trend in the state of Kerala during
these stations fall near the transition zone (from sub-humid S–W monsoon. As the southern part of the study area has
common boundary with the state of Kerala and the monsoon

13
Spatio‑temporal analysis of rainfall pattern in the Western Ghats region of India 1097

Table 2  Statistics of rainy days in the study area

ID Mean SD PSD ID Mean SD PSD

Northern 91 118.63 11.09 9.35 Central 43 32.69 10.82 33.09

92 115.50 16.24 14.06 44 39.95 16.97 42.47
93 109.58 16.47 15.03 45 50.08 21.03 42.00
94 89.13 21.32 23.92 46 40.12 14.68 36.58
96 111.54 16.23 14.55 47 119.09 14.79 12.42
97 106.89 17.72 16.58 48 56.36 13.98 24.81
98 106.00 23.13 21.82 49 107.38 19.00 17.69
99 103.52 13.24 12.78 50 115.13 30.79 26.75
101 100.90 16.90 16.75 70 130.88 25.26 19.30
104 114.54 13.14 11.47 71 121.58 13.49 11.09
106 121.08 19.65 16.23 72 102.05 16.74 16.41
107 124.35 16.21 13.04 73 112.71 15.30 13.58
108 127.19 19.56 15.38 74 106.13 14.18 13.36
109 102.97 19.06 18.51 75 75.03 19.03 25.37
111 100.41 14.43 14.37 76 112.32 16.86 15.01
113 90.96 17.34 19.06 77 96.36 16.09 16.70
114 104.04 11.85 11.39 78 102.22 15.33 15.00
115 129.73 23.90 18.42 79 124.09 13.24 10.67
117 110.19 13.34 12.11 80 92.91 18.40 19.81
118 106.37 14.38 13.52 Southern 1 59.00 19.19 32.52
119 118.70 26.91 22.67 2 53.60 17.25 32.19
120 75.42 22.23 29.47 3 70.57 25.56 36.22
121 114.20 15.56 13.63 4 78.00 17.11 21.93
122 118.47 16.68 14.08 5 45.96 12.16 26.47
Central 13 120.86 16.95 14.03 6 69.94 16.23 23.20
14 47.82 11.52 24.08 7 54.93 12.76 23.23
16 54.79 19.37 35.36 8 46.26 11.64 25.16
17 54.34 13.75 25.31 9 44.66 17.16 38.44
18 132.39 13.91 10.51 10 95.59 19.83 20.74
19 109.01 14.86 13.64 11 57.18 12.71 22.22
21 40.32 18.14 45.00 12 52.11 13.30 25.52
22 57.83 17.93 31.01 51 141.52 19.22 13.58
23 31.17 11.65 37.37 52 133.60 19.96 14.94
24 126.95 14.31 11.27 53 146.10 17.53 12.00
25 125.14 16.83 13.45 54 132.30 17.62 13.32
37 115.06 25.49 22.15 55 114.71 16.70 14.56
38 91.89 20.82 22.66 56 128.44 17.56 13.68
39 55.62 10.89 19.58 57 113.37 18.97 16.73
40 35.93 14.76 41.08 58 127.96 19.49 15.23
41 79.96 17.97 22.47 59 51.15 11.91 23.28
42 54.50 18.14 33.28 60 73.16 19.03 26.01

13
1098 B. Venkatesh et al.

Table 2  (continued)
ID Mean SD PSD ID Mean SD PSD

Southern ID Mean SD PSD Coastal ID Mean SD PSD

61 58.42 21.28 36.43 81 125.53 14.46 11.52
63 70.81 14.90 21.04 82 125.12 13.49 10.78
64 94.41 19.28 20.42 83 133.00 12.42 9.34
65 63.79 19.28 30.22 84 118.60 15.95 13.45
66 66.36 17.09 25.75 85 120.72 14.34 11.88
67 60.42 17.71 29.31 86 117.48 13.30 11.32
68 55.95 11.07 19.79 87 129.82 26.75 20.61
69 58.21 15.87 27.26 88 130.00 28.67 22.05
Coastal 26 135.50 14.08 10.39 89 131.03 12.12 9.25
27 131.95 15.35 11.64 90 149.20 13.52 9.06
28 137.27 14.04 10.23 95 120.31 15.93 13.24
29 138.31 13.26 9.59 100 110.74 14.64 13.22
30 143.93 15.25 10.59 102 120.86 13.09 10.83
31 144.79 16.51 11.41 103 124.33 13.84 11.13
32 154.48 18.16 11.75 105 124.05 15.44 12.45
33 130.03 19.79 15.22 110 119.37 11.17 9.36
35 124.64 12.25 9.83 112 109.07 14.13 12.96
36 142.17 15.44 10.86

winds are emanating from that direction has more bearing 5% confidence level (Umblebylu and Sakaleshpura). The
on the amount of rain received and their trend. The stations decreasing trend of the annual rainfall is dominated the
which have the close proximity to the maritime system northern region (Table 6) which is bordering the State of
(coastal region) have recorded increasing trend. Goa. However, a significant increasing trend in the rainfall
The stations which are located further inward from the at 5% confidence level has been observed in certain pockets,
mountains have registered decreasing trend. Further it can i.e., northern-most part (Jamgoan, Chapoli, Kanakumbi, and
be noticed that, stations such as Virajpet, Ponnampet, Naga- Khanapur), middle-part (Jagalbet and Joida), and lower-part
rahole, and Kushalnagar stations show a significant decrease (Nilkund, Bandal, and Kundal). Further, it is noticed that
in rainfall at 0.1% confidence level whereas at Napoklu, the these stations are located close to the summit of the Western
decrease in annual rainfall is significant at 0.01% level. It Ghats (Fig. 3a). The results in the present analysis matches
is interesting to note that, these stations are located on the with that of the increasing trend of rainfall in the State of
higher elevation of the Western Ghats and receive very high Goa reported by Nandargi and Mulye (2014), as these sta-
rainfall. The reason could be that some of these may be tions are close to those stations and have a common rain
influenced by the local climate. As Gunnel (1997) reported, producing mechanism. However, within the northern region,
the quantum of rainfall is influenced both by the aspect the stations which are located away from the influence of
and relief in the Western Ghats. On the other hand, further Ghats and closer to the central region have shown decreasing
south in the eastern part of the study area, there is signifi- trend in rainfall but not significant at 5% confidence level.
cant increase in the rainfall (Bailur and Chamarajanagara), The coastal region is dominated by the increasing trend
despite being located fairly at the same elevation range. in rainfall at 5% confidence level (Table 6). The observed
However, the rainfall in the plateau of southern region have rate of increase is as high as 71.50 mm/year at Anasebylu.
non-significant decreasing trend. The stations which are close to the coast show a decreasing
The decreasing trend dominated central region (Table 5) trend, for example Kohila (− 16.34 mm/year), Dharmasthala
and stations bordering the coastal zone. The stations which (− 14.44 mm/year), and Sulya (− 15.06 mm/year), but not
have increasing trend are located mostly in the plateau of significant at 5% confidence level. On the other hand, the
central region and are located far away from the crust of northern side of the coastal region showed a higher increas-
Western Ghats. The stations located in the fringe area on ing trend with a significant increase in rainfall at Ankola
the eastern side recorded significant decreasing trend at (5.08 mm/year).

13
Spatio‑temporal analysis of rainfall pattern in the Western Ghats region of India 1099

Fig. 3  Rainfall and rainy days trend for the study area under different classes a mean annual, b 10–20 mm, c 20–50 mm, and d > 50 mm rainfall
class

4.3 Variation within the selected transects the location and the geographical setting may influence more
orographic precipitation on the windward side of the moun-
Two transects, one in the northern region and another in the tains and hence the higher rainfall quantity and the number
southern region, have been selected to investigate spatial of rainfall days.
variability of rainfall. These transects are selected primar- Figure 4b depicts the variation of mean annual rainfall
ily to capture the nature and characteristics of rainfall in and average rainy days with the elevation in the southern
the southern and northern regions. Figure 4a shows that the region. From Fig. 4b, it is evident that the rainfall reduces
maximum rainfall in the northern transect is > 6000 mm with drastically on the leeward side of the Western Ghats. This
as much as 124 rainy days. Further it is observed that, the may be due to the orographic effect which is responsible for
rainfall decreases and the number of rainy days increases higher rainfall on the windward side and a sharp decline on
moving towards the plains. It is also noticed that the mini- the leeward side (Madavan 2009). It is also noticed that, the
mum number of rainy days in this transect is about 100 days, maximum rainfall and number of rainy days are recorded at
due to very active south–west monsoon season. Further it is locations lower than the maximum elevation and this falls
interesting to note that the maximum rainfall and number of in line with the reported observations of Venkatesh and Jose
rainy days occur at the same elevation, but happen on either (2007). This pattern of variation in rainfall and number of
sides of the mountain summit (at 170 m on windward side rainy days may be due to fact that this region is characterized
and around 970 m on the leeward side).This could possibly by the small hillocks, which may provide requisite condi-
be due to the fact that there are dissected hills on the lee- tions to generate the local rainfall events of lower intensities
ward side of the mountains which may be influencing and and higher durations (Venkatesh et al. 2006).
altering the local climate thereby leading to higher number The plots of mean annual rainfall vis-a-vis average num-
of rainy days. While in the southern transect, the maximum ber of rainy days for northern and southern transects are
rainfall quantum and the number of rainy days are observed shown in Fig. 4b. The plots identify three groups with dis-
at the same elevation. As observed by Taxak et al (2014), tinct characteristics. These groups may be a result of lower

13
1100 B. Venkatesh et al.

Table 3  Sen’s slope and Z (MK test) values for raingauge stations in the southern region (Z values MK test are given in parentheses)
Name Trend analysis of annual rainfall and rainy days (Sen’s slope values)
Annual rainfall Annual rainy Rainfall Rainy days Rainfall Rainy days Rain- Rainy
days 10–20 mm 10–20 mm 20–50 mm 20–50 mm fall > 50 mm days > 50 mm

1 − 16.87** (− − 1.40** (− − 13.54** (− − 0.42* (− − 12.09* (− − 0.25* (− − 5.22** (− − 0.09** (−

3.28) 3.47) 2.68) 2.41) 2.53) 2.32) 2.68) 2.93)
2 0.59 (0.52) 0.48** (4.14) − 0.71 (− 0.00 (0.06) − 1.01 (− − 0.01 (− 0.00 (− 0.61) 0.00 (0.0)
0.57) 0.85) 0.87)
3 0.29 (0.11) 1.27** (3.33) − 1.66 (0.42) 0.09 (0.50) − 2.65 (− 0.00 (− 0.49) − 2.55 (− 0.00 (− 0.60)
0.73) 1.00)
4 − 5.91 (− − 0.44 + (− − 5.92 (− − 0.08 (− − 7.24 (− − 0.12 (− − 2.44 (0.86) 0.00 (− 0.89)
1.19) 1.69) 1.19) 0.91) 1.65) 1.44)
5 11.05* (2.20) − 0.40* (− 12.54* (2.41) 0.50** (3.30) 12.06* (2.43) 0.35** (3.41) 0.68 (1.09) 0.00 (0.94)
2.20)
6 1.71 (0.72) 0.11 (1.02) 1.23 (0.41) 0.06 (1.01) 0.51 (0.12) 0.03 (0.73) − 0.83 (1.70) − 0.01 (− 1.47)
7 2.59 (0.68) 0.37 (1.35) 0.91 (0.40) 0.00 (0.25) 1.97 (0.88) 0.06 (1.34) 0.25 (0.28) 0.00 (0.62)
8 − 5.04 (− − 0.60 (− − 2.84 (− 0.00 (0.17) − 4.17 (− 0.00 (0.00) − 2.50 (− − 0.04 (− 1.44)
0.83) 1.69) 0.79) 1.00) 0.90)
9 0.46 (0.61) 0.15 (2.08) 0.13 (0.25) 0.00 (0.05) 0.14 (0.23) 0.00 (0.16) 0.00 (0.00) 0.00 (0.00)
10 3.81 (0.58) − 1.50** (− 12.12* (2.00) 0.47 (1.63) 13.2** (2.88) 0.42** (3.19) 0.00 (0.04) 0.00 (0.04)
3.19)
11 0.53 − 0.09 (− 0.58 (0.26) 0.04 (0.69) 0.05 (0.03) 0.02 (0.58) 0.00 (0.00) 0.00 (0.49)
(0.12) 0.83)
51 2.20** (− 0.14** (− 1.90** (− 0.06* (− 1.57) 1.72** (− 0.03* (− 1.70) 0.56 + (− 1.20) 0.00 (− 1.30)
2.52) 6.77) 2.06) 1.89)
52 − 4.47* (− − 0.42** (− − 3.69* (1.89) − 0.05 (− − 3.24 + (− − 0.04 + (− − 1.77 (− − 0.01 (− 1.92)
2.26) 2.92) 1.03) 1.97) 2.37) 1.46)
53 − 3.58* (− − 0.16** (− − 2.97 + (− − 0.03 (− − 3.19* (2.26) − 0.05* (1.33) − 1.98 (− − 0.02 + (−
2.69) 2.11) 2.30) 1.22) 2.43) 1.91)
54 − 4.97** (− − 0.12* (− − 4.32* (− − 0.04 (0.46) − 4.01* (− − 0.03 (0.31) − 3.93* (− − 0.03 + (0.00)
0.72) 4.68) 0.26) 0.39) 0.39)
55 − 1.11 (− − 0.27** (− − 0.37 (− 0.01 (− 2.46) − 0.80 (1.74) 0.00 (2.71) − 0.50 (− 0.00 (0.00)
2.55) 0.69) 2.25) 0.49)
56 − 2.41* (− − 0.03 (− − 2.05* (− − 0.08* (− − 1.39 + (− − 0.04* (− 0.00 (− 0.48) 0.00 (− 1.07)
0.89) 1.60) 0.79) 0.62) 0.85) 0.92)
57 − 8.39 (− − 0.50 + (0.37) − 7.64 (− − 0.17 (− − 7.55 (− − 0.18* (− − 4.43 (− − 0.08 (0.00)
2.51) 2.26) 3.08) 2.03) 2.23) 0.07)
58 − 5.36* (− 0.05 (− 0.32) − 5.09* (− − 0.18** (− − 3.45* (− − 0.08 (− − 0.03 (− 0.00 (0.00)
1.36) 1.36) 1,91) 1.31) 1.39) 0.07)
59 − 5.24 (− − 0.02 (0.82) − 5.32 (− − 0.13 + (− − 4.13 (− − 0.08* (− − 4.44 (0.89) − 0.05 (0.00)
0.17) 0.27) 0.98) 0.01) 0.08)
60 − 0.10 (− 0.03 (4.15) − 0.15 (− − 0.01 (− − 0.06 (− 0.00 (− 1.04) 0.15 (− 1.34) 0.00 (0.00)
0.12) 1.45) 1.61) 1.07)
61 − 0.10 (0.87) 0.28** (2.37) − 0.93 (0.44) − 0.04 (0.84) − 0.66 (− − 0.01 (0.05) − 0.31 (− 0.00 (− 0.56)
0.10) 0.61)
62 1.47 (− 0.22) 0.30* (1.38) 0.68 (− 0.60) 0.04 (0.51) − 0.08 (− 0.00 (− 0.34) 0.00 (− 1.66) 0.00 (− 1.44)
0.98)
63 − 7.64* (− 1.00** (3.54) − 10.87** (− − 0.54** (− − 7.61* (− − 0.27** − 0.14 (− 0.00 (− 89)
2.32) 3.13) 3.31) 2.29) (2.91) 0.29)
64 − 7.44 (− 2.0) 0.00 (− 0.04) − 7.05* (− − 0.31* (− − 5.05 (− − 0.12 (− − 1.93 (− 0.00 (− 1.75)
2.07) 2.54) 1.39) 1.56) 1.50)
65 0.60 (0.85) 0.13 + (1.92) 0.44 (0.60) 0.00 (− 0.26) 0.86 (1.30) 0.01 (1.02) 0.50 (1.44) 0.00 (0.00)
66 − 1.69 (− 0.10 + (1.72) − 2.03** (− − 0.05** (− − 1.93** − 0.04** − 0.46 (1.47) 0.00 (0.00)
2.51) 3.05) 2.59) (3.03) (2.80)
67 0.05 (0.10) − 0.09 (− 0.67 (1.21) 0.00 (0.11) 1.07 + (1.90) 0.00 (0.93) 0.64* (2.30) 0.00
1.28) (0.00)

13
Spatio‑temporal analysis of rainfall pattern in the Western Ghats region of India 1101

Table 3  (continued)
Name Trend analysis of annual rainfall and rainy days (Sen’s slope values)
Annual rainfall Annual rainy Rainfall Rainy days Rainfall Rainy days Rain- Rainy
days 10–20 mm 10–20 mm 20–50 mm 20–50 mm fall > 50 mm days > 50 mm

68 0.70 (1.12) 0.02 (0.64) 0.95 (1.27) 0.02 (0.97) 1.09 (1.58) 0.01 (0.99) 0.70* (2.22) 0.00 (0.00)
69 − 0.19 (− − 0.03 (− 0.02 (0.03) 0.00 (0.39) 0.09 (0.16) 0.00 (0.45) 0.00 (0.35) 0.00 (0.00)
0.42) 0.66)

** α = 0.01; * α = 0.05; + α = 0.1

precipitation rate in the plains and proximity to maritime thereby, despite of having an average rainfall between 550
system as it supplies more moisture to the rain producing and 1000 mm, they are, however, characterized by the lesser
system in the coastal region and on the windward side. In rainy days.
the southern transect, the low rainfall amount and number
of rainy days are on the plains and higher values are close to 4.4 Trends analysis of number of rainy days
the mountain region. Whereas, in the northern transect, these
groups characterize the coastal area, mountain area, and the Analysis has also been carried out to study the variation in
eastern part of the mountains where the rainfall decreases the number of rainy days and their trends (Fig. 3e). A rainy
drastically. As described by Venkatesh and Jose (2007), day for the purpose of present analysis is defined as the day
these groups are characterized by higher mean annual rain- with a minimum rainfall of 0.2 mm and extracted the data
fall around the peak of the mountain and lower mean on of rainy days for stations in the study area. The Sen’s slope
the eastern side. Further, the significant difference between was computed and is given in Table 4 and Fig. 3e. From the
these two intersects is that the variation in the number of Fig. 3e, it is observed that number of rainy days is decreas-
rainy days between the groups, lower values are observed ing in most part of the coastal region, which is very impor-
in the northern transect as compared to that in the southern tant finding and crucial for water availability in the coastal
transect, where the variation is much higher. Longobardi and region. There are number of locations where decreasing
Villani (2009) reported a similar observation while studying trend is significant at 0.01% confidence level (Belthangadi,
the rainfall pattern of Campania and Lazio regions of south- Mulky, and Byndur). Further, few stations that have signifi-
ern Italy. They opined that, the mountain range induces the cant decreasing trend in number of rainy days have increas-
barrier effect which will influence the rainfall variability on ing trend in annual rainfall (Mulky, Bantwal, and Byndur).
the leeward side. This implies that, at some stations, the rainfall has become
The other noticeable difference between the northern and more intense over the time which may be indicative of
southern transects is that of the occurrence of maximum change in rainfall pattern influence by the climate change.
rainfall. In the northern transect, the maximum rainfall is The southern region witnessed mixed trends in number of
observed around the crust on leeward side of the Western rainy days (Table 3). The stations which are located close to
Ghats, whereas in the southern transect, the maximum rain- the mountainous region have recorded significant decreasing
fall has been observed on the mountain region. In all, these trends at 0.01% confidence level (Bandiur, Suvarnavathy,
plots (Fig. 4a, b) indicate the influence of Western Ghats on Periyapatna, and Somwarpet). However, stations in central
the spatio-temporal distribution of rainfall pattern, as it cre- region are not experiencing any significant changes. This
ates an orographic obstacle to the monsoon clouds. As it is indicates that stations which are close to mountain fringe
seen in the Fig. 3c, a clustering of few stations falling in the have distinct trends, whereas, stations away from the fringe
plains shows a considerable deviation both in mean rainfall do not show any trend and are not significant (− 0.01 mm/
and rainy days and forms another distinct group. This group year to + 0.08 mm/year).
is characterized by lower mean rainfall and lower number A significantly increasing trend dominated the northern
of rainy days, as compared to other two groups. This could region (Table 5). The stations which are close to the summit
be due to the following reasons: (i) these stations fall on have significant increasing trend (Fig. 3e). There are few
the rain shadow area of the Western Ghats; and (ii) they are stations which have recorded significant increasing rainfall
influenced by weaker south–west and stronger north–east amount with significant decrease in number of rainy days,
monsoon. As Gunnel (1997) observed, this region is pro- as noticed in southern and coastal regions of the study area.
tected from direct monsoon influenced by the elevated pla- On the contrary, there are number of stations with significant
teau rim, and reflects the pattern of convective precipitation, decreasing annual rainfall totals and significant increase in

13
1102 B. Venkatesh et al.

Table 4  Sen’s slope and Z (MK test) values for raingauge stations in the Central Region (Z values of MK test are given in parentheses)
Name Trend analysis of annual rainfall and rainy days (Sen’s slope values)
Annual rainfall Annual rainy Rainfall Rainy days Rainfall Rainy days Rain- Rainy
days 10–20 mm 10–20 mm 20–50 mm 20–50 mm fall > 50 mm days > 50 mm

70 28.31 (0.92) 0.21 (0.45) 31.02 (0.92) 0.17 (0.45) 26.30 (0.97) 0.14 (0.47) 26.29 (0.97) 0.25 (0.92)
71 − 9.14 (− 0.17 (0.86) − 11.15 (− − 0.21 (− − 12.56 (− − 0.25 (− − 9.16 (− − 0.16 (− 1.14)
0.50) 0.50) 1.18) 0.41) 1.35) 0.57)
72 − 7.60 (− 0.39* (3.07) − 8.67 (− 0.02 (0.32) − 11.90* (− − 0.14 (− − 10.87* (− − 0.10 (− 1.94)
1.52) 1.76) 2.08) 1.51) 2.31)
73 − 8.72 (− 0.355 + (1.92) − 9.96 (− − 0.11 (− − 9.53 (− − 0.19 (− − 12.16 (0.60) − 0.12 (− 0.96)
0.57) 0.79) 0.50) 0.66) 0.67)
74 − 16.42 (− − 0.23 (− − 15.69 (− − 0.06 (− − 16.34 (− − 0.05 (− -18.59* (− − 0.16 (− 1.90)
1.47) 1.42) 1.49) 0.37) 1.45) 0.40) 2.08)
75 1.11 (0.64) 0.11 (1.35) 0.24 (0.23) 0.02 (0.48) 0.06 (0.03) 0.00 (0.25) 0.01 (0.30) 0.00 (0.42)
76 0.90 (0.53) 0.08 + (1.68) 1.02 (0.63) 0.03 (0.97) 0.80 (0.46) 0.00 (0.29) 0.84 (0.57) 0.00 (0.41)
77 − 11.48 (− 0.21 (− 0.76) − 11.68 (− − 0.34 (− − 11.40 (− − 0.27 (− − 6.35 (− − 0.62 (− 0.86)
1.10) 1.16) 1.21) 1.27) 1.36) 0.88)
78 − 0.66 (− − 0.01 (− − 0.64 (− − 0.03 (− − 0.08 (− 0.00 (0.15) − 0.29 (− 0.00 (− 0.59)
0.48) 0.41) 0.48) 0.92) 0.05) 0.35)
80 − 10.79* (− − 0.88 (− − 11.90* (− − 0.33 (− − 12.91* (− − 0.36 (− − 1.84 (− 0.00 (− 1.08)
2.18) 1.75) 2.07) 1.46) 2.21) 2.38) 1.19)
13 − 8.15 − 0.04 (− − 7.84 (− − 0.12 (− − 7.70 (− − 0.11 (3.65) − 5.33 (− − 0.05 (− 2.79)
(− 3.45) 0.76) 3.30) 2.90) 3.26) 2.80)
14 0.00 (0.00) 0.12* (2.46) − 0.54 (− 0.00 (− 0.28) − 0.71 (− 0.00 (− 0.59) 0.00 (− 0.73) 0.00 (0.00)
0.60) 0.48)
16 0.57 (0.91) 0.08* (1.96) 0.42 (0.69) 0.00 (0.71) 0.30 (0.60) 0.00 (0.25) 0.01 (0.80) 0.00 (0.00)
17 − 0.07 (-.04) 0.10* (2.11) 0.04 (0.05) 0.01 (0.46) 0.55 (0.29) 0.04 + (1.60) − 0.43 (− 0.00 (0.55)
0.25)
18 − 2.24 (− 0.11* (2.40) − 2.30 (− − 0.03 (− − 2.33 (− − 0.02 (− − 1.53 (− − 0.01 (− 0.93)
1.42) 1.62) 1.22) 1.50) 1.13) 1.11)
19 − 4.16 (− 0.03 (0.78) − 4.02 (− − 0.07 (− − 3.42 (− − 0.04 (− − 2.48 (− − 0.02 (2.61)
3.35) 3.12) 2.37) 2.74) 1.69) 2.95)
24 2.60 (1.20) 0.20** (4.05) 2.62 (1.10) 0.05 (1.41) 2.29 (1.09) 0.02 (0.86) 3.42 (1.52) 0.03 + (1.80)
21 2.42 (2.03) 0.10 (1.18) 1.68 + (1.91) 0.00 (0.35) 2.14* (2.31) 0.03 (1.40) 1.66** (3.24) 0.02** (2.91)
22 − 1.10 (− − 0.11 (− − 0.71 (− 0.00 (− 0.29) − 0.70 (− 0.00 (− 0.33) − 0.34 (− 0.00 (0.00)
1.59) 2.06) 1.25) 1.24) 1.74)
23 − 4.15 (− 0.03 (0.75) − 4.02 (− − 0.07 (− − 3.42 (− − 0.04 (− − 2.48 (− − 0.02 (− 2.51)
3.35) 3.12) 2.37) 2.74) 1.69) 2.95)
25 2.52 (0.70) − 0.03 (− 2.31 (0.90) 0.05 (0.96) 3.10 (0.91) 0.06 (1.22) 2.56 (0.81) 0.03 (1.19)
0.55)
37 − 5.35 (− − 0.16 (− − 3.16 (− 0.11 (0.62) − 5.14 (− 0.16 (1.07) − 8.96 (− − 0.05 (− 1.05)
0.84) 0.73) 0.63) 0.01) 1.74)
38 − 1.96* (− 0.21** (− − 2.36** (− − 0.04 (0.62) − 2.13** (− − 0.03 (1.07) − 1.16** (− 0.00 ( − 1.05)
3.23) 0.73) 0.63) 0.61) 1.74)
39 0.70 (1.18) 0.03 (0.85) 0.98 (1.60) 0.01 (0.65) 1.25* (2.06) 0.03* (2.55) 0.10 (1.15) 0.00 (0.00)
40 − 1.14 (− − 0.11* (− − 0.81 (− − 0.05 (1.94) − 0.37 (− 0.00 (− 0.85) − 0.01 (− 0.00 (0.00)
1.24) 2.05) 1.10) 0.55) 0.70)
41 − 0.94 (− 0.04 (0.88) − 0.79 (− − 0.01 (− − 0.55 (− 0.00 (− 0.34) − 0.02 (− 0.00 (0.00)
1.31) 0.93) 0.46) 0.76) 0.31)
42 0.69 (1.08) 0.16* (2.10) 0.29 (0.40) 0.00 (0.18) 0.22 (0.33) 0.00 (0.24) 0.13 (1.37) 0.00 (0.00)
43 0.48 (0.61) − 0.03 (− 0.59 (0.79) 0.00 (0.24) 0.73 (1.10) 0.00 (0.59) 0.00 (1.03) 0.00 (0.0)
0.87)
44 − 0.56 (− − 0.07 (− − 0.44 (− − 0.03 (− − 0.15 (− 0.00 (− 0.22) 0.45 (0.99) 0.00 (0.0)
0.35) 0.85) 0.32) 0.93) 0.11)
45 2.38 (1.35) 0.50** (3.28) 1.13 (1.03) 0.05 (0.74) 0.00 (0.00) 0.00 (0.00) 0.63 (1.21) 0.00 (0.0)
46 0.69 (0.83) 0.03 (0.77) 0.69 (0.86) 0.00 (0.33) 0.74 (1.02) 0.00 (0.63) 0.00 (0.72) 0.00 (0.00)

13
Spatio‑temporal analysis of rainfall pattern in the Western Ghats region of India 1103

Table 4  (continued)
Name Trend analysis of annual rainfall and rainy days (Sen’s slope values)
Annual rainfall Annual rainy Rainfall Rainy days Rainfall Rainy days Rain- Rainy
days 10–20 mm 10–20 mm 20–50 mm 20–50 mm fall > 50 mm days > 50 mm

47 − 3.51* (− − 0.04 (− − 2.66 (− − 0.03 (− − 2.86 (− − 0.02 (− − 3.22** (− − 0.02* (− 2.08)

2.16) 1.05) 0.49) 0.07) 0.27) 0.40) 2.58)
48 − 0.15 (− 0.10* (2.35) − 0.28 (− 0.00 (− 0.07) − 0.16 (− 0.00 (− 0.40) 0.00 (− 0.51) 0.00 (0.0)
0.27) 0.49) 0.27)
49 − 0.11 (− 0.15 + (1.91) − 0.43 (− 0.04 (1.38) − 0.06 (− 0.04 (1.11) − 2.20 (− − 0.02 (− 1.41)
0.06) 0.81) 0.06) 1.46)
50 − 12.77* (− − 0.03 (− − 12.29* (− − 0.07 (− -13.51* (− − 0.08 (− − 11.84* (− − 0.07* (− 2.24)
2.49) 0.41) 2.23) 1.12) 2.24) 1.21) 2.44)

** α = 0.01; * α = 0.05; + α = 0.1

number of rainy days. This suggests that a confined smaller Further, it has been observed that, the number of rainy
region receives high intensity rainfall whereas at other parts days corresponding to these rainfall classes has approxi-
of the study region are receiving low intensity rainfall. These mately similar trends across the entire study area. However,
two phenomena have altogether different hydrological conse- there are few stations, where the 10–20 mm and > 50 mm
quences and have not been detected for the other region. The rainfall classes (Fig. 3f, h) have registered significant
higher intensity may cause flood and landslides, whereas increasing trends in the central and coastal regions, whereas
the lower intensity rainfall may not be yielding the runoff. the southern region recorded significant decreasing trends
The farmers in the coastal area are completely dependent for the same class (Napoklu). The northern part of the study
on the monsoon rainfall for their farming activities. These area recorded no trend in the number of rainy days for any of
changes would impact greatly on these activities and the these rainfall classes. Furthermore, it needs to be highlighted
water availability. here, that these stations are located on the leeward side of the
Western Ghats. The stations on windward side have mixed
4.5 Analysis of variation in rainfall class trends, viz., many stations have recorded significant increas-
ing trends in number of rainy days for the rainfall classes
Many studies elsewhere (Sharma and Singh 2014; Ibra- such as 10 mm and 10—20 mm (Fig. 3f, g), with decreas-
him et al. 2014; Iqbal et al. 2014; Ahmed et al. 2019; ing trend in > 50 mm (Fig. 3h) class (Mulky, Kota, Ankola
Sabrina et al. 2019) studied the spatio-temporal variation and Bandal). These changes may be due to the latitudinal
of various percentile rainfall class. However, Putty et al. influence or may be location specific. Roberto et al. (2012)
(2000) and Gunnel (1997) pointed out that the annual observed that the rainfall intensities and mean annual rainfall
rainfall total in the Western Ghats region are dominated are influenced by the latitudinal variation. This observation
by the higher daily rainfall amount. In line with the may hold good for the Western Ghats also, as it is spread
above, the present study classify the daily rainfall into across 11° 20′ to 16° 00′ N latitude. Other than these factors,
four classes namely, 0.2–10 mm, 10–20 mm, 20–50 mm it is also possible that the cascading nature and width of the
and > 50 mm aimed to represent the moderate, medium, Western Ghats in Karnataka may be the responsible for such
and extreme rainfall events in line with the classifica- spatial variation (Tawade 2013).
tion of Bhatla et al. (2019). Furthermore, the number of
days of rainfall events falling in these three classes has 4.6 Change point analysis in rainfall series
also been extracted. The subsequent trend analysis and
its results are given in Tables 3, 4, 5 and 6 and Fig. 3b, d The change point probability and detection of homogeneity
for 0.2–10 mm, 10–20 mm, and > 50 mm rainfall classes, of the data series using Pettitt’s test and SNHT test were
respectively. carried out. From the analysis, it is found that at annual time
The trend analysis of rainfall classes such as 10–20 mm, scale, there is a significant increase in rainfall in few stations
20–50 mm, and > 50 mm show decreasing trend across the under the northern region. Whereas, in the central region,
study region. However, there are few stations which have only few stations have shown decrease in the rainfall. There
registered increasing trend (Fig. 3b, c). This implies that the are no significant changes in the rainfall during the mon-
higher intensity rainfall (> 50 mm/day) is decreasing over soon in the study area except in the northern region. These
most of these regions and such events have now become observed changes have happened during the decade of 1970.
occasional (Fig. 3d). However, by and large the rainfall has been homogeneous

13
1104 B. Venkatesh et al.

Table 5  Sen’s slope and Z (MK test) values for raingauge stations in the northern region (Z values of MK test are given in parentheses)
Name Trend analysis of annual rainfall and rainy days (Sen’s slope values)
Annual rainfall Annual rainy Rainfall Rainy days Rainfall Rainy days Rain- Rainy
days 10–20 mm 10–20 mm 20–50 mm 20–50 mm fall > 50 mm days > 50 mm

91 4.12 (0.36) 0.07 (0.78) 2.66 0.12 3.26 0.10 (0.95) 0.53 (0.07) 0.00 (0.23)
(0.19) (1.15) (0.34)
92 16.93 (1.13) 0.64** (3.10) 15.49 (1.02) 0.28* (1.98) 16.45 (1.05) 0.30 + (1.80) 4.50 0.11
(0.26) (0.66)
93 37.34 (1.56) 0.33 (1.54) 37.37 (1.60) 0.60* (2.56) 34.43 (1.22) 0.40* (2.03) 31.24 (1.07) 0.17
(0.84)
94 − 7.05 0.75* (2.29) − 9.09 (− 1.36) 0.05 − 8.08 − 0.09 (0.83) − 3.55 (1.01) − 0.05
(− 0.92) (0.19) (− 1.41) (− 1.12)
96 − 7.09 0.60* (2.37) − 9.40 0.06 (0.54) − 14.91 − 0.11 − 15.02 (− − 0.17
(− 0.86) (− 1.13) (− 1.29) (− 0.82) 1.35) (− 1.60)
97 − 12.85 0.03 (0.31) − 11.70 0.30 + (1.19) − 13.01 0.15 (1.60) − 15.80 (1.28) 0.00 (0.11)
(− 0.88) (− 0.82) (− 1.16)
98 17.43 (0.70) 0.84** (2.62) 15.66 (0.51) 0.22 (0.68) 15.07 (0.51) 0.16 (0.95) 13.93 (0.54) 0.29 (1,35)
99 − 3.80 0.33 + (1.93) − 3.76 0.00 (0.02) − 2.55 (− 0.00 − 1.81 0.00
(− 0.82) (− 0.71) 0.43) (− 0.04) (− 0.61) (− 0.80)
101 8.24 (1.13) 0.36* (2.11) 7.27 (0.97) 0.50** (3.10) 2.77 (0.45) 0.17 (1.52) 1.87 (0.18) 0.00 (0.14)
104 − 4.76 0.25 (0.84) − 3.64 (0.36) − 0.05 − 1.79 − 0.04 0.48 (0.25) 0.00 (0.27)
(− 0.60) (− 0.20) (− 0.17) (− 0.45)
106 0.21 (0.24) − 0.06 0.54 (0.53) 0.00 (0.38) 0.27 (0.36) 0.00 (0.18) 0.57 (1.30) 0.00 (0.60)
(− 1.29)
107 − 0.19 (− 0.04 (0.08) 0.17 (0.13) 0.00 (0.36) 0.41 (0.20) 0.01 (0.71) − 0.24 (− 0.00 (− 0.58)
0.09) 0.16)
108 − 0.95 (− − 0.16** (− − 0.84 (− 0.51) − 0.03 (− − 0.61 (− 0.00 (− 0.48) − 0.03 (− 0.00 (− 0.64)
0.56) 2.93) 1.41) 0.35) 0.02)
109 3.99 (0.20) 1.56** (3.86) − 2.11 (− 0.11) 0.00 (0.12) − 3.71 (− − 0.01 (− − 2.81 (− 0.00 (− 0.34)
0.23) 0.14) 0.26)
111 − 5.28 (− 0.62* (2.50) − 8.60 (− 0.62) 0.27 (1.35) -14.14 (− 1.02) 0.00 (− 0.01) − 15.67 (− − 0.09 (− 0.73)
0.31) 0.99)
113 3.77 (0.43) 0.16 (1.15) 5.15 (0.60) 0.29 (1.42) 2.20 (0.26) 0.00 (0.11) − 0.56 (− 0.00 (− 0.56)
0.17)
114 − 2.17 (− 0.36 + (1.67) − 1.46 (− 0.14) 0.05 (0.43) − 1.65 (− 0.00 (0.10) − 3.35 (− 0.00 (− 0.30)
0.40) 0.17) 0.51)
115 − 1.94 (− − 0.08 (− − 1.59 (− 0.70) − 0.03 (− − 2.03 (− − 0.03 (− − 1.17 (− − 0.01 (− 0.78)
0.84) 1.64) 0.89) 0.82) 1.05) 0.55)
117 66.64* (2.55) 0.22 66.56* (2.46) 0.90** (3.14) 64.36* (2.43) 0.80* (2.57) 46.84* (2.26) 0.51* (2.12)
(1.53)
118 82.42** (2.260 0.52* (2.06) 85.78** (2.70) 0.91** (2.44) 85.46** (2.70) 1.00** (2.87) 65.16* (2.36) 0.00
(2.12)
119 96.95** (3.02) 0.76** (2.62) 99.47** (3.05) 1.30** (3.95) 99.51** (3.0) 1.33** (3.59) 87.13* (2.50) 0.78* (2.41)
120 0.32 (0.07) − 0.66* (− 2.88 − 0.06 4.78 0.03 6.48 (1.33) 0.06
2.47) (0.44) (− 0.41) (0.54) (0.22) (1.46)
121 4.08* (2.50) 0.01 (0.21) 4.17* (2.55) 0.06 4.35* (2.56) 0.04 + (1.70) 3.73** (2.50) 0.03* (2.41)
(1.60)
122 0.3 − 0.05 0.2 0.00 0.1 0.00 0.6 (1.11) 0.00
(0.34) (− 1.24) (0.14) (0.09) (0.12) (0.22) (0.42)

** α = 0.01; * α = 0.05; + α = 0.1

in the study area. It is reported elsewhere that, the major Kafatos 2004), or weakening global monsoon circulation
reasons for these types of changes could be in general, (Pant 2003). The stations which have recorded the signifi-
human activities (Adeyeri et al. 2017), reduction in forest cant change in the northern region are close to the mountain
cover (Nair et al. 2003), anthropogenic activities (Sarkar and peak and are located on leeward side of the mountain. The

13
Spatio‑temporal analysis of rainfall pattern in the Western Ghats region of India 1105

Table 6  Sen’s slope and Z (MK test) values for raingauge stations in the Coastal Region (Z values of MK test are given in parentheses)
Name Trend analysis of annual rainfall and rainy days (Sen’s slope values)
Annual rainfall Annual rainy Rainfall Rainy days Rainfall Rainy days Rain- Rainy
days 10–20 mm 10–20 mm 20–50 mm 20–50 mm fall > 50 mm days > 50 mm

81 − 9.67 (− − 0.11 (− − 8.53 (− 0.64) 0.09 (− 0.46) − 12.80 − 0.10 (− − 10.95 (− − 0.13 (− 0.13)
0.82) 0.62) (− 1.01) 0.56) 0.75)
82 4.90 (0.08) 0.45* (2.47) 3.72 (0.19) 0.05 (0.40) 0.80 (0.02) 0.00 (0.19) − 2.59 (− − 0.09 (− 0.78)
0.11)
83 − 12.67 (− 0.00 (0.17) − 12.22 (− 0.07 (0.31) − 17.19 (− − 0.20 (− − 18.62 (− − 0.27 (0.03)
0.67) 0.61) 0.88) 0.93) 0.38)
84 11.29 (1.11) 0.47* (2.05) 7.93 (0.90) 0.12 (0.76) 8.03 (0.91) 0.10 (0.64) 9.29 (1.31) 0.14 (1.40)
85 71.50 (1.35) − 0.35 (− 71.51 (1.52) 0.45 (1.22) 69.07 (1.58) 0.50 (0.68) 80.20 (1.64) 0.78 (1.61)
0.73)
86 − 2.98 (− − 0.18 (− − 3.64 (− 0.40) − 0.40 (− − 2.06 (− − 0.28 (− 13.99 (1.13) 0.08 (0.61)
0.20) 0.79) 1.95) 0.26) 1.40)
87 3.41 (1.02) − 0.30** (− 4.11 (1.17) 0.06 (1.07) 3.34 (1.13) 0.04 (1.01) 1.98 (0.65) 0.00 (0.35)
4.37)
89 2.75 (1.30) − 0.09* (− 3.22 (1.51) 0.06 (1.94) 3.25 (1.47) 0.04 (1.35) 2.39 (1.14) 0.00 (0.56)
2.29)
90 0.43 (0.19) − 0.11** (− 0.75 (0.28) 0.00 (− 0.21) 1.05 (0.33) 0.00 (− 0.16) 2.00 (0.88) 0.03 (1.42)
2.86)
26 − 26.34 + (− 0.00 (0.02) − 27.54* (− − 0.10 (− − 29.13* (− − 0.25 (− − 32.24* (− − 0.29 +
1.90) 1.96) 0.73) 2.09) 1.31) 2.06) (− 1.69)
27 4.07 (0.44) 0.00 (0.16) 4.52 (0.48) − 0.22 (− 5.04 (0.61) − 0.10 (− 10.36 (1.00) 0.04 (0.88)
1.55) 0.83)
28 0.42 (0.20) − 0.13** (− 0.55 (0.28) − 0.04 (− 1.09 (0.55) 0.00 (0.33) 0.53 (0.29) 0.00 (− 0.35)
3.14) 1.33)
29 0.16 (0.05) − 0.09* (− 0.40 (0.21) − 0.02 (− 0.87 (0.35) 0.00 (− 0.28) 0.21 (0.10) − 0.01 (− 0.76)
2.21) 0.83)
30 − 15.06 (− − 0.32 (− − 13.82 (− − 0.18 (− − 14.47 (− − 0.22 (− − 11.47 (− − 0.06 (-091)
1.33) 1.63) 1.25) 1.12) 1.01) 1.22) 1.13)
31 − 14.44 (− 0.18 (1.11) − 17.44 (− − 0.16 (− − 16.98 (− − 0.17 (− -18.81 (− 1.27) − 0.11 (0.80)
1.27) 1.32) 0.72) 1.32) 0.73)
32 − 3.67 (− 0.12 (0.37) − 4.85 (− 0.23) − 0.07 (− − 5.11 (− 0.00 (− 0.14) − 8.11 (− − 0.12 (− 0.89)
0.23) 0.36) 0.36) 0.39)
33 4.79 + (1.03) − 0.26** (− 5.19* (1.28) 0.00 (1.11) 5.91* (1.28) 0.03 (0.98) 5.38* (0.99) 0.03 + (1.00)
1.70)
35 6.97 (0.36) 0.21 (0.02) 9.46 (0.47) − 0.33 (− 10.30 (0.74) − 0.07 (− 12.59 (0.89) 0.06 (0.76)
1.32) 0.37)
36 − 2.63 (− − 0.17** (− − 2.27 (− 1.11) − 0.03 (− − 2.39 (− − 0.03 (− − 1.05 (− 0.00 (0.00)
1.22) 3.32) 0.84) 1.07) 1.31) 0.38)
100 6.03 (0.28) 0.33 (1.37) 4.37 (0.24) 0.00 (− 0.04) 11.53 (0.32) 0.05 (0.28) 9.28 (0.39) 0.03 (0.34)
103 3.29 (1.36) 0.00 (− 0.14) 3.56 (1.55) 0.04 (1.34) 3.84 (1.59) 0.04 (1.44) 4.02 + (1.87) 0.04 (1.59)
105 − 1.57 (− 0.09 (2.15) − 1.83 (− 0.68) 0.01 (0.40) − 1.96 (− 0.00 (− 0.22) − 2.51 (1.00) − 0.02 (− 1.06)
0.61) 0.77)
95 − 0.67 (− − 0.12** (− − 2.37 (− 0.27) − 0.50* (− 1.22 (0.03) − 0.25 (− 15.65 (1.39) 0.04 (0.36)
0.03) 1.07) 2.20) 1.59)
110 12.88 (1.28) 0.37 (1.56) 11.90 (1.22) 0.04 (0.41) 12.70 (1.25) 0.08 (0.44) 11.96 (1.48) 0.10 (1.20)
112 − 5.07 (− − 0.02 (− − 5.27 (− 0.57) 0.00 (0.01) − 8.18 (− − 0.11 (− − 8.18 (0.91) − 0.13 (− 1.45)
0.45) 0.21) 0.65) 0.61)
102 5.08* (2.47) 0.05 (1.21) 4.96* (2.34) 0.08* (2.32) 4.78* (2.24) 0.06* (1.97) 4.21* (2.15) 0.03 (1.58)

**α = 0.01; *α = 0.05; + α = 0.1

13
1106 B. Venkatesh et al.

Fig. 4  Plot showing a variation

of mean rainfall and rainy days
with elevation in the northern
and southern transects; b num-
ber of rainy days against mean
annual rainfall

location of these stations may have a significant bearing on significant level at annual scale. However, the second part of the
these changes in rainfall pattern. As reported by Pant (2003) data shows more area under the higher PCI values covering the
and Varikoden et al. (2019), the weak monsoon winds have northern, central, and parts of coastal area. Similar is the case
significantly influence the rainfall pattern in the Western during the monsoon (wet) season. Also, it observed that, the
Ghats region. This might be one of the possible reasons for southern region has recorded higher values in the wet season
these significant changes in rainfall at these locations. compared to the first period (Fig. 5b).
It is seen from the Fig. 5a that, the PCI values are more
4.7 Estimation of precipitation concentration index than 20 in the northern region along with the adjoining coastal
(PCI) and central regions. However, PCI values in the southern
region are lower. These values represent a clear distinct rain-
The precipitation concentration index (PCI) calculated on an fall phenomenon as reported by the Gunnel (1997), wherein
annual scale varies across the area under study from values he observed that, the area covering the Mysore, Kodagu, and
lower than 15 in the southern region to higher than 32 in the adjoining areas (within the southern region of this study)
northern region. In general, lower values are observed in the are influenced by the bi-modal rainfall, receiving the rain-
southern region, while higher values are found in the northern fall during both south–west and north–east monsoons, hence
and coastal regions (Fig. 5a). To evaluate the temporal differ- the lower values of PCI. The monsoon PCI values provide
ence in the PCI, at annual and seasonal scale, the data have altogether a new picture, wherein the second part of the data
been divided into two periods, i.e., 1951–1981 and 1983–2012. (1983–2014) has higher PCI values in comparison to the first
This analysis revealed that, the higher values are more concen- part of the data (1951–1981). This indicates that there is more
trated in the northern and central regions with very few sta- concentration of rainfall in the second part of the study period
tions showing the significant change in the PCI at P < 0.01% and become more inhomogeneous.

13
Spatio‑temporal analysis of rainfall pattern in the Western Ghats region of India 1107

Fig. 5  Mean values of precipitation concentration index for a annual scale and b monsoon season

The aforementioned results indicate that, over all the 5 Summary and conclusions
rainfall in the Western Ghats region on declining trend at
annual time scale with increasing number of rainy days. Western Ghats, the primary catchment of many peninsular
On the other hand, the region is experiencing several high rivers, and the people of peninsular states are dependent on
intensity rainfall (> 50 mm/day) with decreasing rainy days, these rivers for water. Any changes in the rainfall pattern in
especially in the southern region. These rainfall events may this region would impact greatly on the economy of these
results in high inflow into the reservoir in the region and states. Therefore, the present study was initiated with an aim
subsequent flooding in the downstream of reservoir. How- to analyse rainfall, for detecting spatio-temporal trends their
ever, the coastal region shows a completely different trend, significance over the Western Ghats of Karnataka, India,
wherein, the > 50 mm/day events are increasing with the using rainfall data of more than 100 years.
number of rainy days resulting in the flooding of the region. The variation of mean annual rainfall over the study area
In the recent times, these two regions have been experi- was found to be 4000 mm in the coastal areas to as high
encing the floods on annual basis with adjoining Kerala as > 6000 mm at the mountain crust. The variation of mean
state. These changes and increase in the rainfall trend may annual rainfall with elevation in the northern transect is more
be attributed to: (i) irregular rainfall features and complex than 6000 mm, with as much as 110 rainy days in the southern
with respect to time and space (Venkatesh and Jose 2007); transect, and close to mountain crest on the leeward side on
(ii) negative correlation with Arabian Sea SST (Revadekar the northern transect and drastically reduces on the leeward
et al. 2018), and tropospheric temperature of the north India side of the Western Ghats in the both transects. This is due
(Preethi et al. 2016; Varikoden et al. 2019); (iii) availability to the orographic effect which is responsible for higher rain-
of moisture, wind velocity, wind direction, and orography of fall on the windward side and sharp decline in rainfall on
the region (Patwardhan and Asnani 2000); (iv) change in the the leeward side. MK test based trend analysis for the annual
low level westerlies will lead to change in the rainfall pattern rainfall shows that decreasing trends are dominating in the
over the Western Ghats region (Joseph and Simon 2005); central region along with the stations bordering the coastal
and (v) and wind direction and speed (Rajendran 2012). zone, whereas the coastal region is dominated by the increas-
ing trend in annual rainfall. It has also been observed that the
number of rainy days has been decreasing in most parts of

13
1108 B. Venkatesh et al.

the coastal region whereas the stations located away from the pertinent to understand the synoptic parameters and their rela-
fringe in the southern region do not show any trend. In the tionship with very high rainfall amount and their concentration
northern region, there are many stations which have signifi- in few location in the Western Ghats. Further, it also invites to
cant increasing and decreasing trends in rainfall, but stations re-look at the policies in which the water resources projects are
located close to the mountain depict significant decrease in being operated presently, to accommodate the large amount of
the number in rainy days. The rainfall in 10–20 mm rainfall water in the dams to reduce the flooding in the downstream.
class showed a decreasing trend in the central region (plains)
as compared to the coastal region. The rainfall in the rainfall
class > 50 mm depicted decreasing trends in the southern and References
northern regions. However, mixed trends in number of rainy
days have been observed for all three rainfall classes for the Adeyeri OE, Lamptey BL, Lawin AE, Sanda IS (2017) Spatio-temporal
precipitation trend and homogeneity analysis in Komadugu-Yobe
stations located on the windward side of the Western Ghats. Basin, Lake Chad Region. J Climatol Weather Forecasting 5:214.
Moreover, it is interesting to note that the number of low rain- https://​doi.​org/​10.​4172/​2332-​2594.​10002​14
fall events is increasing with decrease in the intense rainfall Ahmed A, Deb D, Mondal S (2019) Assessment of rainfall variability and
events. These observed changes could be due to the latitudinal its impact on groundnut yield in Bundelkhand region of India.Curr
Sci 117(5):794–803. https://​doi.​org/​10.​18520/​cs/​v117/​i5/​794-​803
influence or it may be location specific, as the rainfall intensi- Bharati V, Singh C, Ettema J et al (2016) Spatiotemporal characteristics of
ties and mean annual rainfall are influenced by the latitudinal extreme rainfall events over the Northwest Himalaya using satellite
variation and other weather parameters. Further homogeneity data. Int J Climatol 36:3949–3962. https://​doi.​org/​10.​1002/​joc.​4605
and the PCI results indicate that few stations have significant Bhatla R, Shruti Verma, Pandey R, Tripathi A (2019) Evolution of
extreme rainfall events over Indo-Gangetic plain in changing cli-
changes in the rainfall amount in the decade of 1970. The mate during 1901–2010. Earth Syst Sci 128:120. https://​doi.​org/​
PCI result shows that the southern region is influenced by 10.​1007/​s12040-​019-​1162-1
the bi-modal rainfall with significant contribution during the Bhowmik RSK, Durai VR (2008) Multi-model ensemble forecasting
N–E monsoon. of rainfall over Indian monsoon region. Atmosfera 21(3):225–239
Chandrashekar VD, Shetty A, Singh BB, Sharma S (2017) Spatio-
From the results and its discussion, the following conclu- temporal precipitation variability over Western Ghats and Coastal
sions have been drawn: region of Karnataka, envisaged using high resolution observed
gridded data. Model Earth Syst Environ 3:1611–1625
1. The raingauge stations in the coastal region show an Connor GJ, Bonell M (1998) Air mass and dynamic parameters affect-
ing trade wind precipitation on the northeast Queendsland tropical
increase in annual rainfall at many places. However, sig- coast. Int J Climatol 18:1357–1372
nificant increasing trends are noticed in the northern region De Lima MIP, Marques ACP, De Lima JLMP et al (2007) Precipitation
and significant decreasing trend in the southern region. trends in Mainland Portugal in the period 1941–2000. Water in celtic
2. It is observed that, the rainfall event > 50 mm/day is countries: quantity, quality and climate variability. In: Proceedings of
the fourth inter celtic colloquium on hydrology and management of
decreasing across the Western Ghats, except at a few water resources, Guimarães, July 2005. IAHS Publ. 310
places in the northern region. Gadgil M (1987) Depleting renewable resources: a case study from
3. The rainy days in the rainfall classes of 10–20 mm and Karnataka Western Ghats. Indian J Agric Econ 42:376–387
20–50 mm are increasing markedly in the study area. Gopalkrishnan R, Jayaraman M, Bala G et al (2011) Climate change and
Indian forests. Curr Sci 101(3):348–355
Goswami BN, Venugopal V, Sengupta D et al (2006) Increasing trend of
In summary, it can be concluded that the Western Ghats extreme rain events over India in a warming environment. Science
area of Karnataka is receiving more rainfall amount from the 314(5804):1442–1445. https://​doi.​org/​10.​1126/​scien​ce.​11320​27
rainfall events falling between > 0.2 mm/day and < 50 mm/ Gunnel Y (1997) Relief and climate in South Asia: the influence of the
Western Ghats on the current climate pattern of Peninsular India.
day which occurred more frequently as compared to the rain- Int J Climatol 17:1169–1182
fall events of > 50 mm/day. The lower rainfall contribution Ibrahim B, Karambiri H, Polcher J et al (2014) Changes in rainfall
from > 50 mm/day along with their fewer occurrences are a regime over Burkina Faso under the climate change conditions
matter of concern for the water resources managers and deci- simulated by 5 regional climate models. Clim Dyn 42:1363–138.
https://​doi.​org/​10.​1007/​s00382-​013-​1837-2
sion makers, as the higher rainfall events are mostly respon- Iqbal M, Wen J, Wang S et al (2018) Variations of precipitation char-
sible for the production of runoff from the catchments of the acteristics during the period 1960–2014 in the source region of
rivers originating in the Western Ghats. the Yellow River, China. J Arid Land 10(3):388–401. https://​doi.​
The present study details about the spatio-temporal variation org/​10.​1007/​s40333-​018-​0008-z.
Jackson IJ, Weinand H (1994) Towards a classification of tropical rain-
in the rainfall of Western Ghats of Karnataka. However, the fall stations. Int J Climatol 14:263–286
annual maximum rainfall has not been included in the present Joseph PV, Simon A (2005) Weakening trend of the southwest mon-
study. In recent times, the region is witnessing high to very soon current through peninsular India from 1950 to present. Curr
high one day rainfall causing flooding and landslides result- Sci 89:687–694
Kendall MG (1975) Rank correlation methods, 4th edn. Charles Grif-
ing in heavy loss of life and property. Such incidents in this fin, London, p 6
region were not reported earlier. In view of such incidents, it is

13
Spatio‑temporal analysis of rainfall pattern in the Western Ghats region of India 1109

Kharol SK, Kaskaoutis DG, Sharma AR et al (2013) Long-term (1951– Revadekar JV, Varikoden H, Murumkar PK, Ahmed SA (2018) Lati-
2007) rainfall trends around six Indian cities: current state, mete- tudinal variation in summer monsoon rainfall over Western Ghat
orological, and urban dynamics. Adv Meteorol. https://​doi.​org/​ of India and its association with global sea surface temperatures.
10.​1155/​2013/​572954 Sci Total Environ 613–614:88–97
Krishnakumar KN, Prasada Rao GSLHV, Gopakumar CS (2009) Rain- Roberto P, Valdés R, García-Chevesich P et al (2012) Latitudinal analy-
fall trends in twentieth century over Kerala, India. Atmos Environ sis of rainfall intensity and mean annual precipitation in Chile.
43:1940–1944 Chilean J Agric Res 72(2):252–261
Krishnamurthy CKB, Lall U, Kwon HH (2009) Changing frequency Sabrina T, Mohamed M, Gil M (2019)Seasonal rainfall variability in
and intensity of rainfall extremes over India from 1951 to 2003. J the southern Mediterranean border: observations, regional model
Clim 22. https://​doi.​org/​10.​1175/​2009J​CLI28​96.1 simulations and future climate projections, Atmósfera 32(1):39–
Kumar V, Jain SK (2010) Trends in seasonal and annual rainfall and 54. https://​doi.​org/​10.​20937/​ATM.​2019.​32.​01.​04
rainy days in Kashmir Valley in the last century. Quat Int 212:64– Sarkar S, Kafatos M (2004) Interannual variability of vegetation over
69. https://​doi.​org/​10.​1016/j.​quaint.​2009.​08.​006 the Indian sub-continent and its relation to the different meteoro-
Kundu SK, Mondal TK (2019) Analysis of long-term rainfall trends logical parameters. Remote Sens Environ 90:268–280
and change point in West Bengal, India. Theor Appl Climatol Sen PK (1968) Estimates of the regression coefficient based on Kend-
138:1647–1666. https://​doi.​org/​10.​1007/​s00704-​019-​02916-7 all’s tau. J Am Stat Assoc 63:1379–1389
Lacombe G, McCartney M (2014) Uncovering consistencies in Indian Sharma D, Singh MB (2014) Trends in extreme rainfall and tempera-
rainfall trends observed over the last half century. Clim Change ture indices in the Western Thailand. Int J Climatol 34:2393–
123(2):287–299. https://​doi.​org/​10.​1007/​s10584-​013-​1036-5 2407. https://​doi.​org/​10.​1002/​joc.​3846
Longobardi A, Villani P (2009) Trend analysis of annual and seasonal Shrestha D, Singh P, Nakamura K (2012) Spatiotemporal variation of
rainfall time series in the Mediterranean area. Int J Climatol rainfall over the central Himalayan region revealed by TRMM
30:1538–1546. https://​doi.​org/​10.​1002/​joc.​2001 precipitation radar. J Geophys Res 117:D22106. https://​doi.​org/​
Lyons WF, Bonell M (1992) Daily mesoscale rainfall in the tropical 10.​1029/​2012J​D0181​40
wet/dry climate of the Townsville area, North-East Queensland Singh P, Kumar V, Thomas T et al (2008) Changes in rainfall and
during the 1988-1989 wet season: synoptic-scale airflow consid- relative humidity in different river basins in the northwest and
eration. Int J Climatol 12:655–684. https://​doi.​org/​10.​1002/​joc.​ central India. Hydrol Process 22(16):2982–2992. https://​doi.​org/​
33701​20702 10.​1002/​hyp.​6871
Madhavan V (2009) The interplay of climate and landscape evolu- Subash N, Gangwar B (2014) Statistical analysis of Indian rainfall and
tion along the Western Ghats of India. Dissertation, University rice productivity anomalies over the last decades. Int J Climatol
of llinois, Illinois 34:2378–2392. https://​doi.​org/​10.​1002/​joc.​3845
Malik A, Kumar A (2020) Spatio-temporal trend analysis of rainfall Suji Kumar S, John L, Manjusha K (2013) Sensitivity study on the
using parametric and non-parametric tests: case study in Uttara- role of Western Ghats in simulating the Asian summer monsoon
khand, India. Theor Appl Climatol 140(1):183–207. https://​doi.​ characteristics. Meteorol Atmos Phys 120(1–2):53–60
org/​10.​1007/​s00704-​019-​03080-8 Sumner GN, Bonell M (1986) Circulation and daily rainfall in the north
Malik A, Kumar A, Guhathakurta P, Kisi O (2019) Spatial-temporal Queensland wet season 1979–1982. Int J Climatol 6:531–549
trend analysis of seasonal and annual rainfall (1966–2015) using Tawade SA (2013) Investigation of orographically induced rainfall over
innovative trend analysis method with significance test. Arab J Western Ghats and its association with other monsoon parameters.
Geosci 12:328. https://​doi.​org/​10.​1007/​s12517-​019-​4454-5 Dissertation, Andhra University, Visakhapatnam
Mann HB (1945) Nonparametric tests against trend. Econometrica Taxak AK, Murumkar AR, Arya DS (2014) Long term spatial and
13(3):245–259 temporal rainfall trends and homogeneity analysis in Wainganga
Modarres R, Silva VPR (2007) Rainfall trends in arid and semi-arid basin Central India. Weather Clim Extrem 4:50–61
regions of Iran. J Arid Environ 70:344–355 Thomas T, Gunthe S, Ghosh NC (2015) Analysis of monsoon rainfall
Nair US, Lawton RO, Welch RM, Pielke RA (2003) Impact of land use variability over Narmada basin in central India: implication of
on Costa Rican tropical montane cloud forests: sensitivity of cumu- climate change. J Water Clim Change 6(3):615–627. https://​doi.​
lus cloud field characteristics to low land deforestation. J Geophys org/​10.​2166/​wcc.​2014.​041
Res 108:4206–4219 Tirkey N, Parhi PK, Lohani AK, Chandniha SK (2020) Analysis of precipi-
Nandargi S, Mulye SS (2014) Spatio-temporal rainfall variability over tation variability over Satluj Basin, Himachal Pradesh, India: 1901–
Goa, India. Int J Meteorol 39(385):99–121 2013. J Water Clim Chang. https://​doi.​org/​10.​2166/​wcc.​2020.​136
Pant GB (2003) Long-term climate variability and change over mon- Turner AG, Annamalai H (2012) Climate change and the South Asian
soon Asia. J IndGeophys Union 7(3):125–134 summer monsoon. Nat Clim Change. https://​doi.​org/​10.​1038/​
Patwardhan SK, Asnani GC (2000) Meso-scale distribution of summer Nclim​ate14​95
monsoon rainfall near the Western Ghats (India). Int J Climatol Varikoden H, Revadekar JV, Kuttippurath J, Babu CA (2019) Contrast-
20:575–581 ing trends in southwest monsoon rainfall over the Western Ghats
Preethi B, Mujumdar M, Kripalani RH et al (2017) Recent trends and region of India. Clim Dyn 52:4557–4566
tele-connections among South and East Asian summer monsoons Venkatesh B, Jose MK (2007) Identification of homogeneous rainfall
in a warming environment. Clim Dyn 48:2489–2505 regimes in parts of Western Ghats Region of Karnataka. J Earth
Putty MR, Prasad VSR, Ramaswamy R (2000) A study of the rainfall Syst Sci 116(4):321–330
intensity pattern in Western Ghat, Karnataka. In: Varadan KM (ed) Venkatesh B, Purandara BK, Bonell M, Jayakumar R (2006) Study
Proceedings of the workshop on watershed development in Western of rainfall intensity-duration frequency relationships for parts
Ghats Region of India, Centre for Water Resources Development of Western Ghats in Karnataka, India. In Krishnaswamy J, Lele
and Management, Kozhikode, Kerala India, pp 44–51 S, Jayakuamr R (eds) Hydrology and Watershed Services in the
Rajendran K (2012) Will the South Asian monsoon overturning circu- Western Ghats of India; effects of land use and land cover change.
lation stabilize any further? Clim Dyn 40:187–211 Tata-McGraw Hill, New Delhi, pp 45–64
Rakhecha PR, Soman MK (1994) Trends in the annual extreme rainfall
events of 1 to 3 days duration over India. Theor Appl Climatol Publisher’s Note Springer Nature remains neutral with regard to
48(4):227–237 jurisdictional claims in published maps and institutional affiliations.

13