Madras Agric. J.

, 98 (1-3): 69-73, March 2011

Probability Analysis for Estimating Annual One Day Maximum

Rainfall in Tamil Nadu Agricultural University

M. Manikandan*, G. Thiyagarajan** and G. Vijayakumar*

*Department of Soil & Water Conservation Engineering,
AEC& RI, Tamil Nadu Agricultural University, Coimbatore - 641 003
**Agricultural Research Station, Bhavanisagar-638 451

The daily rainfall data of 37 years were collected and annual one day maximum rainfall was
sorted to estimate the probable one day maximum rainfall for different return periods by
using probability distribution function at Tamil Nadu Agricultural University (TNAU) Campus.
The mean value of annual one day maximum rainfall was found to be 77.79 mm with standard
deviation and coefficient of variation and skewness of 28.56, 0.37 and 0.71 respectively. October
month received the highest amount of one day maximum rainfall and the highest number of
one day maximum rainy days. Five probability distributions such as Normal, Log Normal,
Gumbel, Pearson Type-III and Log Pearson Type-III distribution have been tested to determine
the best fit probability distribution that describes the annual one day maximum rainfall by
comparing with the Chi-square value. The results revealed that the log-normal distribution
was the best fit probability distribution for annual one day maximum rainfall. Based on the
best fit probability distribution, the minimum rainfall of 33.99 mm in a day can be expected to
occur with 99 per cent probability and one year return period and maximum of 173.8 mm
rainfall can be received with one per cent probability and 100 year return period. The results
from the study could be used as a rough guide by design engineers and hydrologists for
appropriate planning and design of small soil and water conservation structures, irrigation
and drainage systems.

Keywords: Return period, frequency, probability distribution, one day maximum rainfall.

Rainfall is one of the most important natural crop failure due to deficit or excess rainfall. To exploit
resource input to crop production and its occurrence the available rainfall effectively, crop planning and
and distribution is erratic, temporal and spatial management practices must be followed based on
variations in nature. The analysis of rainfall data the rainfall amount and distribution at a place
strongly depends on its distribution pattern. It has (Prakash and Rao. 1986). Probability analysis helps
long been a topic of interest in the fields of in developing cropping plan and estimating the
meteorology in establishing a probability distribution design flow rate for maximizing crop production
that provides a good fit to daily rainfall (Barkotulla et (Tomar and Ranade, 2001). Scientific prediction of
al. 2009). Analysis of rainfall and determination of rains and crop planning done analytically may prove
annual maximum daily rainfall would enhance the a significant tool in the hands of farmers for better
management of water resources applications as economic returns (Bhakar et al., 2008).
well as the effective utilization of water resources
The probability distribution functions most
(Rajendra Subudhi, 2007). Probability and frequency
commonly used to estimate the rainfall frequency
analysis of rainfall data enables us to determine
are Normal, Log Normal, Gumbel, Pearson Type-III
the expected rainfall at various chances (Bhakar et
and Log Pearson Type-III distributions. Kumar
al., 2008). Such information can also be used to
(2000) and Singh (2001) concluded that the log
prevent floods and droughts, and applied to planning
normal distribution is the best probability model for
and designing of water resources related to
predicting annual maximum daily rainfall for
engineering such as reservoir design, flood control
Ranichauri (Tehri-Garhwal) and Tandong (sikkim),
work and soil and water conservation planning
respectively. Ajay kumar et al. (2007) predicted annual
(Agarwal et.al., 1988, Dabral et al., 2009).
maximum rainfall and inferred that Log Pearson
Though the rainfall is erratic and varies with time Type III probability distribution function can be used
and space, it is commonly possible to predict return to design hydraulic and soil and water conservation
periods using various probability distributions structures at Almora and similar places in
(Upadhaya and Singh, 1998). Therefore, probability Uttarakhand. Rajendra Subudhi (2007) found that
analysis of rainfall is necessary for solving various normal distribution is the best fit for predicting the
water management problems and to access the annual maximum daily rainfall of Chakapada block
*Corresponding author email: muthiahmanikandan29@gmail.com of Kandhamal district in Orissa. There is no widely
70

accepted procedure to forecast the one day The weibull's method was used for computation
maximum rainfall (Barkotulla et al., 2009). In this of observed annual one day maximum rainfall
present study, an attempt is made to determine the amounts at the return periods of 1.01, 1.05, 2, 5, 10,
statistical parameters and prediction of annual one 25 and 38 years.
day maximum rainfall data of TNAU at various return
Frequency analysis
periods using five probability distribution functions,
viz., Normal, Log Normal, Gumbel's, Pearson Type- Determination of the frequency of occurrence of
III and Log Pearson Type-III distribution. extreme hydrologic events like floods, droughts and
severe storms are important in water resource
Materials and Methods planning and management. There is a definite
The analysis was done at Tamil Nadu Agricultural relation between the frequency of occurrences and
University, Coimbatore which is located at 11N magnitude; the ordinary events occur almost
latitude and 77E longitude with an elevation of 426.72 regularly than the severe storms. Frequency or
m above mean sea level covering an area of 323.88 probability distribution helps to relate the magnitude
ha. The mean annual rainfall is 657 mm distributed of these extreme events with their number of
in 47 rainy days. Heavy rains are likely to occur during occurrences such that their chance of occurrence
North East Monsoon. The mean maximum and with time can be predicted successfully.
minimum temperature is 31.5C and 21.4C, Chow (1964) has shown that most frequency
respectively. The mean relative humidity ranges distribution functions applicable in hydrologic
between 61 to 91 per cent during 07:22 hours and studies can be expressed by the following equation.
ranges between 41 to 68 per cent during 14:22 hours.
XT = X m + K
Data Collection and analysis
Where, XT-variate X of a random hydrologic series
Daily rainfall data collected from Agro Climatic
with a return period of T years; Xm - mean of the
Research Centre in TNAU campus for a period of
variate; - standard deviation of the variate; K -
37 years (1970-2007) were used in the present study.
frequency factor which depends upon the return
Annual maximum daily rainfall was sorted out from
period T and the assumed frequency distribution.
these data. The statistical behavior of any
hydrological series can be described on the basis The theoretical value of annual maximum daily
of certain parameters. Generally, mean, standard rainfall for the same return periods were computed
deviation, coefficient of variation and coefficient of for determining the best fit probability distributions,
skewness were taken as measures of variability of five distributions (normal, lognormal, Pearson type
hydrological series. All the parameters have been III, log Pearson type III and Gumbel) are tried.
used to describe the variability of rainfall in the Continuous distributions
present study.
In hydrology many events are considered as part
Particulars Formula
of continuous processes. For such events,
Mean (Xm) continuous distributions like normal, lognormal,
Gumbel, Pearson type III and log Pearson type III
Standard deviation ( )
may be applied to the observed hydrologic variable.
Distribution Probability density function
Coefficient of variation (Cv)
Normal

Coefficient of Skewness (Cs) Log normal

Glumbel
Return period

Recurrence interval or return period is the

Pearson Type III
average interval of time within which any extreme
event of given magnitude will be equaled or
Log Pearson Type III
exceeded at least once (Patra, 2001). Return period
was calculated from Weibull's method (Chow, Testing the goodness of fit
1964) by arranging one day maximum daily rainfall
Comparing the observed (Oi) and estimated (Ei)
in descending order giving their respective rank as:
values of the relative frequency of the cumulative
T=M+1 frequency function can test the goodness of fit of a
N probability distribution. In case of the relative
frequency function, the Chi-Square test is used. The
Where, N - the total number of years of record
and M - the rank of observed rainfall values arranged Chi-square variates (2) is expressed as:
in descending order.
 71

where, and minimum (30.5 mm) annual one day maximum

k = No. of observations. rainfall was recorded during the year 1977 (Oct-17)
and 1981 (Nov-19), respectively. The average for
The distribution that gives the smallest Chi- these 37 years rainfall was found to be 77.79 mm. It
square valve (Agarwal et al., 1988) is selected are was also observed that 18 years (47%) received
recommended probability distribution function for one day maximum daily rainfall above the average,
the study area. however, no general trend in rainfall occurrence was
Development of regression model observed during these years.
Regression models were developed for Distribution of one day maximum rainfall and
estimating the one day annual maximum rainfall to
return periods in the present study and found the
coefficient of determination (R2)
Results and Discussion
One day maximum daily rainfall

One day maximum daily rainfall corresponding

date for the period of 37 years (1971 to 2007) is
presented in Table 1. The maximum (157.9 mm)
Table 1. One day maximum daily rainfall for the
Figure 1. Distribution of one day maximum
period of 1971 to 2007
annual rainfall in a year
SI. No. Year Date Rainfall (mm)
1. 1971 Sep-19 97.0 rainy days received during different months in a year
2. 1972 Dec-09 143.0 are presented in figure 1 and 2. From the figure, it
3. 1973 Dec-04 106.6 can be seen that October received the highest
4. 1974 Sep-17 38.8 amount of one day maximum rainfall (44%) and
5. 1975 Nov-07 59.8 highest number of one day maximum rainy days
6. 1976 Nov-19 51.1 (43%) followed by November (rainfall - 28% and
7. 1977 Oct-17 157.9 rainy days - 30%) and December (rainfall - 11% and
8. 1978 Nov-05 84.0 rainy days - 8%). This is due to the fact that the study
9. 1979 Nov-21 107.5
10. 1980 Oct-09 94.2
11. 1981 Nov-19 30.5
12. 1982 Oct-14 59.0
13. 1983 Dec-26 73.0
14. 1984 Oct-04 98.0
15. 1985 Oct-01 57.0
16. 1986 Nov-05 49.0
17. 1987 Oct-21 53.5
18. 1988 Apr-18 62.6 Figure 2. Distribution of one day maximum rainy
19. 1989 Nov-06 48.5 days in a year
20. 1990 Oct-31 39.0
area received most of its rain from Northeast
21. 1991 Apr-13 43.5
monsoon (57 per cent of annual rainfall).
22. 1992 Nov-15 112.5
23. 1993 Nov-12 90.0 Statistical parameters
24. 1994 Nov-04 79.5
The mean, standard deviation, coefficient of
25. 1995 May-06 56.0
variation and skewness of annual one day daily
26. 1996 Oct-11 74.0
maximum rainfall for 37 years and their respective
27. 1997 Oct-30 62.5
logarithmic transformation is given in Table 2. These
28. 1998 Nov-10 85.0
statistical parameters can be used to find the
29. 1999 Oct-16 69.0
30. 2000 Aug-15 89.0
Table 2. Statistical parameters of annual one day
31. 2001 Oct-20 103.5
maximum rainfall
32. 2002 Oct-10 69.0
33. 2003 Oct-13 55.6 Sl. No. Statistical parameters Values Logarithmic
transformation
34. 2004 Oct-18 100.2
1. Mean 77.79 1.86
35. 2005 Oct-21 89.2
2. Standard deviation 28.56 0.16
36. 2006 Mar-03 82.4
3. Coefficient of variation 0.37 0.087
37. 2007 Oct-23 107.5
4. Skewness 0.71 -0.204
72

estimated one day maximum rainfall from different The expected annual one day maximum rainfall
probability distribution functions. for different probability distributions such as Normal,
Log Normal, Gumbel's, Pearson type-III and Log
Computation of observed and expected one day
maximum rainfall Pearson Type-III are calculated and presented in
Table 3 for different return periods. The expected
Annual one day maximum rainfall for the period annual one day maximum rainfalls for different
of 37 years was plotted against return period in years probabilities are graphically represented in figure
which was calculated from weibull's method and 4. From the figure, it can be observed that the
presented in figure 3. estimated annual one day maximum rainfall for
different probability distributions are following the
same trend of observed rainfall for different
probabilities.
Chi-square test

All five probability distribution functions were

compared by Chi-square test of goodness of fit and
then selecting the function that gave the smallest
Chi-square value determined the best probability
distribution function. Chi-square values of five
Figure 3. Annual one day maximum rainfall vs probability distribution functions are presented in
return period by Weibulls method Table 4. The Chi-square values for Normal, Log
Normal, Gumbel's, Pearson Type-III and Log
Observed rainfall were found for return periods
Pearson Type-III distributions were 14.33, 2.29, 2.64,
of 1.01, 1.05, 2, 5, 10, 25 and 38 year and presented
4.55 and 2.98. Log normal distribution gave the
in Table 3.

Table 3. Different probability distribution functions of one day maximum expected rainfall against observed
rainfall

Sl. No. Probability Return Observed Expected Rainfall (mm)

(%) period Rainfall Normal Log Normal Gumbal Pearson log
(Years) (mm) Pearson
1. 99 1.0101 29.62 19.63 33.99 30.94 26.49 28.86
2. 95 1.0526 38.80 33.36 40.68 40.51 37.29 38.54
3. 50 2 74.00 77.79 72.81 73.10 74.42 73.74
4. 20 5 101.42 101.84 99.77 98.34 100.32 100.07
5. 10 10 108.29 114.40 117.62 115.05 115.87 116.55
6. 4 25 147.71 127.80 140.17 136.16 134.06 136.43
7. 2.63 38 157.90 132.82 149.70 145.64 140.25 143.63
8. 2 50 136.45 157.00 151.82 146.70 150.61
9. 1 100 144.22 173.81 167.37 158.68 164.29
10. 0.5 200 151.36 190.85 182.86 170.16 177.66

lowest calculated Chi-square value among the five and one year return period and maximum of 173.81
probability distributions. Hence, Log normal mm rainfall can be received with one per cent
distribution has been found the best probability probability and 100 year return period. A maximum
distribution for predicting annual one day maximum of 72.81 mm rainfall is expected to occur at every 2
rainfall for TNAU. According to this distribution, in a years which is approaching mean annual one day
day the minimum rainfall of 33.99 mm rainfall can maximum rainfall. It is generally recommended that
be expected to occur with 99 per cent probability 2 to 100 years is sufficient return period for soil and
Table 4. Chi-square test of goodness of fit for water conservation measures, construction of dams,
different probability distribution for one day irrigation and drainage works (Bhakar et al., 2006).
maximum daily rainfall Regression model
SI. Probability Return Normal Log Gumbal Pearson Log Regression model was developed from the
No. (%) period Normal Pearson
(Years) observed annual one day maximum rainfall against
1. 99 1.0101 5.09 0.56 0.06 0.37 0.02 different return period by using Weibull's method.
2. 95 1.0526 0.89 0.09 0.07 0.06 0.00 The trend analysis (figure 4.) for prediction of one
3. 50 2 0.19 0.02 0.01 0.00 0.00
day maximum rainfall for different return period was
4. 20 5 0.00 0.03 0.10 0.01 0.02
5. 10 10 0.33 0.74 0.40 0.50 0.59 carried out and it is found that the exponential trend
6. 4 25 3.10 0.40 0.98 1.39 0.93 line gives better coefficient of determination (R2) =
7. 2.63 38 4.74 0.45 1.03 2.22 1.42 0.9614 and the equation is: Y = 38.25e 0.0339x
2Cal 14.33 2.29 2.64 4.55 2.98
 73

where, This study gives an idea about the prediction of

annual one day maximum rainfall to design the small
Y - Annual one day maximum rainfall, mm
and medium hydraulic and soil and water
X - Return period, Year conservation structures, irrigation, drainage works,
vegetative waterways and field diversions. This study
also helps in developing cropping plan and
estimating design flow rate for maximizing crop
production.
References
Agarwal, M.C., Katiyar, V.S. and Ramu Babu. 1988.
Probability analysis of annual maximum daily rainfall
of U.P. Himalayas. Indian J. Soil Cons., 16: 35-42.
Ajay kumar, Kaushal.K.K and Singh,R.D. 2007. Prediction
of annual maximum daily rainfall of Almora based on
probability analysis. Indian J. Soil Cons., 35: 82-83.
Figure 4. Estimated annual one day maximum Barkotulla, M.A.B., Rahman, M.S. and Rahman, M.M. 2009.
rainfall for different return period Characterization and frequency analysis of consecutive
days maximum rainfall at Boalia, Rajshahi and
Bangladesh. J. Develop. Agric. Eco. 1: 121-126.
Conclusions
Bhakar, S.R., Bansal, A.N, Chhajed, N. Purohit, R.C. 2006.
Rainfall is a renewable resource, highly variable Frequency analysis of consecutive days maximum
in space and time and subject to depletion or rainfall at Banswara, Rajasthan India. ARPN J. Eng.
enhancement due to both natural and anthro Appl. Sci., 1: 64-67.
pogenic causes. The frequency analysis of annual Bhakar,S.R., Mohammed Iqbal, Mukesh Devanda, Neeraj
one day maximum rainfall for identifying the best fit Chhajed and Anil K. Bansal. 2008. Probability analysis
of rainfall at kota. Indian J. Agric. Res., 42: 201 -206,
probability distribution can be studied for five
probability distributions such as Normal, Log Chow,V.T.,1964. Hand book of applied hydrology. McGraw-
Hill Book company, Newyork.
Normal, Gumbel's, Pearson Type-III and Log
Pearson Type-III by using Chi-square goodness of Dabral, P.P., Mautushi Pal and Singh, R.P. 2009. Probability
analysis for one day to seven consecutive days
fit test. The results of study were, the mean value of
annual maximum rainfall for Doimukh(Itanagar),
annual one day maximum rainfall was found to be Arunachal Pradesh. J. Indian Water Res. 2: 9-15.
77.79 mm with standard deviation and coefficient of
Kumar, Anil, 2000. Prediction of annual maximum daily rainfall
variation of 28.56 and 0.37, respectively. The of Ranichauri (Tehri Garhwal) based on probability
coefficient of skewness was observed to be 0.71. analysis. Indian J. Soil Cons., 28: 178-180.
October month received the highest amount of one Patra, P.C. 2001. Hydrology and water resources engineering.
day maximum rainfall (44%) and highest number of Narosa Publishing House Pvt. Ltd., New Delhi.
one day maximum rainy days (43%) followed by Prakash,C. and Rao, D.H. 1986. Frequency analysis of
November and December. rain data for crop planning, Kota. Indian J. Soil Cons.,
14: 23-26.
It was observed that all the five probability
distribution functions fitted significantly. Log normal Rajendra Subudhi. 2007. Probability analysis for prediction
of annual maximum daily rainfall of Chakapada block
distribution was found to be the best fitted to annual
of Kandhamal district in Orissa. Indian J. Soil Cons.,
one day maximum rainfall data by Chi-square test 35: 84-85.
for goodness of fit. A maximum of 72.81 mm rainfall
Singh, R.K.2001. Probability analysis for prediction of annual
is expected to occur at every 2 years and 50 per cent maximum rainfall of Eastern Himalaya (Sikkim mid hills).
probability which is approaching mean annual one Indian J. Soil Cons., 29: 263-265.
day maximum rainfall. For a recurrence interval of Tomar, A.S. and Ranade, 2001. Predicting rainfall probability
100 years and one per cent probability, the annual for irrigation scheduling in black claysoil of indore
one day maximum rainfall is 173.81 mm. region of madyapradesh. Indian J. Soil Cons., 29 :
Regression model for annual one day maximum 82-83.
rainfall was developed by using Weibull's method Upadhaya, A. and Singh, S.R. 1998. Estimation of
to predict the rainfall for different return period. The consecutive day's maximum rainfall by various
coefficient of determination (R2) is 0.9614. methods and their comparison. Indian J. Soil Cons.,
26: 193-201

Received: October14, 2010; Accepted: February 20, 2011