MAUSAM, 56, 4 (October 2005), 785-794

551.577.21 : 551.507.777 (548.23)

Extreme rainfall analysis of Andhra Pradesh using a probability

distribution model : A regional estimate

P. GUHATHAKURTA, P. A. MENON, S. K. DIKSHIT and S. T. SABLE

India Meteorological Department, Pune – 411 005, India
(Received 30 September 2003, Modified 6 January 2005)

lkj & bl 'kks/k&i= esa ekulwu vkSj ekulwuksRrj _rqvksa ds ifj.kkeLo:Ik gksus okyh izpjq o"kkZ okys
izk;}hih; Hkkjr ds rVh; jkT; vka/kz izns’k esa vR;kf/kd o"kkZ gksus dh ?kVukvksa dk fo’ys"k.k djus dk iz;kl
fd;k gSA bl v/;;u ds fy, 1901&2000 rd dh vof/k ds nkSjku ds lRrj o"kksaZ vFkok mlls Hkh vf/kd o"kksZa
ds vk¡dM+ksa okys 155 LFkkuksa ds latky dk mi;ksx fd;k x;k gSA ;s LFkku iwjs jkT; esa QSys gq, gSaA
,d & fnu] nks & fnu vkSj rhu & fnu dh pje okf"kZd o"kkZ J`a[kyk,a rS;kj dh xbZ vkSj ;s leo`f"V js[kk
fo’ys"k.k rhu Hkkjh o"kkZ okys {ks=ksa dks fu:fir djrh gSaA laHkkO;rk forj.k iz.kkyh izR;sd LFkku dh vR;kf/kd
o"kkZ J`a[kykvksa esa tyok;q ifjorZuksa ds {ks=h; vkdyuksa ds vuq:Ik ikbZ XkbZ gSA pje eku forj.k uker% xEcy
vkSj ykWx lkekU; forj.k ekulwuksRrj ds vuq:Ik ikbZ gS tks ekulwu _rq ls dqN esy [kkrh gSA
dkWyeksxksoZ&fLejukWo vk¡dM+kas ls forj.k dh vuq:irk dh xq.kork dh tk¡p dh xbZ gSA

ABSTRACT. The study is an attempt to analyze the extreme rainfall events of Andhra Pradesh, a coastal state of
Peninsular India, where both monsoon and post-monsoon seasons contribute significant rain. A network of 155 stations
having data of seventy years or more during the period 1901-2000 has been used for the study. These stations are well
distributed over the state. One-day, two-day and three-day extreme annual rainfall series are made and the isohyetal
analysis demarcates three heavy rainfall receiving zones. Probability distribution functions have been fitted for the
regional estimates of climate changes in extreme rainfall series of each station. Both the extreme value distribution viz.,
Gumbel and log normal distribution fit well with latter one giving slightly better fit over the former. Goodness of fit of
the distribution is tested with Kolmogorov - Smirnov Statistic.

Key words – Extreme rainfall, Probability distribution, Climate change, Regional estimate.

1. Introduction Also short period heavy falls occur with intense

convective activities during the pre-monsoon season.
Modelling of extreme rainfall is essential in the
designing of water related structure, in agriculture
planning, in weather modification, water management and A large amount of the variability of rainfall is related
also in monitoring climate changes. Moreover, knowledge to the occurrence of extreme rainfall events and their
of spatial and temporal variability of extreme rainfall intensities. Therefore, there is a need to know the
events is very much useful for the design of dam and magnitudes of extreme rainfall events over different parts
hydrological planning. of the area under study. The study of spatial variability of
extreme rainfall events helps to identify the zone of high
Andhra Pradesh is situated on the east coast of and low value of ever extreme rainfall events. A detailed
Peninsular India. It experiences heavy to very heavy regionalized study is practically useful for the planners
rainfall associated with the meteorological systems during and other users. However, not much work has been done
pre-monsoon, south-west monsoon and also during post in India examining the same, although in some studies
monsoon / northeast monsoon seasons. Heavy rains occur viz., Rakhecha and Pisharoty (1996) authors have studied
over the region due to the passage of westerly moving the heavy rainfall events during the southwest monsoon
tropical disturbances and due to strong monsoon currents. season for some selected stations over the country.

(785)
786 MAUSAM, 56, 4 (October 2005)

TABLE 1

Rainfall normals in mm for the three meteorological sub-divisions of Andhra Pradesh and for the state

Met. sub div. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Annual

Coastal A.P. 6.8 8.9 13 22.8 60.3 106.5 169.8 164.1 168 195.1 94.6 21.7 1031

Telangana 4.5 5.7 9.8 16.3 29.1 136.1 244 221.8 172.9 83.5 20.5 4.2 948.4

Rayalseema 3.5 3.4 6.7 18.1 54.8 59.8 91.1 97.3 132.3 122.2 67.8 25.2 682.2

Andhra Pradesh 5.1 6.2 9.8 19.1 48.2 100.2 167.2 160.3 157.9 135 61.9 17.7 888.6

In modelling extreme rainfall/flood frequency, cyclonic storms/depressions form over south and central
selection of distribution to be used is an important task Bay of Bengal and move in a northwesterly direction
(WMO, 1989). The choice of suitable distribution function causing heavy to very heavy rainfall over the region.
has attracted considerable interest (Gumbel, 1958; During the southwest monsoon period, monsoon system
Hosking, 1990; Boughton 1980). References to exhaustive brings persistent copious rain through out the season. Due
literatures are also available in (WMO, 1989). Recently to the formation of tropical disturbances/storms over south
Parida (1999), Unkasevic and Radiomovic (2000), Park and central Bay of Bengal and their movement over the
and Jung (2002) have tried to fit distribution for extreme region bring incessant rain during post monsoon season.
rainfall while Katz and Acero (1994) have done the Northeast monsoon also some times brings heavy rainfall
sensivity analysis of extreme precipitation events. over the region.
Regional analysis of extreme values (temperature) in
terms of spatial analogue for climate change was done by Monthly rainfall normals (Table 1) over the three
Brown & Katz (1995). Regional analysis has also been sub-divisions and the state based on the period 1941-90
applied in hydrology for better estimates of flood show that average monthly rainfall for three different sub-
frequencies (Chowdhury et al., 1991; Hosking et al., divisions is maximum in three different months. Coastal
1985). Andhra Pradesh which is the most affected by the cyclonic
storms gets maximum average monthly rainfall during the
In this study, spatial distribution of one-day, two- month of October. This sub-division receives more than
day and three-day extreme rainfall over Andhra Pradesh 30% of annual rainfall during the post-monsoon season.
are presented. It is found that the annual extreme rainfall While for the sub-divisions Telangana and Rayalseema,
values generally follow a positively skewed distribution average rainfall is maximum during July and September
function. Individual distribution can differ significantly respectively.
and may vary noticeably from one another in the values
estimated for large return period. Since the hydraulic Andhra Pradesh has a good network of rainfall
design is often based on estimate of large recurrences observation stations. Rainfall data of large number of
interval events, it is important to determine the stations are available in the archive of the India
distribution as accurately as possible. In this study several Meteorological Department. We have considered the data
standard probability distribution functions are tested to of the complete 20th century i.e., 1901-2000. Out of all the
find the best fit for the spatial distributions of one-day, rain gauge stations, only those stations are considered
two-day & three-day extreme rainfall. which are evenly distributed and have 70 years or more
data. Table 2 shows the list of 155 stations, their
2. Data longitude, latitude and the availability of the data. Out of
155 stations, 115 stations have the data of 90 years or
Andhra Pradesh consists of three meteorological more.
homogeneous sub-divisions viz., Coastal Andhra Pradesh,
Telangana and Rayalseema. Climatologically heavy 3. Regional analysis of extreme daily rainfall events
rainfall over Andhra Pradesh is mainly associated with
three different synoptic/meso-scale systems. Intense 3.1. Descriptive rainfall statistics
convection leads to the development of thunder storm
during pre-monsoon season which causes heavy rainfall in 155 stations selected for the study have a relatively
short duration. Sometimes in pre-monsoon season, long series of data. From the daily rainfall series, Annual
 GUHATHAKURTA et al. : DISTRIBUTION MODEL FOR RAINFALL OVER A.P. 787

TABLE 2

Station name, Latitude, Longitude, number of years of data available, 1-day extreme rainfall in cm and the corresponding
date of occurrence and the Kolmogorov-Smirnov's statistic D computed from the time series of annual maxima of
one-day precipitation fitted with Log normal distribution

S. Lat. Long. No. of 1-day S. Lat. Long. No. of 1-day

Station name Date D Station name Date D
No. Deg. Min. Deg. Min. Years extreme No. Deg. Min. Deg. Min. Years extreme
1 Kuppam 12 45 78 21 94 18.5 26 Sep ′78 0.0420 47 Isakapally 14 44 80 7 82 34 02 Oct ′77 0.0823
2 Madakasira 13 56 77 16 94 19.4 31 Oct ′91 0.0424 48 Krishnapatnam 14 16 80 7 86 43.1 23 Oct ′69 0.0359
3 Hindupur 13 49 77 29 94 15.6 15 Sep ′75 0.0388 49 Gooty 15 7 77 38 94 20.1 08 Jul ′90 0.0583
4 Venkatagirikota 13 0 78 30 79 20.1 5 Nov ′03 0.0348 50 Yadki 15 3 77 52 88 19.7 28 Jul ′81 0.0764
5 Palamner 13 12 78 45 94 21.7 20 May ′43 0.0397 51 Yemiganur 15 47 77 29 89 21.9 31 Oct ′16 0.0403
6 Punganur 13 22 78 35 93 19.9 07 Jul ′94 0.0510 52 Adoni 15 38 77 17 91 40 14 Jun ′89 0.0481
7 Madanapalli 13 33 78 30 84 40 13 Aug ′90 0.0434 53 Alur 15 23 77 14 91 17.8 13 Jul ′84 0.0490
8 Vayalpad 13 37 78 38 84 18.2 19 May ′43 0.0276 54 Dhone 15 24 77 52 87 19.2 31 Oct ′16 0.0510
9 Piler 13 39 78 57 83 14.5 22 Oct ′54 0.0646 55 Peapalli 15 14 77 44 91 18 25 Jul ′93 0.0478
10 Tanakal 13 55 78 12 81 19.2 30 Oct ′91 0.0772 56 Pattikonda 15 23 77 31 90 23.2 01 Sep ′65 0.0238
11 Venkatagiri 13 58 79 35 94 27.6 15 Nov ′91 0.1012 57 Gudur 15 47 77 48 94 15 02 Aug ′74 0.0575
12 Satyavedu 13 26 79 58 94 28 24 Dec ′83 0.0458 58 Kurnool Obsy 15 50 78 4 91 16.6 23 May ′52 0.0939
13 Puttur 13 27 79 33 94 25.4 17 Jan ′06 0.0365 59 Giddalore 15 22 78 55 92 18.2 23 May ′52 0.0333
14 Kalahasti 13 46 79 43 88 53 17 Oct ′76 0.0650 60 Aloograda 15 9 78 31 88 33.1 18 Oct ′75 0.0266
15 Tirupati 13 38 79 24 94 26.5 17 Jan ′06 0.0425 61 Nandyal 15 30 78 30 89 38.9 23 May ′52 0.0572
16 Chandragiri 13 36 79 20 94 27.2 17 Jan ′06 0.0369 62 Owk 15 13 78 8 88 15 09 Jul ′89 0.0829
17 Pakala 13 28 79 8 72 25.2 22 Nov ′15 0.0734 63 Koilkuntla 15 15 78 19 91 20.3 31 Oct ′60 0.0764
18 Chittoor 13 13 79 7 92 22.7 30 Nov ′30 0.0213 64 Atmakur 15 53 78 35 93 46.4 06 Oct ′94 0.0738
19 Sullurpet 13 47 80 1 93 42.3 14 Nov ′84 0.0307 65 Nandikottur 15 52 78 16 91 25.8 23 May ′52 0.0478
20 Tada 13 36 80 2 90 42 08 Oct ′43 0.0985 66 Addanki 15 48 79 58 93 29.7 18 May ′69 0.0261
21 Rayadurg 14 42 76 51 94 24.6 25 Jul ′56 0.0356 67 Kanigiri 15 25 79 31 93 43 13 May ′79 0.0323
22 Penukonda 14 5 77 35 94 19 31 Oct ′91 0.0545 68 Podili 15 38 79 37 93 32.2 22 Jul ′84 0.0689
23 Bukkapatnam 14 12 77 47 94 20.6 07 Jul ′89 0.0207 69 Darsi 15 47 79 41 93 28.4 20 Dec ′06 0.0532
24 Dharmavaram 14 25 77 43 94 21.2 12 Sep ′83 0.1372 70 Kandukur 15 13 79 54 93 28.7 01 Nov ′94 0.0445
25 Kalyandrug 14 33 77 7 92 14.1 20 May ′43 0.0813 71 Markapur 15 45 79 18 93 40 13 May ′79 0.0438
26 Anantapur Obsy 14 41 77 37 89 16.8 27 Sep ′74 0.0304 72 Cumbum 15 35 79 7 76 20.3 23 Sep ′49 0.0731
27 Urvakonda 14 57 77 15 94 24.6 10 Nov ′56 0.0800 73 Bapatla 15 54 80 28 94 48.9 10 May ′90 0.1023
28 Pulivendla 14 25 78 14 92 16.3 06 Oct ′44 0.0445 74 Chinaganjam 15 42 80 15 85 31.2 18 May ′69 0.1077
29 Jammalamadugu 14 51 78 24 93 18.7 08 Oct ′58 0.0273 75 Pakala 15 15 80 0 82 30.5 29 Oct ′39 0.0872
30 Proddatur 14 45 78 34 92 24.8 08 Jun ′34 0.0519 76 Mahbubnagar 16 44 77 59 92 17.1 14 May ′79 0.0354
31 Kamalapuram 14 35 78 40 92 21.7 12 Jun ′06 0.0322 77 Gurjala 16 33 79 38 94 17.2 19 Oct ′45 0.0552
32 Cuddapah Obsy 14 29 78 50 97 27 08 Oct ′58 0.0411 78 Macharla 16 28 79 26 92 19.1 19 Oct ′45 0.0681
33 Sidhout 14 28 78 58 94 16.4 06 Jun ′91 0.0336 79 Vinukonda 16 3 79 45 90 32 20 Dec ′06 0.0315
34 Rayachoti 14 3 78 45 93 24.8 06 Nov ′03 0.0195 80 Avanigadda 16 2 80 55 94 50 17 Oct ′69 0.0543
35 Kadiri 14 7 78 10 94 17.3 06 Nov ′03 0.0612 81 Gudivada 16 26 80 59 93 23.2 17 Oct ′69 0.0309
36 Tadpatri 14 55 78 2 94 18.9 20 Jul ′77 0.0417 82 Gannavaram 16 33 80 48 92 20.4 31 Jul ′33 0.0419
37 Lakkireddipalam 14 12 78 42 74 16.5 22 Oct ′54 0.0426 83 Nuzvid 16 48 80 52 79 27.5 23 Jul ′89 0.0348
38 Udayagiri 14 53 79 18 93 32 01 Nov ′94 0.0511 84 Vijayawada 16 31 80 37 88 32.4 17 May ′25 0.0355
39 Nellore Obsy 14 27 79 59 97 52.3 03 Nov ′87 0.0437 85 Jaggiayapet 16 53 80 6 94 19.2 23 Jul ′89 0.0330
40 Gudur 14 9 79 52 87 40 14 Nov ′88 0.0575 86 Nandigama 16 47 80 17 93 25.7 29 Oct ′36 0.0607
41 Rapur 14 12 79 32 94 25.2 06 Dec ′46 0.0254 87 Guntur 16 18 80 27 94 40 20 Nov ′77 0.0804
42 Badvel 14 45 79 4 94 24.5 23 May ′52 0.0280 88 Mangalgiri 16 25 80 35 92 22.4 18 May ′69 0.0359
43 Chitvel 14 10 79 20 80 22.2 08 May ′30 0.1162 89 Sathenapalli 16 23 80 8 94 23.6 02 Nov ′27 0.0749
44 Rajampet 14 12 79 10 94 20.2 06 Nov ′03 0.0641 90 Tenali 16 14 80 39 94 25.8 18 Oct ′82 0.0398
45 Atmakur 14 37 79 38 89 23.9 27 Nov ′07 0.0266 91 Repalle 16 2 80 51 94 47.5 20 Nov ′77 0.0369
46 Kavali 14 55 80 0 94 40.6 17 Oct ′76 0.0238 92 Ponnuru 16 3 80 33 93 36 20 Nov ′77 0.0814
788 MAUSAM, 56, 4 (October 2005)

TABLE 2 (Contd.)

S. Lat. Long. No. of 1-day S. Lat. Long. No. of 1-day

Station name Date D Station name Date D
No. Deg. Min. Deg. Min. Years extreme No. Deg. Min. Deg. Min. Years extreme

93 Narasaraopet 16 14 80 3 94 23.3 08 Oct ′30 0.0421 125 Yelamanchili 17 33 82 52 91 35.6 22 Oct ′28 0.0806
94 Alamur 16 46 81 53 94 38.8 26 Sep ′08 0.0213 126 Polavaram 17 24 82 49 73 35.9 20 Oct ′58 0.1120
95 Kothapeta 16 43 81 54 92 36 23 Jul ′89 0.0459 127 Tuni 17 21 82 33 94 35.5 10 May ′90 0.1022
96 Razole 16 28 81 50 94 27.1 16 Nov ′23 0.0626 128 Pithapuram 17 7 82 15 94 24.2 22 Oct ′28 0.0269
97 Eluru 16 42 81 7 93 23.5 22 Sep ′91 0.0516 129 Prathipadu 17 14 82 12 94 29.9 10 May ′90 0.0684
98 Tadepalligudem 16 50 81 31 93 27.2 23 Jul ′89 0.0408 130 Peddapuram 17 5 82 8 94 26.9 11 May ′90 0.0295
99 Tanuku 16 45 81 43 92 32.1 23 Jul ′89 0.0654 131 Yellavaram 17 26 82 2 76 32.7 26 Sep ′49 0.1044
100 Penugonda 16 39 81 45 80 32.5 23 Jul ′89 0.0415 132 Bhimunipatnam 17 53 83 26 93 52 17 Oct ′82 0.0385
Visakhapatnam
101 Bhimavaram 16 32 81 33 93 33.4 31 Aug ′64 0.0558 133 Ap 17 43 83 14 97 29.3 20 Oct ′58 0.0316
102 Narsapur 16 26 81 42 93 28 16 Oct ′44 0.0549 134 Anakapalli 17 41 83 0 94 28.2 22 Oct ′28 0.0375
103 Manginapudi 16 14 81 11 88 29.8 29 Oct ′02 0.1239 135 Nizamabad Obsy 18 40 78 6 97 35.5 06 Oct ′83 0.0803
104 Musulipatnam Obsy 16 11 81 8 90 46.5 06 Nov ′68 0.0159 136 Karimnagar Rev 18 25 79 9 88 19.5 24 Jul ′89 0.0672
Hanamkonda
105 Kaikalur 16 33 81 12 91 24.1 19 Oct ′33 0.0890 137 Obsy 18 1 79 34 97 22.8 13 Jul ′03 0.0832
106 Ramachandrapuram 16 50 82 2 88 36.9 27 Sep ′08 0.1001 138 Nuguru 18 20 80 33 72 44.2 29 Jul ′88 0.0434
107 Kakinada Obsy 16 57 82 14 97 50.1 02 Jun ′41 0.0828 139 Srikakulam 18 18 83 54 94 26.7 15 Nov ′01 0.0250
108 Coringa 16 48 82 14 83 39.8 30 Oct ′02 0.0788 140 Parvatipuram 18 47 83 26 91 27.4 25 Jun ′14 0.0406
109 Mummidivaram 16 39 82 7 91 36.9 02 Jun ′41 0.0577 141 Palakonda 18 36 83 45 93 27.9 18 Nov ′23 0.0497
110 Amalapuram 16 34 82 0 94 27.7 10 Sep ′50 0.0368 142 Bobbili 18 34 83 22 91 28.2 14 Oct ′31 0.0823
111 Biccavole 16 57 82 3 76 43.2 20 Oct ′58 0.0633 143 Salur 18 31 83 12 91 29.2 14 Oct ′31 0.0792
112 Sangareddi Rev 17 38 78 5 93 30.7 27 Sep ′08 0.0302 144 Gajapathinagaram 18 18 83 20 89 51.1 14 Oct ′31 0.0534
113 Begumpet Obsy 17 27 78 28 97 19.1 01 Aug ′54 0.0403 145 Chipurupally 18 18 83 34 90 35.4 14 Oct ′31 0.0351
114 Janwada 17 24 78 13 85 23.1 27 Sep ′08 0.0642 146 Vijayanagaram 18 7 83 25 90 40.3 24 Oct ′90 0.0963
115 Nalgonda 17 3 79 16 90 20.5 27 Aug ′49 0.0140 147 Srungavarapukota 18 7 83 8 90 26.3 11 May ′90 0.0689
116 Bhadrachalam 17 40 80 54 93 28.6 05 Oct ′83 0.0914 148 Konda 18 1 83 35 76 38.7 18 Nov ′23 0.0743
117 Chintalapudi 17 4 80 59 93 37 23 Jul ′89 0.1173 149 Sompeta 18 56 84 36 94 28.6 23 Sep ′72 0.0783
118 Tiruvur 17 7 80 37 94 37.8 23 Jul ′89 0.0457 150 Pundi 18 40 84 22 82 25.4 01 Nov ′43 0.0911
119 Rajahmundry 17 0 81 46 94 33.4 26 Sep ′08 0.0741 151 Tekkali 18 37 84 15 94 29 18 Nov ′23 0.0591
120 Rampachodavaram 17 27 81 47 86 25.8 19 Oct ′33 0.1211 152 Narasannapeta 18 25 84 3 93 36.3 08 Sep ′55 0.0421
Kalingapatnam
121 Kovvur 17 1 81 44 93 24.8 23 Jul ′89 0.0570 153 Obsy 18 20 84 8 92 33.8 17 Oct ′82 0.0532
122 Polavaram 17 15 81 39 93 25.9 19 Oct ′33 0.0722 154 Asifabad Rev 19 22 79 18 87 24 12 Aug ′86 0.0329
123 Chidavaram 17 50 82 56 93 32.3 17 Oct ′04 0.0528 155 Itchapuram 19 7 84 42 93 35.1 19 Nov ′23 0.0518
124 Narsipatnam 17 40 82 37 93 44.3 23 Sep ′88 0.0300

Maxima of daily Precipitation (AMP1), Annual Maxima Kalahasti. Similarly in the case of 2-day annual maxima,
of two - day Precipitation (AMP2) and Annual Maxima of lowest value of 15.4 cm is over Kalyandurg and highest
three - day Precipitation (AMP3) are constructed for each value of 81.28 cm is over Biccavole. In case of 3-day
of the 155 stations. Figs. 1-3 depict different statistics maxima, lowest value of 17.27 cm is also over
used in data analysis. Sample statistics including Kalyandurg and highest value of 96.66 cm is over
maximum, minimum of AMP, sample median, computed Sulurpet. Minimum of 1-day annual maxima over the
from the time series of AMP1, AMP2 and AMP3 are stations varies from 0.8 cm over Hanamkonda to 6.1 cm
analyzed for variability. Highest ever recorded one day over Vijayanagaram. The average value of 1-day extreme
precipitation alongwith the corresponding date of rainfall series is not highest over Kalahasti where 1-day
occurrence for each of the 155 stations are also given in annual maximum reaches 53 cm but over Nellore.
Table 2. The highest 1-day annual extreme value over Nellore is
52.3 cm.
Maximum of 1-day annual maxima over the stations
has a large variation from the lowest value of 14.1 cm The variability of extreme rainfall can also be
over Kalyandurg to the highest value of 53 cm over measured from the standard deviation. Standard deviation
 GUHATHAKURTA et al. : DISTRIBUTION MODEL FOR RAINFALL OVER A.P. 789

60.0
58.0
MEAN
56.0
MAX
54.0
52.0 MIN

50.0 MEDIAN

48.0 SD
46.0
44.0
42.0
40.0
38.0
36.0
RAINFALL IN MM

34.0
32.0
30.0
28.0
26.0
24.0
22.0
20.0
18.0
16.0
14.0
12.0
10.0
8.0
6.0
4.0
2.0
0.0
1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97
101
105
109
113
117
121
125
129
133
137
141
145
149
153
STATION ACCORDING TO SERIAL NO. IN TABLE 2

Fig. 1. Descriptive statistics of the 1-day extreme rainfall series (AMP1)

90
88
86
84 MEAN
82 MAX
80
78 MIN
76 MEDIAN
74 SD
72
70
68
66
64
62
60
58
56
54
RAINFALL IN MM

52
50
48
46
44
42
40
38
36
34
32
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
1

11

16

21

26

31

36

41

46

51

56

61

66

71

76

81

86

91

96

101

106

111

116

121

126

131

136

141

146

151

STATION ACCORDING TO SERIAL NO. IN TABLE 2

Fig. 2. Descriptive statistics of the 2-day extreme rainfall series (AMP2)

is high in one-day annual maximum precipitation series of 3.2. Spatial analysis of extreme daily rainfall
Nellore where coefficient of variability is 52%. Lowest
value of standard deviation in one-day extreme rainfall Extreme rainfall over Andhra Pradesh exceeds even
series is observed in Kalyandurg where coefficient of 50 cm. From Fig. 1 and Table 2 it can be seen that there
variability is only 32%. Therefore, variability of one-day are 6 stations which are having maximum one day rainfall
extreme rainfall is high in areas of high value of extreme of over 50 cm. One - day, two - day and three - day
rainfall and vice-versa. highest rainfall values of 155 stations have been plotted
790 MAUSAM, 56, 4 (October 2005)

100
MEAN
95
MIN
90
MAX
85 MEDIAN
80 SD

75
70

65
60
RAINFALL IN MM

55
50

45
40

35
30

25
20

15
10

5
0
1

11

16

21

26

31

36

41

46

51

56

61

66

71

76

81

86

91

96

101

106

111

116

121

126

131

136

141

146

151
STATION ACCORDING TO SERIAL NO. IN TABLE 2

Fig. 3. Descriptive statistics of the 3-day extreme rainfall series (AMP3)

Fig. 4. Isohyetal analysis of one-day extreme rainfall of Andhra Pradesh Fig. 5. Isohyetal analysis of two-day extreme rainfall of Andhra Pradesh

separately on three maps. Isohyetal analysis are then done First maximum zone which is in the region of north-
on each of these maps. The resulting patterns are shown in eastern parts of Chittor district consists of values even
Figs. 4 to 6. Mostly three zones of maxima are seen in higher than 50 cm. It may be mentioned that this region is
Fig. 4 and all of these are along the coast. frequently affected by the cyclonic storms which form
 GUHATHAKURTA et al. : DISTRIBUTION MODEL FOR RAINFALL OVER A.P. 791

3.3. Spatial distribution of highest rainfall events :

Regional analysis

Probabilistic extreme value theory, which primarily

deals with the stochastic behaviour of the maximum and
minimum random variables, extreme and intermediate
order statistics and exceedance over (below) high (low)
thresholds are determined by the underlying distribution.
This section highlights the selection of a suitable extreme
value distribution.

Mathematical forms of the most commonly used

probability distributions frequently applied in the extreme
value theory are given below :

(i) Extreme value type 1 (Gumbel or EV1)

  x − u  − ∝≤ x <∝ ,
f ( x) = exp − exp − ,
  α  α >0

Fig. 6. Isohyetal analysis of three-day extreme rainfall of Andhra (ii) Extreme value type 2 (EV 2)
Pradesh

  u − e k 
over the Bay of Bengal and move over this region in a f ( x) = exp −   , k > 0, e ≤ x, 0 ≤ e < u
north-westerly direction. The extreme value of 52.96 cm   x − e  
occurred on 17th October 1976. The low pressure area
over south Bay concentrated in to a deep depression on
(iii) Extreme value type 3 (Weibull)
the morning of 15th October, 1976 with its centre near
12° N and 83° E. It continued to move north-westward
b −1   x − m b 
and lay 50 km to the north east of Chennai on 17th. Later it b  x−m m ≤ x, if a > 0
f ( x) =   exp −   ,
recurved north-eastwards and crossed Bangladesh coast a  a    a   x ≥ m, if a < 0
on 21. Due to its influence, Nellore, Kalahasti, Sullerpet
recorded exceptionally heavy rain on 17th.
(iv) General extreme value (GEV)
The second highest value of 1-day annual maximum
also occurs over the same area slightly north of the first    x − u  
1/ k

maxima. It is 52.34 cm over Nellore on 3rd November, f ( x) = exp − 1 − k   , α > 0

   α  
1987. The synoptic situation for this is as follows:
α
u+ ≤ x ≤ ∞, if k < 0
A low pressure area was developed over southeast k
and adjacent southwest Bay of Bengal on 30th October, α
1987. It intensified into a cyclonic storm by the evening of −∞ < x ≤u+ , if k > 0
k
31st October. Moving west ward, it further intensified into
a severe cyclonic storm on 2nd November and lay centered (v) 2 Parameter Lognormal (LN 2)
at 0830 hrs of IST near Lat. 13.5° N and Long. 82.0° E.
Moving west-north-westwards, the severe cyclonic storm
1 1  1  log x − b  2 
crossed south Andhra coast just north of Nellore around f ( x) = exp −    0<x
0530 hrs of IST of 3rd and weakened and lay at 0830 hrs 2π ax  2  a  
of IST as cyclonic storm about 150 km west-north-west of
Nellore.
Regionalization of 1 - day, 2 - day and 3 - day
extreme rainfall are tested with all of the five distributions
Low values of highest 1-day rainfall are seen all over for selecting the best distribution that can be used for
Telangana. Almost similar types of patterns can be seen regional estimates. Though the procedure of checking the
in Fig. 5 and Fig. 6. goodness of fit of candidate distribution to extreme value
792 MAUSAM, 56, 4 (October 2005)

Bhimunipatnam Gajapatinagaram Musulipatnam

y = 93 * 50 * lognorm (x, 4.729234, 0.469651) y = 89 * 50 * lognorm (x, 4.524125, 0.427285) y = 90 * 50 * lognorm (x, 4.715224, 0.451951)
42
52
39 39
48
36 36
44
33 33
40
30 30
36
27 27
24 32
24
28

No of obs

No of obs
No of obs

21 21
18 24 18
15 20 15
12 16 12
9 12 9
6 8 6
3 4 3
0 0 0

<= 0

<= 0
<= 0

(0,50]
(50,100]

(0,50]
(50,100]

(0,50]

(50,100]
> 550

> 500
> 550
(100,150]

(150,200]
(200,250]
(250,300]

(300,350]
(350,400]

(400,450]
(450,500]
(500,550]

(100,150]
(150,200]
(200,250]

(250,300]
(300,350]
(350,400]
(400,450]
(450,500]
(500,550]

(100,150]

(150,200]

(200,250]

(250,300]

(300,350]

(350,400]

(400,450]

(450,500]
1-day extreme rainfall in mm 1-day extreme rainfall in mm 1-day extreme rainfall in mm

Nellore Repalle Visakhapatnam

y y==97*50*
97 * 50 * lognorm
lognorm (x, (x, 4.90497,
4.90497, 0.475999)
0.475999) y = 94 * 50 * lognorm (x, 4.690924, 0.485196) y = 77 * 20 * lognorm (x, 4.664093, 0.367925)
36 42
18
33 39
36 16
30
33
27 14
30
24
27 12
21 24
10
No of obs

No of obs

No of obs
18 21
15 18 8

12 15
6
12
9
9 4
6
6
2
3 3
0 0 0
(100,150]

(150,200]

(200,250]

(250,300]

(300,350]

(350,400]

(400,450]

(450,500]
(50,100]
(0,50]
<= 0

> 500

<= 40
<= 0
(0,50]

(40,60]
(60,80]

> 300
> 550
(50,100]
(100,150]
(150,200]
(200,250]
(250,300]
(300,350]
(350,400]
(400,450]
(450,500]
(500,550]

(80,100]
(100,120]
(120,140]
(140,160]
(160,180]
(180,200]
(200,220]
(220,240]
(240,260]
(260,280]
(280,300]
1-day extreme rainfall in mm 1-day extreme rainfall in mm 1-day extreme rainfall in mm

Fig. 7. Histogram and the fitted log normal distribution of one-day extreme rainfall series for the six coastal stations

series is the sole criteria, we have done the Kolmogorov - The quantile (or design value) corresponding to a
Smirnov goodness of fit test to examine the best fit. The return period of T years (abbreviatedly, T years return
EV 1 and LN 2 are widely used for extreme value value) is defined by a magnitude x(F), with F = 1-1/T.
analysis. However, LN 2 distribution can show a Return period analysis for indicating distribution of the
reasonable fit to a wider variety of observed data compare maximum expected daily precipitation in regional
to the EV 1 distribution (WMO, 1989). modeling of extreme rainfalls has been done by Garcia-
Ruitz et al. (2003). Isopluvial maps of the estimated
Table 2 also shows the Kolmogorov-Smirnov's design values corresponding to the return periods of 5, 10,
goodness of fit statistic D computed from the time series 25 and 100 years for AMP have been produced. Fig. 8
of annual maxima of one-day precipitation fitted with Log shows the distribution of the 24 hour maximum
normal distribution. The values are compared with the precipitation corresponding to a return period of 100
tabular values of D for the corresponding number of years. The results obtained show a relatively close
sample size and these show that the log normal similarity between Fig. 4 & Fig. 8. The values are high
distribution can not be rejected at 5% level of significance along the coast and also in the north-east part of the state.
for each of stations. The log normal distribution is a best The lowest values are seen in the south-west parts of the
fit for the time series of annual maxima of one-day state.
precipitation of Andhra Pradesh. Histogram and the fitted
Log normal distribution curve for the one-day annual 4. Implications on climate change
maximum series for six coastal stations viz.,
Bhimunipatnam, Gajapatinagaram, Musulipatnam, According to the report of Intergovernmental Panel
Nellore, Repalle and Visakhapatnam are shown in Fig. 7. on Climate Change, extreme weather events are becoming
 GUHATHAKURTA et al. : DISTRIBUTION MODEL FOR RAINFALL OVER A.P. 793

5. Conclusions

Extreme rainfall analyses are presented in two parts.

We have selected 155 stations from the large network of
more than 1700 stations using the criteria of more than 70
years of data availability. First part consists of basic
statistics, foremost important in the data analysis are
calculated and discussed. One-day extreme rainfall is high
in along the coastal areas and low in the western and the
southwestern regions. Also, the variability is high along
the coastal areas and low in the western and southwestern
regions. There are some stations where in a particular year
highest daily rainfall was significantly less than even
1 cm.

Spatial analysis of extreme rainfall events have been

Fig. 8. Distribution of the 24 hour maximum precipitation carried out for regional estimates in the second part. Both
corresponding to a return period of 100 years (CM)
methods of isohyetal analysis and modelling by a suitable
probability distribution function approaches have been
followed in this study. From the isohyetal analysis zones
of high extreme values are explained. The results of this
more intense and causing more losses. We have studied
regional analysis of extreme rainfall events have
the temporal and the spatial distribution of changes in
significant potential implications concerning the extreme
extreme rainfall. Trend analyses are done for one-day
rainfall events change as a part of changing climate. The
extreme rainfall series of all the 155 stations.
proposed statistical model for spatial differences in
climate is consistent with satisfactory fit of two parameter
Trends are estimated by fitting linear regression
lognormal distribution. Significant changes in extreme
equation and the slope co-efficient is tested with
rainfall are noticed in the high rainfall receiving zones of
t-statistic.
coastal Andhra Pradesh. These support the views of
Intergovernmental Panel on Climate Change, regarding
It is noticed that out of 155 stations only 13 stations
global warming that extreme weather events have a
reported significant changes (95%) in one-day extreme
significant increasing trends and these extreme events all
rainfall. Interestingly, most of these stations are situated in
go into calculating the monthly and annual averages.
coastal Andhra Pradesh where average rainfall as well as
one-day extreme rainfall is higher than the other regions.
Three stations viz., Coringa, Visakhapatnam and Chitvel References
show significant decreasing trend in one-day extreme
rainfall. Coringa is in east Godavari district which is Boughton, W. C., 1980, “A frequency distribution for annual floods”,
adjacent to Visakhapatnam district respectively while Water Resour. Res., 16, 2, 347-354.
Chitvel is in Cuddapa district. It may be noticed that the
second highest one-day extreme rainfall zone is over Brown, B. G. and Katz, R. W., 1995, “Regional analysis of temperature
extremes : spatial analog for climate change”, J. Climate, 8, 1,
Visakhapatnam district (Fig. 4). Not a single station in 108-119.
Telangana sub-division, where one-day extreme rainfall is
less than 40 cm (Fig. 4) shows any significant trend. This Choudhury, J. U., Stedinger, J. R. and Lu, L. N., 1991, “Goodness of fit
infers even the view of World Meteorological tests for regional generalized extreme value flood distribution”,
Organization (WMO) that there is a clear and alarming Water Resour. Res., 27, 1765-1776.
trend towards wilder weather conditions. The frequency
Garcia-Ruiz, J. M., Lorente-Grima, A., Begueria-Portugues, S., Marti-
and intensity of extreme weather events increases due to a Bono, C., Valero-Garces, B. and Lopez-Moreno, J. I., 2003,
change in the distribution of heat, which disrupts the flow Damoclees Project Report, Instituto Pirenaico de Ecologia,
of energy through the climate system, altering the CSIC, Zaragoja, Spain.
circulation patterns of the atmosphere and oceans and also
modifying the Earth’s hydrological cycle. High Gumbel, E. J., 1958, “Statistics of extremes”, Colombia University
Press, p375.
temperature increases evaporation and transpiration and
raises the air capacity to hold moisture, making more of it
Hosking, J. R. M., 1990, “L – moments : Analysis and estimation of
available to fall as rain. Thus it increases the occurrences distributions using linear combinations of order statistics”,
of heavy rainfall incidences. J. Roy States soc., Ser B 52, 105-124.
794 MAUSAM, 56, 4 (October 2005)

Hosking, J. R. M., Wallis, J. R. and Wood, E. F., 1985, “An appraisal of Rakhecha, P. R. and Pisharoty, P. R., 1996, “Heavy rainfall during
the regional flood frequency procedure in the UK Flood studies monsoon season : point and spatial distribution”, Current
Report”, Hydrol. Sci., 30, 85-109. Science, 71, 3, 179-186.

Katz, R. W. and Acero, J. G., 1994, “Sensitivity analysis of extreme

precipitation events”, Int. J. Climatol., 14, 985-999.
Unkasevic, M. and Radinovic, D., 2000, “Statistical analysis of daily
Park, J. S. and Jung, H. S., 2002, “Modelling Korean extreme rainfall maximum and monthly precipitation of Belgrade”, Theor. Appl.
using a Kappa distribution and maximum likelihood”, Theor. Climatol., 66, 241-249.
Appl. Climatol., 72, 1-2, 55-64.

Parida, B. P., 1999, “Modelling of Indian summer monsoon rainfall

using a four parameter Kappa distribution”, Int. J. Climatol., 19, World Meteorological Organization, 1989, “Statistical distributions for
1389-1398. flood frequency analysis”, OHR No. 33, WMO No. 718.