MethodsX 10 (2023) 102058

Contents lists available at ScienceDirect

MethodsX
journal homepage: www.elsevier.com/locate/mex

Study the seasonal trend analysis and probability distribution

functions of rainfall for atmospheric region of Pakistan
Muhammad Yonus a, Bulbul Jan b,∗, Hamza Khan c, Faisal Nawaz b, Muhammad Ali d
a
Department of Computing, Indus University Karachi, Pakistan
b
Department of Mathematics, Dawood University of Engineering and Technology, Karachi, Pakistan
c
Department of Textile and Clothing, National Textile University Karachi Campus, Pakistan
d
Department of Artificial Intelligence and Mathematical Sciences (AIMS), Faculty of Information Technology, Sindh Madressatul Islam
UniversityAiwan-e-Tijarat Road, Karachi, Pakistan

a r t i c l e i n f o a b s t r a c t

Method name: It is known that the rainfall is one of the major aspects of the climate change and hydrological
Trend Analysis and Probability Distribution processes. Its impact on the resources of water has always been considered very essential. In the
Functions countries like Pakistan having severe weather conditions the effect of rainfall can be more cru-
Keywords: cial. Due to the geographical differences, the impact of sudden rainfall can be more disastrous
Trend analysis in Pakistan as in 2011 and 2022 only the few days’ extreme rainfall in summer monsoon dis-
Hydrological resources turbed many regions of Pakistan especially southern region of Pakistan and lost hundreds of lives
Rivers and homes. In addition, there are two methods applied for this study (1) trend analysis and (2)
Flow probability distribution functions. Our results revealed that only three cities i.e. Chitral, Skardu
Precipitation and Gilgit having increasing as well as the station Sialkot have increasing trend pattern in only
summer rainfall, and other all have shown decreasing trend overall. Furthermore, key extreme
weather events over Pakistan have changed in frequency and intensity during the past decades
due to significant increase in global warming. The methods and results will be useful to study the
historical rainfall data for projecting future rainfall variations impact on hydrological processes
due to climate change.

• The Data, methods and results of this study can be useful to government officials for protective
measures and future developments.
• Detecting trend analysis was applied to point out how strong the trend in the rainfall is and
whether it is increasing or decreasing.
• Additionally, the probability distribution was applied to indicate changed in frequency and
intensity during the past decades due to significant variations in rainfall data.

∗
Corresponding author.
E-mail address: bulbul@duet.edu.pk (B. Jan).

https://doi.org/10.1016/j.mex.2023.102058
Received 30 December 2022; Accepted 2 February 2023
Available online 4 February 2023
2215-0161/© 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/)
M. Yonus, B. Jan, H. Khan et al. MethodsX 10 (2023) 102058

Specification table

Subject area: Earth and Planetary Sciences

More specific subject area: Climate Change
Name of your method: Trend Analysis and Probability Distribution Functions
Name and reference of original method: Babar and Ramesh,. [22] Analysis of south west monsoon rainfall trend using statistical techniques over
Nethravathi basin. International journal of advanced technology in civil engineering, 2(1), (2013), 130–136.
Bhakar et al., [14] Probablity analysis of rainfall at Kota. Indian Journal of Agricultural Research, 42(3), (2008),
201–206.
Karmeshu, [17] Trend detection in annual temperature & precipitation using the Mann Kendall test–a case study
to assess climate change on select states in the northeastern United States, (2012).
Ali et al., [25] Estimation of return levels against different return periods of extreme annual rainfall over
Baluchistan. The Nucleus, 49(3), (2012), 221–230.
Yonus, & Hassan [23] Probabilistic Flood Analysis of Indus River Flow. Journal of Mathematics (ISSN
1016–2526), 51(8), (2019), 129–140.
Resource availability: Nil

Introduction

The effect of climate change on the precipitations is one of the important aspects to understand the scenario of weather conditions
in areas where high mountains and permanent glaciers occur and act as a source of water whole year. To measure those affects many
studies were conducted and produced several papers [1]. Robust increase in temperature directly effects the weather and turns the
circumstances to intensify the evaporation rate [2], rainfall intensity [3], uprising the rivers level and so on. Such varying aspects
of weather and climate related things results disastrous circumstances, e.g., strong flood magnitude, flood events can be occurred,
drought intensity and prolong the drought period [4].
The importance of rainfall in whole the world is prime, as a key source of providing water, but in Pakistan its countless need due
to high distress of climate shift/change makes it a lifeline for all the stake holders. The variation of flow in Indus River is particularly
due to the impact of climate changes [5]. A major part of the economy of Pakistan depends on agricultural growth [6] and power
generation partly depends on the flow of water which further depends on temperature as well as rainfall. Furthermore, a requirement
of water to the agricultural purposes and power generation in Pakistan is gotten by permanently running rivers, glaciers & perennial
ice, and summer monsoonal rainfalls [7].
It has been concluded by Naheed et al., [8] and Sarfaraz [9] that Pakistan faces two major kinds of rainfall seasons summer
monsoon and winter monsoon. According to them the summer monsoon is intensively dominating shown in (Fig. 2.5). It can be
factual for the downward regions of Pakistan where winter temperature does not fall below 0 °C mostly and summer monsoonal
rainfall mostly is observed high, whereas for the Upper Indus Basin (UIB) region of Pakistan where high mountainous ranges occur,
winter precipitation is dominated than summer monsoonal rainfall [5]. Their trend analysis can be a well addition in study of the
water resources and effect of rainfall on them. In every summer, from 2010 to 2015 a high and intense summer monsoon with high
melting glaciers and perennial ice on mountains happened in winter, badly affected the downward regions of Pakistan ruining millions
of acres of standing crops and hundreds of lives [7,10–13]. Their comparative studies in several cities provide the opportunity for
planning and would better confer national benefits.
The analysing trend scheme of the rainfalls can provide the opportunity for the batter understanding of the effects of the rainfalls in
Pakistan. Regarding the probability distribution in [14] weekly and monthly rainfall data was employed using four kind of probability
distributions Normal, Log Normal, Weibull and Gumbel distributions. Out of them the Gumbel distribution was found best fit for
monthly maximum rainfall and Weibull’s distribution was found best fit distribution for weekly maximum rainfall in rainy season at
specific station of Kota. The best fit was carried out through Chi- square test. However, by studying the peak daily rainfall distribution
characteristics [15] in Nigeria conducted several analyses to construct mathematical equation for each station. The use of six statistical
analysis Gumbel, Log-Gumbel, Normal, Log-Normal, Pearson and Log-Pearson distributions they concluded that log-Pearson type III
distribution performed the best by occupying 50% of the total station number, while Pearson type III performed second best by
occupying 40% of the total stations and lastly by log-Gumbel occupying 10% of the total stations. Furthermore, the assessment of the
maximum rainfall in Malaysia using the considered models Gamma, Generalized Normal, Generalized Pareto, Generalised Extreme
Value, Gumbel, Log Pearson Type III, Pearson Type III and Wake by results, based on the comparative study, that the Generalised
Extreme Value (GEV) distribution is the most appropriate distribution for describing the annual maximum rainfall series [16]. It is
observed that different schemes for trend analysis of the rainfall can be conducted by different researchers where the best scheme
depends on the particular circumstances of that region and a specific one can be significant. In this paper, rather a daily or monthly
data trend, annual, seasonal (summer or monsoon (JAS) & winter (DJFM)) rainfall (1976–2013) data is taken to examine the trend
analysis of selected cities in Pakistan using the non-parametric Mann–Kendall method and Sen’s slope estimator method. These
methods are used to detect the significance and the magnitude of linear trend of rainfall change respectively. The commonly used
modified Mann-Kendall test was run at 5% significance level on all three types of time series data. Its resultant statistic (S) of Mann-
Kendall test shows how strong trend appeared in rainfall data regarding their decreasing or increasing values. Further, the behavior
of annual and seasonal rainfall analysed for accurate and reliable prediction of selected cities through candidate distributions namely
the Normal (NOR), Pearson (PER), Gamma (GAM), Generalized-extreme value (GEV), Weibull (WEI), generalized Pareto (G.PAR),
Gumble Max (G.MAX) and Gumble Min (G.MIN). Goodness of fit of these distributions was tested by Kolmogorov–Smirnov (KS) test.

2
M. Yonus, B. Jan, H. Khan et al. MethodsX 10 (2023) 102058

Table 1
The Geographical characteristics of the selected meteorological stations.

Name of Station Latitude Longitude Elevation/Height a.m.s.l.

Skardu (P.B.O) 35° 18′ 75° 41′ 2317.0 meter

Murree 33° 54′ 73° 26′ 2127 meter
Chitral 35° 51′ 71° 50′ 1497.8 meter
Drosh 35° 34′ 71° 47′ 1463.90 meter
Gilgit (P.B.O) 35° 55′ 74° 20′ 1460.0 meter
Balakot 34° 33′ 72° 21′ 995.40 meter
Kotli 33° 31′ 73° 54′ 614.0 meter
Sialkot 32° 31′ 74° 32′ 255.1 meter
Lahore PBO. 31° 33′ 74° 20′ 214.00 meter

Maximum likelihood estimation (MLE) was applied to estimate the parameters of each distribution. For selection of best fit probability
distribution function that describes the annual, seasonal rainfall by comparing with the values of Kolmogorov–Smirnov.

Method details

Data collection

Monthly accumulated rainfall records of nine (9) selected cities of Pakistan for 37 years (1976–2013) are collected from Pakistan
Meteorological Department (PMD) for the study. The geographical location of rainfall recording cities is presented in Table 1. For the
said purpose this study considers two series of season wise date namely summer season (July, August and September) (JAS), winter
season (December, January, February and March) (DJFM) and third series as annual rainfall data for all cities. These rainfall series
of annual and seasonal were formed by using monthly rainfall series.

Mann-Kendall (MK) test

A statistical test Mann-Kendall (MK) is commonly used for the investigation of trend in hydrologic and climatologic time series
[17]. It has two advantages; it is a nonparametric test and does not need the data distribution to be normal and second is they have
low sensitivity to sudden breaks due to in-homogeneous time series. According to MK test, the null hypothesis Ho of no trend tested
against the alternative hypothesis HA of the trend exists (decreasing or increasing) in the data values [1,17].
The Mann-Kendall (S) statistic is computed as follows:
𝑛−1 ∑
∑ 𝑛
( )
𝑠= 𝑠𝑖𝑔𝑛 𝑌𝑗 − 𝑌𝑖
𝑖=1 𝑗=𝑖+1

⎧ 1 𝑖𝑓 𝑌 − 𝑌 > 0
( ) ⎪ 𝑗 𝑖
𝑠𝑖𝑔𝑛 𝑌𝑗 − 𝑌𝑖 = ⎨ 0 𝑖𝑓 𝑌𝑗 − 𝑌𝑖 = 0
⎪−1 𝑖𝑓 𝑌𝑗 − 𝑌𝑖 < 0
⎩
where Yi and Yj are the yearly data values in i and j, j>i correspondingly.
The value of |𝑆| is evaluated directly to the theoretical distribution of S resulting by Mann-Kendall if n < 10. Although the two
tailed test is used. The null hypothesis Ho is rejected in favor of HA (alternate hypothesis) at certain probability level if the absolute
value of S equals or increases a specified value S𝛼 /2 where S𝛼 /2 is the minimum value of S which has the probability less the 𝛼/2
to appear in case of no trend. A negative or positive value of S indicates a downward or upward trend respectively. For n ≥ 10, the
statistics Sis distribution approximately normal with mean and variance as follow [18].
𝐸 (𝑆 ) = mean = 0
The 𝜎2(variance) for the S-statistic is defined by:
∑
𝑛(𝑛 − 1)(2𝑛 + 5) − 𝑡𝑖 (𝑖)(𝑖 − 1)(2𝑖 + 5)
σ2 =
18
Where 𝑡𝑖 symbolize the number of ties to extent i. The summation term in the numerator is used. only if the data series contains tied
values [19].

Sen’s slop estimation

Sen. (1968) developed a simple nonparametric procedure to determine the magnitude of linear trend in a time series [20]. The
slope (Qk) for N pairs of data is computed as
xi − xj
Qk = , 𝑘 = 1, … , N
𝑖−𝑗

3
M. Yonus, B. Jan, H. Khan et al. MethodsX 10 (2023) 102058

Fig. 1. Linear trends in total annual, summer or monsoon (JAS) and winter(DJFM) Rainfall (1976- 2013) for (a) Balakot (b) Drosh (c) Chitral (d)
Skardu (e) Sialkot (f) Murree (g) Lahore (h) Gilgit (i) Kotli cities.

Where, xi and xj are data values at times i and j(i> j), respectively. The median of the N values of ‘Qk ’ is represented as Sen’s estimator
of slope which is computed as:

⎧ 𝑄 𝑁+1 if 𝑁 is odd
⎪ ( 2 )
𝑄𝑚𝑒𝑑 = ⎨1
⎪2 𝑄 𝑁 + 𝑄 𝑁+2 if 𝑁 is even
⎩ 2 2

To test the null hypothesis (i.e., no trend) of zero slope, Qmed is computed by a two-sided test at 100(1-𝛼)% confidence interval
for true slope by the non-parametric test. Negative value of Qi indicates a downward or decreasing trend and a positive value of Qi
indicates an upward or increasing trend in the time series [20–22].

Probability distribution functions

The probability distributions namely Normal (NOR), Pearson (PER), Gamma (GAM), Generalized-extreme value (GEV), Weibull
(WEI), generalized Pareto (G.PAR), Gumble Max (G.MAX) and Gumble Min (G.MIN) were considered as potential candidates. In order
to determine the best fit probability distribution function that describes the annual, seasonal rainfall were subjected to a statistical

4
M. Yonus, B. Jan, H. Khan et al. MethodsX 10 (2023) 102058

Table 2
Trend detection of Rainfall data series (1976–2013) based on Mann-Kendall test and Sen’s slope estimator (mm/year) for annual and seasonal
rainfall.

Mann Kendall Test

(Significance level 𝜶=0.05)

Stations Rainfall Totals Sen’s slope M-K Statistic(S) P-Value (two Test Interpretation
(SS) tailed test)

Skardu Annual 2.16 139.0000 0.0829 Accept H0 (NSTD)

Summer (JAS) 0.42 130.0000 0.1048 Accept H0 (NSTD)
Winter(DJFM) 1.35 125.0000 0.1047 Accept H0 (NSTD)
Murree Annual −14.45 −243.0000 0.0020 Reject H0 (STD)
Summer (JAS) −6.33 −169.0000 0.0341 Reject H0 (STD)
Winter(DJFM) −5.01 −124.0000 0.1082 Accept H0 (NSTD)
Chitral Annual 3.13 135.0000 0.0923 Accept H0 (NSTD)
Summer (JAS) 0.21 86.0000 0.2852 Accept H0 (NSTD)
Winter(DJFM) 1.09 52.0000 0.5076 Accept H0 (NSTD)
Drosh Annual 0.69 31.0000 0.7081 Accept H0 (NSTD)
Summer (JAS) −0.21 −37.0000 0.6532 Accept H0 (NSTD)
Winter(DJFM) −0.51 −20.0000 0.8052 Accept H0 (NSTD)
Gilgit Annual 1.15 117.0000 0.1458 Accept H0 (NSTD)
Summer (JAS) 0.13 49.0000 0.5489 Accept H0 (NSTD)
Winter(DJFM) 0.36 107.0000 0.1656 Accept H0 (NSTD)
Balakot Annual −16.88 −253.0000 0.0013 Reject H0 (STD)
Summer (JAS) −6.7 −195.0000 0.0141 Reject H0 (STD)
Winter(DJFM) −4.39 −130.0000 0.0919 Accept H0 (NSTD)
Kotli Annual −9.2 −133.0000 0.0974 Accept H0 (NSTD)
Summer (JAS) −2.79 −61.0000 0.4535 Accept H0 (NSTD)
Winter(DJFM) −3.53 −76.0000 0.3293 Accept H0 (NSTD)
Sialkot Annual −2.56 −49.0000 0.5489 Accept H0 (NSTD)
Summer (JAS) 0.84 17.0000 0.8418 Accept H0 (NSTD)
Winter(DJFM) −1.60 −81.0000 0.2954 Accept H0 (NSTD)
Lahore Annual −3.73 −79.0000 0.3294 Accept H0 (NSTD)
Summer (JAS) −1.72 −55.0000 0.5000 Accept H0 (NSTD)
Winter(DJFM) −2.28 −172.0000 0.0247 Reject H0 (STD)

Test interpretation: H0 : There is no trend in the series; Ha: There is a trend in the series; As the computed p-value is lower than the significance level
alpha=0.05, one should reject the null hypothesis H0 , and accept the alternative hypothesis Ha .STD significant trend detected, NSTD no significant
trend detected.

test Kolmogorov–Smirnov (KS). The KS test statistic D is defined as:

[ ]
( ) 𝑖−1 𝑖 ( )
𝐷 = max 𝐺 𝑥𝑖 − , − 𝐺 𝑥𝑖
1≤𝑖≤𝑁𝑜 𝑁𝑜 𝑁𝑜
Where a random sample, 𝑥1 , … , 𝑥𝑛 from some distribution with the CDF is, 𝐺(𝑥𝑖 ) and 𝑁𝑜 is the number of observations.
If 𝐷 > 𝐷𝑛α (where 𝛼 is the significance level of critical value), in that case the null hypothesis 𝐻𝑜 ∶ the data follow a particular
distribution is rejected [23,24].

Method validation

Trend analysis

Firstly, the linear trends of annual, summer or monsoon and winter rainfall time series date (1976–2013) of the all nine stations
(Balakot, Drosh, Chitral, Skardu, Sialkot, Murree, Lahore, Gilgit and Kotli) is presented in Fig. 1. Most of the stations Balakot, Drosh,
Sialkot, Murree, Lahore and Kotli (6 out of 9) showed decreasing trend for all date series (summer, winter and annual) and the rest of
the stations (Chitral, Skardu and Gilgit) showed increasing trend for all date series as well as the station Sialkot have increasing trend
pattern in only summer (JAS) rainfall. The presented summary in Fig. 1 further verified by MK and SS statistical results (Table 2).
Secondly, this study employed Mann-Kendall (MK) test to determine whether the time series of a variable of interest mostly increase
or decrease over chosen period of time statistically significant (at Significance level 𝛼=0.05) or not, and Sen’s slope estimator (SS) for
magnitude (mm/year) of the trends. In MK test the negative sign (-) of S statistics indicate decreasing trend and a positive sign (+) an
increasing trend. In sen’s slope estimation the SS value is used to figure out the change per unit time of the trends observed in time
series, in which a positive sign indicate an upward slope and a negative sign downward one. The results of the MK and SS presented
in Table 2 indicated that, only three stations have statistical significant decreasing rainfall trends: Murree in annual (p = 0.002;
S = −243.0) and summer (p = 0.0341; S = −169.0); Balakot in annual (p = 0.0013; S = −253.0) and summer (p = 0.041; S = −195.0);
and Lahore in winter (p = 0.0247; S = −172.0). The decreasing trends: Murree in winter (p = 0.1082; S = −124.0); Drosh in summer
(p = 0.6532; S = −37.0) and winter (p = 0.8052; S = −20.0); Balakot in winter (p = 0.0919; S = −130.0); Kotli in annual (p = 0.0974;
S = −133.0), summer (p = 0.4535; S = −61.0) and winter (p = 0.3293; S = −76.0); Sialkot in annual (p = 0.5489; S = −49.0) and

5
M. Yonus, B. Jan, H. Khan et al. MethodsX 10 (2023) 102058

Table 3
Parameters of probability distributions at rainfall gaging stations. Results of goodness of fit test (alpha level=0.05) Critical Value (∗) and P-value
=∗∗.

Distribution
(Kolmogorov Smirnov test)

Stations Rainfall Totals NOR PER GAM GEV WEI G.PAR G.MAX G.MIN

Skardu Annual 0.133 0.105 0.105 0.100 0.124 0.106 0.110 0.186
(0.215∗ ) 0.47∗∗ 0.74∗∗ 0.74∗∗ 0.80∗∗ 0.55∗∗ 0.74∗∗ 0.69∗∗ 0.12∗∗
Summer (JAS) 0.148 0.149 0.105 0.108 0.126 0.152 0.100 0.212
(0.215∗ ) 0.33∗∗ 0.33∗∗ 0.74∗∗ 0.72∗∗ 0.53∗∗ 0.30∗∗ 0.80∗∗ 0.05∗∗
Winter(DJFM) 0.157 0.140 0.116 0.121 0.125 0.092 0.120 0.222
(0.218∗ ) 0.28∗∗ 0.42∗∗ 0.65∗∗ 0.60∗∗ 0.56∗∗ 0.87∗∗ 0.61∗∗ 0.04∗∗
Murree Annual 0.139 0.1 0.118 0.101 0.143 0.085 0.102 0.204
(0.215∗ ) 0.40∗∗ 0.80∗∗ 0.61∗∗ 0.79∗∗ 0.37∗∗ 0.92∗∗ 0.77∗∗ 0.07∗∗
Summer (JAS) 0.099 0.084 0.079 0.072 0.103 0.061 0.091 0.166
(0.215∗ ) 0.80∗∗ 0.92∗∗ 0.95∗∗ 0.97∗∗ 0.77∗∗ 0.99∗∗ 0.87∗∗ 0.21∗∗
Winter(DJFM) 0.146 0.110 0.102 0.106 0.108 0.091 0.110 0.209
(0.218∗ ) 0.367∗∗ 0.71∗∗ 0.79∗∗ 0.75∗∗ 0.74∗∗ 0.88∗∗ 0.72∗∗ 0.06∗∗
Chitral Annual 0.102 0.163 0.124 0.103 0.108 0.139 0.159 0.154
(0.215∗ ) 0.78∗∗ 0.23∗∗ 0.55∗∗ 0.77∗∗ 0.72∗∗ 0.40∗∗ 0.25∗∗ 0.29∗∗
Summer (JAS) 0.199 0.156 0.152 0.075 0.065 0.092 0.191 0.251
(0.215∗ ) 0.08∗∗ 0.28∗∗ 0.31∗∗ 0.96∗∗ 0.99∗∗ 0.87∗∗ 0.10∗∗ 0.01∗∗
Winter(DJFM) 0.109 0.122 0.091 0.092 0.108 0.136 0.101 0.174
(0.218∗ ) 0.72∗∗ 0.59∗∗ 0.89∗∗ 0.88∗∗ 0.73∗∗ 0.45∗∗ 0.80∗∗ 0.18∗∗
Drosh Annual 0.110 0.167 0.138 0.113 0.115 0.116 0.180 0.101
(0.215∗ ) 0.69∗∗ 0.21∗∗ 0.41∗∗ 0.66∗∗ 0.65∗∗ 0.63∗∗ 0.15∗∗ 0.78∗∗
Summer (JAS) 0.098 0.126 0.084 0.082 0.092 0.083 0.085 0.162
(0.215∗ ) 0.82∗∗ 0.53∗∗ 0.92∗∗ 0.94∗∗ 0.87∗∗ 0.93∗∗ 0.92∗∗ 0.24∗∗
Winter(DJFM) 0.117 0.113 0.084 0.095 0.093 0.110 0.115 0.180
(0.218∗ ) 0.64∗∗ 0.68∗∗ 0.93∗∗ 0.85∗∗ 0.87∗∗ 0.71∗∗ 0.66∗∗ 0.16∗∗
Gilgit Annual 0.123 0.087 0.083 0.068 0.096 0.094 0.062 0.174
(0.215∗ ) 0.56 0.91 0.93 0.98 0.84 0.85 0.99 0.17∗∗
Summer (JAS) 0.164 0.119 0.121 0.124 0.138 0.084 0.137 0.211
(0.215∗ ) 0.22∗∗ 0.61∗∗ 0.58∗∗ 0.55∗∗ 0.42∗∗ 0.92∗∗ 0.42∗∗ 0.05∗∗
Winter(DJFM) 0.185 0.312 0.094 0.113 0.116 0.089 0.135 0.246
(0.218∗ ) 0.13∗∗ 0.001∗∗ 0.86∗∗ 0.68∗∗ 0.65∗∗ 0.90∗∗ 0.46∗∗ 0.01∗∗
Balakot Annual 0.130 0.101 0.101 0.093 0.125 0.085 0.107 0.201
(0.215∗ ) 0.49∗∗ 0.79∗∗ 0.79∗∗ 0.86∗∗ 0.54∗∗ 0.92∗∗ 0.73∗∗ 0.08∗∗
Summer (JAS) 0.116 0.084 0.090 0.083 0.104 0.096 0.074 0.173
(0.215∗ ) 0.63∗∗ 0.92∗∗ 0.88∗∗ 0.93∗∗ 0.75∗∗ 0.83∗∗ 0.97∗∗ 0.18∗∗
Winter(DJFM) 0.097 0.118 0.077 0.087 0.067 0.101 0.093 0.165
(0.218∗ ) 0.83∗∗ 0.63∗∗ 0.96∗∗ 0.91∗∗ 0.99∗∗ 0.80∗∗ 0.87∗∗ 0.23∗∗
Kotli Annual 0.080 0.077 0.072 0.079 0.085 0.123 0.076 0.134
(0.215∗ ) 0.94∗∗ 0.96∗∗ 0.98∗∗ 0.95∗∗ 0.92∗∗ 0.56∗∗ 0.96∗∗ 0.45∗∗
Summer (JAS) 0.097 0.090 0.081 0.077 0.078 0.090 0.094 0.163
(0.215∗ ) 0.82∗∗ 0.88∗∗ 0.94∗∗ 0.96∗∗ 0.95∗∗ 0.89∗∗ 0.85∗∗ 0.23∗∗
Winter(DJFM) 0.130 0.084 0.077 0.076 0.084 0.087 0.087 0.195
(0.218∗ ) 0.50∗∗ 0.93∗∗ 0.96∗∗ 0.97∗∗ 0.93∗∗ 0.91∗∗ 0.91∗∗ 0.10∗∗
Sialkot Annual 0.131 0.096 0.098 0.093 0.128 0.106 0.099 0.190
(0.215∗ ) 0.48∗∗ 0.83∗∗ 0.82∗∗ 0.86∗∗ 0.51∗∗ 0.73∗∗ 0.81∗∗ 0.10∗∗
Summer (JAS) 0.127 0.084 0.082 0.080 0.111 0.078 0.088 0.198
(0.215∗ ) 0.52∗∗ 0.92∗∗ 0.94∗∗ 0.95∗∗ 0.69∗∗ 0.96∗∗ 0.90∗∗ 0.08∗∗
Winter(DJFM) 0.104 0.128 0.113 0.091 0.085 0.074 0.123 0.160
(0.218∗ ) 0.77∗∗ 0.52∗∗ 0.68∗∗ 0.89∗∗ 0.93∗∗ 0.97∗∗ 0.58∗∗ 0.26∗∗
Lahore Annual 0.142 0.061 0.099 0.063 0.120 0.102 0.074 0.212
(0.215∗ ) 0.38∗∗ 0.99∗∗ 0.81∗∗ 0.99∗∗ 0.60∗∗ 0.78∗∗ 0.97∗∗ 0.05∗∗
Summer (JAS) 0.132 0.119 0.082 0.086 0.089 0.099 0.081 0.195
(0.215∗ ) 0.47∗∗ 0.60∗∗ 0.94∗∗ 0.91∗∗ 0.89∗∗ 0.80∗∗ 0.95∗∗ 0.09∗∗
Winter(DJFM) 0.111 0.1591 0.096 0.090 0.094 0.076 0.102 0.179
(0.218∗ ) 0.70∗∗ 0.27∗∗ 0.84∗∗ 0.89∗∗ 0.86∗∗ 0.96∗∗ 0.79∗∗ 0.16∗∗

Note: Normal(NOR), Pearson (PER), Gamma (GAM), Generalized-extreme value (GEV), Weibull (WEI), generalized Pareto (G.PAR), Gumble Max
(G.MAX) and Gumble Min (G.MIN).

winter (p = 0.2954; S = −81.0); and Lahore in annual (p = 0.3294; S = −79.0) and summer (p = 0.500; S = −55.0) were found not
significant (at Significance level 𝛼=0.05). An increasing trends for: Skardu in annual (p = 0.0829; S = 139.0), summer (p = 0.1048;
S = 130.0) and winter (p = 0.1047; S = 125.0); Chitral in annual (p = 0.0923; S = 135.0), summer (p = 0.2852; S = 86.0) and
winter (p = 0.5076; S = 52.0); and Gilgit in annual (p = 0.1458; S = 117.0), summer (p = 0.5489; S = 49.0) and winter (p = 0.1656;
S = 107.0) were found not significant, also Sialkot in summer (p = 0.8418; S = 17.0) and Drosh in annual (p = 0.7081; S = 31.0) have
insignificant increasing trends (Table 2).

6
M. Yonus, B. Jan, H. Khan et al. MethodsX 10 (2023) 102058

Furthermore, Table 2 revealed from Son’s slope estimated values that the maximum increase in annual rainfall was of the order
of 3.13 mm/year for Chitral followed by skardu (2.16 mm/year) and Gilgit (1.15 mm/year) and Drosh (0.69 mm/year), whereas
the maximum decrease was for Balakot (−16.88 mm/year) followed by Murree (−14.45 mm/year), kotli (−9.2 mm/year), La-
hore (−3.73 mm/year) and Sialkot (−2.56 mm/year). For summer rainfall we found maximum increase in Sialkot (0.84 mm/year)
followed by Skardu (0.42 mm/year), Chitral (0.21 mm/year) and Gilgit (0.13 mm/year), whereas the maximum decrease was
for Balakot (−6.7 mm/year) followed by Murree (−6.33 mm/year), Kotli (−2.70 mm/year), Lahore (−1.72 mm/year) and Drosh
(−0.21 mm/year). For winter rainfall we found maximum increase in Skardu (1.35 mm/year) followed by Chitral (1.09 mm/year)
and Gilgit (0.36 mm/year), whereas maximum decrease was for Murree (−5.01 mm/year) followed by Balakot (−4.39 mm/year),
Kotli (−3.53 mm/year), Lahore (−2.28 mm/year), Sialkot (−1.60 mm/year) and Drosh (−0.51 mm/year).

Analysis of probability distributions

Major extreme weather events over Pakistan have changed in frequency and intensity during the past decades due to significant
increase in global warming [25]. Therefore, it is needed to study probability distributions for selected stations. On the basis of goodness
of the fit test results, the best fit probability distribution of annual, summer and winter rainfall at each city are shown in Table 3. It
indicates the NOR distribution provided the best-fit probability distribution only for Chitral annual rainfall. The PER distribution is
best for only Lahore annual rainfall. The GAM distribution is best for Chitral winter (DJFM); Drosh winter and Kotli annual rainfall.
The GEV distribution is best for Skardu annual; Drosh summer (JAS); Kotli summer & winter and Sialkot annual rainfall. The WEI
distribution is best for Chitral summer; Balakot winter and Sialkot winter rainfall. The G.PAR is the best for Skardu winter; Murree
annual, summer & winter; Gilgit summer & winter; Balakot annual; Sialkot summer and Lahore winter rainfall. The G.MAX is the
best for Skardu summer; Gilgit annual; balakot summer; Lahore summer. The G.MIN is the best for Drosh annual.

Conclusion

It is known that the climate change over the past several decades has shown in shifting rainfall pattern and transforming rainfall
intensity, which has increased the hydrological processes and added the uncertainty and instability to these phenomena [26]. The
trends of natural resources change with time and become hazardous sometimes. The trend analysis provides opportunity to understand
the overall scenario of those natural resources. Therefore, in this study we have applied trend analysis and probability distribution
to study the variability of rainfall for time interval 1976 to 2013. Following are some key findings throughout the study area;

1. Three cities Chitral, Skardu and Gilgit having increasing trend while the Sialkot station have increasing trend for summer
rainfall and other all cities have shown decreasing trend.
2. Furthermore, key extreme weather events over Pakistan have changed in frequency and intensity during the past decades due
to significant increase in global warming.
3. NOR is distribution the best fitted for the annual rainfall of Chitral and PER distribution is best-fit for only Lahore.
4. The PER distribution is best for only Lahore annual rainfall.
5. The GEV distribution is best for Skardu annual, Drosh summer (JAS), Kotli summer & winter and Sialkot annual rainfall.
6. Different cities of Pakistan indicate different probability distributions which increased the uncertainty and instability to the
hydrological processes.
7. Finally, the study therefore, proposed incredible insights and new context for policy makers and stakeholders in helping them
to take proactive measures in the perspective of climate change-based river flow variations.Their comparative study in several
cities provides the opportunity for planning and would better confer national benefits.

CRediT author statement

Muhammad Younus, Bulbul Jan and Hamza Khan contribute in the concept of the manuscript, Methodology, and Software usage,
writing original manuscript draft preparation. Visualization and Investigation. Faisal Nawaz and Muhammad Ali participated writing-
Reviewing and Editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to
influence the work reported in this paper.

Data availability

Data will be made available on request.

Acknowledgment

We are thankful Pakistan Meteorological Department (PMD) for providing precious data for this study.

7
M. Yonus, B. Jan, H. Khan et al. MethodsX 10 (2023) 102058

References

[1] N. Karmeshu, Trend Detection in Annual Temperature & Precipitation Using the Mann Kendall Test–A Case Study to Assess Climate Change On Select States, the
northeastern United States, 2012.
[2] K.E. Trenberth, Changes in precipitation with climate change, Clim. Res. 47 (1–2) (2011) 123–138.
[3] C. Lepore, D. Veneziano, A. Molini, Temperature and CAPE dependence of rainfall extremes in the eastern United States, Geophys. Res. Lett. 42 (1) (2015) 74–83.
[4] Z.M. Zahudi, M.S. Adnan, N.F. Amat, Y. Erfen, N.C. Ali, Climate change impacts on rainfall distribution at TelukIntan catchment, in: Applied Mechanics and
Materials, 773, Trans Tech Publications Ltd, 2015, pp. 1296–1300.
[5] H. Khan, S. Hassan, Stochastic river flow modelling and forecasting of Upper Indus Basin, J. Basic Appl. Sci. 11 (2015) 630–636, doi:10.6000/1927-5129.
2015.11.84.
[6] A. Piracha, &.Z. Majeed, Water use in Pakistan’s agricultural sector: water conservation under the changed climatic conditions, Int. J. Water Res. Arid Environ.
1 (3) (2011) 170–179.
[7] B. Khan, M.J. Iqbal, M.A.K. Yosufzai, Flood risk assessment of river Indus of Pakistan, Arab. J. Geosci. 4 (2011) 115–122.
[8] G. Naheed, D.H. Kazmi, G. Rasul, Seasonal variation of rainy days in Pakistan, Pak. J. Meteorol. 9 (18) (2013).
[9] S. Sarfaraz, The sub-regional classification of Pakistan’s winter precipitation based on principal components analysis, Pak. J. Meteorol. 10 (20) (2014) 57–66.
[10] A. Ali, Indus Basin Floods Mechanisms, Impacts, and Management, Asian Development Bank, Mandaluyong City,Philippines, 2013 https://www.adb.org/
publications/indus-basin-floods-mechanisms-impacts-and-management.
[11] U. Khalil, N.M. Khan, Assessment of flood using geospatial technique for Indus River Reach, ChashmaTaunsa.Sci. Int. J. 27 (3) (2015) 1985–1991.
[12] A.N. Laghari, D. Vanham, W. Rauch, The Indus basin in the framework of current and future water resources management, Hydrol. Earth Syst. Sci. 16 (2012)
1063–1083, doi:10.5194/hess-16-1063-2012.
[13] K.M.W. Lau, K. Kyu-Myong, The 2010 Pakistan flood and Russian heat wave: teleconnection of hydrometeorological extremes, J. Hydrometeorol. 13 (1) (2012)
392–403, doi:10.1175/JHM-D-11-016.1.
[14] S.R. Bhakar, M. Iqbal, M. Devanda, N. Chhajed, A.K. Bansal, Probablity analysis of rainfall at Kota, Ind. J. Agricult. Res. 42 (3) (2008) 201–206.
[15] O.O. Olofintoye, B.F. Sule, A.W. Salami, Best–fit Probability distribution model for peak daily rainfall of selected Cities in Nigeria, New York Sci. J. 2 (3) (2009)
1–12.
[16] M.D. Zalina, M.N.M. Desa, V.T.A. Nguyen, &A.H.M. Kassim, Selecting a probability distribution for extreme rainfall series in Malaysia, Water Sci. Technol. 45
(2) (2002) 63–68.
[17] M.K. Patil, &.D.N. Kalange, Detection of trend in rainfall data: a case study of Sangli District, Rev. Res. 4 (6) (2015) 1–15.
[18] K. Drápela, &.I. Drápelová, Application of Mann-Kendall test and the Sen’s slope estimates for trend detection in deposition data from BílýKříž (Beskydy Mts.,
the Czech Republic) 1997-2010, Beskydy 4 (2) (2011) 133–146.
[19] T. D.Roy, K.K. Das, Temperature trends at four stations of Assam during the period 1981-2010, Int. J. Sci. Res. Publ 3 (6) (2013) 1–3.
[20] H. Tabari, S. Marofi, A. Aeini, P.H. Talaee, K. Mohammadi, Trend analysis of reference evapotranspiration in the western half of Iran, Agric. For. Meteorol. 151
(2) (2011) 128–136.
[21] S. Mukherjee, R. Joshi, R.C. Prasad, S.C. Vishvakarma, K. Kumar, Summer monsoon rainfall trends in the Indian Himalayan region, Theor. Appl. Climatol. 121
(3) (2015) 789–802.
[22] S.F. Babar, &.H. Ramesh, Analysis of south west monsoon rainfall trend using statistical techniques over Nethravathi basin, Int. J. Adv. Technol. Civil Eng. 2 (1)
(2013) 130–136.
[23] M. Yonus, & S.A. Hassan, Probabilistic flood analysis of Indus river flow, J. Math. Tokushima Univ. 51 (8) (2019) 129–140 (ISSN 1016-2526).
[24] M.A. Sharma, J.B. Singh, Use of probability distribution in rainfall analysis, New York Sci. J 23 (1968) (2010) 40–49.
[25] M. Ali, B. Jan, M.J. Iqbal, Estimation of return levels against different return periods of extreme annual rainfall over Baluchistan, The Nucleus 49 (3) (2012)
221–230.
[26] Z. Zhou, Y. Ouyang, Y. Li, Z. Qiu, M. Moran, Estimating impact of rainfall change on hydrological processes in Jianfengling rainforest watershed, China using
BASINS-HSPF-CAT modeling system, Ecol. Eng. 105 (2017) 87–94.