Cc

CEa

b

c

d

e

a

ARA

KWTGCWI

1

cuptc2Ir2sw

Pf

h0

Agricultural Water Management 148 (2015) 52–62

Contents lists available at ScienceDirect

Agricultural Water Management

jou rn al hom epage: www.elsev ier .com/ locat e/agwat

limate-smart tank irrigation: A multi-year analysis of improvedonjunctive water use under high rainfall variability

. Sideriusa,∗, H. Boonstraa, V. Munaswamyb, C. Ramanab, P. Kabatc,

. van Ierlandd, P. Hellegersd,e

Alterra, Wageningen University and Research Centre, Wageningen, The NetherlandsAcharya N.G. Ranga Agricultural University, Hyderabad, Andhra Pradesh, IndiaInternational Institute for Applied Systems Analysis, Laxenburg, AustriaEnvironmental and Natural Resources Group, Wageningen University, The NetherlandsWater Resources Management Group, Wageningen University, The Netherlands

r t i c l e i n f o

rticle history:eceived 13 March 2014ccepted 7 September 2014

eywords:ater harvesting

ank rehabilitationroundwater

a b s t r a c t

Although water harvesting is receiving renewed attention as a strategy to cope with increasing seasonaland inter-annual rainfall variability, many centuries-old local water-harvesting reservoirs (tanks) in Indiaare rapidly deteriorating. Easy access to groundwater is seen as one of the major threats to their mainte-nance and functioning. Potentially, however, conjunctive use of water from rain, tanks and groundwaterreserves, supported by proper monitoring, could improve the resilience and productivity of traditionaltank irrigation systems. To date, few quantitative multi-annual analyses of such climate-smart systemshave been published. To redress this, we assess the sustainability of a rehabilitated tank irrigation system,

onjunctive useater productivity

ndia

by monitoring all inputs and outputs over a period of six years (12 cropping seasons). Our results showthat during the period considered, improved conjunctive use resulted in a more stable cropping intensity,increased economic water productivity and higher net agricultural income. Groundwater tables were notnegatively affected. We argue that improved conjunctive use can considerably reduce the vulnerabilityof tank irrigation to rainfall variability and thus is a valuable strategy in light of future climate change.

© 2014 Elsevier B.V. All rights reserved.

. Introduction

India faces severe seasonal and regional water shortages in theoming decades. Demand from agriculture, by far the biggest waterser, is increasing, to support the growing and increasingly affluentopulation (National Academy of Agricultural Sciences, 2009). Athe same time, availability of water is under pressure due to climatehange and overexploitation of groundwater resources (Biemans,012; Rodell et al., 2009). Although average total rainfall over the

ndian subcontinent is likely to remain unchanged, the variability inainfall is expected to increase (Kumar et al., 2011; Mathison et al.,013). In a monsoonal climate that is already erratic and highly

easonal in nature, this increased variability due to climate changeill further impact water availability.

∗ Corresponding author at: Alterra, Wageningen University and Research Centre,O Box 47, 6700 AA Wageningen. The Netherlands. Tel.: +31 317 486448;ax: +31 317 419000.

E-mail address: Christian.siderius@wur.nl (C. Siderius).

ttp://dx.doi.org/10.1016/j.agwat.2014.09.009378-3774/© 2014 Elsevier B.V. All rights reserved.

In order to cover periods of shortages, farmers in India havefor thousands of years been constructing so-called tanks1 to har-vest and store rainfall and surface runoff (Gunnell et al., 2007; VonOppen and Subba Rao, 1987). Serving more than 20% of croppedarea in southern states, tank irrigation is still one of the majorstrategies for coping with rainfall variability. In tank irrigation sys-tems, water is harvested during the monsoon and used during thesubsequent dry season. It is a flexible system, in which the volumeof water stored in the tank at the end of the monsoon determineswhat and how much area farmers crop. Although this does not guar-antee a stable year-to-year production and income, farmers preventloss of investments by making timely adjustments to the croppingplan and allocation of resources. Besides their primary purpose as asource of water for irrigation, tanks also have important secondary

purposes, such as the provision of drinking water, flood mitigationand water for livestock and fish production (Palanisami and Easter,1983; Palanisami and Meinzen-Dick, 2001).

1 Artificial lakes, generally with earthen embankment dams, for harvesting andstoring surface runoff after heavy rainfall.

ater M

f22immcswRbuG2gPgoiotiui

mcaumatetf2VgtDmtsljht

usapcsTcwstatpicr

support from the FAO.3 The rehabilitation entailed both institu-tional improvements, such as enhancing the empowerment of theWUA, and technical and agronomical improvements. To augment

C. Siderius et al. / Agricultural W

Despite their advantages, many tank irrigation systems haveallen into disrepair during the past 50 years (Kajisa et al.,007; Palanisami and Meinzen-Dick, 2001; Sakurai and Palanisami,001): throughout India, the cropped area supported by tank

rrigation has declined from 19% in the 1950s to 4% at present. Theain causes of this decline have been (i) centralization of wateranagement, whereby the state took over the responsibility of

ommunal tanks, which led to an institutional breakdown withevere implications for maintenance schemes and the collection ofater charges (Palanisami and Easter, 1983; Von Oppen and Subbaao, 1987); (ii) siltation and encroachment of farming onto the tanked, both symptoms of institutional breakdown and a higher pop-lation pressure (Dasog et al., 2012; Easter and Palanisami, 1985;unnell and Krishnamurthy, 2003; Palanisami and Meinzen-Dick,001) and (iii) access to cheap and easily available canal water androundwater (Dasog et al., 2012; Kajisa et al., 2007; Sakurai andalanisami, 2001). While tank irrigation declined, irrigation withroundwater rose sharply in India: it now accounts for almost 60%f the irrigated area. Farmers with access to groundwater have lessncentive to contribute to the communal maintenance of the tanknce it has deteriorated (Sakurai and Palanisami, 2001). They preferhe rapid return on investments in boreholes rather than contribut-ng to the rehabilitation of the tank system. The resulting free-ridingndermines the runoff harvesting, storage and distribution capac-

ty of the tank system.Groundwater also offers opportunities to enhance the perfor-

ance of the tank system, however, by providing additional storageapacity to buffer seasonal and inter-annual shortages in rainfallnd tank water (Ranganathan and Palanisami, 2004). Conjunctivese – maximizing the yield of water resources by the coordinatedanagement of supplies of surface water and groundwater – is

well described concept in large-scale surface water supply sys-ems (Bredehoeft and Young, 1983; Burt, 1964; Tsur, 1990), butmpirical evidence on its benefit for tank irrigation and rehabilita-ion is limited. Early evaluation of tank rehabilitation programmesocussed largely on the merits of participatory execution (ADB,006; Gunnell & Krishnamurthy, 2003; Palanisami & Easter, 1983;on Oppen & Subba Rao, 1987). It mainly described how well pro-rammes were executed and their internal efficiency, rather thanheir efficacy in terms of achieving the desired effect. Recently,asog et al. (2012) and Reddy and Behera (2009) followed aore quantitative approach, comparing yields and improvements

o livelihoods before and after rehabilitation, but without payingpecific attention to changes in water use. To our knowledge, noongitudinal empirical studies have been reported in which con-unctive use of water from rain, tanks and groundwater reservesas been monitored over several years, thus taking into accounthe high inter-annual variability in rainfall.

Our aim is to assess the sustainability of improved conjunctivese of rainfall, tank water and groundwater in a tank irrigationystem. We base our assessment on primary data collected over

period of 6 years, comprising 12 cropping seasons. During thiseriod, all water inputs and yield outputs of a single tank irrigationommand area were measured at farm and tank level in an exten-ive monitoring campaign as part of a tank rehabilitation project.he performance of the tank system was assessed using three indi-ators; cropping intensity, net agricultural income and economicater productivity. Whether groundwater resources were used

ustainably was assessed by groundwater level observations. Sec-ion 2 explains the monitoring approach and three indicators usednd gives a short background description of the study site andhe rehabilitation measures implemented during the monitored

eriod. In Section 3 we present the annual performance of the tank

rrigation and the impact on groundwater levels. The paper con-ludes with a discussion on the observed changes and the widerelevance of our findings.

anagement 148 (2015) 52–62 53

2. Methodology

2.1. Study area

Our case study area is a tank irrigation site near the villageof Musilipedu, approximately 45 km east of the town of Tiru-pati, in the Yerpedu Mandal2 of Chittoor District, in the state ofAndhra Pradesh, India (79◦42′E and 13◦36′N). The region is mostlyinfluenced by the north-east monsoons (October–December)and, to a much lesser extent, by the south-west monsoons(June–September). Average annual rainfall at Tirupati is 988 mm(1975–2006 period) with a high inter-annual variability not onlyin quantity (238 mm standard deviation), but also in the number oflow and high-intensity rainfall events.

The Musilipedu tank is a non-system tank, fed solely by rainfallin its catchment area, with no connections to other tanks, upstreamor downstream. The area upstream of the tank, i.e. the tank’s catch-ment area, is approximately 740 ha. When full, the tank covers54 ha; the irrigated area is 188 ha (Fig. 1). The tank has two com-partments, separated by a low bund. When both compartments arefull, excess water can flow over two surplus weirs into the Swarna-mukhi River. Irrigation water from the tank can be diverted into thetank command area through a culvert, closed by a gate. It is thendiverted by gravity from the main channels into a tertiary systemconsisting of field channels dug by the farmers.

The distribution of the tank water is managed by the WUA. Onlyfarmers who own land can become members. The WUA farmersnumber 223, with an average land holding of 0.6 ha. In the Kharifcropping season (1 June–15 October) only a portion of the com-mand area is cultivated, mainly with groundnuts and rice, and thelimited rainfall during the south-west monsoon is supplemented bygroundwater irrigation. During this season the tank remains empty.The area cropped in the Rabi cropping season (15 October–15March) largely depends on how much water has accumulated inthe tank during the north-east monsoon (September–November)at the start of the season. The main crop cultivated is paddy rice.Supplementary irrigation from groundwater is applied, especiallyat the end of the growing season and in the tail ends of the irrigationcanals. From April to June the entire command area is left fallow,except for a small area cropped with sugarcane.

In common with the trend throughout India, groundwater use inthe Musilupedu tank irrigation site has steadily increased in recentdecades. Groundwater is abstracted from a shallow aquifer whichis replenished during the monsoon, after which groundwater lev-els rise to near the soil surface. Although the initial investmentrequired is substantial, boreholes are cheap to exploit (fuel andelectricity are subsidized), reliable (under the farmer’s own con-trol) and efficient (water is available when and where needed). InAndhra Pradesh, electricity is provided for free to farmers, thoughonly for several hours a day, with power cuts occurring regu-larly. 54% of the 223 farmers had access both to groundwaterand tank water, 42% relied on tank water alone and 8% used onlygroundwater. The average farm size of farmers with access to bothgroundwater and tank water was, at approximately 1 ha, more thantwice that of farmers with only tank water.

During the monitoring period the Musilipedu tank was rehabil-itated using standard funds from the District Collector with further

2 A mandal is an administrative division in India, above which is the district andbelow which are the villages.

3 FAO, with the Dutch Government, funded the APWAM project, an 8-year projecton improving water productivity in irrigated agriculture in Andhra Pradesh.

54 C. Siderius et al. / Agricultural Water Management 148 (2015) 52–62

F on of

t .

rcriarwie

TC

ig. 1. Location of the Musilipedu tank, with catchment boundaries (general directio color in this figure legend, the reader is referred to the web version of this article

unoff into the tank through the four supply channels, farmersonstructed revetments from boulders. Sediment in the tank wasemoved with earth-moving equipment and a programme of chem-cal treatment each summer was instigated, to control noxiousquatic weeds. In the tank command area, the defective sluice waseplaced by a new one by the Irrigation Department. A new gate

as installed to regulate and control the total outflow of tank water

nto the tank command area. The two main irrigation channels werequipped with lock gates to measure and control the distribution

able 1osts of technical interventions during the Musilipedu tank rehabilitation.

Measure Number of items Cost per

Sluice gates 2 60,000Lock gates 2 29,000Renovation sluice

Division boxes 5 40,000Lining irrigation channels* 1350 1050Construction RBC flumes 3 5722Weed removal tank bed

Total

* Length in metres.

slope and flow of water is from south to north). For interpretation of the references

of tank water to the fields. The main irrigation channels, with atotal length of 1350 m including five division boxes, were graduallylined by the farmers, using cement and bricks. Costs for these meas-ures are given in Table 1. The chronological order of the variousinterventions was mainly determined by the farmers.

On farmers’ fields, several agronomic interventions were

tested. Alternative rice crop water management packages wereintroduced, such as system of rice intensification (SRI) and alterna-tive wetting and drying (AWD). Different crops for green manuring,

item (INR) Total costs (INR) Total costs (USD)

120,000 2727 58,000 1318

270,000 6136 200,000 4545

1,417,500 32,216 17,166 390

70,000 1591

2,152,666 48,924

ater M

wAidoa

2

owtcaectrscfitoa‘

tcgf

Wwosutct9t

bPtsmBrlstr

tli(m

C. Siderius et al. / Agricultural W

hich increases soil fertility and crop yields, were demonstrated.nother groundnut variety better suited to local conditions was

ntroduced to the farmers. Also, an improved tillage implement wasesigned. None of the agronomic measures was actively promotedr supported by financial incentives. Farmers had total freedom todopt or ignore the measures demonstrated.

.2. Methods

To evaluate conjunctive use in a tank irrigation site we devel-ped an approach to monitor performance in terms of land andater use, based on low-cost and low-maintenance monitoring

echniques. Using such techniques, farmers and WUA membersould themselves do the monitoring, requiring only limited guid-nce from external agricultural extension workers or irrigationxperts. The performance monitoring was primarily based on arop-specific water budgeting method: for each cropping season,he actual volume of water supplied was compared with crop waterequirements (Fig. 2). Based on this comparison, land and water usetrategies could be adapted in the subsequent cropping season. Theycle was repeated for several consecutive years, which improvedarmers’ insight into their resource use and generated an overallnsight into the effectiveness of the various improvements in theank irrigation system. This integration of performance monitoringf land and water use to support conjunctive use, combined with

range of technical innovations in a participatory setting, we callimproved conjunctive use’.

Actual water supplied was monitored from three sources: effec-ive rainfall, tank water and groundwater. Effective rain within theommand area, i.e. the amount of rain actually benefitting croprowth, i.e. not contributing to runoff or seepage, was calculatedrom total monthly rainfall as (Dastane, 1978):

if P ≤ 16.7 mm/month Pef = 0

if P > 16.7 and < 75 mm/month Pef = P ∗ 0.6 − 10

if P > 75 mm/month Pef = P ∗ 0.75 − 25

(1)

ith P as rainfall and Pef as effective rainfall. Effective rainfallas furthermore assumed not to exceed the total monthly evap-

transpirative demand of the major crops, rice, groundnut andugarcane. Actual volume of tank water supplied was calculatedsing a standard tank conveyance efficiency of 70% in the initialhree years and an estimated 80% efficiency after rehabilitation,orrecting for losses between the tank outlet measurement loca-ion and farmer fields. Groundwater irrigation efficiency was set at0%, as minimal losses are expected over short distances within theank command area.

Crop water requirements for each of the major crops wereased on the crop factor (Kc) method that uses the Modifiedenman–Monteith equation (Allen et al., 1998) to estimate poten-ial evapotranspiration. The meteorological data input were dailyunshine hours, humidity, wind speed and minimum and maxi-um temperature from the Tirupati weather station 40 km away.

ased on the comparison between actual supply and potentialequirements, strategies to modify water allocations were formu-ated and discussed with the WUA members. If agreed upon, thesetrategies were then implemented by the WUA members volun-arily in the subsequent cropping season, and the above cycle wasepeated.

To evaluate the impact of improved conjunctive use we usedhree indicators: two are related to the basic production factors

and (cropping intensity) and water (economic water productiv-ty), whereas the third is related to the main output for a farmernet agricultural income). The production factor of labour was not

onitored, because it was assumed it would be constant during

anagement 148 (2015) 52–62 55

the monitoring period, as sufficient labour was available. However,unsolicited comments from farmers on the rising cost of labourshowed that changes can be significant over several years, sug-gesting that in future analyses, labour should indeed be takeninto account more explicitly. In areas of high rainfall variability,analysing these indicators for six years rather than conducting animpact assessment by comparing two years (before and after reha-bilitation) provides a better comparison. The three initial yearsbroadly represent the initial situation, during which rehabilitationmeasures were gradually being implemented, while the last 3 yearsrepresent the “after” situation, with improved land and water man-agement. Both periods contained years of drought and years ofabundant rainfall.

Annual total cropped area was calculated for the two croppingseasons of Kharif and Rabi individually for the major crops rice,groundnut, sugarcane and for sunflower. The cropped area of sug-arcane, which has a growing period of 11–12 months, was addedto the seasonal cropped area of both Kharif and Rabi. The crop-ping intensity (CI) was derived for each season by dividing seasonalcropped area by the total command area. To calculate overall annualnet agricultural income (Inet), cropped area was multiplied by yieldand market prices per crop, after which crop-specific total costsof cultivation were subtracted. This was done annually, and foreach farmer, and for the tank command area as a whole. Annualeconomic water productivity (WPecon) at tank command level wascalculated as:

WPecon = Net agricultural incomeWater available

in INR or USD/m3 (2)

Water available was calculated at tank command level by totallingsupplied tank water and groundwater, before subtracting anyconveyance losses, and effective rainfall. Effective rainfall wasincluded, as we considered it a resource that can be managed orused effectively in combination with targeted tank and ground-water applications. The inclusion of effective rainfall also preventsannual fluctuations in productivity that result from variations ineffective rainfall from showing up in the indicator. By focussingon the tank command level, WPecon is able to show productiv-ity gains as a result both of crop water management practicesat field level and improved distribution and timing of deliverywithin the command area, i.e. the quality of irrigation management.Improvements at both field and command area scale are often inter-linked (van Halsema and Vincent, 2012). We avoid the term WaterUse Efficiency, as the numerous, often value-laden, interpretationscomplicate its use (Perry, 2007; van Halsema and Vincent, 2012).

2.3. Data collection

The monitoring period covered six consecutive years of the tankrehabilitation, starting with the Kharif season of 2004. Primary dataon water use, crop production and market prices were collectedat intervals ranging from days to years. To derive water use, rain-fall was measured daily using a rain gauge installed in the tankcommand area. Tank outflow into the command area was derivedfrom daily gauge readings of three installed RBC flumes, fromNovember 2006 onwards. Before 2006, stage-discharge relationswere used. Groundwater use was based on daily interviews withfarmers on their pumping hours. For each bore hole, pumped vol-ume was calculated multiplying pumping hours with each pump’sdesign capacity based on factory specifications; the accuracy ofthis method was checked by periodically inserting a commercialwater metre in the discharge pipe. Groundwater levels were mon-

itored weekly at 14 observation wells inside and outside the tankcommand area.

During each cropping season, the various types of crops grownwere identified and their areal extent recorded per land holding.

56 C. Siderius et al. / Agricultural Water Management 148 (2015) 52–62

Fig. 2. Flow diagram of the performance monitoring approach, with a cycle repeated after each year i.

ater M

Trptlwp

piz(dlf

3

3

ics2nar2iihaitemdca2

6pbtwdAfiawlcft

3

ttygi

C. Siderius et al. / Agricultural W

he dates on which crops were sown, planted or transplanted wereecorded, and so were the harvest dates. At the end of each crop-ing season, crop cuttings experiments were made to determinehe crop yield of certain preselected land holdings; one third of alland holdings were sampled each time. In addition, each farmer

as interviewed to ascertain production in terms of kilograms ofroduce sold on the local market.

In 2005, a one-off socio-economic survey focussing on costs ofroduction was conducted. Farmers in the command area were

nterviewed, to collect data on the operational costs (seed, fertili-ers, chemicals, machine maintenance), labour costs and fixed coststaxes, rental values of owned land), and from these the total pro-uction costs for each crop were estimated. In addition, prices in

ocal and regional markets were collected, in order to calculate thearmer’s income from each crop.

. Results

.1. Impact on cropping intensity

Fig. 3 shows the variation in total cropped area and the availabil-ty of rainfall, tank water and groundwater. Traditionally, the arearopped in Kharif depends on the onset of the south-west mon-oon. If rains are insufficient or too late, planting is cancelled. In004/05 this led to a sharp decrease in area cropped with ground-ut. During Rabi, the extent of cropped area depends mainly on thevailability of tank water. If the tank is not full, the cropped area iseduced, as happened to the area cropped with rice in 2004/05 and006/07. Water supplied to the fields thus follows the fluctuations

n the amount of water required during most seasons (Table 2). Dur-ng years of relatively abundant rainfall, water supplied is 10–30%igher than water required, while during years of shortage it ispproximately 10–30% less. When cropping strategies of all 223ndividual farmers for the dry year of 2006/07 were compared withheir strategies for 2007/08 and 2008/09, it was found that farm-rs who had access only to tank water reduced their cropped areaost: over these three years, the average variation in cropped area

uring Rabi was more than 60% (relative standard deviation). Inontrast, for those farmers with access to groundwater water, theverage variation in cropped area over these three years was only5% and half maintained a stable cropping pattern during Rabi.

Over the course of the monitoring period, CI stabilized to over0% during Kharif and almost 100% in Rabi. In particular, the arealanted with rice, a crop with a high water demand, increased inoth the Kharif and Rabi cropping seasons. This high CI was main-ained during the dry year of 2009/10, the last year of monitoring,hen the amount of effective rainfall and tank water was as low asuring the first year of the study: the dry year of 2004/05 (Table 2).

better conjunctive use of rain, tank and groundwater proved suf-cient to buffer the rainfall shortage. Variations in cropped areare not only a resultant of water availability: sugarcane growingas actively promoted in the beginning of the study period by

ocal sugar factories, and the area cropped reached up to 17% of theommand area. However, a sharp increase in labour costs causedarmers to lose interest in growing sugarcane, with the result thathe area under this crop shrank.

.2. Impact on yield, income and water productivity

In addition to a more stable CI, an increase in productivity inerms of yield per hectare was observed. Table 3 shows yields of

he main crops in the Musilipedu tank command area. The riceield in the Kharif season rose gradually during the 6 years, whereasroundnut yields in both Rabi and Kharif seasons improved mostn the last three years, after all rehabilitation measures had been

anagement 148 (2015) 52–62 57

implemented. Rice yields in the Rabi season also increased, butdropped again in the dry year of 2009/10. For sugarcane, no trendswere observed. Sunflower was only grown for 2 seasons and theyields were 1.5 and 1.4 t/ha. The yields of all the crops in bothseasons are considerably higher than regional or national yields.

The increase in CI and yields did not lead to a higher wateruse (Table 4). Improved conjunctive use of rainfall, tank water andgroundwater led to a considerable increase in WPecon, especiallyduring dry years. An important effect of the yearly performancemonitoring was a gradual adjustment of the water needed forpaddy rice: from the national advised 1200 mm to approximately800 mm per year, an amount sufficient under local climatic circum-stances. As a result, the WPecon increased by almost 40%, from US$0.050/m3 for the first three years of monitoring to US$ 0.069/m3

cents per m3 for the last three years (at 2005 exchange rates).As a result of higher yields and a more stable CI, overall Inet

increased (Table 4). When average Inet per farm, expressed in USDper hectare, is plotted as a function of water availability, a cleardifference is apparent between the initial three years and the lastthree years (Fig. 4A). An indication of the increased buffer capacityof the tank system is the continued high Inet in the very dry seasonof 2009/10. The increase in average farm income seems to be theresult of an increase in both maximum and minimum Inet (Fig. 4B),though the spread between farmers’ Inet remains large in later yearstoo. Farmers with access only to tank water remain more likely toleave land fallow during the Kharif season, and to have a lower Inet.

3.3. Sustainability of conjunctive use of tank water andgroundwater

Conjunctive use of rainfall, tank water and groundwaterrequires land for seasonal storage of water and a groundwa-ter aquifer for inter-annual storage. Groundwater depletion andencroachment of farming on the tank and upstream catchment areaare two risks for a sustainable tank irrigation system.

Over the whole monitoring period, the current use of groundwa-ter did not result in continuing groundwater depletion (Fig. 5). Thesystem is recharged every year by high-intensity rainfall events inthe command area and by infiltration from the tank, which usuallyfills in October. During the Kharif and Rabi seasons, the system isdepleted by natural lateral groundwater outflow towards the riverand groundwater pumping. However, sustainable use of ground-water is not guaranteed. The performance monitoring revealedthat farmers owning bore wells were over-irrigating their cropsin the initial period, especially in the third year (2006/07, Table 5).The installation of automatic power switches resulted in pumpsrunning whenever the power supply was on. This led to a notice-able reduced recovery of groundwater levels during the 2006/07Rabi season (Fig. 5). In reaction, specific pumping schedules wereintroduced, based on site-specific crop water requirements calcu-lated using the available meteorological data. One of the elderlyfarmers was selected to act as a special pump operator, to ensurethat the principles of these schedules were complied with. Thepartial groundwater recovery in 2006/07 was compensated in thesubsequent two years, which indicates that the average annualgroundwater recharge is just sufficient to cover incidental highpumping rates.

Although groundwater availability reduces the relevance of thetank for farmers owning pumps, switching to a system that is fedonly by rainfall and groundwater is unlikely to be sustainable in theMusilipedu area. Conjunctive use of tank and groundwater to sup-plement rainfall allows farmers to cope with the erratic behaviour

of the north-eastern monsoon: low- to medium-intensity rainfallevents contribute mostly to effective rainfall, medium-intensityrainfall events lead to most groundwater replenishment, and onlythe high-intensity events fill the tank. Table 5 shows how rainfall

58 C. Siderius et al. / Agricultural Water Management 148 (2015) 52–62

Fig. 3. Cropped area per cropping season per year as a percentage of the total command area, and per crop. Crop water supply from different sources per year (blue circlerepresents the maximum supplied amount of water, in the agricultural year 2005/06). Tank water and groundwater are net figures (1000 m3) after subtracting conveyancelosses. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

C. Siderius et al. / Agricultural Water Management 148 (2015) 52–62 59

Table 2Cropped area, Water available, water supplied – after subtracting conveyance losses – and water required based on potential evaporation per crop for Kharif (K) and Rabi (R)season. Tank water could not be attributed to individual crops as it was measured at the tank outlet only. It is primarily used during the Rabi season.

Crop Crop area(ha)

Tank water(1000 m3)

Effective rain(1000 m3)

Groundwater(1000 m3)

Water available(1000 m3)

Water supplied(1000 m3)

Water required(1000 m3)

2004/05 Rice K 8 10 55a 49Groundnut K 20 25 75a 94Rice R 113 61 220 843Groundnut R 38 7 16 170Sugarcane 3 16 65 33

Total 182 670 120 431 1221 976 1187

2005/06 Rice K 10 22 86 54Groundnut K 72 161 129 317Rice R 145 306 31 1266Groundnut R 10 2 34 41Sugarcane 33 152 286 393

Total 269 1033 643 565 2241 1875 2071

2006/07 Rice K 19 21 199 108Groundnut K 75 57 271 352Rice R 82 67 603 671Groundnut R 28 0 102 124Sunflower 7 NA NA NASugarcane 32 123 493 362

Total 242 446 269 1667 2382 2081 1617

2007/08 Rice K 15 31 120 98Groundnut K 81 238 89 379Rice R 147 211 0 1028Groundnut R 16 6 0 71Sunflower 8 NA NA NASugarcane 16 96 165 178

Total 282 1294 582 374 2250 1954 1754

2008/09 Rice K 37 27 317 235Groundnut K 71 51 247 333Rice R 152 163 383 901Groundnut R 14 0 24 69Sugarcane 9 29 79 101

Total 283 784 269 1051 2103 1842 1638

2009/10 Rice K 34 46 301 207Groundnut K 79 104 143 363Rice R 167 22 383 1247Groundnut R 6 3 17 27Sugarcane 3 12 31 37

Total 289 515 187 873 1576 1385 1883

S

depih

TA

a Based on estimates of pumping hours.ource: village secretary

uring the north-eastern monsoon was distributed over differ-

nt intensity categories during the six years of the monitoringrogramme. During the study period it was observed that signif-

cant tank fillings took place for rainfall events with intensitiesigher than 60 mm/day. From this dataset it can be seen that such

able 3verage crop yields (tonnes/ha) in the tank command area for each year and the Andhra

Agricultural years Rice Kharif Groundnut Khari

2004/05 4.4 2.0

2005/06 4.7 2.0

2006/07 4.4 2.0

2007/08 4.8 2.6

2008/09 5.1 2.6

2009/10 5.4 2.6

Andhra Pradesh averageb 2.8 1.0

All-India averageb 2.0 0.7

a Average of the 2010/11 and 2011/12 cropping seasons.b Includes crop yields of all rainfed and irrigated systems, not only tank irrigation.

rainfall events occurred in 3 of the 6 years. In years with high-

intensity rainfall, these events contributed to more than 40 to totalrainfall (Table 4) and the tank provided up to 40% to total water used(Fig. 3). In years with fewer high-intensity events the tank fills onlypartly and the absolute contribution of tank water to irrigation is

Pradesh and all-India averages (tonnes/ha) for the 6-year period (GoI, 2012, 2013).

f Rice Rabi Groundnut Rabi Sugarcane

5.3 2.5 976.0 1.9 1176.0 2.4 916.4 2.4 1006.6 2.7 765.7 3.3 993.7 1.8 80a

3.1 1.9 70a

60 C. Siderius et al. / Agricultural Water Management 148 (2015) 52–62

Fig. 4. Average net agricultural income (Inet) as a function of water availability for the initiin USD/ha (B).

Table 4Performance indicators for the Musilipedu tank command area. Water availableis calculated as effective rainfall plus water supplied from tank and groundwaterreserves, before subtracting any conveyance losses. Inet is the total for the wholetank command area. WPecon was calculated as an average for the tank commandarea.

Crop season Wateravailable(103 m3)

Net income (Inet)(106 Indian Rupee(INR))

Economic waterproductivity (WPecon)INR/m3 (USD cents/m3)

2004/05 1221 2.82 2.31 (5.2)2005/06 2241 5.71 2.55 (5.8)2006/07 1903 4.37 2.30 (4.1)2007/08 2251 6.11 2.72 (6.2)2008/09 2268 5.98 2.64 (6.5)

FR

2009/10 1576 5.52 3.51 (8.0)

ig. 5. Composite groundwater level of the Musilipedu tank command area, based on theabi and Kharif cropping periods is based on annual farmer interviews.

al three years and last three years (A), with spread in individual farm Inet, expressed

less. But in these years, effective rainfall is also less, which meansthat the contribution from the tank remains relatively important.Gradual encroachment of farming onto the tank bed or a deliberateconversion of the tank and part of the catchment area to croplandwould result in an important source of water being lost. Such aloss of water cannot be compensated from the shallow ground-water aquifer, except by tapping deeper groundwater aquifers.Moreover, tank water and groundwater are highly connected andencroachment of farming on the tank area will also reduce ground-water replenishment, leading to a reduction in both tank andgroundwater resources.

4. Discussion and conclusion

Traditionally, tank irrigation is a dynamic form of irrigationin India, in areas where high variability in rainfall leads to

average of four observation bore wells in the tank command area. Duration of the

C. Siderius et al. / Agricultural Water M

Tab

le

5N

orth

-eas

t

mon

soon

rain

fall

(Oct

ober

–Nov

embe

r–D

ecem

ber)

for

the

Mu

sili

ped

u

tan

k

site

in

mm

per

inte

nsi

ty

cate

gory

(wit

h

the

nu

mbe

r

of

even

ts

in

brac

kets

)

and

as

tota

l of a

ll

cate

gori

es, a

nd

tota

l an

nu

al

rain

fall

.

Yea

r

0–20

mm

/day

inte

nsi

tym

m

(#

of

even

ts)

21–4

0

mm

/day

inte

nsi

tym

m

(#

of

even

ts)

41–6

0

mm

/day

inte

nsi

tym

m

(#

of

even

ts)

61–8

0

mm

/day

inte

nsi

tym

m

(#

of

even

ts)

81–1

00

mm

/day

inte

nsi

tym

m

(#

of

even

ts)

>100

mm

/day

inte

nsi

tym

m

(#

of

even

ts)

Tota

lO

ctob

er–D

ecem

ber

mm

Tota

l an

nu

alra

infa

llm

m

2004

184

(15)

125

(4)

45

(1)

130

(2)

–

–

485

864

2005

162

(21)

333

(12)

157

(3)

131

(2)

86

(1)

733

(5)

1601

2120

2006

114

(20)

217

(7)

116

(2)

124

(2)

87

(1)

–

658

1049

2007

68

(12)

182

(6)

162

(3)

140

(2)

–

441

(3)

993

1534

2008

127

(12)

305

(10)

–

–

100

(1)

270

(2)

802

1083

2009

53

(6)

176

(6)

155

(3)

124

(2)

–

–

509

920

anagement 148 (2015) 52–62 61

considerable inter-annual fluctuations in cropping intensity,income and water productivity. In our multi-annual analysis wehave shown that improved conjunctive use of rainfall, tank waterand groundwater can reduce these fluctuations and lead to higherand more stable cropping intensity, economic water productivityand net agricultural income. These increases appear sustainable,with groundwater being able to recover annually.

Whether improved conjunctive use can be successfully scaledout to other tank sites will depend on the local availability anddistribution of land and water resources. A threat to most tankirrigation sites is the smallness of landholdings, with ‘marginal’and ‘small’ landholding classes being dominant, and farmers havinglimited opportunities for expansion. There is therefore a contin-ued risk of encroachment of farming on the tank and upstreamcatchment areas, and of overuse of groundwater in order to furtherincrease cropping intensity. However, with yields almost doubleregional and all-India yields, improved conjunctive tank irrigationin our case study site appears to be economically viable. And fur-ther improvements in water productivity, using more advanced soilmoisture monitoring and rainfall forecasting, seem feasible, whichwould preserve more groundwater for use in the Kharif periodwhen at present 40% of the area is still left fallow. Linking theseimprovements to a better understanding of the resource base, aswas done in the performance monitoring method presented here,can help limit the risk of overexploitation.

When upscaling these results, higher water productivity at tankcommand level will, in general, not linearly lead to higher waterproductivity at the larger basin scale. There is an on-going debatein the literature on the merits of efficiency improvements in basinswhere losses and return flows are fully used again by downstreamusers (Perry, 2007; Seckler et al., 2003; van der Kooij et al., 2013;van Halsema and Vincent, 2012). Often, higher water productivityleads to an increase in cropping intensity or cropped area, as in ourcase study, and to these losses being limited. However, whetherthis impacts downstream users will depend on the period in whichlosses occur and the location of the tank within a tank system orthe river basin. In our case study site, an improved capture andstorage of runoff mainly reduced losses during the monsoon, whenwater is abundant anyway. Outside the monsoon, return flows andconveyance losses contribute to the local groundwater system. Asthe area is close to the river mouth, any reduction in losses has littleimpact downstream.

Finally, whether tank irrigation is able to adapt to a future cli-mate in which rainfall variability is expected to increase over India(Kumar et al., 2013) cannot be answered merely by an empiricalstudy. But our results do show that improved conjunctive use insmall-scale tank systems is able to buffer the kind of inter- andintra-annual variability in rainfall expected in future. More gen-erally, decentralized systems with a potential for self-organizationare considered to be very adaptable to change and to be less affectedby sudden change or failure in parts of the system. Improved con-junctive tank irrigation, in which farmers have control over theirland and water resources and pro-actively adapt to each season,shows clear characteristics of such a system. The lessons go beyondsmall-scale systems: decentralized storage and smart conjunctiveuse of water resources could also be considered in the case oflarge-scale canal irrigation, where the combination of higher rain-fall variability, groundwater depletion, over-allocation of irrigationwater and increasing competing claims from industrial and domes-tic uses jeopardizes the stable supply of water.

Improved tank rehabilitation, as presented in this study,requires little additional investment compared to traditional tank

rehabilitation with its exclusive focus on technical interventions.One prerequisite is the availability of a local organization that candisseminate the required knowledge on crop-specific irrigationwater requirements and water supply monitoring to WUAs. In the

6 ater M

prtRhtst

A

PNiPat(UTepiKsnart

R

A

A

B

B

B

D

2 C. Siderius et al. / Agricultural W

resent study, the regional office of a State university fulfilled thisole. Throughout India there is an extensive network of agricul-ural extension services under the Indian Council of Agriculturalesearch, which could do likewise. Strengthening their capacity toelp WUAs should be promoted as part of a climate-smart agricul-ure. The benefits – a more stable income for farmers and a moretable food production to support a growing population – are likelyo exceed the costs.

cknowledgements

Data supporting this study were collected as part of the Andhraradesh Water Management Project (APWAM) which ran from 1ovember 2003 to 31 October 2010 with the main objective of

mproving water productivity in irrigated commands in Andhraradesh, India. The executive authority of the project was the Foodnd Agriculture Organization (FAO), New Delhi. The implemen-ing agencies were the Acharya N.G. Ranga Agricultural UniversityANGRAU), Hyderabad, Andhra Pradesh and Alterra, Wageningenniversity and Research Centre, Wageningen, The Netherlands.his joint applied research study was conducted in farmers’ fields inight different pilot areas throughout Andhra Pradesh; one of theseilot areas was located in the Musilipedu tank command. The writ-

ng of this paper was supported by the strategic research programBIV “Sustainable spatial development of ecosystems, landscapes,eas and regions” which was funded by the Dutch Ministry of Eco-omic Affairs. Herco Jansen, Jochen Froebrich and Henk Ritzemare thanked for commenting on earlier versions. Joy Burrough car-ied out substantive and language editing of a near-final draft ofhe paper.

eferences

DB, 2006. Rehabilitation and management of tanks in India: a study of selectedstates, ADB; Manilla.

llen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop Evapotranspiration – Guide-lines for Computing Crop Water Requirements. FAO – Food and AgricultureOrganization of the United Nations, Rome.

iemans, H., 2012. Water Constraints on Future Food Production, Earth System Sci-ence and Climate Change Group. Wageningen UR, Wageningen.

redehoeft, J.D., Young, R.A., 1983. Conjunctive use of groundwater and surfacewater for irrigated agriculture: risk aversion. Water Resour. Res. 19, 1111–1121.

urt, O.R., 1964. The economics of conjunctive use of ground and surface water.Hilgardia 36, 31–111.

asog, G.S., Lingappa, S., Kadekod, G.K., Radder, B.M., 2012. Agricultural Production,Resource Conservation and Farm Income Elevation in Tank Commands of NorthKarnataka. University of Agricultural Sciences, Dharwad, pp. 130.

anagement 148 (2015) 52–62

Dastane, N.G., 1978. Effective Rainfall in Irrigated Agriculture. FAO – Food and Agri-culture Organization of the United Nations, Rome.

Easter, K.W., Palanisami, K., 1985. Tank Irrigation in India: An Example of CommonProperty Resource Management, Conference on Common Property ResourceManagement , pp. 215–230.

GoI, 2012. Agricultural Statistics at a Glance. Government of India, Ministry of Agri-culture, New Delhi.

GoI, 2012–2013. State of Indian Agriculture. Government of India, Ministry of Agri-culture, New Delhi.

Gunnell, Y., Anupama, K., Sultan, B., 2007. Response of the South Indian runoff-harvesting civilization to northeast monsoon rainfall variability during thelast 2000 years: instrumental records and indirect evidence. The Holocene 17,207–215.

Gunnell, Y., Krishnamurthy, A., 2003. Past and present status of runoff harvestingsystems in dryland peninsular india: a critical review. AMBIO 32, 320–324.

Kajisa, K., Palanisami, K., Sakurai, T., 2007. Effects on poverty and equity of the declinein collective tank irrigation management in Tamil Nadu, India. Agric. Econ. 36,347–362.

Kumar, K., Kamala, K., Rajagopalan, B., Hoerling, M., Eischeid, J., Patwardhan, S., Srini-vasan, G., Goswami, B., Nemani, R., 2011. The once and future pulse of Indianmonsoonal climate. Clim. Dynam. 36, 2159–2170.

Kumar, P., Wiltshire, A., Mathison, C., Asharaf, S., Ahrens, B., Lucas-Picher, P., Chris-tensen, J.H., Gobiet, A., Saeed, F., Hagemann, S., Jacob, D., 2013. Downscaledclimate change projections with uncertainty assessment over India using a highresolution multi-model approach. Sci. Total Environ. 468–469 (Suppl.), S18–S30.

Mathison, C., Wiltshire, A., Dimri, A., Falloon, P., Jacob, D., Kumar, P., Moors, E., Ridley,J., Siderius, C., Stoffel, M., 2013. Regional projections of North Indian climate foradaptation studies. Sci. Total Environ. 468, S4–S17.

Palanisami, K., Easter, K.W., 1983. The Tanks of South India (A Potential for FutureExpansion in Irrigation), Economic Reports. University of Minnesota, St. Paul.

Palanisami, K., Meinzen-Dick, R., 2001. Tank performance and multiple uses in TamilNadu, South India. Irrig. Drain. Syst. 15, 173–195.

Perry, C., 2007. Efficient irrigation; inefficient communication; flawed recommen-dations. Irrig. Drain. 56, 367–378.

Ranganathan, C.R., Palanisami, K., 2004. Modeling economics of conjunctive surfaceand groundwater irrigation systems. Irrig. Drain. Syst. 18, 127–143.

Reddy, V.R., Behera, B., 2009. The economic and ecological impacts of tank restora-tion in South India. Eur. J. Dev. Res. 21, 112–136.

Rodell, M., Velicogna, I., Famiglietti, J.S., 2009. Satellite-based estimates of ground-water depletion in India. Nature 460, 999–1002.

Sakurai, T., Palanisami, K., 2001. Tank irrigation management as a local commonproperty: the case of Tamil Nadu, India. Agric. Econ. 25, 273–283.

National Academy of Agricultural Sciences, 2009. State of Indian Agriculture, NewDelhi.

Seckler, D., Molden, D., Sakthivadivel, R., 2003. The concept of efficiency in waterresources management and policy. In: Water Productivity in Agriculture: Limitsand Opportunities for Improvement., pp. 37–51.

Tsur, Y., 1990. The stabilization role of groundwater when surface water suppliesare uncertain: the implications for groundwater development. Water Resour.Res. 26, 811–818.

van der Kooij, S., Zwarteveen, M., Boesveld, H., Kuper, M., 2013. The efficiency of dripirrigation unpacked. Agric. Water Manage. 123, 103–110.

van Halsema, G.E., Vincent, L., 2012. Efficiency and productivity terms for water

management: a matter of contextual relativism versus general absolutism. Agric.Water Manage. 108, 9–15.

Von Oppen, M., Subba Rao, K.V., 1987. Tank Irrigation in Semi-Arid Tropical India:Economic Evaluation and Alternatives for Improvement, Research Bulletin No.10. Patancheru, ICRISAT.