arabian journal paper - autor copy final

21
1 23 Arabian Journal of Geosciences ISSN 1866-7511 Arab J Geosci DOI 10.1007/s12517-014-1696-0 Spatio-temporal analysis of droughts in the semi-arid Pedda Vagu and Ookacheti Vagu watersheds, Mahabubnagar District, India Sreedhar Ganapuram, R. Nagarajan, G. Chandra Sehkar & V. Balaji

Upload: sreedhar-ganapuram

Post on 16-Apr-2017

38 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Arabian journal paper - autor copy final

1 23

Arabian Journal of Geosciences ISSN 1866-7511 Arab J GeosciDOI 10.1007/s12517-014-1696-0

Spatio-temporal analysis of droughts in thesemi-arid Pedda Vagu and Ookacheti Vaguwatersheds, Mahabubnagar District, India

Sreedhar Ganapuram, R. Nagarajan,G. Chandra Sehkar & V. Balaji

Page 2: Arabian journal paper - autor copy final

1 23

Your article is protected by copyright and

all rights are held exclusively by Saudi

Society for Geosciences. This e-offprint is

for personal use only and shall not be self-

archived in electronic repositories. If you wish

to self-archive your article, please use the

accepted manuscript version for posting on

your own website. You may further deposit

the accepted manuscript version in any

repository, provided it is only made publicly

available 12 months after official publication

or later and provided acknowledgement is

given to the original source of publication

and a link is inserted to the published article

on Springer's website. The link must be

accompanied by the following text: "The final

publication is available at link.springer.com”.

Page 3: Arabian journal paper - autor copy final

ORIGINAL PAPER

Spatio-temporal analysis of droughts in the semi-arid Pedda Vaguand Ookacheti Vagu watersheds, Mahabubnagar District, India

Sreedhar Ganapuram & R. Nagarajan & G. Chandra Sehkar & V. Balaji

Received: 21 May 2014 /Accepted: 29 October 2014# Saudi Society for Geosciences 2014

Abstract This paper presents spatio-temporal meteorologicaldrought analysis of Pedda Vagu and Ookacheti Vagu water-sheds of Mahabubnagar and Ranga Reddy Districts ofTelangana state, South Central India. Rainfall anomaly index(RAI) and run analysis have been leveraged to assess droughtcharacteristics at different stations in the basin. The study alsopresents the interpolation of RAI values using spline tech-nique in a geographic information system (GIS) environmentto map the spatial extent and variation of drought se-verity in different time steps. The study reveals that theoccurrence, magnitude, and recurrence of drought variedamong the stations in the basin during an observed timeframe, i.e., 1986 to 2013. Significant variations in theoccurrences of number of drought events are observedamong the stations in the basin. The spline interpolatedrainfall anomaly index maps illustrated that some re-gions experienced more severe drought while other re-gions were well-off. This uncertainty in rainfall essen-tially indicates that a finer scale of drought vulnerabilityassessment is highly necessary for better drought man-agement practices. Furthermore, empirical relationships

were developed between drought duration and magni-tude to support decision-making during various agricul-tural practices and water management.

Keywords Semi-arid tropics . Drought . Pedda Vagu .

Ookacheti Vagu . RAI . Run analysis

Introduction

Drought is a typical climatic natural disaster that occurs in anyclimatic conditions. Drought is caused primarily due to defi-ciency in precipitation over a span of time especially for aseason or more (Iglesias et al. 2009). It has critical impact onthe socio-economic aspects of the rural communities mainlythose dependent on agriculture, as it may last for few monthsto several years with varying intensity and spatial extent. Indiahas a long history of drought events, with 22 major droughtyears faced during the period 1871–2002 (Prabhakar andShaw 2008). The 2002 and 2004 droughts show clear evi-dence of the inherent vulnerability of the Indian monsoonsystem to the El Niño phenomenon, which was also demon-strated with the linkage between El Niño and SouthernOscillation and Indian food grain production (Saith andSlingo 2006; Selvaraju 2003). Consequently, it is also evidentthat agriculture is at the mercy of monsoon rainfall occurrenceand failure. India is the second most populated country in theworld, with over 69 % of the populations’ livelihood depen-dent on agriculture and allied activities. India has a totalgeographical area of 328 million hectares (Mha), out of whichthe total cropped area is 174 Mha including 142 Mha ofrainfed area (Murali Krishna et al. 2009). Population growthand the expansion of irrigation led to the scarcity of the waterin the Krishna river basin water (Gaur et al. 2007).Additionally, climate variability adds pressure on the availablewater resources making the basin much more vulnerable to

S. Ganapuram (*) :R. NagarajanCentre of Studies in Resources Engineering, Indian Institute ofTechnology Bombay, Mumbai, India 400076e-mail: [email protected]

R. Nagarajane-mail: [email protected]

G. C. SehkarInfosys Technology Limited, Bangalore, Indiae-mail: [email protected]

V. BalajiTechnology & Knowledge Management, Commonwealth ofLearning, Vancouver, Canadae-mail: [email protected]

Arab J GeosciDOI 10.1007/s12517-014-1696-0

Author's personal copy

Page 4: Arabian journal paper - autor copy final

drought. Between the years 2001 and 2004, Krishna basinexperienced severe droughts causing acute water shortages inlower Krishna basin (Gaur et al. 2007; Biggs et al. 2007).

Although drought and variability in rainfall are notpredictive as most of its causes are natural, but theimpacts could be mitigated with prior awareness aboutthe possible vulnerable regions. The disaster risk miti-gation (DRM) program initiated by the Government ofIndia in collaboration with the United NationsDevelopment Program (UNDP) envisages preparing di-saster management plans for effective preparednessagainst disasters at village, block, district, and provinciallevels (Prabhakar and Shaw 2008). Drought is mostlyanalyzed using point rainfall data at different timescaleswhich is mapped at different spatial scales. TheSoutheast Asia Drought Monitoring system developedby the International Water Management Institute(IWMI) provides drought information at the regional,district/provincial, and pixel level and helps decision-makers to monitor and mitigate the impact of drought.Remote sensing-based applications invariably needground information such as meteorological and

agricultural data to make them more dependable(Thenkabail et al. 2004). Suresh et al. (1993) studiedrainfall data of 26 years at Pusa, Bihar, by analyzingthe characteristics and variation in rainfall data withrespect to normal, abnormal, and drought months in ayear. It was reported that at 90 % probability level,these regions’ expected annual rainfall obtained wasbelow the drought level and during rabi season thesituation was terrible. Several other studies were con-ducted to analyze the rainfall data for drought assess-ments and the variability and trends on annual, monthly,seasonal, and weekly basis (Ankegowda et al. 2010;Kwarteng et al. 2009; Srivastava et al. 2000; Raoet al. 1998; Subudhi et al. 1996). Ankegowda et al.(2010) analyzed rainfall data of Karnataka region for23 years (1986–2008) and showed that 80.94 % ofrainfall occurred during June to September, and thereis no significant trend in mean annual rainfall. Kwartenget al. (2009) analyzed the characteristics of rainfall inthe semi-arid Sultanate of Oman using 27-year (1977–2003) rainfall data. Statistics show a negative but insig-nificant rainfall trends in this region.

Fig. 1 Location map of Pedda Vagu and Ookacheti Vagu watersheds

Arab J Geosci

Author's personal copy

Page 5: Arabian journal paper - autor copy final

Some studies concerned to Mahabubnagar district locatedin lower Krishna basin include drought vulnerability assess-ment using water deficit/surplus details (Sreedhar et al. 2013)and physiographic parameters like rainfall, slope, drainagedensity, etc. (Sreedhar et al. 2012). But detailed analysis ofmeteorological droughts of this region using standard droughtindices are not available. RAI is a standardized drought indexused to recognize temporal droughts at various times scales(Van Rooy 1965) and can be interpolated to assess spatialextent of the droughts. Run analysis (Yevjevich 1967) helps inthe objective identification and characterization of droughtevents (Sirdas and Sen 2003; Mishra and Nagarajan 2011).The present study is conducted with an objective to determinethe spatial and temporal patterns of meteorological droughts inPedda Vagu and Ookacheti Vagu watersheds of Krishna riverbasin. Additionally, empirical relationships are developedusing drought magnitude and length for objective identifica-tion of droughts.

Research method

Study area

The study area (Fig. 1) consists of four watersheds of thelower Krishna basin, located in Southern Telangana Agro-climatic zone, India. The four watersheds consist of two

Pedda Vagu watersheds and two Ookacheti Vagu watersheds.The study area lies between 77° 28′ 33.799″ to 78° 13′31.134″ east longitude and 16° 11′ 45.63″ to 17° 8′ 23.744″north latitude. The total geographical area of the basin is4353 km2 spread in 31 mandals of Mahabubnagar districtand 3 mandals of Ranga Reddy district. The altitude of thebasin ranges from 191 to 637 m. The basin belongs to thesemi-arid tropics with distribution of rainfall mainly duringsouth-west (June–September) monsoon season. The averageannual rainfall of the basin is around 663 mm. The basinconsists of two medium reservoirs, namely Koil Sagar andSarla Sagar, and two small reservoirs Kanayapalli Cheruvuand Raman Pahad. The climate of the area transits fromtropical to subtropical climate. The region has four distinctclimatic seasons like summer, winter, and south-west andnorth-east monsoon. The summers are relatively hot, and theperiod is from March to May with temperature ranging from16.9 to 41.5 °C. The winter temperature ranges from 16.9 to19.1 °C, i.e., from November to January. Agriculture andlivestock are the main livelihood opportunities of the ruralfamilies in the basin. The region follows two agriculturalseasons, viz, kharif (June to October) and rabi (November toMarch). Paddy is widely cultivated in the basin. Apart frompaddy, crops like sorghum, pearl millet, finger millet, maize,groundnut, castor, vegetables, sunflower, chili, and red gramare also being cultivated. Kharif crop cultivation is mostlydependent on rainfall, whereas rabi crop is dependent on

Table 1 Details of location, observation period, and statistics of annual rainfall at various stations of the basins

S.no

Station (observationyears)

Lat. Long. Altitude(m)

Period Mean(mm)

Standarddeviation

Minimum(mm)

Median Maximum CS CK CV

1 Adakkal (23) 16.50 77.93 356 1989–2013 654.5 146.84 332 643.4 1001.1 0.129 0.563 0.224

2 Atmakur (18) 16.32 77.81 310 1996–2013 785.2 253.93 396.5 749.2 1242 0.246 −0.730 0.323

3 Bhoothpur (26) 16.71 78.05 444 1988–2013 617.4 140.02 369 604 925.3 0.565 0.035 0.227

4 C.C.kunta (28) 16.43 77.8 330 1986–2013 602.6 153.41 327.6 649.8 1002.6 0.060 0.285 0.255

5 Devarkadra (18) 16.60 77.85 370 1996–2013 646.5 137.71 477.5 588.15 903.8 0.679 −0.754 0.213

6 Dhanwada (18) 16.65 77.67 436 1996–2013 675.6 150.79 352.2 696.1 936.6 −0.291 0.223 0.223

7 Doulatabad (18) 17.01 77.58 532 1996–2013 757.7 238.79 447.2 690.35 1350.8 1.025 0.813 0.315

8 Gandeed (28) 16.91 77.81 510 1986–2013 621.7 156.83 351 619.4 927.7 0.115 −0.455 0.252

9 Ghanpur (28) 16.55 78.06 448 1986–2013 600.8 160.96 240 612.6 855.6 −0.337 −0.600 0.268

10 Gopalpet (28) 16.38 78.14 418 1986–2013 607.2 149.05 317 619 899.4 −0.157 −0.137 0.245

11 Hanwada (18) 16.80 77.91 442 1996–2013 678.7 156.58 346.2 685.7 917.4 −0.419 −0.010 0.231

12 Kodangal (28) 16.50 77.93 356 1986–2013 744.0 172.06 382 757 1000.8 −0.325 −0.782 0.231

13 Koilkonda (18) 16.75 77.79 442 1996–2013 555.6 141.27 346.8 561.6 916.2 0.758 1.093 0.254

14 Kosgi (27) 16.99 77.75 512 1986–2013 651.1 184.52 355.3 665.6 1046.4 −0.161 −0.617 0.283

15 Kothakota (23) 16.37 77.93 344 1991–2013 685.9 161.44 424.5 653 942.6 −0.125 −1.226 0.235

16 Kulkacherla (28) 17.01 77.86 566 1986–2013 792.3 180.98 283 838 1084.8 −1.254 1.862 0.228

17 Maddur (18) 16.85 77.63 488 1996–2013 501.8 126.54 268 496 733 −0.055 −0.670 0.252

18 Mahabubnagar (18) 16.72 77.99 474 1996–2013 803.5 196.79 473 814.45 1140.7 −0.253 −0.074 0.245

19 Peddamandadi (22) 16.41 78.03 383 1992–2013 561.7 139.51 244.7 547.05 833.2 −0.001 0.073 0.248

20 Wanaparthy (28) 16.34 78.07 445 1986–2013 718.0 163.36 395.9 710.9 982.1 −0.073 −0.902 0.228

Arab J Geosci

Author's personal copy

Page 6: Arabian journal paper - autor copy final

groundwater due to the depletion of water in surface waterbodies. Irrigation water during rabi season is obtained fromeither canals or groundwater pumped from open wells that are10 to 20 m deep or bore wells which are 80 to 100 m deepinstalled with submersible pumps. The predominant soils inthe basin are clayey soils, cracking clay soils, gravelly claysoils, gravelly loam soils, and loamy soils. Soil types includeEntisols and Vertisols (black cotton soils) and Alfisols (redsoils) with low water holding capacity.

Data

Rainfall data for 20 meteorological stations available for theperiod 1986 to 2013 is collected from District Collectorateoffice, Mahabubnagar District, and Directorate ofEconomics and Statistics, Hyderabad, India (refer toTable 1). The daily rainfall data analyzed in this studyis avai lable for the period 1999 to 2013 forMahabubnagar and Ranga Reddy Districts. Several stan-dard statistical parameters like mean, median, minimum(Min), maximum (Max), standard deviation (SD), skew-ness, kurtosis, and coefficient of variability mentionedin Kwarteng et al. (2009) and Ankegowda et al. (2010)are estimated for all the stations and presented inTable 1.

Computation of rainfall anomaly index (RAI)

Rainfall measure is used in drought index calculations as it isthe most vital hydrological variable generally the only mete-orological measurement made in semi-arid areas (Oladipo1985; Tilahun 2006). Several indices are available to calculatetemporal meteorological droughts. In this study, RAI is mod-ified to account for non-normality like SPI which is usedfor the assessment of both temporal and spatial droughtsas it is independent of time and space. Hence, it is moreuseful in semi-arid regions particularly India since at manymeteorological stations, the recorded rainfall data availableis less than 30 years, while most of the meteorologicaldrought assessment indices require more than 30 years ofdata (Van Rooy 1965; Loukas et al. 2003). Additionally,as the rainfall of the present study area is not normallydistributed, RAI is used in drought assessment of thisregion. One more utility of this index is that it can becomputed at different time scales (1, 2, 3, 6, 9, and12 months); short duration is useful in agriculture droughtmonitoring while longer duration helps in water resourceassessment (Loukas et al. 2003). The recommended lengthof data for any drought index is 30 years; however, RAIis considered in this study due to above reasons. Theyearly and kharif rainfall variability is pursued using therainfall data details given in Table 1. RAI is used toidentify droughts by establishing some arbitrary values

for drought identification. The RAI is used to assess andidentify droughts, drought severity, and variability by com-paring with some established arbitrary value. It is reportedthat this index based on only rainfall as input performedcomparatively better than more complicated indices likePalmer and Bholme-Mooley in depicting periods and den-sity of droughts (Oladipo 1985; Tilahun 2006). Rainfallanomaly index (RAI) is described as rainfall variabilityover a time (Van Rooy 1965) and is estimated as belowfor positive anomalies

RAI ¼ þ3 RF−MRF=MH10−MRF½ �

and for negative anomalies

RAI ¼ −3 RF−MRF=ML10−MRF½ �

where RAI represents the annual RAI,

RF is the actual rainfall for a given year,MRF is mean rainfall of the total length of record,MH10 is the mean of the 10 highest values of rainfall on

record, andML10 is the mean of the 10 lowest values of rainfall on

record.

A ranking of nine classes of rainfall abnormality rangingfrom extremely wet to extremely dry and range of eachclass is shown in Table 2 (Keyantash and Dracup 2002;Roshan et al. 2012). If the purpose of the study is toinvestigate dry periods, the negative prefixed RAI isused, while positive RAI is used to study wet periods(Hansel and Matschullat 2006). In the present study, negativeRAI is calculated for annual and kharif time scales for droughtassessment of all the stations.

Spatial drought severity assessment

Mapping of spatial extent of drought severity is essential toknow the drought severity of adjacent regions where point

Table 2 Classification of RAI ranges into drought classes

S. No. RAI Drought class

1 >3 Extreme wet

2 2.1 to 3 Severe wet

3 1.2 to 2.1 Medium wet

4 0.3 to 1.2 Weak wet

5 +0.3 to −0.3 Normal

6 −0.3 to −1.2 Weak drought

7 −1.2 to −2.1 Medium drought

8 −2.1 to −3 Severe drought

9 <−3 Extreme drought

Sources: Keyantash and Dracup 2002; Roshan et al. 2012

Arab J Geosci

Author's personal copy

Page 7: Arabian journal paper - autor copy final

estimates are not available. Interpolation is the most common-ly used method to predict the values of attributes at unknownpoints using measured points within the same area. The inter-polationmethods commonly used are inverse distance weight-ed (IDW), co-kriging, and thin-plate smoothing splines;among the three methods, thin-plate smoothing splines isrecommended for interpolating climate variables. Thin-platesmoothing splines can be used for exact interpolation or forsmoothing to generate spatially coherent surface. Comparedto IDW, splining and co-kriging provides better spatial qualityof the prediction surfaces, but spline interpolation is preferredover co-kriging as it is faster and easier to use (Hutchinson andGessler 1994; Hartkamp et al. 1999). The spatial extent of

drought severity is mapped by using annual and kharif seasonRAI values in GIS environment using spline interpolationmethod for years 1996 to 2013.

Objective identification of drought parameters using “runtheory”

Run theory is proposed and applied for the objective assess-ment of drought parameters like drought duration, magnitude,and intensity (Yevjevich 1967). To derive these parameters,the threshold level approach is used, which is a constant or afunction of time. In run theory, a run is defined as a portion oftime series of drought variable which is either below or above

Fig. 2 Mean monthly rainfall distribution at different stations

Table 3 Descriptive statistics of monthly rainfall data of kharif (June–October) season

Station Maddur Mahabubnagar

Statistics Jun Jul Aug Sept Oct Jun Jul Aug Sept Oct

Mean 60.94 104.6 117.4 105.7 69.23 108.51 156.4 201.9 150.2 115.62

Median 49.70 75.20 115.8 101.5 65.20 108.60 139.80 206.90 136.50 94.40

Mini 11.00 8.20 14.00 36.00 0.00 17.00 41.60 45.30 31.60 36.60

Max 172.00 212.0 231.0 229.0 175.2 258.10 333.20 344.60 256.00 208.20

SD 40.74 64.21 65.25 54.29 54.59 62.04 83.57 96.19 70.05 59.72

CS 1.19 0.33 0.24 0.69 0.61 0.71 0.51 −0.13 −0.04 0.32

CK 1.81 −1.35 −0.88 0.07 −0.59 0.45 −0.48 −1.01 −1.09 −1.40CV 0.67 0.61 0.56 0.51 0.79 0.57 0.53 0.48 0.47 0.52

Arab J Geosci

Author's personal copy

Page 8: Arabian journal paper - autor copy final

the threshold level, characterized as a negative or positive run,respectively (Sirdas and Sen 2003). The threshold level isachieved by dividing the average annual precipitation by themean number of rain days of the basin. Based on the thresholdvalue, a dry spell is recognized if seven consecutive daysreceive rainfall lesser than the threshold value (Mishra andNagarajan 2011). The duration of a drought event is determinedas the time taken by a drought from the time it initiates to until itterminates. Drought magnitude/severity is estimated as the sumof cumulative deficiencies of precipitation that occurred belowthreshold level, while drought intensity is obtained by dividingthe drought magnitude with drought duration.

Analysis and findings

Annual rainfall distribution and variability

Annual rainfall is the most vital climatic indicator of waterdeficit or surplus in any region. The mean annual rainfall of

the basin is 663 mm, with mean kharif, rabi, and summerrainfall of 599, 27.4, and 36.4 mm, respectively. The maxi-mum annual rainfall of 1350 mm is recorded in 2010 atDoulatabad and minimum of 240 mm in 2004 at Ghanpur.The highest mean annual rainfalls are observed atMahabubnagar (803 mm), Kulkacherla (792 mm), Atmakur(785 mm), and Doulatabad (757 mm) meteorological stations,whereas the lowest mean annual rainfalls are observed atMaddur (501 mm) , Ko i lkonda (555 mm) , andPeddamandadi (561 tm) stations. The lowest mean rainfallsof the basin which resulted in severe droughts are recorded in1994, 1997, 1999, 2002, and 2004 (Gaur et al. 2007). Otherlow rainfall years that resulted in various levels of droughtinclude 1992, 2001, 2003, 2006, 2008, 2011, and 2012 (referto Table 1). Themean annual rainfalls at all the meteorologicalstations are quiet variable and irregular as compared to meanannual rainfall of the basin. For instance, in 2010, Doulatabadstation recorded the highest annual rainfall of 1350 mm, butother stations located at Ghanpur (499 mm), Gopalpet(500 mm), Peddamandadi (568 mm), Gandeed (481 mm),and Maddur (573 mm) have received very less rainfall

Fig. 3 Temporal variation of annual and kharif RAI of different stations

Arab J Geosci

Author's personal copy

Page 9: Arabian journal paper - autor copy final

accounting severe drought. Furthermore, the descriptive sta-tistics like coefficients of variation (CV), skewness (CS), andkurtosis (CK) are presented in Table 1; these coefficients arehighly variable from one station to another and infer highannual rainfall variability among stations. The positive skew-ness and kurtosis values indicate frequency of low precipita-tions at Addakal and Koilkonda stations. A comparative

assessment of the probability distribution of rainfall was car-ried out using Kolmogorov-Smirnov test hypothesis, and thedistribution with least p value is chosen as best fit. It isobserved that Gumbel distribution is the best fit for ninestations while exponential distribution is the least suitable fitfor all stations. The details of the test data are provided asadditional data in Table 6 and from Figs. 13 and 14.

Fig. 4 Temporal variation of annual and kharif RAI of different stations

Fig. 5 Temporal variation of annual and kharif RAI of Peddamandadi and Wanaparthy stations

Arab J Geosci

Author's personal copy

Page 10: Arabian journal paper - autor copy final

Tab

le4

Distributionof

occurrencesof

droughteventsunderdifferentcategories

Drought

severity

(range)

Extremedrought(<−3

)Severe

drought(−2

.1to

−3)

Medium

drought(−1

.2to

−2.1)

Station\scale

Annual

Kharif

Annual

Kharif

Annual

Kharif

Addakal

1999,2002,2004

1997,1999,2002,2004,2006

1994,2003

1992,1994,2008

1992,1997,2001,2008,

2011

2003,2011

Atm

akur

1997,1999,2003,2004

1997,1999,2002,2003,2004,

2006

2002

2008

1996,2006

Bhoothpur

1997,1999,2004,2007

1997,2004,2006,

1994,1996,2002

1994,1999,2002,2012

1988,2001,2003,2006

1988,1996,2007,2008

C.C.Kunta

1986,1999,2004,2011,

2012

1986,1992,1999,2004,2006,2012

1992,1997,2013

2011

1994,2002

1987,1997,2008,2013

Devarkadra

1997,2004,2011,2012

1997,2004,2006,2008,2012

1999,2007,2008

2007,2011

1996,2000,2006

Dhanw

ada

1997,1999,2004,2009

1997,1999,2002,2004,2006

2002,2004

2009

2008

Doulatabad

1999,2004,2012,2013

2004,2006,2008,2012

2001,2003

1997,1999,2013

1997,2000,2002

2000,2001,2002,2003

Gandeed

1986,2001,2002,2004,

2007

2002,2004,2006,2007,2008

2006,2010

1986,2001

1994,1999,2008

2010

Ghanpur

1986,1997,2004

1986,1997,2004

1999,2002,2005,2011,2012

1994,1999,2002,2011,

2012

1994,2010

1992,2005,2008

Gopalpet

1986,1999,2004,2008

1986,1999,2004,2008,2011

2011

1992,2001,2010,2012

1987,1992,2006,2010,

2012

Hanwada

1999,2004,2008

1997,1999,2004,2006,2008

1996,1997

2001,2006,2007

2001,2007

Kodangal

1986,1994,2004,2013

1986,1992,1994,2004,2013

1992,1999,2002

2006

1993,1997,2000,2003

1997,1999,2002,2008

Koilkonda

1997,2002,2004,2006,

2008

1997,2006,2008

1999,2009

2002,2004

2012

1999,2009,2011,2012

Kosgi

1993,1994,2002,2006

1993,1994,2002,2006

1997,1999,2007

1992,1997,2007

1992

1987,1999,2008

Kothakota

1992,1997,1999,2004

1992,1997,1999,2004,2006,

2008

1994,2002,2008

1994,2002,2007

2003,2007,2012

2011,2012

Kulkacherla

1986,1994,2002,2004

1986,1994,2002,2004,2008

1997

1997

1992

1992,1999,2006

Maddur

1999,2003,2004,2005,

2006

2003,2004,2006,2008

2000

2002,2005

2002

1999,2000

Mahabubnagar

1996,1997,1999,2004

1997,1999,2004,2006

1996

2006

2007

Peddam

andadi

1994,1999,2003,2011

1994,1999,2008

2004,2008,2012

2003,2004,2006,2011,

2012

2002,2009

1997

Wanaparthy

1986,1997,1999,2004

1986,1987,1997,1999,2004,2008

1987,1992,2008,2012

1992

2002

1993,2011,2012

Arab J Geosci

Author's personal copy

Page 11: Arabian journal paper - autor copy final

Mean monthly rainfall distribution

The mean monthly rainfall distribution over the basin is quitevariable from station to station as shown in Fig. 2. During themonths of January and February, most of the regions arecompletely dry without any rainfall. Monthly rainfall slightlyincreases from March with nil or very less rainfall duringJanuary and February. The rainfall increased gradually fromMay to October, and again, there is a decrease in rainfall fromNovember to December. The monthly average rainfall account-ing nearly 90 % is recorded mainly during June to October(Kharif season) of the year at various stations. Major rainfall isfrom south-west monsoon with August and September being thepeak months of rainfall. The lowest monthly rainfall is observedfrom January toMay andNovember to December accounting forless than 10 % of annual rainfall. Hence, the months of June toOctober are significant rainfall-contributing months to the annualrainfall. Similar pattern of monthly rainfall distribution is ob-served in other locations (Kwarteng et al. 2009; Ankegowdaet al. 2010). The descriptive statistics like mean, median, maxi-mum, minimum, standard deviation (SD), coefficients of varia-tion (CV), skewness (CS), and kurtosis (CK) of monthly rainfall(June–October) for lowest mean rainfall region and highest meanannual rainfall are presented in Table 3. The coefficients of

Table 5 Occurrences of number of drought events under different categories

Drought severity (range) Number of extremedrought (<−3)

Number of severedrought (−2.1 to −3)

Number of mediumdrought (−1.2 to −2.1)

Total number ofdrought events (n)

Frequency=n/N

Station\scale Annual Kharif Annual Kharif Annual Kharif Annual Kharif Annual Kharif

Addakal 3 5 2 3 5 2 10 10 0.43 0.43

Atmakur 4 6 1 1 2 0 7 7 0.39 0.39

Bhoothpur 4 3 3 4 4 4 11 11 0.42 0.42

C.C.Kunta 5 6 3 1 2 4 10 11 0.36 0.39

Devarkadra 4 5 3 2 3 0 10 7 0.56 0.39

Dhanwada 4 5 2 1 0 1 6 7 0.33 0.39

Doulatabad 4 4 2 3 3 4 9 11 0.50 0.61

Gandeed 5 5 2 2 3 1 10 8 0.36 0.29

Ghanpur 3 3 5 5 2 3 10 11 0.36 0.39

Gopalpet 4 5 1 0 4 5 9 10 0.32 0.36

Hanwada 3 5 2 0 3 2 8 7 0.44 0.39

Kodangal 4 5 3 1 4 4 11 10 0.39 0.36

Koilkonda 5 3 2 2 1 4 8 9 0.44 0.50

Kosgi 4 4 3 3 1 3 8 10 0.30 0.37

Kothakota 4 6 3 3 3 2 10 11 0.43 0.48

Kulkacherla 4 5 1 1 1 3 6 9 0.21 0.32

Maddur 5 4 1 2 1 2 7 8 0.39 0.44

Mahabubnagar 4 4 0 1 1 1 5 6 0.28 0.33

Peddamandadi 4 3 3 5 2 1 9 9 0.41 0.41

Wanaparthy 4 6 4 1 1 3 9 10 0.32 0.36

Table 6 Rainfall distri-bution details Station Distribution

1 Addakal Gumble

2 Atmakur Rayleigh

3 Bhoothpur Gumble

4 Chinnachintakunta Log normal

5 Devarkadra Gumble

6 Dhanwada Gumble

7 Doulatabad Rayleigh

8 Gandeed Gumble

9 Ghanpur Weibull

10 Gopalpet Gumble

11 Hanwada Log normal

12 Kodangal Gamma

13 Koilkonda Gumble

14 Kosgi Rayleigh

15 Kothakota Log normal

16 Kulkacherla Gamma

17 Maddur Rayleigh

18 Mahabubnagar Log normal

19 Peddamandadi Gumble

20 Wanaparthy Gumble

Arab J Geosci

Author's personal copy

Page 12: Arabian journal paper - autor copy final

variation show that rainfall distribution is quite variable from onemonth to another in both regions. The positive skewness andkurtosis values of monthly rainfall indicate that low precipitationfrequency is observed in June and September at Maddur stationwhile only June in Mahabubnagar station. It can be inferred thatregions which have good distribution of monthly rainfall likeMahabubnagar have comparatively better annual rainfall as 90%rainfall occurs in kharif season.

Temporal drought severity assessment using RAI

The RAI values were computed for kharif (June–October) andannual time scales for all the 20 meteorological stations of thebasin. The time series of RAIs computed for the two kharifand annual scales are depicted in Figs. 3, 4, and 5. From the

figures, it is evident that annual and kharif rainfall varied at allstations with time. In the RAI time series, the positiveranges of RAI correspond to wet periods while thenegative ranges correspond to dry periods, i.e.,droughts. The RAI time series show that different re-gions experienced varying magnitudes (severity) ofdroughts over the time. Based on the magnitude rangesof RAI as shown in Table 2, the droughts are classifiedinto extreme, severe, and medium droughts. Table 4shows the occurrences of drought under categories ex-treme, severe, and medium drought at different meteo-rological stations. The annual and kharif RAI rangesless than −3 are categorized as extreme droughts, −3to −2.1 as severe droughts, and −2.1 to −1.2 as mediumdroughts. Visual interpretation of annual and kharif RAI

Fig. 6 Spatial variation of annual (top) and kharif (bottom) RAI interpolated drought severity, 1996 to 1998

Arab J Geosci

Author's personal copy

Page 13: Arabian journal paper - autor copy final

time series shows that both have the same pattern, butthe magnitude varied over the time at all the stations.From the interpretation of graphs as well as fromTable 4, it is evident that all the regions experiencedextreme droughts of more than −3 magnitude in 1997,1999, 2002, 2004, 2006, and 2008; besides this,Chinnachintakunta, Peddamandadi, Doulatabad, andDevarkadra regions have experienced extreme droughtsbetween 2011 and 2013. Moreover, all the regions havealso experienced either severe or medium droughts inother years, and the details of occurrences of thesedroughts are shown in Table 4.

Furthermore, Table 5 shows a matrix of number of droughtevents that occurred under different drought categories at allthe stations. On the whole, there are around 3 to 6 drought

events under extreme drought category, while the occurrencesof number of drought events under severe and medium haveranged from 0 to 5 at all the station. Comparison of annual andkharif time series shows that kharif time scale experiencedmore extreme and medium category drought events, whereasannual time scale experienced more severe droughts. It isobserved that Addakal, Bhoothpur, Chinnachintakunta,Ghanpur, Gopalpet, Kodangal, Kothakota, and Wanaparthyregions experienced more number of drought events thanother regions. The frequency of droughts gives informationabout the recurrence of a drought at a station over time.The frequency of droughts is estimated by dividing thetotal drought events with observation period (N) of thestation and presented in Tables 5 and 6. Frequenciesvaried from 0.21 to 0.56 per year in annual time scale

Fig. 7 Spatial variation of annual (top) and kharif (bottom) RAI interpolated drought severity, 1999 to 2001

Arab J Geosci

Author's personal copy

Page 14: Arabian journal paper - autor copy final

and 0.29 to 0.61 per year for kharif time scale. It can beinferred that Doulatabad, Devarkadra, and Koilkonda sta-tions experience more frequently than other regions. Someregions which experienced most vicious droughts of mag-nitude ranging between −9 and −6 are in 1997 atDhanwada ; 1999 a t Addaka l , Hanwada , andPeddamandadi; 2002 at Kosgi and Kulkacherla; 2004 atMaddur, Mahabubnagar, Kodangal, and Kulkacherla; and2006 at Devarkadra and Koilkonda stations. It is observedthat in many regions, there has been recurrence of droughtevents every 2 or 3 years. The above results show that thedroughts varied in space with time in the basin; as theregion is solely dependent on agriculture for livelihood, aspatial assessment of drought extent and recurrence isenvisaged to provide better understanding about regionaldrought development.

Spatial drought severity assessment spline interpolation RAIvalues

Spatial variation of drought severity maps are derived fromthe interpolated RAI time series data using spline interpolationmethod in ArcGIS. Spatial drought severity maps were gen-erated for kharif and annual RAI time series data only from1996 to 2013 as the time series data for all the station isavailable from 1996. Drought severity of kharif and annualtime series is classified into nine categories as per Table 2(Keyantash and Dracup 2002; Roshan et al. 2012) and arepresented in Figs. 6, 7, 8, 9, 10, and 11. The tone of droughtseverity are categorized as follows: red indicates extremedroughts, orange indicates severe drought, yellow indicatesmoderate drought, and light yellow indicates weak drought,while light blue to dark blue indicates varying wet periods.

Fig. 8 Spatial variation of annual (top) and kharif (bottom) RAI interpolated drought severity, 2002 to 2004

Arab J Geosci

Author's personal copy

Page 15: Arabian journal paper - autor copy final

The colors indicate that red to yellow color regions wouldsuffer varying levels of droughts, while light blue to dark bluewould be comparatively wet.

From the visual interpretations of time series of RAI maps(Figs. 6, 7, 8, 9, 10, and 11), it is obvious that the droughtseverity varied in space and time from 1996 to 2013.Moreover, the visual observation of these maps show thatthe spatial extent and pattern of drought severity categoriesvaried in annual and kharif timescales to a certain extent dueto differences of magnitudes of two RAI time scales. From thevisual observation of maps, years 1997, 1999, 2002, 2004,2006, and 2008 are identified to be extreme drought years andyears 2011 and 2012 as severe droughts. Similarly, the years2001 to 2004 have been reported as critical drought years inlower Krishna basin by Gaur et al. (2007), and the reason forthe droughts has been associated with the failure of Indian

monsoon system due to the El Niño phenomenon (Saithand Slingo 2006; Selvaraju 2003). Even during 1996,2001, 2003, 2007, 2009, and 2013, some regions haveexperienced severe to extreme droughts. From the maps,it is also evident that even during the years 1998, 2000,2005, and 2010 when the rainfall is above mean rainfallof the basin, some regions experienced medium to se-vere droughts. The drought severity maps show a lot ofvariation both in space and time from one region toanother as well as in the recurrence of droughts.

The study area is classified into southern Telangana (700 –900 mm) and scarce rainfall (500–700) agro-climatic zonesbased on rainfall (Valli et al. 2013). Atmakur, Doulatabad,Kodangal, Kulkacherla, Mahabubnagar, and Wanaparthy re-gions are categorized under the southern Telangana agro-climatic zone; these regions experienced droughts only during

Fig. 9 Spatial variation of annual (top) and kharif (bottom) RAI interpolated drought severity, 2005 to 2007

Arab J Geosci

Author's personal copy

Page 16: Arabian journal paper - autor copy final

critical drought years like 2002 and 2004. Hence, these re-gions are comparatively less prone to drought than the scarcerainfall zones. The remaining 15 regions are categorized underscarce rainfall zone; among these 15 stations, Maddur,Koilkonda, Peddamandadi, Ghanpur, Goplapet, andChinnachintakunta regions are more prone to droughts. Thedrought severity of both kharif and annual time scales showedvariations in the north, the center, and the south due to eitherchanges in the topography. The rainfall decreased from northto center and to the south, similarly the drought severityincreased from north to center and towards south. It impliesthat droughts increase with decrease in rainfall. The stationsDoulatabad, Kodangal, Kulkacherla, Mahabubnagar, andGandeed regions located in higher-elevation regions havereceived better rainfall and experienced less drought events.

Central and lower part of the basin which is characterized withlower elevations than the upper basin received lower rainfalland is prone to more droughts. On the whole, spatial droughtassessment provides regional assessment and characterizationof drought events only, but the estimation of magnitude andlength of drought events are crucial for water resource assess-ment and micro-level planning. For this purpose, run analysisis carried out for objective assessment of droughts magnitudeand duration.

Objective assessment of drought magnitude and durationusing run analysis

Run analysis is used to study magnitude, duration, and inten-sity of droughts in the present study. The mean rainy days of

Fig. 10 Spatial variation of annual (top) and kharif (bottom) RAI interpolated drought severity, 2008 to 2010

Arab J Geosci

Author's personal copy

Page 17: Arabian journal paper - autor copy final

the basin was observed to be 41 days. The threshold value of16 mm for the basin was obtained by dividing mean annualrainfall of 663 mm with mean rainfall days. Based on thethreshold value and rainfall, the water deficit and surplusrainfall of region is determined for kharif season, i.e., Juneto October only as 90 % of the rainfall occurs during thisperiod. The maximum duration of dry spells was 70 daysobserved at Koilkonda and Ghanpur stations; consecutively,63 days dry spells were observed at Atmakur, Devarakadra,and Peddamandadi stations. Other dry spells of 56 days werefound at Chinnachintakunta, Kothakota, Peddamandadi, andGopalpet.

Additionally, it is observed that 42- and 35-day dry spellsare very common at all the stations in the basin. The years withmore number of dry spells during kharif season have resulted

in severe drought years like 1999, 2002, 2004, 2006, and2008. The maximum drought magnitude of 126 mm wasobserved at Ghanpur in 2009. Drought magnitudes of morethan 80 mm occurred mostly during drought years 2002,2004, and 2006. The empirical relationships developed be-tween drought magnitude and duration are presented usingscatter plots shown in Fig. 12. Such objective assessment ofpast drought characteristics using statistical relationships pro-vides valuable insights for what-if analysis of present andfuture assessment of drought magnitude and helps in waterresource planning. Using these relationships, one can find thewater required for particular drought duration and can planwater resource allocation accordingly from the water storedduring wet season or by transferring water from other catch-ments (Figs. 13 and 14).

Fig. 11 Spatial variation of annual (top) and kharif (bottom) RAI interpolated drought severity, 2011 to 2013

Arab J Geosci

Author's personal copy

Page 18: Arabian journal paper - autor copy final

Conclusions

In this paper, the assessment of duration, magnitude, andtemporal and spatial drought occurrence of Pedda Vagu andOokacheti Vagu watersheds is presented. The spatial andtemporal characterization of droughts achieved can be usedin highlighting drought-prone regions or vulnerable regions.The identification of highly vulnerable regions helps in plan-ning drought-proofing measures. The drought events werecharacterized by using rainfall anomaly index (RAI) appliedto kharif and annual time scales using monthly precipitationdata of 20 meteorological stations. Spline interpolation tech-nique is used in identifying the spatial pattern of the RAI timeseries. The study showed that the drought severity of the twotime scales varied at all the regions, north to south of the basin.

Years 1997, 1999, 2002, 2004, 2006, 2008, and 2011 haveshowed high negative RAI values which varied in space and

time. It is also observed that droughts occur after every 2 or3 years at all stations with varying frequencies. It is alsoobserved that the central part of the basin is more prone todroughts. The main reason for the variation in the droughtseverity is attributed to the topographical variation and varia-tion in the rainfall abnormalities. From the time series plots ofRAI, it is observed that drought of either extreme or severe ormedium severity occurs every 2 or 3 years at all stations. Thespline interpolated rainfall anomaly index maps are useful forregionalization of droughts which can be of high value inclassification of drought-prone areas as well as in planningmanagement measures. They also have shown that someregions experienced more severe drought while other regionswere well-off. Furthermore, run analysis is employed on dailyrainfall for developing drought duration and magnitude em-pirical relationships. These empirical relations are useful indetermining the water deficits based on duration during a

Fig. 12 Empirical relationships of drought magnitude and duration of some stations in the basin

Arab J Geosci

Author's personal copy

Page 19: Arabian journal paper - autor copy final

Fig. 13 Cumulative distribution function of different stations

Arab J Geosci

Author's personal copy

Page 20: Arabian journal paper - autor copy final

Fig. 14 Cumulative Distribution Function of different stations

Arab J Geosci

Author's personal copy

Page 21: Arabian journal paper - autor copy final

particular drought event and eventually help in planning thewater resources.

Acknowledgments The authors are grateful to Indian Institute of Tech-nology Bombay, for the support and encouragement. The authors are alsograteful to District planning office, Mahabubnagar District, Regionalagriculture research station, Hyderabad and Directorate of Economicsand Statistics, Hyderabad for providing rainfall data. The authors are alsothankful to the valuable suggestions provided by the two anonymousreviewers for the improvement of the manuscript.

References

Ankegowda SJ, Kandiannan K, Venugopal MN (2010) Rainfall andtemperature trends—a tool for crop planning. J Plant Crop 38(1):57–61

Biggs TW, Gaur A, Scott CA, Thenkabail P, Parthasaradhi G, GummaMK, Acharya S, Turral H (2007) Closing of the Krishna Basin:stream flow depletion and macro scale hydrology. IWMI ResearchReport No. 111, Colombo

Gaur A, McCornick PG, Turral H, Acharya S (2007) Implications ofdrought and water regulation in the Krishna Basin, India. Int J WaterResour Dev 23(4):583–594

Hansel S, Matschullat J (2006) Drought in a changing climate, Saxon dryperiods. Bioclimatological Conference 2006. Bioclimatology andwater in the land. International scientific conference, 11–14September 2006, Strecno

Hartkamp AD, De Beurs K, Stein A, White JW (1999) Interpolationtechniques for climate variables. NRG-GIS Series 99–01.CIMMYT, Mexico

Hutchinson MF, Gessler PE (1994) Splines more than just a smoothinterpolator. Geoderma 62:45–67

Iglesias A, Garrote L, Cancelliere A, Cubillo F, Wilhite AD (2009) Copingwith drought risk in agriculture and water supply systems, droughtmanagement and policy development in the Mediterranean, Advancesin Natural and Technological Hazards Research, Volume 26

Keyantash J, Dracup JA (2002) The quantification of drought: an evalu-ation of drought indices. Bull Am Meteorol Soc 1167–1180

Kwarteng AY, Dorvlo AS, Vijaya Kumar GT (2009) Analysis of a 27-year rainfall data (1977–2003) in the Sultanate of Oman. Int JClimatol 29(4):605–617

Loukas A, Vasiliades L, Dalezios N R (2003) Intercomparison of mete-orological drought indices for drought assessment and monitoring inGreece. 8th International Conference on Environmental Science andTechnology, Lemnos island, Greece, 8–10, Sept, 2003

Mishra S S, Nagarajan R (2011) Drought assessment in tel watershed: anintegrated approach using run analysis and SPI. Earthzine (IEEE)

Murali Krishna T, Ravikumar G, Krishnaveni M (2009) Remote sensingbased agricultural drought assessment in Palar Basin of Tamil NaduState, India. J Indian Soc Remote Sens 37:9–20

Oladipo EO (1985) A comparative performance analysis of three mete-orological drought indices. J Climatol 5:655–664

Prabhakar SVRK, Shaw R (2008) Climate change adaptation implica-tions for drought risk mitigation: a perspective for India. ClimateChange 88:113–130

Rao B, Sreenivas Prasad P, Ahmed Iftekhar S (1998) Watershed manage-ment and consequential conservation/augmentation to groundwaterresources. In Proceedings of International Conference onWatershedmanagement and Conservation 259–268

Roshan G, Mirkatouli G, Ali S (2012) A new approach to technique fororder-preference by similarity to ideal solution (TOPSIS)method fordetermining and ranking drought: a case study of Shiraz station. Int JPhys Sci 7(23):2994–3008. doi:10.5897/IJPS12.308

Saith N, Slingo J (2006) The role of the Midden-Julian Oscillation in theEl Nino and Indian drought of 2002. Int J Climatol 26:1361–1378

Selvaraju R (2003) Impact of El Nino-Southern Oscillation on Indianfoodgrain production. Int J Climatol 23:187–206

Sirdas S, Sen Z (2003) Spatio-temporal drought analysis in the Trakyaregion, Turkey. Hydrol Sci J 48(5)

Sreedhar G, Mishra S, Nagarajan R, Balaji V (2012) Micro-level droughtvulnerability assessment in Peddavagu basin, a Tributary of KrishnaRiver, Andhra Pradesh, India. Earthzine (IEEE)

Sreedhar G, Nagarajan R, Balaji V (2013) Village-level drought vulner-ability assessment using geographic information systems (GIS). Int JAdv Res Comp Sci Softw Eng 3(3)

Srivastava SK, Upadhyay AP, Sahu AK, Dubey AK (2000) Rainfallcharacteristics and rainfall based cropping strategy for Jabalputrregion. J Soil Conserv 28(3):204–211

Subudhi CR, Pradhan PC, Behara B, Senapati PC, Singh GS (1996)Rainfall characteristics at Phulani. Indian J Soil Conserv 24(1):41–43

Suresh R, Singh NK, Prasad P (1993) Rainfall analysis for drought studyat Pusa, Bihar. Indian J Agric Eng 3(1–2):77–82

Thenkabail PS, Gamage MSDN, Smakhtin VU (2004) The use of remotesensing data for drought assessment and monitoring in south westAsia. Research report. 85. International Water ManagementInstitute, Sri Lanka

Tilahun K (2006) Analysis of rainfall climate and evapo-transpiration inarid and semi-arid regions of Ethiopia using data over the last half acentury. J Arid Environ 64:474–487

Valli M, Shanti Sree K, Murali Krishna IV (2013) Analysis of precipita-tion concentration index and rainfall prediction in various agro-climatic zones of Andhra Pradesh, India. Int Res J Environ Sci2(5):53–61

van Rooy MP (1965) A rainfall anomaly index independent of time andspace. Notos 14:43–48

Yevjevich V (1967) An objective approach to definition and investigationof continental hydrologic droughts. Colorado State Univ., FortCollins, Colorado, USA. Hydrology Paper 23

Arab J Geosci

Author's personal copy