researcharticle spi-based spatiotemporal drought over sri

Research ArticleSPI-Based Spatiotemporal Drought over Sri Lanka

N S Abeysingha and U R L N Rajapaksha

Department of Agriculture Engineering and Soil Science Faculty of Agriculture Rajarata University of Sri LankaAnuradhapura Sri Lanka

Correspondence should be addressed to N S Abeysingha nabeysinghagmailcom

Received 23 October 2019 Revised 7 December 2019 Accepted 12 December 2019 Published 21 January 2020

Academic Editor Gabriele Buttafuoco

Copyright copy 2020 N S Abeysingha and U R L N Rajapaksha is is an open access article distributed under the CreativeCommons Attribution License which permits unrestricted use distribution and reproduction in any medium provided theoriginal work is properly cited

Drought is one of the most significant hazards in Sri Lanka Status of drought in Sri Lanka was assessed using StandardizedPrecipitation Index (SPI) at 3 6 and 12 monthsrsquo time scales using monthly rainfall (1970 to 2017) data of 54 weather stations efrequency of drought events was evaluated using SPI and trend of SPI was also detected using the MannndashKendall (MK) test andSenrsquos slope estimator e result based on SPI identified hydrological years 1975-76 1982-83 1986-87 1988-89 2000-01 2001-022013-14 and 2016-17 as drought years for 52 32 35 33 33 31 31 and 31 of tested stations (54) respectively at annual timescale Comparison of the SPI at different time scales revealed that more drought events (SPIle minus 1) occurred during Yala seasonthanMaha cropping season Considering theiessen polygon average rainfall more frequent drought events occurred in the dryzone (57) than the wet (49) and intermediate zone (47) at the annual time scale SPI trend results showed greater increase indrought (59 of stations) during Yala seasons as compared to the Maha cropping season (15 of stations) in Sri Lanka

1 Introduction

Drought is one of the natural disasters which can causehuge damage to agriculture and other economic and socialactivities of the human system and considerable damage isalso caused to the ecosystem It fundamentally occurs as aresult of weather extremes that are driven by naturalvariability and climate change and also stimulated byanthropogenic influences [1] Drought occurs naturally innearly all climatic zones as a result of the reduction ofprecipitation from normal amounts for an extended periodof time [2]

Drought is categorized into four types namely me-teorological agricultural hydrological and socioeconomicdroughts [3] Meteorological droughts are mainly deter-mined on the basis of the degree of dryness in comparisonto some normal quantity and the duration of the dryperiod

Drought indices are a helpful tool for monitoring andevaluating the different kinds of drought since they facilitatecommunication of climate anomalies to numerous user

audiences Many indices have been developed to identify andquantify drought events based on different types of variables[4] e Standardized Precipitation Index (SPI) is one of themost applied indices to analyze meteorological drought [5]as most other indices demand a number of other parametersother than rainfall SPI was developed by McKee et al [6]primarily for defining and monitoring droughts at differenttime scales e main advantage of the application of thisindex is its versatility SPI uses only rainfall data to deliverfive major dimensions of a drought ie duration intensityseverity magnitude and frequency Also based on SPIdrought can be calculated at different time scales [3] and itcan also be considered as an agricultural drought indicator[7] A number of published studies are available in theliterature on the use of SPI in assessing drought intensity inmany countries [4 8 9] However application of SPI toassess the drought during recent past over entire Sri Lankahas not been reported in the literature to the best of authorsrsquoknowledge

Sri Lanka is an island at the southern tip of India with aland area of nearly 65610 km2 and population of 214 million

HindawiAdvances in MeteorologyVolume 2020 Article ID 9753279 10 pageshttpsdoiorg10115520209753279

[10] and is divided into three climatic zones ie dry wetand intermediate based on the total annual rainfall [11]According to the Global Climate Risk Index 2019 Sri Lankabecame the most affected countries along with Puerto Ricoand Dominica [12] us Sri Lanka is extremely vulnerableto climate change impacts In terms of people affected andrelief provided drought has been the most significant hazardin Sri Lanka and Sri Lanka also serves as a case study intropical regions for analysis of drought hazard and riskassessment [13] Drought has affected many parts of thecountry intermittently as one of the serious natural hazardsin Sri Lanka [14] It is reported that 2001 drought severelyaffected dry and intermediate zone of the country whileHambantota area experienced prolonged severe droughtduring 2001 to 2002 [15] Prolonged drought occurred in2001ndash02 led to 1 drop in the GDP growth rate in thecountry by particularly affecting hydropower generation andagriculture sector [13]

Drought studies in Sri Lanka so far have particularlyfocused on analysis of the spatial variations [16 17] severityand duration of drought mostly in part of the country [18]and spatial extent and temporal evolution of drought [19]Most of these studies are in relation to dry zone areas of SriLanka However Lyon et al [13] studied the relationshipbetween SPI and drought relief payments in the country andfound the strongest relationships with a 9-month cumulativedrought index Herath et al [20] analyzed monthly rainfalldata using SPI to identify possible drought conditions onlyfor the year 2015

Moreover it is rather rare to find drought trend analysisover entire Sri Lanka It is worth to investigate the trend ofdrought over the entire country specially for drought mit-igation and agricultural planning It is also projected thatmost of the districts in the dry zone in Sri Lanka will facesevere seasonal or year-round water scarcity by 2025 withpresent level of irrigation efficiency [21] Hence this paperanalyzes the severity frequency and trend of meteorologicaldrought over Sri Lanka during the recent past (1970 to 2017)using 48 years of rainfall data at 54 stations

2 Material and Methods

21 Description of Data Monthly precipitation data of 48years (1970ndash2017) were collected from 54 meteorologicalstations (Table 1) from theMeteorological Department of SriLanka e Meteorological Department of Sri Lanka is thenodal agency in Sri Lanka to record quality check andarchive all meteorological data e monthly rainfall valuesused in the study were prepared by the MeteorologicalDepartment of Sri Lanka and data quality is maintained bythem [22 23]

Although there are only 54 stations used in this studythe stations are well distributed across the three climaticzones ie dry intermediate and wet zones of Sri Lanka(Figure 1) and hence adequately showed the climatic spatialvariability in the country ere were missing values in fewstations and Table 1shows the missing values as a per-centage and missing values were filled using the regressionand normal ratio method

22 Meteorological Drought Analysis Standardized Precip-itation Index (SPI) was used to evaluate the drought both theshort-term (3 and 6 months) and the long-term (12 months)time scales In order to calculate the SPI index for each timescale the variability of precipitation totals is described by agamma distribution and then transformed to a normaldistribution [6] e gamma distribution is defined by itsfrequency or probability density function

g(x) 1

βaΓ(a)x

aminus 1e

minus (xβ) forxgt 0 (1)

where a and β are the shape and scale parameters respec-tively x is the precipitation amount and Γ(a) is the gammafunction Maximum likelihood estimations of a and β are

a 14A

1 +

1 +4A

3

1113970

1113890 1113891

β x

a whereA (x) minus

1113936 ln(x)

n

(2)

where n number of observationse subsequent parameters are then used to find the

cumulative probability of a recorded precipitation amountfor the given month and time scale for the given locationSince the gamma function is undefined for x 0 and aprecipitation distribution may contain zeros the cumulativeprobability is computed as

H(x) q +(1 minus q)G(x) (3)

where q is the probability of zero precipitation and G(x) isthe cumulative probability of the incomplete gammafunction e cumulative probability H(x) is then trans-formed to the standard normal random variable z with meanand variance of zero and one respectively which is the valueof the SPI [24]

Analysis was done in different time sequence and timeduration October to September was used as the hydrologicalyear (SPI12) and October to March and April to Septemberwere used as 6-month time scale (SPI6) as these periods arecropping seasons of Maha and Yala respectively in SriLanka October to December January to March April toJune and July to September were used as 3-month time scale(SPI3) Calculated SPI values were compared with the SPIclassification values for drynesswetness category to indicatethe status of the drought (Table 2)

23 Drought Frequency Analysis Drought frequencies werecalculated for dry wet and intermediate zones separatelyFor this purpose average rainfall for three climate zoneswas calculated separately using the iessen polygonmethod first en these average rainfall was used incomputing SPI for three climatic zones separately and theresulted SPI was used in frequency calculation For thisfrequency calculation the drought was defined as SPI lt 0[25] Also the frequency of occurrence of drought eventsmore or less than 50 was calculated based on the cal-culated SPI in each station

2 Advances in Meteorology

24 StatisticalMethods for Trend Analysis In order to detectthe drought trend using SPI as drought indicator over SriLanka we used the most commonly used nonparametric

MannndashKendall (MK) test [26 27] together with Senrsquos slopeestimator e MannndashKendall test statistic S can beexpressed as

Table 1 Geographical coordinates of meteorological stations used for the analysis

No Name of the station Latitude Longitude Missing value ()Dry zone1 Allai 84 8132 482 Ampara 728 8167 123 Angamedilla 785 8092 214 Anuradhapura 835 8038 05 Bakamuna 777 8082 126 Batticaloa 772 817 07 Diyabeduma 793 8087 118 Hambanthota 612 8113 029 Hingurakgoda 805 8095 4410 Iranamadu 935 804 4111 Jaffna 968 8003 1412 Kalawewa 8 8053 2713 Kannukkeni 92 808 6614 Kanthale 835 8098 1615 Mahailuppallama 812 8047 1616 Mannar 898 7992 017 Maradankadawala 813 8057 3918 Medawachchiya 855 8048 2119 Minneriya 805 809 3520 Murunkan 883 8005 0921 Pavatkulam 868 8043 3222 Pothuvil 688 8183 0923 Puttalam 803 7983 0224 Tissamaharama 628 813 1625 Trincomalee 858 8125 14Intermediate zone26 Badulla 698 8105 027 Bandarawela 683 8098 028 Chilaw 758 7978 1629 Dandeniya 6 8065 1430 Kurunegala 747 8037 031 Nikaweratiya 775 8012 9632 Wellawaya 673 811 21Wet zone33 Alupolla 672 8058 0434 Ambepussa 728 8017 1235 Avissawella 692 8018 3736 Chesterford 707 8018 4337 Colombo 69 7987 038 Deniyaya 635 8062 1839 Galle 603 8022 040 Galthura 67 8028 1641 Geekiyanakanda 66 8012 1642 Halwathura 672 802 1143 Hanwella 688 8012 0444 Henarathgoda 71 7998 1445 Kaluthara 658 7995 3446 Katugastota 733 8063 047 Katunayaka 717 7988 048 Maliboda 688 8043 7149 Matale 747 8062 0250 Mawarella 62 8058 4451 Nuwara Eliya 697 8077 052 Pasyala 715 8013 1153 Peradeniya 727 806 7954 Ratnapura 668 804 0

Advances in Meteorology 3

S 1113944

nminus 1

i11113944

n

ji+1sign Xj minus Xi1113872 1113873

sign Xj + Xi1113872 1113873

+1 Xj ltXi1113872 1113873

0 Xj Xi1113872 1113873

minus 1 Xj ltXi1113872 1113873

⎧⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎩

(4)

where n is the number of data and x is the data point at timesi and j (jgt i) e variance of S is as follows

var(S) n (n minus 1)(2n + 5) minus 1113936

mi tii (i minus 1)(2i + 5)1113858 1113859

18 (5)

where ti is the number of ties of extent i and m is the numberof tied groups for n larger than 10

e MannndashKendall test would moreover confirm theexistence of a positive or negative trend for a confidence levelof 005

Finally Senrsquos method [28] was applied to estimate thetrue slope of a linear trend To derive an estimate of the slopeQ the slopes all data pairs are calculated as follows

Qi xj minus xk1113872 1113873

j minus k i 1 2 N jgt k (6)

where xj and xk are data values at time j and ke medianof the N values of Qi is Senrsquos estimator of the slope [29]

However for autocorrelated data the modifiedMannndashKendall test by Hamed and Ramachandra Rao [30]was used in this study by checking the lag one autocorre-lation is method is robust in presence of autocorrelationand is based on modified variance of the MK test

As the MK test indicates only the trend directionmagnitude of the trend was detected using Senrsquos slope es-timator If a linear trend is present in a time series then thetrue slope (change per unit time) can be estimated by using asimple nonparametric procedure developed by Sen [28]Senrsquos slope for a monotonic increasing or decreasing time (t)series f(t) is computed as

f(t) Qt + B (7)

where Q is the slope of the trend f(t) and B is the interceptTo determine Q in equation (7) slopes between each datapair are calculated using the following equation

Qi Xj minus Xk1113872 1113873

j minus k⎡⎣ ⎤⎦ where jlt k (8)

If there are n values of xj in the time series then therewill be N n(n minus 1)2 slope estimates of Qi Senrsquos estimatorof slope is the median of these N values of Qi

3 Results and Discussion

31 Meteorological Drought Analysis at Different RainfallStations Using SPI Being an island of 65610 km2 the cli-mate of Sri Lanka is represented well by spatially distributed54 rainfall stations and we assessed drought events in 54meteorological stations using SPI at 3 6 and 12 monthsrsquotime scales using 1970ndash2017 rainfall data

Annual rainfall of each of the 54 stations was subjected toSPI analysis and we identified nine hydrological years 1975-76 1982-83 1986-87 1988-89 2000-01 2001-02 2003-042013-14 and 2016-17 as drought years (SPIlt minus 1) for 52 3235 33 33 31 30 31 and 31 of tested stations respectively(Figure 2) Among them the 1975-76 hydrological yearshowed the maximum number of drought events in differentstations (52 of stations) than other years (Figure 2)

Previous studies showed 2014 [18] and 2001-02 [13] asdrought years in Sri Lanka Impact of disasters in Sri Lankapublished in 2016 [31] reported severe drought events in2001 2004 2012 and 2014us SPI analysis confirms these

Table 2 Drynesswetness category according to SPI value [6]

SPI value ClassSPIge 200 Extremely wet150le SPIlt 200 Severely wet100le SPIlt 150 Moderately wet000le SPIlt 100 Mildly wetminus 100le SPIlt 000 Mild droughtminus 150le SPIlt minus 100 Moderate droughtminus 200le SPIlt minus 150 Severe droughtSPIlt minus 200 Extreme drought

80deg0prime0PrimeE

10deg0prime0Prime

N9deg

0 prime0Prime

N8deg

0prime0Prime

N7deg

0prime0Prime

N6deg

0prime0Prime

N

10deg0prime0Prime

N9deg

0prime0Prime

N8deg

0prime0Prime

N7deg

0prime0Prime

N6deg

0prime0Prime

N

81deg0prime0PrimeE 82deg0prime0PrimeE


0 25 50 100Kilometers

82deg0prime0PrimeE

Wet zoneRain gauge stations

Dry zoneIntermediate zone

N

Figure 1 Geographical distribution of meteorological stationsused in the study


historical drought situations over the country though SPIuses only rainfall as an input

If the SPI values are below minus 200 then those events arereferred to be extreme drought events [6] Accordingly theyear 1975-76 can be considered as extreme drought year foreight out of 54 stations (Figure not shown) ese stationswere Angamedilla (minus 223)Anuradhapura (minus 209)Hanwella(minus 232) Hingurakgoda (minus 244) Mannar (minus 226) Matale(minus 26)Murunkan (minus 203) and Puttalam (minus 226) Analysingthese extreme drought events is of great importance for thedesign and management of water resources systems

Figure 3 shows the occurrence of drought events(SPIle minus 1) during Yala and Maha seasons in Sri Lankaduring 1970 to 2017 as a percentage considering 54 rainfallstations Comparatively more number of stations showedthe occurrence of drought events during Yala than Mahacropping seasons (Figure 3) Moreover there is moderate(minus 150le SPIlt minus 100) severe (minus 200le SPIlt minus 150) or ex-treme drought event (SPIlt minus 200) during Yala season of1970-71 1985-86 1993-94 1999-00 2010-11 and 2012-13hydrological years Main rain source during this time isSouthwest Monsoon (SWM) and frequent occurrence ofdrought events during these seasons indicates the uncer-tainty and variability of rainfall during SWM is studyconfirms that Yala cropping season is more prone todrought thanMaha season In general there is relatively lowrainfall during the Yala season compared to the Mahacropping season [32] in Sri Lanka Furthermore it is shownthat average seasonal rainfall in Kurunegala Anuradhapuraand Polonnaruwa districts during the Yala was in decreasingtrend over 1982 to 2012 period [33] Rainfall trends of theSouthwest Monsoon (SWM) which is the Yala season werealso shown to be deceasing in trend over Sri Lanka during1981 to 2010 [34] ese drying tendencies during Yalaseason may result in more number of drought events even infuture during Yala season However 1973-74 and 1987-88drought events have happened only during Maha season inmost of the stations during which Northeast Monsoonprevails over the country According to the analysis 1988-89Maha season became drought prone by 57 of stationswhile 57 and 50 of stations during 2011-12 and 1975-76respectively were drought prone during Yala seasons in SriLanka It is observed that a few consecutive years showed theoccurrence of drought events during the Yala season such as2000 to 2002 2004 to 2006 and 2010 to 2013 Consecutive

drought events definitely affect the potable water require-ments and the economy of the farming community

Analysis of SPI at 3-months (OctoberndashDecember) timescale (SPI3) revealed drought in 78 43 37 57 31 43 43 57and 67 of tested stations during the hydrological years1974-75 1984-85 1986-87 1988-89 1995-96 2000-012003-04 2013-14 and 2016-17 respectively (Figure 4)Chesterford Colombo and Diyabeduma stations located inwet zone of the country showed more drought events thanother stations during OctoberndashDecember (SPI3) time scaleAs Maha cropping season (September to March) startsduring this time the crop water requirements increasespecially for paddy field preparation Similarly drought in50 70 37 87 76 and 57 of tested stations were foundduring the hydrological years 1971-72 1979-80 1981-821982-83 1991-92 and 1996-97 respectively for SPI atJanuaryndashMarch time scale (Figure 4) e number ofdrought events was higher for both Badulla and Kalutharastations during this period SPI at AprilndashJune time scaleidentified years 1975-76 1982-83 and 1999-00 2011-12 and2016-17 as drought years for 43 39 39 52 and 33 of testedstations respectively For AprilndashJune time scale Hamban-tota station showed more drought events than other testedstations April to August is generally the Yala croppingseason to the country and occurrence of drought situationduring this season is likely to influence the food security ofthe country Hydrological years 1975-76 1977-78 1989-902001-02 and 2015-16 were found as drought years with 4139 39 54 and 83 of tested stations showing droughtrespectively during July to September time scale (Figure 4)However all the stations in the 2015-16 hydrological year

028

06

5630

74

613

1743

92

130

4320

5713

239

1300

2646

420

2030

048

44

96

911

013

052

24

26

42

200

450

730

119

76

374

1141

130

720

91113

312246

1517

2422

62

3526

1515

222

2857

227

617

9

1971-19721973-19741975-19761977-19781979-19801981-19821983-19841985-19861987-19881989-19901991-19921993-19941995-19961997-19981999-20002001-20022003-20042005-20062007-20082009-20102011-20122013-20142015-2016

Maha seasonYala season

Figure 3 Occurrence of drought events (SPIle minus 1) during Yala andMaha seasons as a of stations

0

10

20

30

40

50

60

Dro

ught

affe

cted

of

stat

ions

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

Figure 2 No of stations ()showing drought events in eachhydrological year (1970 to 2017)


showed the SPI values below zero while 26 out of 54 stationsshowed the occurrence of extreme drought events during theend of the Yala season Also both Kalawewa and MahaI-luppallama stations showed more drought events than othertested stations during the JulyndashSeptember time scale duringthe last 47 years

32 Frequency ofOccurrence ofDrought Events (SPIlt 0) in theClimatological Zones of Sri Lanka Considering the calcu-lated =iessen polygon average rainfall for wet dry andintermediate zones frequencies of occurrence of droughtevents in the climatological zones were separately cal-culated based on 6-month and annual SPI time scalesAccordingly there were more frequent drought eventsthat occurred in the dry zone (57) than the wet (49)and intermediate zone (47) at the annual time scale(Figure 5) At 6-month time scale there were morefrequent drought events that occurred in the dry zone(55) than the wet (49) and intermediate zone (53)

during OctoberndashMarch (Maha season) time scale (Fig-ure 5) During AprilndashSeptember (Yala season) time scaleboth dry and wet zones showed a similar frequency ofdrought (48) but it was less than the intermediate zone(51) (Figure 5)

33 Trend Analysis of SPI for Rainfall Stations We used themost commonly used MannndashKendall test (MK) with Senrsquosslope estimator to detect the trend of meteorological droughtin annual and 6-month SPI time scales during 1970 to 2017over Sri Lanka Figure 6 shows only the significant trendresults of the MannndashKendall test for annual and 6-monthtime scales (Maha and Yala)

Trend analysis of SPI at annual scale revealed decreasingtrend of SPI only at two stations (Avissawella and Pavat-kulam) exhibiting significant decreasing trend However 12stations exhibited wetting tendency showing significantincreasing trend for annual SPI us considering overalltrend over the country the country witnesses wetting

020406080

100

Dro

ught

affe

cted

o

f sta

tions

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(a)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(b)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(c)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(d)

Figure 4 Percentage of stations showing drought events in SPI3 (a) OctoberndashDecember (b) JanuaryndashMarch (c) AprilndashJune(d) JulyndashSeptember

0

10

20

30

40

50

60

70

Dry zone Wet zone Intermediatezone

Freq

uenc

y of

dro

ught

(a)

0

10

20

30

40

50

60

Dry zone Wet zone Intermediate zone

Freq

uenc

y of

dro

ught

Oct-MarApr-Sep

(b)

Figure 5 Frequency of drought in climatological zones at the (a) annual time scale and (b) 6-month time scale based on iessen polygonaverage rainfall for wet dry and intermediate zones


ndash5

ndash4

ndash3

ndash2

ndash1

0

1

2

3

4

Ang

amed

illa

Anu

radh

apur

aAv

issaw

ella

Baka

mun

aBa

ttica

loa

Ches

terfo

rdG

eeki

yana

kand

aJa

ffna

Kant

ale

Mat

ale

Med

awac

hchi

yaPa

vatk

ulam

Poth

uvil

Wel

law

aya

Am

para

Ang

amed

illa

Anu

radh

apur

aBa

kam

una

Batti

calo

aD

iyab

edum

aH

ingu

rakg

oda

Kant

ale

Mah

ailu

ppal

lam

aM

atal

eM

edaw

achc

hiya

Poth

uvil

Wel

law

aya

Alla

iAv

issaw

ella

Ches

terfo

rdCh

ilaw

Gee

kiya

naka

nda

Hin

gura

kgod

aJa

ffna

Nuw

ara E

liya

Annual Maha season Yala season

Z va

lue o

f MK

test

Positive trendNegative trend

Figure 6 Summarized results of the MK test for SPI12 (Annual) and SPI6 (Maha and Yala seasons) Stations showing a significant trend at005 level of significance are only shown

400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0

Rain gauge stationsClimate zonesndash0004 to ndash0002ndash0002 to ndash0001

ndash0001-00-001001-002002-003

N


(a)

400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(b)

Figure 7 Continued


tendency (decrease in drought events) in terms of SPI12(Figure 6) Most of these stations showing wetting ten-dencies are located in dry zone of the country exceptChesterford Geekiyanakanda and Matale Nisansala et al[35] analyzing the rainfall trend at 37 stations during therecent 31 years (1987ndash2017) showed an increasing annualrainfall trend over the country and further they showed asignificant increase at four stations located in the dry zonebut only one station in the wet zone

Considering SPI6 during Yala season significant de-creasing trends were observed in 5 stations whereas none ofthe stations recorded decreasing trend duringMaha croppingseasons Only 3 stations showed significant wetting tendenciesduring Yala seasons but 13 stations out of 54 stations dis-played significant increasing trend (wetting) during Mahaseasons in Sri Lanka during 1970 to 2017ese results in-dicate that rainfall during Yala seasons is in decreasing trendwhile it increases during Maha seasons Wickramagamage[34] also showed a decreasing trend of SWM which bringsrainfall during Yala season to the entire island and observedcomparatively increasing trend during NEM (January toFebruary) from a study conducted for the period 1981 to 2010

To understand the spatial variation of drought trend atannual SPI time scale Senrsquos slope values were interpolatedfor the entire country using Ordinary Kriging in ArcGIS

(102) (Figure 7) Figure 7 (a) shows that Eastern part of thecountry (part of dry and intermediate zone) getting wetterwhereas some part of dry intermediate and wet zone gettingdryer according to the annual SPI time scale Similarlyeastern segment of the country is getting wetter during theMaha cropping season However during Yala season mostparts of the country are getting dryer except far north andsouthwestern part A similar trend was observed for therainfall trend analysis for Yala and Maha season during the1987ndash2017 period using MK test [35]

Previous studies have revealed that droughts are relatedto cyclic global teleconnections such as El Nino SouthernOscillation (ENSO) and Indian Ocean Dipole (IOD)[36 37] For example De Silva and Hornberger [38] showedthat there is high probability of occurrence of drought whenboth IOD and MEI (multivariate ENSO index) are positiveand nonoccurrence of drought when both IOD and MEI arenegative [38] However understanding the spatial variationof drought and their trends and frequency as recorded in thepresent study is also useful for the water resources planningin the country Water resources and agricultural plannersneed to analyze cyclic teleconnections drought trend andfrequency in developing water resources and allocation ofwater for different sectors and planning crops for differentzones of the country for the food security

400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(c)

Figure 7 Drought trend over Sri Lanka in terms of the Sensrsquo slope of SPI (a) Annual (b) Yala and (c) Maha season


4 Conclusions

SPI was calculated at different time scales but the analysisshowed that more drought events (SPIle minus 1) occurred atAprilndashSeptember time scale (Yala season) than the 3-month 6-month (Maha season) and annual SPI timescales During 1975-76 hydrological year 52 of stationsshowed drought more than other years while 8 out of 54stations showed extreme drought events during annual SPItime scale Drought analysis based on crop season therewere moderate severe or extreme drought events occurredin 1970-71 1985-86 1993-94 1999-00 2010-11 and 2012-13 hydrological years during Yala season (April-Septem-ber) only During JulyndashSeptember SPI time scale all thestations during the 2015-16 hydrological year showed theSPI values below zero while 26 out of 54 stations showedthe occurrence of extreme drought events

Based on iessen polygon average rainfall there weremore frequent drought events in the dry zone (57) than thewet (49) and intermediate zones (47) at the annual timescale According to annual SPI time scale stations in dryintermediate and wet zones showed more than 50 of thefrequency of drought events at 72 71 and 68 of stationsrespectively During Maha season stations in dry and in-termediate zones showed more than 50 of the frequency ofdrought events at 76 and 58 of stations respectivelywhile the wet zone showed less than 50 of the frequency ofdrought at 59 of stations

Based on the MannndashKendall trend test Allai Avissa-wella Chilaw Hingurakgoda and Nuwara Eliya stationsshowed a statistically significant decreasing trend during theYala season e spatial variation of drought trend showedthat part of the dry wet and intermediate zone located inWestern and South-western getting dryer Particularly dryand part of the intermediate zone are getting dryer duringthe Yala cropping seasone results of this study suggest animmediate drought mitigation plan for drought-prone areasespecially for the Yala cropping season for the reduction ofthe disaster of droughts

Data Availability

e data are available at the Meteorological department ofSri Lanka data portal

Conflicts of Interest

e authors declare that they have no conflicts of interest

References

[1] J Sheffield and E F Wood ldquoGlobal trends and variability insoil moisture and drought characteristics 1950ndash2000 fromobservation-driven simulations of the Terrestrial HydrologicCyclerdquo Journal of Climate vol 21 no 3 pp 432ndash458 2008

[2] A Belayneh and J Adamowski ldquoStandard precipitation indexdrought forecasting using neural networks wavelet neuralnetworks and support vector regressionrdquo Applied Compu-tational Intelligence and Soft Computing vol 2012 pp 1ndash132012

[3] A D Malakiya and T M V Suryanarayana ldquoAssessment ofdrought using standardized precipitation index (SPI) andreconnaissance drought index (RDI) a case study of amrelidistrictrdquo International Journal of Science and Research vol 5no 8 pp 1995ndash2002 2016

[4] C A Karavitis S Alexandris D E Tsesmelis andG Athanasopoulos ldquoApplication of the standardized pre-cipitation index (SPI) in Greecerdquo Water vol 3 no 3pp 787ndash805 2011

[5] B Habibi M Meddi P J J F Torfs M Remaoun andH A J Van Lanen ldquoCharacterisation and prediction ofmeteorological drought using stochastic models in the semi-arid Cheliff-Zahrez basin (Algeria)rdquo Journal of HydrologyRegional Studies vol 16 pp 15ndash31 2018

[6] T B McKee N J Doesken and J Kleist ldquoe relationship ofdrought frequency and duration to time scalesrdquo in Pro-ceedings of the 8th conference on Applied Climatology vol 17pp 179ndash184 AmericanMeteorological Society Anaheim CAUSA January 1993

[7] N R Patel P Chopra and V K Dadhwal ldquoAnalyzing spatialpatterns of meteorological drought using standardized pre-cipitation indexrdquo Meteorological Applications vol 14 no 4pp 329ndash336 2007

[8] R Meenakshi M Navamuniyammal and S MahalingamldquoAssessment of meteorological drought using Drinc and GISin Tiruttani block of iruvallur district Tamil Nadu IndiardquoInternational Journal of Engineering Research and Technologyvol 6 no 5 pp 1003ndash1011 2017

[9] T Caloiero P Caloiero and F Frustaci ldquoLong-term pre-cipitation trend analysis in Europe and in the Mediterraneanbasinrdquo Water and Environment Journal vol 32 no 3pp 433ndash445 2018

[10] Central Bank of Sri Lanka ldquoCentral bank reportrdquo 2018httpswwwcbslgovlk

[11] M Domroes ldquoMonsoon and land use in Sri Lankardquo GeoJournal vol 3 no 2 pp 179ndash192 1979

[12] Germanwatch ldquoGlobal climate risk index 2019rdquo 2019 httpswwwgermanwatchorgen16046

[13] B Lyon L Zubair V Ralapanawe and Z Yahiya ldquoFinescaleevaluation of drought in a tropical setting case study in SriLankardquo Journal of Applied Meteorology and Climatologyvol 48 no 1 pp 77ndash88 2009

[14] R K W G Nianthi ldquoMitigation and adaptation strategies setup a drought task force (DTF) in Sri Lankardquo in Proceedings ofthe National Conference on Understanding and ManagingUnsettled Drought pp 26ndash32 Rajarata University of SriLanka Anuradhapura Sri Lanka July 2017

[15] S Manesha S Vimukthini and K H M S Premalal ldquoDe-velop drought monitoring in Sri Lanka using standardizedprecipitation index (SPI)rdquo Sri Lanka Journal of Meteorologyvol 1 pp 64ndash71 2015

[16] N Piratheeparajah and S Raveendran ldquoSpatial variations ofthe flood and drought in the Northern Region of Sri LankardquoInternational Research Journal of Earth Sciences vol 2 no 6pp 2321ndash2527 2014

[17] K Navarathinam M A Gusyev A Hasegawa andJ Magome ldquoAgricultural flood and drought risk reduction bya proposed multi-purpose dam a case study of the malwa-thoya river basin Sri Lankardquo in Proceedings of the 21st In-ternational Congress on Modelling and Simulation GoldCoast Australia December 2015

[18] E Ekanayake and K Perera ldquoAnalysis of drought severity andduration using copulas in Anuradhapura Sri Lankardquo British


Journal of Environment and Climate Change vol 4 no 3pp 312ndash327 2014

[19] R Lokuhetti L Zubair J Visvanathan and A NijamdeenldquoDrought monitoring for Sri Lanka spatial extent and tem-poral evolution during the 2016-17 drought 13ndash15rdquo inProceedings of the International Roundtable on the Impact ofExtreme Natural EventsScience and Technology for Mitiga-tion-2017 Colombo Sri Lanka December 2017

[20] H M R C Herath K H M S Premalal A Kaumadee andSanjeewani ldquoAnalysis of standardized precipitation indices toidentify for drought condition in 2015rdquo Sri Lanka Journal ofMeteorology vol 1 pp 20ndash31 2015

[21] M Samad M Aheeyar J Royo-Olid and I Arulingam =ePolitical and Institutional Context of theWater Sector in Sri LankaAn Overview p 92 European Union Luxembourg Europe 2017

[22] B A Malmgren R Hulugalla Y Hayashi and T MikamildquoPrecipitation trends in Sri Lanka since the 1870s and rela-tionships to El Nintildeo-southern oscillationrdquo InternationalJournal of Climatology vol 23 no 10 pp 1235ndash1252 2003

[23] H Jayawardene D Sonnadara and D Jayewardene ldquoTrendsof rainfall in Sri Lanka over the last centuryrdquo Sri LankanJournal of Physics vol 6 pp 7ndash17 2005

[24] D Tigkas H Vangelis and G Tsakiris ldquoDrinC a software fordrought analysis based on drought indicesrdquo Earth ScienceInformatics vol 8 no 3 pp 697ndash709 2015

[25] B Liu Z Yan J Sha and S Li ldquoDrought evolution due toclimate change and links to precipitation intensity in the haiheriver basinrdquo Water vol 9 no 11 p 878 2017

[26] H B Mann ldquoNonparametric tests against trendrdquo Econo-metrica vol 13 no 3 pp 245ndash259 1945

[27] M G Kendall Rank Correlation Methods Charles GriffinLondon UK 1975

[28] P K Sen ldquoEstimates of the regression coefficient based onkendallrsquos taurdquo Journal of the American Statistical Associationvol 63 no 324 pp 1379ndash1389 1968

[29] S N Rahmat N Jayasuriya and M A Bhuiyan ldquoTrendanalysis of drought using Standardized Precipitation Index inVictoria Australiardquo in Proceedings of the 34th Hydrology andWater Resources Symposium pp 441ndash448 Sydney AustraliaNovember 2012

[30] K H Hamed and A Ramachandra Rao ldquoA modified Mann-Kendall trend test for autocorrelated datardquo Journal of Hy-drology vol 204 no 1ndash4 pp 182ndash196 1998

[31] Consortium of Humanitarian Agencies (CHA) Impact ofDisasters in Sri Lanka e Consortium of HumanitarianAgencies (CHA) Colombo Sri Lanka 2016

[32] K S Sanjaya T Priyadarshana and N Wijayarathna ldquoEffectof rainfall abnormalities on rice yield in Hambanthota districtSri Lankardquo Asian Journal of Agriculture and Food Sciencesvol 2 no 6 pp 494ndash499 2014

[33] H G A S Sathischandra B Marambe and R PunyawardenaldquoSeasonal changes in temperature and rainfall and its rela-tionship with the incidence of weeds and insect pests in rice(Oryza sativaL) cultivation in Sri Lankardquo Climate Change andEnvironmental Sustainability vol 2 no 2 pp 105ndash115 2014

[34] P Wickramagamage ldquoSpatial and temporal variation ofrainfall trends of Sri Lankardquo =eoretical and Applied Cli-matology vol 125 no 3-4 pp 427ndash438 2015

[35] W D S Nisansala N S Abeysingha A Islam andA M K R Bandara ldquoRecent rainfall trend over Sri Lanka(1987 to 2017)rdquo International Journal of Climatology 2019

[36] C F Ropelewski and M S Halpert ldquoQuantifying southernoscillation-precipitation relationshipsrdquo Journal of Climatevol 9 no 5 pp 1043ndash1059 1996

[37] J Chandimala and L Zubair ldquoPredictability of stream flowand rainfall based on ENSO for water resources managementin Sri Lankardquo Journal of Hydrology vol 335 no 3-4pp 303ndash312 2007

[38] T De Silva and G M Hornberger ldquoIdentifying El NintildeondashSouthern Oscillation influences on rainfall with classificationmodels implications for water resource management of SriLankardquo Hydrology and Earth System Science vol 23pp 1905ndash1929 2019


[10] and is divided into three climatic zones ie dry wetand intermediate based on the total annual rainfall [11]According to the Global Climate Risk Index 2019 Sri Lankabecame the most affected countries along with Puerto Ricoand Dominica [12] us Sri Lanka is extremely vulnerableto climate change impacts In terms of people affected andrelief provided drought has been the most significant hazardin Sri Lanka and Sri Lanka also serves as a case study intropical regions for analysis of drought hazard and riskassessment [13] Drought has affected many parts of thecountry intermittently as one of the serious natural hazardsin Sri Lanka [14] It is reported that 2001 drought severelyaffected dry and intermediate zone of the country whileHambantota area experienced prolonged severe droughtduring 2001 to 2002 [15] Prolonged drought occurred in2001ndash02 led to 1 drop in the GDP growth rate in thecountry by particularly affecting hydropower generation andagriculture sector [13]

Drought studies in Sri Lanka so far have particularlyfocused on analysis of the spatial variations [16 17] severityand duration of drought mostly in part of the country [18]and spatial extent and temporal evolution of drought [19]Most of these studies are in relation to dry zone areas of SriLanka However Lyon et al [13] studied the relationshipbetween SPI and drought relief payments in the country andfound the strongest relationships with a 9-month cumulativedrought index Herath et al [20] analyzed monthly rainfalldata using SPI to identify possible drought conditions onlyfor the year 2015

Moreover it is rather rare to find drought trend analysisover entire Sri Lanka It is worth to investigate the trend ofdrought over the entire country specially for drought mit-igation and agricultural planning It is also projected thatmost of the districts in the dry zone in Sri Lanka will facesevere seasonal or year-round water scarcity by 2025 withpresent level of irrigation efficiency [21] Hence this paperanalyzes the severity frequency and trend of meteorologicaldrought over Sri Lanka during the recent past (1970 to 2017)using 48 years of rainfall data at 54 stations

2 Material and Methods

21 Description of Data Monthly precipitation data of 48years (1970ndash2017) were collected from 54 meteorologicalstations (Table 1) from theMeteorological Department of SriLanka e Meteorological Department of Sri Lanka is thenodal agency in Sri Lanka to record quality check andarchive all meteorological data e monthly rainfall valuesused in the study were prepared by the MeteorologicalDepartment of Sri Lanka and data quality is maintained bythem [22 23]

Although there are only 54 stations used in this studythe stations are well distributed across the three climaticzones ie dry intermediate and wet zones of Sri Lanka(Figure 1) and hence adequately showed the climatic spatialvariability in the country ere were missing values in fewstations and Table 1shows the missing values as a per-centage and missing values were filled using the regressionand normal ratio method

22 Meteorological Drought Analysis Standardized Precip-itation Index (SPI) was used to evaluate the drought both theshort-term (3 and 6 months) and the long-term (12 months)time scales In order to calculate the SPI index for each timescale the variability of precipitation totals is described by agamma distribution and then transformed to a normaldistribution [6] e gamma distribution is defined by itsfrequency or probability density function

g(x) 1

βaΓ(a)x

aminus 1e

minus (xβ) forxgt 0 (1)

where a and β are the shape and scale parameters respec-tively x is the precipitation amount and Γ(a) is the gammafunction Maximum likelihood estimations of a and β are

a 14A

1 +

1 +4A

3

1113970

1113890 1113891

β x

a whereA (x) minus

1113936 ln(x)

n

(2)

where n number of observationse subsequent parameters are then used to find the

cumulative probability of a recorded precipitation amountfor the given month and time scale for the given locationSince the gamma function is undefined for x 0 and aprecipitation distribution may contain zeros the cumulativeprobability is computed as

H(x) q +(1 minus q)G(x) (3)

where q is the probability of zero precipitation and G(x) isthe cumulative probability of the incomplete gammafunction e cumulative probability H(x) is then trans-formed to the standard normal random variable z with meanand variance of zero and one respectively which is the valueof the SPI [24]

Analysis was done in different time sequence and timeduration October to September was used as the hydrologicalyear (SPI12) and October to March and April to Septemberwere used as 6-month time scale (SPI6) as these periods arecropping seasons of Maha and Yala respectively in SriLanka October to December January to March April toJune and July to September were used as 3-month time scale(SPI3) Calculated SPI values were compared with the SPIclassification values for drynesswetness category to indicatethe status of the drought (Table 2)

23 Drought Frequency Analysis Drought frequencies werecalculated for dry wet and intermediate zones separatelyFor this purpose average rainfall for three climate zoneswas calculated separately using the iessen polygonmethod first en these average rainfall was used incomputing SPI for three climatic zones separately and theresulted SPI was used in frequency calculation For thisfrequency calculation the drought was defined as SPI lt 0[25] Also the frequency of occurrence of drought eventsmore or less than 50 was calculated based on the cal-culated SPI in each station







S 1113944

nminus 1

i11113944

n


sign Xj + Xi1113872 1113873

+1 Xj ltXi1113872 1113873

0 Xj Xi1113872 1113873

minus 1 Xj ltXi1113872 1113873

⎧⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎩

(4)



mi tii (i minus 1)(2i + 5)1113858 1113859

18 (5)




Qi xj minus xk1113872 1113873





f(t) Qt + B (7)


Qi Xj minus Xk1113872 1113873









80deg0prime0PrimeE

10deg0prime0Prime

N9deg

0 prime0Prime

N8deg

0prime0Prime

N7deg

0prime0Prime

N6deg

0prime0Prime

N

10deg0prime0Prime

N9deg

0prime0Prime

N8deg

0prime0Prime

N7deg

0prime0Prime

N6deg

0prime0Prime

N




82deg0prime0PrimeE



N








028

06

5630

74

613

1743

92

130

4320

5713

239

1300

2646

420

2030

048

44

96

911

013

052

24

26

42

200

450

730

119

76

374

1141

130

720

91113

312246

1517

2422

62

3526

1515

222

2857

227

617

9

1971-19721973-19741975-19761977-19781979-19801981-19821983-19841985-19861987-19881989-19901991-19921993-19941995-19961997-19981999-20002001-20022003-20042005-20062007-20082009-20102011-20122013-20142015-2016



0

10

20

30

40

50

60

Dro

ught

affe

cted

of

stat

ions

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017








020406080

100

Dro

ught

affe

cted

o

f sta

tions

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(a)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(b)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(c)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(d)


0

10

20

30

40

50

60

70


Freq

uenc

y of

dro

ught

(a)

0

10

20

30

40

50

60


Freq

uenc

y of

dro

ught

Oct-MarApr-Sep

(b)



ndash5

ndash4

ndash3

ndash2

ndash1

0

1

2

3

4

Ang

amed

illa

Anu

radh

apur

aAv

issaw

ella

Baka

mun

aBa

ttica

loa

Ches

terfo

rdG

eeki

yana

kand

aJa

ffna

Kant

ale

Mat

ale

Med

awac

hchi

yaPa

vatk

ulam

Poth

uvil

Wel

law

aya

Am

para

Ang

amed

illa

Anu

radh

apur

aBa

kam

una

Batti

calo

aD

iyab

edum

aH

ingu

rakg

oda

Kant

ale

Mah

ailu

ppal

lam

aM

atal

eM

edaw

achc

hiya

Poth

uvil

Wel

law

aya

Alla

iAv

issaw

ella

Ches

terfo

rdCh

ilaw

Gee

kiya

naka

nda

Hin

gura

kgod

aJa

ffna

Nuw

ara E

liya


Z va

lue o

f MK

test



400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(a)

400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(b)

Figure 7 Continued







400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(c)



4 Conclusions




Data Availability




References















































S 1113944

nminus 1

i11113944

n


sign Xj + Xi1113872 1113873

+1 Xj ltXi1113872 1113873

0 Xj Xi1113872 1113873

minus 1 Xj ltXi1113872 1113873

⎧⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎩

(4)



mi tii (i minus 1)(2i + 5)1113858 1113859

18 (5)




Qi xj minus xk1113872 1113873





f(t) Qt + B (7)


Qi Xj minus Xk1113872 1113873









80deg0prime0PrimeE

10deg0prime0Prime

N9deg

0 prime0Prime

N8deg

0prime0Prime

N7deg

0prime0Prime

N6deg

0prime0Prime

N

10deg0prime0Prime

N9deg

0prime0Prime

N8deg

0prime0Prime

N7deg

0prime0Prime

N6deg

0prime0Prime

N




82deg0prime0PrimeE



N








028

06

5630

74

613

1743

92

130

4320

5713

239

1300

2646

420

2030

048

44

96

911

013

052

24

26

42

200

450

730

119

76

374

1141

130

720

91113

312246

1517

2422

62

3526

1515

222

2857

227

617

9

1971-19721973-19741975-19761977-19781979-19801981-19821983-19841985-19861987-19881989-19901991-19921993-19941995-19961997-19981999-20002001-20022003-20042005-20062007-20082009-20102011-20122013-20142015-2016



0

10

20

30

40

50

60

Dro

ught

affe

cted

of

stat

ions

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017








020406080

100

Dro

ught

affe

cted

o

f sta

tions

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(a)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(b)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(c)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(d)


0

10

20

30

40

50

60

70


Freq

uenc

y of

dro

ught

(a)

0

10

20

30

40

50

60


Freq

uenc

y of

dro

ught

Oct-MarApr-Sep

(b)



ndash5

ndash4

ndash3

ndash2

ndash1

0

1

2

3

4

Ang

amed

illa

Anu

radh

apur

aAv

issaw

ella

Baka

mun

aBa

ttica

loa

Ches

terfo

rdG

eeki

yana

kand

aJa

ffna

Kant

ale

Mat

ale

Med

awac

hchi

yaPa

vatk

ulam

Poth

uvil

Wel

law

aya

Am

para

Ang

amed

illa

Anu

radh

apur

aBa

kam

una

Batti

calo

aD

iyab

edum

aH

ingu

rakg

oda

Kant

ale

Mah

ailu

ppal

lam

aM

atal

eM

edaw

achc

hiya

Poth

uvil

Wel

law

aya

Alla

iAv

issaw

ella

Ches

terfo

rdCh

ilaw

Gee

kiya

naka

nda

Hin

gura

kgod

aJa

ffna

Nuw

ara E

liya


Z va

lue o

f MK

test



400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(a)

400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(b)

Figure 7 Continued







400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(c)



4 Conclusions




Data Availability




References















































028

06

5630

74

613

1743

92

130

4320

5713

239

1300

2646

420

2030

048

44

96

911

013

052

24

26

42

200

450

730

119

76

374

1141

130

720

91113

312246

1517

2422

62

3526

1515

222

2857

227

617

9

1971-19721973-19741975-19761977-19781979-19801981-19821983-19841985-19861987-19881989-19901991-19921993-19941995-19961997-19981999-20002001-20022003-20042005-20062007-20082009-20102011-20122013-20142015-2016



0

10

20

30

40

50

60

Dro

ught

affe

cted

of

stat

ions

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017








020406080

100

Dro

ught

affe

cted

o

f sta

tions

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(a)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(b)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(c)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(d)


0

10

20

30

40

50

60

70


Freq

uenc

y of

dro

ught

(a)

0

10

20

30

40

50

60


Freq

uenc

y of

dro

ught

Oct-MarApr-Sep

(b)



ndash5

ndash4

ndash3

ndash2

ndash1

0

1

2

3

4

Ang

amed

illa

Anu

radh

apur

aAv

issaw

ella

Baka

mun

aBa

ttica

loa

Ches

terfo

rdG

eeki

yana

kand

aJa

ffna

Kant

ale

Mat

ale

Med

awac

hchi

yaPa

vatk

ulam

Poth

uvil

Wel

law

aya

Am

para

Ang

amed

illa

Anu

radh

apur

aBa

kam

una

Batti

calo

aD

iyab

edum

aH

ingu

rakg

oda

Kant

ale

Mah

ailu

ppal

lam

aM

atal

eM

edaw

achc

hiya

Poth

uvil

Wel

law

aya

Alla

iAv

issaw

ella

Ches

terfo

rdCh

ilaw

Gee

kiya

naka

nda

Hin

gura

kgod

aJa

ffna

Nuw

ara E

liya


Z va

lue o

f MK

test



400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(a)

400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(b)

Figure 7 Continued







400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(c)



4 Conclusions




Data Availability




References















































020406080

100

Dro

ught

affe

cted

o

f sta

tions

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(a)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(b)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(c)

Dro

ught

affe

cted

o

f sta

tions

020406080

100

1970

-197

119

72-1

973

1974

-197

519

76-1

977

1978

-197

919

80-1

981

1982

-198

319

84-1

985

1986

-198

719

88-1

989

1990

-199

119

92-1

993

1994

-199

519

96-1

997

1998

-199

920

00-2

001

2002

-200

320

04-2

005

2006

-200

720

08-2

009

2010

-201

120

12-2

013

2014

-201

520

16-2

017

(d)


0

10

20

30

40

50

60

70


Freq

uenc

y of

dro

ught

(a)

0

10

20

30

40

50

60


Freq

uenc

y of

dro

ught

Oct-MarApr-Sep

(b)



ndash5

ndash4

ndash3

ndash2

ndash1

0

1

2

3

4

Ang

amed

illa

Anu

radh

apur

aAv

issaw

ella

Baka

mun

aBa

ttica

loa

Ches

terfo

rdG

eeki

yana

kand

aJa

ffna

Kant

ale

Mat

ale

Med

awac

hchi

yaPa

vatk

ulam

Poth

uvil

Wel

law

aya

Am

para

Ang

amed

illa

Anu

radh

apur

aBa

kam

una

Batti

calo

aD

iyab

edum

aH

ingu

rakg

oda

Kant

ale

Mah

ailu

ppal

lam

aM

atal

eM

edaw

achc

hiya

Poth

uvil

Wel

law

aya

Alla

iAv

issaw

ella

Ches

terfo

rdCh

ilaw

Gee

kiya

naka

nda

Hin

gura

kgod

aJa

ffna

Nuw

ara E

liya


Z va

lue o

f MK

test



400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(a)

400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(b)

Figure 7 Continued







400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(c)



4 Conclusions




Data Availability




References










































ndash5

ndash4

ndash3

ndash2

ndash1

0

1

2

3

4

Ang

amed

illa

Anu

radh

apur

aAv

issaw

ella

Baka

mun

aBa

ttica

loa

Ches

terfo

rdG

eeki

yana

kand

aJa

ffna

Kant

ale

Mat

ale

Med

awac

hchi

yaPa

vatk

ulam

Poth

uvil

Wel

law

aya

Am

para

Ang

amed

illa

Anu

radh

apur

aBa

kam

una

Batti

calo

aD

iyab

edum

aH

ingu

rakg

oda

Kant

ale

Mah

ailu

ppal

lam

aM

atal

eM

edaw

achc

hiya

Poth

uvil

Wel

law

aya

Alla

iAv

issaw

ella

Ches

terfo

rdCh

ilaw

Gee

kiya

naka

nda

Hin

gura

kgod

aJa

ffna

Nuw

ara E

liya


Z va

lue o

f MK

test



400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(a)

400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(b)

Figure 7 Continued







400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(c)



4 Conclusions




Data Availability




References















































400000 500000 600000

400000 500000 600000

7000

0080

0000

9000

0010

0000

011

0000

0

7000

0080

0000

9000

0010

0000

011

0000

0


ndash0001-00-001001-002002-003

N


(c)



4 Conclusions




Data Availability




References










































4 Conclusions




Data Availability




References










































researcharticle spi-based spatiotemporal drought over sri

Documents