ORIGINAL PAPER

Monsoon variability, crop water requirement, and crop planningfor kharif rice in Sagar Island, India

S. Mandal1 & B. U. Choudhury2 & L. N. Satpati1

Received: 4 August 2014 /Revised: 21 March 2015 /Accepted: 6 April 2015# ISB 2015

Abstract In the Sagar Island of Bay of Bengal, rainfed low-land rice is the major crop, grown solely depending on erraticdistribution of southwest monsoon (SM) rainfall. Lack of in-formation on SM rainfall variability and absence of cropscheduling accordingly results in frequent occurrence of inter-mittent water stress and occasional crop failure. In the presentstudy, we analyzed long period (1982–2010) SM rainfall be-havior (onset, withdrawal, rainfall and wetness indices, dryand wet spells), crop water requirement (CWR, by Food andAgriculture Organization (FAO) 56), and probability of week-ly rainfall occurrence (by two-parameter gamma distribution)to assess the variability and impact on water availability,CWR, and rice productivity. Finally, crop planning was sug-gested to overcome monsoon uncertainties on water availabil-ity and rice productivity. Study revealed that the normal onsetand withdrawal weeks for SM rainfall were 22nd±1 and 43rd±2 meteorological weeks (MW), respectively. However, ef-fective monsoon rainfall started at 24th MW (rainfall92.7 mm, p>56.7 % for 50 mm rainfall) and was terminatedby the end of 40th MW (rainfall 90.7 mm, p<59.6 % for50 mm rainfall). During crop growth periods (seed to seed,21st to 45th MW), the island received an average weeklyrainfall of 65.1±25.9 mm, while the corresponding weeklyCWR was 47.8±5.4 mm. Despite net water surplus of353.9 mm during crop growth periods, there was a deficit of159.5 mm water during MW of 18–23 (seedling raising) and

MWof 41–45 (flowering to maturity stages). Water stress wasobserved in early lag vegetative stage of crop growth (32ndMW). The total dry spell frequency during panicle initiationand heading stage was computed as 40 of which 6 dry spellswere >7 days in duration and reflected a significant (p<0.05)increasing trend (at 0.22 days year−1) over the years (1982–2010). The present study highlights the adaptive capacity ofcrop planning including abiotic stress-tolerant cultivars tomonsoon rainfall variability for sustaining rainfed rice produc-tion vis-à-vis food and livelihood security in vulnerableislands of coastal ecosystem.

Keywords Southwest monsoon onset . Rainfall probability .

Cropwater requirement . Rainfed rice . Crop planning .

Sagar Island

Introduction

There are about 54 million ha of rainfed lowlands, whichcontribute 19 % of the world’s total rice production with anaverage yield of 2.3 t ha−1 (Bouman et al. 2007). Asia alone ishome to 95 % of the world’s total rainfed rice area (Swainet al. 2005) and, among the Asian countries, India has thelargest rainfed lowland rice area (16.5 million ha), which isabout 39 % of the total rice area in India (Mandal et al. 2010).

Sagar Island, the largest island in Sunderbans DeltaicComplex sitting in the continental shelf of Bay of Bengal, isone of the most vulnerable deltas to climate change (Mandalet al. 2013; Mandal and Choudhury 2014). Low productiverainfed lowland rice occupies nearly 52 % of the geographicalarea (28, 211.4 ha) and contributes nearly 60 % of the totalfood grain production (44,274 t) of the island (Anonymous2010–2011). Majority of the population (67 %) depends onrice-based production system for their livelihood security.

* B. U. [email protected]; [email protected]

1 Department of Geography, University of Calcutta, 35 BallygungCircular Road, Kolkata 700019, India

2 Division of Natural Resource Management (Soil Science),ICAR Research Complex for NEH Region,Umiam 793 103, Meghalaya, India

Int J BiometeorolDOI 10.1007/s00484-015-0995-9

Water requirement in this rainfed rice system solely de-pends on the erratic distribution of southwest monsoon(SM). Heavy downpour at the start of SM, abrupt mid-monsoon break (Chand et al. 2012; Mishra 2012), and ear-ly withdrawal of monsoon rain increase the frequency ofdry spell (DS) at different growth stages of rice. As a re-sult, rice production often suffers from intermittent waterstresses during peak growing seasons and even leads tooccasional crop failure. Apart from this, the secondaryconsequences of severe cyclonic storms and tidal ingres-sions led to submergence of lowlands and are posing seri-ous threats to the rainfed rice production systems, liveli-hood, and food security of the rapidly growing populationin the island (Sarkar and Bhattacharjee 2011; Anonymous2013; Dhar et al. 2013; Mandal et al. 2013; Bhowmicket al. 2014; Mandal and Choudhury 2014).

Under such circumstances, information on appropriatecrop scheduling is essential to overcome the shortfallsassociated with intermittent droughts during rice-growingperiods. Knowledge of crop water requirement (CWR) atvarious sensitive stages of rice is an important practicalconsideration for crop scheduling (Choudhury et al.2013). CWR means the amount of water required to com-pensate the crop evapotranspiration (ETc.) loss from thecropped field. For accurate estimation of crop ETc., localweather-based atmospheric evaporative demand known asreference evapotranspiration (ET0) is very much importantin determining the exact amount of CWR (Allen et al.1998). Till date, no such efforts were made for cropscheduling through comprehensive analysis of monsoonrainfall variability (onset, withdrawal, dry spells (DSs),and probability of weekly rainfall (PWR)) and CWR dur-ing sensitive growth stages of rainfed rice in the island.Understanding the annual and seasonal variability of mon-soon rainfall is another important aspect for the develop-ment of appropriate strategies to adopt and mitigate thelikely impact of changes in rainfall pattern on sustainabil-ity of rainfed rice-based production system in the island.Aside from that, the sea–land frontier location, coastal-saline agro-climatic environment, and distinguished pat-tern of local weather variables characterized the rice-growing environment differently in this island than therest of India. Because of its geographic location in thecoastal area and away from mainland India, it is one ofthe most unexplored islands in terms of information onmonsoon rainfall variability, water availability, and thelikely impact of changes in rainfall pattern on rice pro-ductivity. Lack of such information on monsoon rainfallvariability and its effect on rice productivity is the majorbottleneck in devising sustainable management ap-proaches (including crop planning) for coping up the ad-verse effect of rainfall variability on rice productivity andlivelihood security in the island.

Keeping these gaps in view and the importance of futurerainfall variability on rice productivity, the present studyaimed at finding out the behavior of monsoon rainfall variabil-ity including anomalies of spell length, start and end of mon-soon spells, dry spell analysis at different rice-growing stages,and the probability of weekly rainfall occurrence by two-parameter gamma distribution. We also estimated stage-wisecrop water requirement of rice, correlated with monsoon rain-fall behavior, and established relationship between annualmonsoon rainfall variability and rice productivity. Finally,crop planning was suggested as an adaptation to monsoonrainfall variability for rainfed lowland rice (Oryza sativa L.)cultivation in the vulnerable islands of coastal ecosystem ofdelta region.

Materials and methods

Characterization of study area

The study area (Sagar Island) extends from 21° 37′ 20″ N to21° 52′ 28″ N latitude and 88° 2′ 17″ E to 88° 10′ 25″ Elongitude, with an elevation range of 2.5–3.5 m from msl(Fig. 1). Total area of the island is 28,211.4 ha, of which only55 % of the area is available for cultivation. Majority of thecultivated areas (97.4 %) are under rainfed lowland rice culti-vation. Rice crop is cultivated during three seasons: Aus,Aman, and Boro. Aman rice dominates the rice-cultivated area(91.0 of 97.4 %) followed by Boro (6.3 of 97.4 % rice area)andmarginal byAus rice. Irrigated rice cultivation is limited toonly 19.8% of the total rice area, and the irrigation sources arefarm ponds and irrigation tanks (Anonymous 2010–2011) .

Meteorological data used

The meteorological data used for climate analysis wererecorded at Sagar IMD station (station index 42,903,∼3 m above mean sea level) and collected from theRegional Meteorological Department, Government ofIndia, Kolkata. The database contained long-term(1982–2010) records of daily minimum and maximumtemperatures, rainfall, evaporation, wind speed, sunshineduration, maximum and minimum relative humidity, andair pressure. The rice yield data (2002–2011) were col-lected from the Sagar Block Agricultural Developmentoffice. For the analysis of monsoon climate change andvariability, daily weather data were converted into stan-dard meteorological weekly (MW) data since weeklyanalysis of weather variables, particularly rainfall is es-sential for planning rainfed agricultural production(Mishra et al. 1999; Jat et al. 2005). Field preparationfor kharif rice in the island starts from May, and cropharvesting begins from October till November (Mandal

Int J Biometeorol

et al. 2013; Mandal and Choudhury 2014). Hence, theweather data from 18th MW (30 April–6 May) to 45thMW (28 October–4 November) were considered for theanalysis.

Estimation of CWR

Followingwater balance approach, CWR at different stages ofrice growth (May–October) was computed from ETc. as wellas seepage and deep percolation (PERC) losses in the ricefields. Water loss through seepage and deep percolation pro-cess in paddy field was considered about 3.5 mm day−1 inSagar Island (Dey et al. 2011). The overland flow (Roff) wasconsidered zero, as the bund heights maintained in the areausually varied from 25 to 35 cm, and fields were mostlyleveled (<3 % slope). Capillary rise (Cr) was also consideredzero, because of the continuous downward flow of water fromthe puddled layer which prevents Cr into the root zone(Bouman et al. 2007). The ETc. was estimated by multiplyingreference evapotranspiration (ETo) with crop coefficient (Kc)of rice. Crop Kc for this region at different stages was takenfrom Food and Agriculture Organization (FAO) 56 method(Allen et al. 1998), and ETo was computed from the dailyweather data by Penman-Monteith FAO 56 method (Allenet al. 1998).

ETO ¼0:408Δ Rn−Gð Þ þ γ

900

T þ 273u2 es−eað Þ

Δþ γ 1þ 0:34ð Þu2where ETo is reference evapotranspiration [mm/day], Rnet isnet radiation at the crop surface [MJ/m2/day], G is soil heatflux density [MJ/m2/day], T is mean daily air temperature at2 m height [°C], u2 is wind speed at 2 m height [m/s], es issaturation vapor pressure [kPa], ea is actual vapor pressure[kPa], es-ea is saturation vapor pressure deficit [kPa], Δ isslope vapor pressure curve [kPa/°C], and γ is psychrometricconstant [kPa/°C] (Zeleke and Wade 2012). Moreover, thenormal water balance during southwest monsoon in differentMW was classified as deficit when total weekly rainfall was<LPA±CV (LPA, long period average; CV, coefficient of var-iation), normal when actual rainfall was within LPA±CV, andexcess when actual rainfall was >LPA±CVover the 29-yearperiod of study (Choudhury et al. 2012).

Computation of onset and withdrawal weeks, spell length,and distribution of dry and wet spells (DSs and WSs)during southwest monsoon

Onset and withdrawal of rainy season were computed fromweekly rainfall data by forward and backward accumulationmethods proposed by Reddy et al. (2008) and Mondal et al.

Fig. 1 Administrative boundary map of West Bengal and location of study site (Sagar Island)

Int J Biometeorol

(2013). In this method, weekly rainfall was summed by for-ward accumulation (20th+21st+…52nd MW) until 75 mmrainfall was accumulated. This quantum of rainfall (75 mm)has been considered as the onset time for sowing of rainfedrice. The withdrawal of monsoon was determined by backwardaccumulation of rainfall (48th+47th+46th+…30th MW) data.Accumulated rainfall of 20mmwas considered as threshold formonsoon withdrawal, which is sufficient for plowing fields forpost-rice crop (Reddy et al. 2008). Weekly rainfall of less than20 mm during the crop growth periods was considered as dryweeks and above 20 mm was considered as wet week (Babuand Lakshminarayana 1997; Reddy et al. 2008).

The starting date of rainy season was affirmed as the daywhen (counting from 1 May) the cumulative rainfall for twoconsecutive day was >20 mm, and there was no dry period oflength >10 days within the following first 30 days after onset(Stern and Cooper 2011; Traore et al. 2013). The 3-, 5-, and7-day dry and wet spells along with extreme wet spells werecalculated in different rice-growing stages. A day was consid-ered to be “dry“ when daily rainfall was <0.1 mm. The dailyobservations were represented as successive sequences of dryand wet periods, and the total number of occurrences of 3, 5,and 7 days DSs within the monsoon spells was also calculated(Traore et al. 2013). Rainfall intensity (RI) in extreme wetspells was also computed to identify the effectiveness ofextreme spells.

Computation of rainfall distribution index (RDI)and wetness index (WI)

Rainfall distribution index for different monsoon weeks wascomputed using following equation:

RDI¼ Weekly rainfallð Þ= Number of days in the weekð Þ

� Number of rainy days

RDI ≤13.5 was considered as land preparation period andRDI 13.5 as cropping weeks (Basavaraju and Joshi 2000).Similarly, WI was calculated using the following equation:

WI ¼ Total rainfall of the weekð Þ= Average rainfall of that weekð Þ

(Panigrahi and Panda 2001).

Probability distribution for describing weekly rainfall

Using incomplete gamma probability distribution, weeklyrainfall for different return periods as well as percent proba-bility of occurrence of desire amount of rainfall was estimated.A random variable x is said to have a gamma probabilitydistribution with parameter and β if its probability densityfunction is equal to zero given by the following equation:

f xð Þ ¼ 1

Γαβα Xα−1e

x

β0 < X < α: ð1Þ

In this distribution, α and β are known as the shape andscale parameters, respectively, and Γ (α) is the gamma func-tion. Maximum likelihood estimation technique wasemployed for obtaining the estimates of α and β. Chi-squaretest was employed for testing the goodness of fit. Chi-squaretest statistic is defined as follows:

x2 ¼ ∑n

i xi−xð Þ2β2α

ð2Þ

This is distributed as χ2 with (n−1) degrees of freedom.The distribution function of gamma probability model is de-fined as follows (Subash et al. 2011):

P X ≤xð Þ ¼ F α;β; xð Þ ¼Z x

0f xð Þdx ¼ 1

Γβα

Z x

0xα−1e

−xβdx

ð3Þ

Methodology adopted in trend analysis

Mann-Kendall trend test (Mann 1945; Kendall 1975) has beenapplied to detect the trends of weekly weather variables andmonsoon rainfall. The magnitudes of such trends also quantifyusing Sen’s slope (Sen 1968).

Methodology adopted in establishing relationshipbetween variation of rainfall and variation of riceproductivity

In order to analyze the relationship between the inter-annualvariation of rainfall with kharif rice production and productiv-ity, we focused on the shorter time scales (9 years) by consid-ering the year to year differences in production (productivity)and their relation to year to year difference in seasonal rainfallfollowing standard procedures for Indian condition as de-scribed by Kumar et al. 2004; Gadgil and Rupa Kumar2006.We detrended each time series by taking the differences,DZi=Zi−Zi−1, between the value Z in each year i and the valuein the previous year i−1 (Box and Jenkins 1976; Kumar et al.2004). These values were then expressed as the percentagechange from the previous year’s value.

Results and discussion

Trends of monsoon week climate variables

During the southwest monsoon weeks (18–45 MW), averageweekly rainfall (AWR) varied widely (CV=38 %) from 15.4±10.2 to 102.3±79.5 mmwith a weekly mean of 61.0±27.0 mm.As the monsoon progressed, AWR increased inconsistently,reached peak, and then declined (Fig. 2). The 31st MW (30

Int J Biometeorol

July–5 August) was the wettest (102±79.5 mm), and 42ndMW (15 October–21 October) was the driest week (15.4±10.2 mm) in the island. Mann-Kendall trend statistics revealedthat a total of 17 (18th–20th, 24th–26th, 29th–32nd, 34th, 36th,40th–41st, and 43rd–45th MW) out of 28 monsoon weeksexperienced non-significant decreasing trend of AWR (at 0.03to 2.33 mm year−1), while the rest of the weeks (except 22ndMW) experienced non-significant (at 0.01 to 1.56 mm year−1)increasing trend. Only 22nd MW exhibited significant(p<0.05) increasing trend (at 1.06 mm year−1) in AWR andwas identified as the normal onset week of monsoon over theisland (discussed in the next section of analysis). Among therice-sensitive growth stages, the seedling preparation (SP),active vegetative stage (AVS), lag vegetative (LVS), andflowering to maturity (FM) stages exhibited non-significant decreasing (Z=−0.51 to −1.14) trend in receiv-ing rainfall at −1.73 to 2.71 mm year−1 (Table 1). This is inconsonance with the findings of WWF-India (2010), which

reported a marginal increasing trend of rainfall overSunderbans deltaic region during 1990–2000 and 2001–2008, respectively.

Climatic water balance revealed that over a span of 29 years(1982–2010), 11 MW (18th to 23rd and 41st to 45th MW)received deficit, and 17 MW (24th to 40th MW) receivedsurplus rainfall in the island (Fig. 2). Weekly rainy daysranged from 1.0±1.0 (42nd to 45th MW) to 5.0±1.0 (28thto 33rd MW) with a considerable variation (CV=18–124.1 %). Non-significant decreasing trend was observed in54 % of the MW except 21st and 38th MW which reflectedsignificant (p<0.1) increasing trend. Therefore, most of theMW experienced lesser number of rainy days. Similar obser-vation of decreasing trend in mean annual rainy days at SagarIsland was made by Mandal et al. (2013).

Weekly minimum temperature varied from 22.5±1.4 °C(45th MW) to 26.9±1.3 °C (28th MW), while maximum tem-perature varied from 29.9±1.4 °C (45th MW) to 33.2±1.5 °C

0

500

1000

1500

2000

2500

3000

3500

0

500

1000

1500

2000

2500

3000

3500

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

Long period total rainfall

Ra

infa

ll, m

m

Rain

fall

, m

m

Meteorological week

Normal

Excess

Defecit

Fig. 2 Characterization of longperiod (1982–2010) weeklyrainfall during monsoon asdeficit, normal, and excess atSagar Island

Table 1 Long period (1982–2010) average weekly weather variables and their annual trends in different phenological stages of kharif rice in SagarIsland

Phenological stages Rainfall, mm Tmin, °C Tmax, °C Evaporation, mm ET0, mm

A Z Q A Z Q A Z Q A Z Q A

DSBP 34 1.41 2.90 27 −1.65 −0.03+ 33 3.42 0.07*** 26 −5.61 −0.47*** 28

SP 76 −0.51 −1.73 27 −0.68 −0.01 32 2.83 0.06** 18 −4.18 −0.23*** 24

AVS 90 −0.56 −2.47 27 −0.30 0 31 4.32 0.07*** 16 −4.63 −0.23*** 25

LVS 71 −1.14 −2.71 27 0.13 0 31 3.60 0.07*** 15 −3.77 −0.14*** 27

PIH 79 0.32 1.90 26 −2.54 −0.02* 31 4.07 0.06*** 15 −5.05 −0.21*** 26

HF 91 0.34 0.19 26 −1.31 −0.04 32 1.63 0.05 15 −3.27 −0.04** 25

FM 28 −1.11 −2.38 24 −3.06 −0.05** 31 3.44 0.08*** 13 −4.78 −0.19*** 14

DSBP dry seedbed preparation, SP seedling preparation, AVS active vegetative stage, LVS lag vegetative stage, PIH panicle initiation to heading, HFheading to flowering, FM flowering to maturity, A weekly absolute value, Z normalized test statistics, Q Sen’s slope, Tmin and Tmax minimum andmaximum temperatures, ET0 reference evapotranspiration+ , at 10 % level of significance; *, significance at 5 % level; **, significance at 1 %; ***, significance at 0.1 %

Int J Biometeorol

(20th MW; Table 1). Significant declining trend (Z=−1.65 to−3.06) in minimum temperature at 0.02 to 0.05 °C year−1 wasobserved during dry seed bed preparation (DSBP), panicleinitiation to heading (PIH), and FM stages of rice crop. Incontrast, maximum temperature exhibited significant increas-ing trend at 0.06 to 0.08 °C year−1 across all the growth stages(Table 1). Though temperature range was within the criticallimit (>8 to <35 °C, FAO 66) for conducive growth of ricecrops, yet rising trend of maximum temperature at 0.05 to0.12 °C year−1 in most of the weeks is the concern since itmay harm different temperature-sensitive growth stages (i.e.,seedling emergence, leaf elongation, and initiation of panicle)of rice, and the situation might be severe during mid-monsoonbreaks and in the occurrence of prolonged dry spells in theisland. The island experienced a maximum evaporation loss of3.9±0.9 mm day−1 in 18th MW and a minimum of 1.7±0.4 mm day−1 in 45th MW. Trend analysis reflected a signif-icant decreasing trend (Z=−2.06 to −4.75) in the evaporationloss (at 0.03 to 0.09 mm day−1) for all MW. Similarly, higherevaporative demand (ETo=27mm) and lesser amount of rain-fall occurrence (71 mm) in LVS compared to active vegetative(AVS=90 mm) and PIH (79 mm) stages of growth made LVSvulnerable to intermittent water stress (Table 1).

Onset, withdrawal, and spell lengths of southwestmonsoon

Analysis of 29 years weekly rainfall data indicated that themonsoon rain arrived earliest (20th MW: 14 to 20 May) dur-ing 1988, 1995, and 2007, while onset was delayed to 25thMW (18 June to 24 June) during 1983. The normal onset ofmonsoon was 22nd±1 MW (28 May to 3 June±7 days).Similarly, the earliest and delayed week of cessation of rainyseason was 40th MW (1 October–7 October) in 1988, 1997,and 2008 and 45th MW (5–11 November) in 1986, 1987,1990, 1993, 1995, 1998, 1999, 2001, and 2009 with a normalcessation in 43rd MW±2 weeks (22 to 28 October±14 days).The longest monsoon spell was 25 weeks (in 1995), and the

shortest was 16 weeks (in 1983) with a long period average of20±2 weeks (140±14 days).

To justify the accuracy of accumulation method of onsetand withdrawal of monsoon spells, Raman’s method was ap-plied to identify the start and end date of the monsoon rainfall.Results revealed that 4 June±11 days was the onset date, and17 October±15 days was the withdrawal date of the monsoonrainfall with a spell length of 135±20 days (Table 2). Hence,the results of both the methods were quite satisfactory withtheir similar range of precession. However, Mishra (2012)reported that 7 June and 12 October are the normal onsetand withdrawal date for the state of West Bengal. SagarIsland being located in the extreme southern land–water tran-sition zone of West Bengal State first welcomes the southwestmonsoon which generally enters the state along a trough ex-tended in the north–south direction. Hence, the onset date atSagar Island might be expected 2–3 days prior to the normalonset date in the state of West Bengal (7 June) and late with-drawal by 5 days (12 October). Formation of low pressure cellin the Bay of Bengal during the second fortnight of Septemberto mid-October (Mishra 2012) might be the cause of late with-drawal of rainfall (by 5 days) in Sagar Island.

The withdrawal date exhibited significant (p<0.1) delayedtrend (Z=+1.92) at 0.60 days year−1. The total rainfall andspell length during southwest monsoon exhibited non-significant increasing trend. During the observation period,the island received an average of 18.6±3.5 mm rainfall peroccurrence of rainy days and 11.6±2.2 mm rainfall per daythroughout the monsoon spell. Both rainfall intensity (RI) andrainfall per day exhibited non-significant decreasing trend(Table 2). Hence, themonsoon spells elongated over the islandwith lesser amount of RI.

Many farmers are aware that rainy seasons with early onsetare generally better for crop production than those with lateonset (Sivakumar 1990; Stewart 1991). Lack of consistentrelationship between the start and end of the rainy seasonrefutes the popular belief that late beginning of rainy seasonis compensated by late ending or that rainy season becomes

Table 2 Characterization of long-term (1982–2010) southwest monsoon rainfall behavior at Sagar Island

Monsoon variables Minimum Maximumx

SD CV (%) Z Q

Rain start, JD 135 182 155 11 7 −0.71 −0.13Rain end, JD 258 319 290 15 5 1.92 0.60*

Spell length 95 168 135 20 14 1.82 0.97*

Rainfall, mm 1,011.9 2,282.3 1,589 329.6 20.7 0.39 2.50

Rainy days 59 111 86 13 15 0.85 0.29

RI, mm 12.4 26.4 18.6 3.5 18.8 −0.77 −0.03Rainfall, day−1 7.5 15.4 11.6 2.2 19.3 −0.81 −0.04

JD Julian day, Z normalized test statistics, Q Sen’s slope, SD standard deviation, RI rainfall intensity

*, at 10 % level of significance

Int J Biometeorol

shorter because of a late onset and early end (Traore et al.2013). In Sagar Island, the retreating of monsoon withdrawalconsiderably delays in kharif rice harvesting, and as a result,sowing of ensuing Rabi crops gets delayed and the conse-quences even extended to pre-kharif season (Dey et al. 2011;Mishra 2012). Delayed withdrawal of monsoon thus affectscrop intensification vis-à-vis food grain production and liveli-hood security in the island (Mishra 2012). Similarly, an earlystart of rainy season poses threat of long dry spell occurrenceafter planting the crops and results in intermittent moisturestress and at extreme, even crop failure. Conversely, a lateend of rainy season makes shorter crop cycle, more chancesof disease and pest incidence, lodging, and other damagesincluding birds (Stewart 1991).

Trends and patterns of dry and wet spell occurrenceduring southwest monsoon

Information on WSs and intervening DSs occurrence is ex-tremely useful for water resourcemanagement and sustainablecrop production. The information takes on greater significancein the wake of global climate change and climate change sce-nario projections (Singh and Ranade 2009). In our study, WSsand intervening DSs with their extremes have been studied fora sustainable crop planning. In the last three decades, the is-land experienced an average number of 21±4 DSs and 30±5WSs with varying duration of variability (18–19 %). Whenspell length was considered, the area had 8±2 days of extremeDS and 12±5 days of extreme WS with RI of 21±15 mmhaving 70 % variability in the monsoon spell. Similarly, themean occurrence of 3, 5, and 7 days DSs andWSs was 3±1, 1±1, and 0.64±1 and 3±2, 2±1, and 1±1, respectively(Table 3). The longest DS observed was 14 days (in 1994;started on 18 September and extended up to 1 October)followed by 12 days (in 2004, extending 30 August to 10September). Similarly, the longest WSwas of 26-day durationduring 1985 (23 June to 18 July), and the second longestextreme WS of 24 days was observed in 2008 (26 July to 18

August). Number of dry spells reflected a significant (p<0.05)increasing (Z=+2.27) trend at 0.22 frequency year−1 (withvarying durations), while WSs showed non-significant in-creasing trend. The occurrence of 7 days DSs increased(Z=+1.81) significantly (p<0.1), but 5-day DSs decreasednon-significantly (Table 3).

The occurrence of DSs in various rice-growing stages re-vealed that during SP stage, rice crop experienced 30 DSevents with varying lengths, while DSs with >7-day durationobserved in 3 years (i.e., 1987, 1991, 2009; Table 4) mighthave affected the seedling emergence (duration of emergence∼5–7 days; total duration=25–30 days) by restricting the ger-mination process and root system development (Brian 1988).Rather, water stress during SP stage affects both cell divisionand enlargement, though cell division appears to be less sen-sitive to water deficit than cell enlargement (Bouman andTuong 2001). In the AVS, frequency of DSs events was 19of which extreme DS duration was of 8 days and the DSs>7 days observed was 3. Hence, water stress in this vegetativestage probably reduced the number and height of effectivetillers and panicles per hill (Kumar et al. 2007). The DS fre-quency during PIH stage was computed as 40 of which 6 DSswas >7-day duration, and this might have reduced the num-bers of fertile spikelet and resulted in decreased number offilled grains per panicle by delayed flowering (Bouman andTuong 2001). Highest frequency (62) of DSs was observedduring flowering to maturity stage (FM, 30–35-day duration)which included 39 DSs of >7-day duration along with DSsextreme of 30 days (Table 4). Severe water stress at this re-productive stage might have resulted in chaffy grains. Fischer(1973) and Tsuda (1993) also reported reduction of grainweight due to water stress at this stage. Frequent occurrenceof DSs with higher evapotranspiration demand may lead to adecrease in yield up to 40 % because of insufficient watersupply during grain filing stage (Traore et al. 2013). Thus,the significant increase in the frequency of DSs during rainfedrice-growing season poses a threat to the sustainability ofrainfed rice production in the island.

Table 3 Dry (DSs) and wet spell (WSs) characterization of long-term (1982–2010) monsoon rainfall at Sagar Island

DSs and WSs parameters DSNUM WSNUM DSDUREXT WSDUR

EXT DSRIEXT DS3 DS5 DS7 WS3 WS5 WS7

Minimum 14 20 4 6 3 1 1 1 1 1 1

Maximum 30 42 14 26 62 7 5 3 7 5 3

x − 21 30 8 12 21 3 1 0.6 3 2 1

SD 4 5 2 5 15 1 1 1 2 1 1

CV (%) 19 18 25 37 70 2 5 4 2 2 3

Z 2.27 1.64 1.00 – −1.59 0.35 −0.02 1.81+ 1.03 0.91 −0.35Q 0.22* 0.19 – −0.36 – – – 0.02 – –

DSs dry spells, WSs wet spells, NUM number, EXT extreme, DUR duration, RI rainfall intensity+ , significance at 10 %; *, significance at 5 % level

Int J Biometeorol

Crop planning in relation to water balance, crop waterrequirement, and rainfall probability

During the monsoon weeks (18th to 45th MW), the islandreceived 61±27 mm AWR, of which CWR for rice growthwas 48.3±5.2 mm. The excess rainfall during 23rd to 41stMW (surplus=27.3 mm week−1), however, compensated thedeficit (18.3 mm week−1) rice water requirement duringDSBP, SDS, and FM stages of crop growth (Fig. 3).

When monsoon spell was considered, the average mon-soon rainfall received was 1,706±27.4 mm, of which 1,353.1±5.2 mm was CWR for rice-growing periods (seed toseed).Water deficit during nursery bed preparation and raisingseedlings (MW 18–23; DSBP and SDS stages) was108.8 mm, while the deficit during flowering to maturitystages of growth (FM; MW 41–45) was 50.7 mm (Fig. 3).Thus, the total water deficit during was 159.5 mm.However, in other stages of growth (SP, AVS, LVS, PIH,and HF), there was a surplus of 518.8 mm water (Table 5).Therefore, the net surplus of water was 353.9 mm during theentire rice growth stages (seed to seed). However, in spite ofthis surplus, crop suffered intermittent moisture stresses dur-ing flowering to maturity (negative balance between rainfalland CWR), mainly because of the erratic distribution of mon-soon rainfall during rice-growing periods. Aside from that, at

the inception of the LVS (MW 32–34), potential water stresscondition (PWSC) was also observed. It was because of thesudden fall in surplus amount (MW32; from 50.5 to 11.2 mm)as well as fall in probability of all expected weekly rainfall.Furthermore, in the climatic water balance study, the 25th and32nd MW were identified as water deficit week (−2.5 to−4.6 mm week−1). Hence, the PWSC in the early LVS mighthave adversely affected the effective tillering process. Theinter-annual change in rice productivity was weakly correlated(r=+0.058) with the inter-annual change in seasonal rainfall(May to October) in Sagar Island. Of the 9 years (2003–2011),rice production in 4 years increased with an increase in sea-sonal rainfall while decreased in 5 years with an increase inrainfall. Similarly, annual rate of changes in area under kharifrice (hectare) and inter-annual change in seasonal rainfall(May to October) exhibited very weakly positive correlation(r=0.052). Relationship between rainfall and production indi-ces revealed that by and large deficit rainfall had a negativeimpact on production, but even positive rainfall anomaly dur-ing 2006 also impacted the rice production negatively. Thismight be due to the occurrence of flood and subsequent sub-mergence of lowland paddy fields the island experienced dur-ing 2006 (Nath et al. 2008). Since most of the rice cultivars inmajor chunk of area were traditionally grown low-yieldingsubmergence susceptible long durations in nature (Sarkar

Table 4 Dry spell statistics during different growth stages of rainfed lowland (Aman) rice in Sagar Island

DS characteristics SP (MW 24–27) AVS (MW 28–31) LVS (MW 32–34) PIH (MW 35–39) HF (MW 40) FM (MW 41–45)

DSNUM 30.0 19.0 14.0 40.0 11.0 62.0

DSDUREXT 10 days 8 days 7 days 13 days 7 days 30 days

DS3 22.0 12.0 9.0 23.0 7.0 9.0

DS5 4.0 3.0 4.0 6.0 2.0 6.0

DS7 1.0 1.0 1.0 5.0 2.0 8.0

>DS7 3.0 3.0 0.0 6.0 0.0 39.0

DS dry spell, MWmeteorological week, SP seedling preparation,AVS active vegetative stage, LVS lag vegetative stage,PIH panicle initiation to heading,HF heading to flowering, FM flowering to maturity

-50

-40

-30

-20

-10

0

10

20

30

40

50

-50

-40

-30

-20

-10

0

10

20

30

40

50

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

Rainfall anomalies

Crop water deficit

Ra

infa

ll a

no

ma

lies

, m

m

Wa

ter

su

rplu

s a

nd

def

icit

, m

m

SMW

Fig. 3 Long period average(1982–2010) crop water balanceduring rice-growing periods (seedto seed) at Sagar Island

Int J Biometeorol

and Bhattacharjee 2011; Dhar et al. 2013; Bhowmick et al.2014; Table 6), therefore, total production declined.Production in 4 out of 9 years decreased with the decrease inrainfall, while in 2 years, it remained unchanged (Fig. 4).During 4 years of negative rainfall anomaly, area under ricecultivation decreased in 3 years, but in positive rainfall anom-aly years, area under rice cultivation increased. In spite ofpositive rainfall anomaly and increase in area under rice cul-tivation, total production decreased irregularly. This wasmostly because of an irregular decrease in productivity (yieldper hectare). The decrease in productivity might be due to theoccurrence of heavy showers caused by deep depression inBay of Bengal in the month of October (locally termed asAshiner jhar) (Mandal and Choudhury 2014) which coincidedwith the maturity to harvesting stages of crop growth.

The climatic water balance indicated surplus rainfall during24th to 40th MWand deficit rainfall from 41st MWonwards,which coincided with the most water stress-sensitive post-an-thesis rice growth stage. Therefore, one to two life-savingirrigations at this stage may improve the existing low produc-tivity of rice in the region. It was observed that the LPAweek-ly rainfall received at Sagar Island gradually increased from23rd MW and continued till 41st MW, then suddenly de-creased. Weekly rainfall (WR) of >50 mmwas observed from23rd MW to 41st MW, while WR >24 mm to <44 mm wasobserved in between 18th to 22nd MW, and the probability ofreceiving >50 mmWRwas very low (p>15 to <33%; Fig. 5).Moreover, the farmers of the island used to take two differentapproaches for seedling preparation. Depending on pre-monsoon rain, they prepare dry tillage seed bed prior to theonset of monsoon (before 22nd MW) and sow the seeds,which germinates after one or two spells of light to moderaterains that come in association with pre-monsoon thunderstorm. So, seedlings are raised in dry tilled nursery seed bed.In the second type, they prepare wet tillage seed beds onlyafter the onset of monsoon followed by broadcasting of pre-germinated seeds in the wet beds and raised seedlings till 21–25 days old for transplanting in the puddled field. Second typeof nursery raising is the preferred one for the rice farmers ofthe island (Mishra 2012). Hence, field preparation, primarytillage operation, and sowing of dry seeds can be initiatedduring 18th–22st MW with AWR less <50 mm. However,insufficient pre-monsoon rain during 18th to 22nd MW aswell as high-intensity rainfall at the start of monsoon (23rdMW) often compelled the farmers to go for wet seed bedpreparation in the island including mainland West Bengal(Panigrahi and Panda 2001; Mishra 2012). Though WR>50 mm started from 23rd MW, the probability of exceedingthe level of >50 %was found on 24th and 25thMW. Thus, thepreparation of wet seed bed and transplanting of kharif paddycan be initiated during that period (24–25 MW). Aside fromthat, balance between AWR and CWR during nursery raisingfrom wet bed assured surplus of 109.3 mm water (Table 5).T

able5

Water

balancecomponentsandcrop

water

requirem

entindifferentstagesof

lowland

rice

cultivatio

nin

Sagar

Island

Water

balanceparameters

DSB

P,SDS(M

W18–23)

SP(M

W24–27)

AVS(M

W28–31)

LVS(M

W32–34)

PIH(M

W35–39)

HF(M

W40)

FM(M

W41–45)

Rainfall,mm

203.4

303.2

360.6

213.4

393.20

90.7

142.50

ETo

179.2

91.7

91.0

67.9

109.9

21.0

96.6

ETc.

165.2

95.9

98.7

80.5

130.9

25.2

70.7

PERC

147.0

98.0

98.0

73.5

122.5

24.5

122.5

CWR

312.2

193.9

196.7

154.0

253.4

49.7

193.2

Balance

−108.8

+109.3

+163.9

+59.4

+139.8

+41.0

−50.7

WSW

18th–22nd

––

32nd

(p)

42nd–45th

SPseedlin

gpreparation,AVSactiv

evegetativ

estage,LV

Slagvegetativ

estage,PIH

panicleinitiationtoheading,HFheadingtoflow

ering,FM

flow

eringtomaturity,E

Toreferenceevapotranspiratio

n,ETc.

crop

evapotranspiratio

n,PERCpercolationandseepageloss,C

WRcrop

water

requirem

ent,WSW

water

stress

week

Int J Biometeorol

The probability of getting >50 mm rainfall further increasedfrom 28th MW (9 July to 15 July) onwards and continued till40th MW (1 October to 7 October; Fig. 5). As a result, therainfall amount and probability of occurrence both were de-creased from 41st MW, and the decreasing trend continued till45th MW. Therefore, transplanting of rice can be done in 29thMW (16 July to 22 July) considering 1-week gap for landpreparation including puddling because seedling transplanta-tion requires spell of heavy rain to allow accumulation of 5 to6 cm depth of standing water in the field (Bouman and Tuong2001). The stage of transplantation of rice seedling thoughcoincides with the peak period of monsoon (July–August);however, transplantation process in the island often hampereddue to prolonged “mid-monsoon break” and intermittent wa-ter stress. From the present study of trend analysis, it might beconcluded that, in Sagar Island, rice transplanting should bedone in the middle of July and continued up to the secondweek of August. Thus, the date of transplanting (DAT) of riceseedlings in Sagar Island can be suggested similar to the

recommended DAT (29th MW, 16–22 July) in mainlandWest Bengal (Dey et al. 2011). Values of WI during 23 to41 MW were lesser than the other weeks (18–22 MW and42–45 MW), and accordingly, the ranges of WI values(0.01–7.2) were less in monsoon weeks. Similarly, percentageof weeks with WI >1.0 were maximum during the foresaidweek range. Thus, compared to other weeks, the rainfall wasless erratic during 25–41 MW which provides favorable con-dition for kharif crop production. The effective monsoon rain-fall starts in 24thMW (rainfall 92.7mm, p<56.7% for 50mmrainfall) and terminates by the end of 40th MW (rainfall90.7 mm, p<59.7 % for 50 mm rainfall; Fig. 5). Thus, theeffective rainfall continued in the island for 17 weeks or119 days.

The productivity of major crop, i.e., rainfed lowland rice inthe island, often suffers from weather-related constraints:flooding and temporary submergence or long periods of stag-nant water in the monsoon season, intermittent droughts atpeak growth periods, daily tidal fluctuations, and seawater

Table 6 Suggested crop planning for kharif rice in Sagar Island

Management parameters Existing practices Suggested practices

1. Cultivar duration Long duration (140–160 days) Medium–short (120–140 days)

2. Cultivar type Traditional (i.e., Bhuri, Dangapatnai,Dharitri, Dudheswar, Langalmura,Malabati, Marichsal, Pankaj,Patnai-23, Sadamota)

Modern (i.e., MTU7029, IR64-Sub1,Bipasa, Sashi, Swarna Sub1,Sambha Mahsuri, SambaMahsuri-Sub1, Sahbhagi Dhan,CSR 36, CSRC(S)-2-1-7

3. Nursery bed Both dry and wet nursery bed Preferably wet nursery bed

4. Nursery raising 22–27 MW (18 May to 8 July) 24–28 MW (11 June–15 July)

5. Transplanting date 28–29 MW (9 July–22 July) 29–30 MW (16 July–29 July)

6. Irrigation and drainage structure Limited and insufficient Irrigation expansion and developmentof drainage structure needed

7. Harvesting, etc. 45th MW (5 November) onwards 43rd MW (22 October)

-50

-40

-30

-20

-10

0

10

20

30

40

-50

-40

-30

-20

-10

0

10

20

30

40

200

3

200

4

200

5

200

6

200

7

200

8

200

9

201

0

201

1

May to October rainfall

Kharif rice production

Linear (Kharif rice production)

Pro

du

ctio

n in

dex

Ra

infa

ll in

dex

Year

(d)

Fig. 4 Correlation betweenrainfall (annual) and kharif riceproductivity in Sagar Island forthe period 2002–2011

Int J Biometeorol

intrusion-led coastal salinity (Dey et al. 2011; Sarkar andBhattacharjee 2011; Anonymous 2013). Cultivation of tradi-tional long-duration (140–160 days), low-yielding genotypes(Table 6), susceptible to temporary or medium-duration(2 weeks and above) flooding, major diseases (mostly bacte-rial blight), coastal salinity, and monsoon rainfall anomaliesfurther reduced the rice productivity in the island. Most of thetraditional cultivars can neither elongate fast nor survive inun-dation and also suffer with lodging when water recedes. Thisfurther increased the vulnerability of stress-prone rainfed riceproductivity, food, and livelihood security in the island. Theselong duration cultivars often experiences sterility due to lowtemperature at flowering periods as well as low solar radiationin cloudy weather during monsoon months and thereby sufferfrom low photosynthetic efficiency and low productivity(Adhya et al. 2008; Adhikari et al. 2010; Anonymous 2013;Dhar et al. 2013; Bhowmick et al. 2014).

Since crop planning including nursery raising, sowingtime, duration, and type of cultivars has great potentiality inenhancing rice productivity by utilizing the monsoon rain ef-fectively while escaping the stresses, hence, a paradigm shiftfrom traditional low-yielding cultivars to adoption of short- ormedium-duration (120–140 days) cultivars with improvedqualities (stress tolerance including submergence anddroughts, disease resistance, high-yielding ability, better grainquality, etc.) is needed. Based on probability of weekly rainfalloccurrence (>50 to >150 mm; Fig. 5), the existing practiceof nursery raising in the month of May (22nd MW) needsto be delayed by at least 2 weeks (24th MW onwards).Similarly, delay in transplanting from the existing 9 July(28th MW) to mid-July (29th MW) till the end of July ispreferable for coping up erratic rainfall distribution as wellas avoidance of damage from heavy downpour during har-vesting in the island (Table 6).

Several short to medium duration improved rainfed low-land rice cultivars tolerant to stresses (flooding/submergence,

drought, salinity, diseases, etc.) have been developed andwidely adopted by the farmers after extensive participatorymode multi-locational field evaluation in rainfed low-lyingareas including coastal region of Orissa and other parts ofIndia, comparable to the agro-ecological condition of SagarIsland (Adhya et al. 2008; Adhikari et al. 2010; Sarkar andBhattacharjee 2011; Bhowmick et al. 2014). Hence, for SagarIsland, we have suggested some of these widely adopted im-proved mega rice cultivars (e.g., MTU7029, Swarna-Sub1,IR64-Sub1, Sambha Mahsuri, Samba Mahsuri-Sub1,Sahbhagi Dhan, CSRC(S)-2-1-7, CSRC-36; Table 6).Swarna (MTU7029) is a popular (30–40 % rainfed lowlandsof India) transgenic fine grain, high-yielding (>3 t/ha) cultivarfor temporary submergence and saline area (Dhar et al. 2013).It is also resistant to serious insect pest (yellow stem borer)(Rao et al. 2008) and diseases (bacterial blight) (Sarkar andBhattacharjee 2011). A flood-resistant gene called Sub1 wasinserted into Swarna rice variety to create the Swarna-Sub1,resilient to flood submergence for up to 2 to 3 weeks withoutany penalty to its high potentials of yield advantage of 1.65 t/ha over Swarna (Anonymous 2013; Sarkar and Bhattacharjee2011). Subsequently, other high-yielding (4.75 to 5.0 t/ha)alternate drought- and submergence-tolerant medium duration(<140 days) mega rice varieties, namely IR64-Sub1, SambaMahsuri-Sub1, and Sahbhagi Dhan, were developed, exten-sively evaluated, and popularized across flood-prone areas ofIndia. They all are quality rice varieties with multi-purposeculinary uses, resistant to major pests (stem borer) and dis-eases (bacterial blights) with premium grain quality(Iftekharuddaula et al. 2011; Sarkar and Bhattacharjee 2011;Anonymous 2013; Dhar et al. 2013). Similarly, CSRC(S)-2-1-7 and CRS-36 are the high-yielding (3.0–4.0 t/ha), saline (6.0–8.0 dsm−1), and submergence-tolerant rice cultivars, resistantto major pests (stem borer, leaf hopper) and diseases (bacterialblights, tungro viruses), especially suited to the coastal salinesoils of India, similar to Sagar Island, West Bengal

0

10

20

30

40

50

60

70

80

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

>75mm >100mm >125mm >150mm >50mm

Meteorological week

Pro

ba

bil

ity

of

wee

kly

ra

infa

ll,%

Fig. 5 Probability of gettingweekly rainfall using gammaprobability distribution functionat Sagar Island

Int J Biometeorol

(Anonymous 2013). Besides improving rice productivity,adoption of these short–medium duration cultivars will facil-itate crop intensification through profile moisture rechargeand its reuse as residual soil moisture for growing post-ricelow water requiring winter crops, such as green chili, watermelon, potato, khesari (Lathyrus sativus), sunflower, mustardseeds, etc. Thus, the risk in food grain vis-à-vis socioeconomyand rural livelihood security of the islanders even under ad-verse climatic aberration will be minimized through resiliencyin agriculture to a great extent.

Conclusions

The study on southwest monsoon rainfall behavior and itsprobability analysis revealed that the effective rainfall forrainfed lowland kharif rice cultivation in Sagar Island wasavailable for 17-week (24th MW–40th MW) duration.During wet land preparation and seedling transplantation(24th to 27th MW), probability of receiving 50 mm WR wasless than 60 %. From 28th MW, probability of 50 mm WRfurther rose at >60 % level, and crop water balance exhibitedsurplus condition. Therefore, transplanting of seedlingsshould be started from 29th MW (16–22 July) with higherprobability (p>60 %) of effective rainfall occurrence.During 18th MW to 23rd MW with less probability (p>15.1to <40.1 %) of WR occurrence, dry tillage and seedbed prep-aration for dry seeded nursery can be initiated, while only afteronset of monsoon (24th MW) farmers are suggested to go forwet nursery raising and wet tillage including puddling follow-ed by transplanting of seedlings. To escape the severe mon-soon rainfall break and subsequent water stress during activevegetative stage (28–31th MW), heavy downpour during latewithdrawal of monsoon in the maturity stage (41–45th MW),and adaptation to other stresses (flooding, submergence,coastal salinity, and temperature change), adoption of high-yielding (4.75 to 5.0 t/ha) short-to medium-duration (120–140 days) popular multiple stress-tolerant rice cultivars (e.g.,IR64-Sub1, Swarna-Sub1, Samba Mahsuri-Sub1, andSahbhagi Dhan) is one of the feasible technical options tosustain rice productivity in the island. Information generatedin the present investigation might be helpful for rainfed kharifrice planning, especially in adoption of sustainable agronomicand water management practices for food grain, socio-econo-my, and livelihood security in Sagar and other vulnerableislands in the coastal areas of India and other parts.

References

Adhikari B, Bag MK, Bhowmick MK, Kundu C (2010) Status paper onrice in West Bengal. Rice Research Station Govt. of West BengalChinsurah – 712102 West Bengal. pp 1–88

Adhya TK, Singh ON, Swain P, Ghosh A (2008) Rice in Eastern India:causes for low productivity and available options. J Rice Res 2(1):1–5

Allen RG, Pereira L S, Raes D, SmithM (1998) Crop evapotranspiration:guidelines for computing crop water requirements. Irrigation andDrainage Engineering Paper 56. United Nations Food andAgriculture Organization, Rome, Italy

Anonymous (2010–2011) Department of Agriculture. Annual ActionPlan on Agriculture South 24 Parganas, Office of the DeputyDirector of Agriculture (Administration), South 24 Parganas,Government of West Bengal, India. 1: 1–450

Anonymous (2013) Cluster demonstrations of stress tolerant rice varietiesin stress prone parts of India. Annual Report (2012–2013).International Rice Research Institute, NASC Complex, Pusa, NewDelhi, India

Babu PN, Lakshminarayana P (1997) Rainfall analysis of a dry landwatershed-Polkepad: a case study. J Indian Water Res Soc 17:34–38

Basavaraju HK, Joshi M (2000) Rainfall distribution index—a new ap-proach to analyze rainfall for better crop management. Indian JAgric Sci 70:806–809

Bhowmick MK, Dhara MC, Singh S, Dar MH, Singh US (2014)Improved management options for submergence-tolerant (Sub1)rice genotype in flood-prone rainfed lowlands of West Bengal. AmJ Plant Sci 5:14–23

Bouman BAM, Lampayan RM, Tuong TP (2007) Water management inirrigated rice: coping with water scarcity. International RiceResearch Institute, Los Banos, pp 1–54

Bouman BAM, Tuong TP (2001) Field water management to save waterand increase its productivity in irrigated rice. AgricWaterManag 49:11–30

Box GEP, Jenkins GM (1976) Time series analysis: forecasting and con-trol, 2nd edn. Holden Day, San Francisco

Brian C (1988) Rainfall, ponding and flood irrigation—their importanceto rice growing in the Adelaide river area, 1–38. http://www.nt.gov.au/d/content/File/p/Tech-Bull/TB126.pdf

Chand BK, Trivedi RK, Dubey SK, Beg MM (2012) Aquaculture inchanging climate of Sundarbans. Survey Report on ClimateChange Vulnerabilities, Aquaculture Practices and CopingMeasures in Sagar and Basanti Blocks of Indian Sundarbans, WestBengal University of Animal and Fishery Sciences, Kolkata, India.Online at http://www.wbuafscl.ac.in/

Choudhury BU, Das A, Ngachan SV, Slong A, Bordoloi LJ, ChoudhwryP (2012) Trend analysis of long term weather variables in mid alti-tude Meghalaya, North-East India. J Agric Phys 12:12–22

Choudhury BU, Singh AK, Pradhan S (2013) Estimation of crop coeffi-cients of dry seeded irrigated rice–wheat rotation on raised beds byfield water balance method in the Indo-Gangetic plains, India. AgricWater Manag 123:20–31

Dhar MH, de Janvry A, Emerick K, Raitzer D, Sadoulet E (2013) Flood-tolerant rice reduces yield variability and raises expected yield, dif-ferentially benefitting socially disadvantaged groups. ScientificReports 3; DOI: 10.1038/srep03315

Dey S, Banerjee S, Saha A (2011)Water deficit patterns for cultivation ofrainfed rice in the lower Gangetic plains of West Bengal. J AgricPhys 11:79–83

Fischer RA (1973) The effect of water stress at various stages of devel-opment on yield processes in wheat. In: Slatyer RO (ed) Plant re-sponses to climate factors. United Nations Educational, Scientificand Cultural Organization, Paris, pp 233–241

Gadgil S, Rupa Kumar K (2006) The Asian monsoon—agriculture andeconomy. B Wang (ed.), The Asian

Iftekharuddaula KM, Newaz MA, Salam MA, Ahmed HU, MahbubMAA, Septiningsih EM, Collard BCY, Sanchez DL, PamplonaAM, Mackill DJ (2011) Rapid and high-precision marker assistedbackcrossing to introgress the SUB1 QTL into BR11, the rainfedlowland rice mega variety of Bangladesh. Euphytica 178:83–97

Int J Biometeorol

Jat ML, Singh RV, Balyan JK, Jain LK (2005) Analysis of weekly rainfallfor crop planning in Udaipur region. J Agric Eng 42:35–41

Kendall MG (1975) Rank correlation methods. Charles Griffin, LondonKumar A, Tripathi P, Singh AK (2007) Effects of dry spell on growth,

development and yield of rice (Oryza sativa). Indian J Agric Sci 76:47–49

Kumar KK, Rupa Kumar K, Ashrit RG, Deshpande NR, Hansen JW(2004) Climate impacts on Indian agriculture. Int J Climatol 24:1375–1393. doi:10.1002/joc.1081

Mandal DK, Mandal C, Raja P, Goswami SN (2010) Identification ofsuitable areas for aerobic rice cultivation in the humid tropics ofeastern India. Curr Sci 99:227–231

Mandal S, Choudhury BU (2014) Estimation and prediction of maximumdaily rainfall at Sagar Island using best fit probability models. TheorAppl Climatol 117(3–4), doi: 10.1007/s00704-014-1212-1

Mandal S, Choudhury BU, Mondal M, Bej S (2013) Trend analysis ofweather variables in Sagar Island, West Bengal, India: a long-termperspective (1982–2010). Curr Sci 105:947–953

Mann HB (1945) Nonparametric tests against trend. Econometrica 13:245–259

Mishra AK, Sarkar JK, Bhattacharya AK (1999) Stocashastic rainfallforecasting for weekly water resources assessment in lowerAssam. Indian J Soil Conserv 27:99–111

Mishra S (2012) Climate change and adaptation strategy in Agriculture—a West Bengal scenario. Geog Rev Ind 74:1–16

Mondal KG, Padhi J, Kumar A, Sahoo DK, Majhi P, Ghosh S, MohantyRK, Roychaudhuri M (2013) Analyzing rainfall events and soilcharacteristics for water resources management in a canal irrigatedarea. J Water Resour Ocean Sci 2:1–8

Nath SK, Roy D, Kiran KST (2008) Disaster mitigation and managementfor West Bengal, India—an appraisal. Curr Sci 94(7):858–864

Panigrahi B, Panda SN (2001) Analysis of weekly rainfall for crop plan-ning in rainfed region. J Agric Eng 38(4):47–57

Rao MVR, Behera KS, Baisakh N, Datta SK, Rao GJN (2008)Transgenic indica rice cultivar ‘Swarna’ expressing a potato chymo-trypsin inhibitor pin2 gene show enhanced levels of resistance toyellow stem borer. Plant Cell Tissue Organ Cult 99:277–285. doi:10.1007/s11240-009-9602-2

Reddy GVS, Bhaskar SR, Purohit RC, Chittora AK (2008)Markov chainmodel probability of dry, wet weeks and statistical analysis of

weekly rainfall for agricultural planning at Bangalore. Karnataka JAgric Sci 21:12–16

Sarkar RK, Bhattacharjee B (2011) Rice genotypes with SUB1 QTLdiffer in submergence tolerance, elongation ability during submer-gence and re-generation growth at re-emergence. Rice 5(7):1–11

Sen PK (1968) Estimates of the regression coefficient based on Kendall’stau. J Am Stat Assoc 39:379–389

Singh N, Ranade A (2009) The wet and dry spells across India during1951–2007. J Hydrometeorol 11:26–45

Sivakumar MVK (1990) Exploiting rainy season potential from the onsetof rains in the Sahelian zone of West Africa. Agric For Meteorol 51:321–332

Stern RD, Cooper PJM (2011) Assessing climate risk and climate changeusing rainfall data—a case study from Zambia. Exp Agric 47:241–266

Stewart JI (1991) Principles and performance of response farming. In:Muchow RC, Bellamy JA (ed), Proceedings of the internationalsymposium on Climatic risk in crop production: Models and man-agement for the semiarid tropics and subtropics, Brisbane, Australia,pp 361–382

Subash N, Rammohan HS, Sikka AK (2011) Integrating rainfall proba-bility and moisture availability index for crop planning during kharifrice (Oryza sativa) in eastern Indo-Gangetic basin. Indian J AgricSci 81:843–851

Swain DK, Herath S, Pathirane A,Mittra BN (2005) Rainfed lowland andflood-prone rice: a critical review on ecology and management tech-nology for improving the productivity in Asia. Online at http://www.mekongnet.org/images/e/e9/Dillip.pdf

Traore B, Corbeels M,Wijkc MTV, RufinoMC, Giller KE (2013) Effectsof climate variability and climate change on crop production insouthern Mali. Eur J Agron 49:115–125

Tsuda M (1993) Grain filling in rice subjected to drought at three stages.Jpn J Crop Sci 62:199–205

WWF-India (2010) Sunderbans: future imperfect, climate adaptationreport, pp 1–29

Zeleke KT, Wade LJ (2012) Evapotranspiration estimation using soilwater balance, weather and crop data. In: Irmak A (ed)Evapotranspiration—remote sensing and modeling. In Tech,ISBN: 978-953-307-808-3 Online at: http://www.intechopen.com/books/evapotranspiration remote-sensing

Int J Biometeorol