Water reuse and cost-benefit of pumping at different spatial levels in a rice irrigation system in UPRIIS, Philippines

Page 1

123

4

5

6

7

8

9

101112

13

14

15

16

17

Q1

AGWAT 2454 1–9

NC

OR

RE

CTE

D P

RO

OFScale effects on water use and water productivity in a

rice-based irrigation system (UPRIIS) in the Philippines

M.M. Hafeez a,b, B.A.M. Bouman c,*, N. Van de Giesen a,d, P. Vlek a

aCenter for Development Research (ZEF), Bonn University, Bonn 53113, GermanybCSIRO Land and Water, Wagga Wagga, LMB 588, NSW 2678, Australiac International Rice Research Institute (IRRI), DAPO Box 7777, Manila, PhilippinesdTechnical University of Delft, PO Box 5048, 2600 GA Delft, Netherlands

a g r i c u l t u r a l w a t e r m a n a g e m e n t x x x ( 2 0 0 7 ) x x x – x x x

a r t i c l e i n f o

Article history:

Accepted 11 May 2007

Keywords:

Rice

Irrigation

Water accounting

Water reuse

Water productivity

Spatial scale

a b s t r a c t

Between 25% and 85% of water inputs to rice fields are lost by seepage and percolation. These

losses can be reused downstream and do not necessarily lead to true water depletion at the

irrigation system level. Because of this potential for reuse, the general efficiency of water use

can increase with increasing spatial scale. To test this hypothesis, a multi-scale water

accounting study was undertaken in District I of the rice-based Upper Pampanga River

Integrated Irrigation System (UPRIIS) in the Philippines. Daily measurements of all surface

water inflows and outflows, rainfall, evapotranspiration, and amounts of water internally

reused through check dams and shallow pumping were summed into seasonal totals for 10

spatial scale units ranging from 1500 ha to 18,000 ha.

The amount of net surface water input (rainfall plus irrigation) per unit area decreased

and the process fraction, depleted fraction, water productivity, and amount of water reuse

increased with increasing spatial scale. In total, 57% of all available surface water was

reused by check dams and 17% by pumping. The amount of water pumped from the

groundwater was 30% of the amount of percolation from rice fields. Because of the reuse

of water, the water performance indicators at the district level were quite high: the depleted

fraction of available water was 71%, the process fraction of depleted water was 80% (close to

the 75% area covered by rice), water productivity with respect to available water was

0.45 kg grain m�3 water, and water productivity with respect to evapotranspiration was

0.8 kg grain m�3 water. Water use in the district can be reduced by cutting down the

49 � 106 m3 uncommitted outflows. The depleted fraction of available water can be

increased to 80% or more by a combination of adopting alternate wetting and drying

(AWD) and increased pumping to capture percolating water. Water productivity with

respect to available water can be increased to 0.83 kg grain m�3 water by a combination

of reduced land preparation time, adoption of AWD, and increased fertilizer N use to

increase yields.

# 2007 Published by Elsevier B.V.

avai lab le at www.sc iencedi rec t .com

journal homepage: www.e lsev ier .com/ locate /agwat

18

19

20

21

U1. Introduction

Rice is eaten by about three billion people and is the most

common staple food in Asia (Maclean et al., 2002). Some 75% of

* Corresponding author. Tel.: +63 2 845 0563; fax: +63 2 845 0606.E-mail address: [email protected] (B.A.M. Bouman).

0378-3774/$ – see front matter # 2007 Published by Elsevier B.V.doi:10.1016/j.agwat.2007.05.006

Please cite this article in press as: Hafeez, M.M. et al., Scale effects on

(UPRIIS) in the Philippines, Agric. Water Manage. (2007), doi:10.1

the world’s annual rice production is harvested from

79 million ha of irrigated lowland rice, mainly in Asia, where

it accounts for 40–46% of the net irrigated area of all crops

(Dawe, 2005). Because of its large area, and because rice

water use and water productivity in a rice-based irrigation system

016/j.agwat.2007.05.006

Page 2

E

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

Fig. 1 – Location map of District I, UPRIIS, and the spatial

units of the case study. The dots indicate the location of

check dams for reuse of surface drainage water.

a g r i c u l t u r a l w a t e r m a n a g e m e n t x x x ( 2 0 0 7 ) x x x – x x x2

AGWAT 2454 1–9

UN

CO

RR

receives relatively much water, Bouman et al. (2006) estimated

that 34–43% of the world’s irrigation water is used to irrigate

rice. However, water resources are getting increasingly scarce

and rice is a main target for water-saving initiatives (Rijsber-

man, 2006).

Total seasonal water input to rice fields (rainfall plus

irrigation) is up to two to three times more than for other

cereals (Tuong et al., 2005). It varies from as little as 400 mm in

heavy clay soils with shallow groundwater tables to more than

2000 mm in coarse-textured soils with deep groundwater

tables (Bouman and Tuong, 2001; Cabangon et al., 2004).

Around 1300–1500 mm is a typical value for irrigated rice in

Asia. Because of these large water inputs, the water produc-

tivity of rice with respect to water inputs is quite low: the

average reported value for rice at the field level of

0.4 kg grain m�3 water is about two times smaller than that

of wheat (Tuong et al., 2005). The large water inputs are mostly

caused by surface drainage and seepage and percolation flows

from the continuously ponded fields into the groundwater,

creeks, and drains. Seepage and percolation flows account for

about 25–50% of all water inputs in heavy soils with shallow

(20–50 cm depth) groundwater tables (Cabangon et al., 2004;

Dong et al., 2004), to 50–85% in coarse-textured soils with

groundwater tables of 1.5 m depth or more (Sharma et al.,

2002; Choudhury et al., 2007). Therefore, most water-saving

technologies developed at the field level aim to reduce seepage

and percolation flows (Bouman and Tuong, 2001; Tuong et al.,

2005). However, though these flows are losses at the field level,

they can be captured and reused downstream and do not

necessarily lead to true water depletion at the irrigation

system level. Therefore, it has been argued that the efficiency

of water use and the water productivity of rice may increase

with increasing spatial scale and may be much higher at the

irrigation system level than at the individual field level (Tuong

et al., 2005). To test this hypothesis, water flows within an

irrigation system would need to be tracked at different spatial

scales. Also, the water flows would need to be separated into

reusable flows and real depletion flows (such as evapotran-

spiration), and amounts of water reuse would need to be

estimated. Loeve et al. (2004) measured water flows at ‘‘micro’’

and ‘‘meso’’ scales in the 467,000 ha Zanghe Irrigation System

(ZIS), in Hubei Province, China, in which 27% of the area was

cropped with lowland rice. The micro-scales consisted of

small sets of farmers’ fields that were together less than 1 ha,

and the meso scales were areas within the irrigation system of

287 ha and 606 ha. Water inflows were also available for main

canal command areas of 28,000–196,000 ha and for the whole

irrigation system. Water productivity decreased from micro to

meso scale, but increased from meso scale to canal command

area and to the whole irrigation system. The number of scale

levels at which detailed flow measurements were made,

however, was not large enough to make solid conclusions on

the relationship between scale and water productivity and

other water use parameters. In this paper, we present results

for a multi-scale study on water use in a rice-based surface

irrigation system in the Philippines. Using the water account-

ing principles of Molden (1997) and Molden and Sakthivadivel

(1999), and computing internal water flows, we calculated

water productivity and various water use indicators for 10

different spatial scale levels. We also estimated formal and

Please cite this article in press as: Hafeez, M.M. et al., Scale effects on

(UPRIIS) in the Philippines, Agric. Water Manage. (2007), doi:10.1

CTE

D P

RO

OF

informal water reuse through surveys and actual water flow

measurements. The ultimate aim was to test the hypothesis

that the efficiency of water use and water productivity of rice

increases with increasing spatial scale because of the reuse of

seepage, percolation, and drainage water. We also used the

water performance indicators to suggest where improvements

in water management in the system could be made.

2. Materials and methods

2.1. Study area

Our study area was District I of the 102,000 ha Upper

Pampanga River Integrated Irrigation System (UPRIIS) in

Central Luzon, Philippines (Fig. 1). UPRIIS is owned and

operated by the National Irrigation Administration (NIA) of

the Philippines with the main purpose of providing irrigation

water to rice fields. District I has a total area of 28,205 ha,

including rice fields (dominant land use), upland crops,

vegetables, roads, settlements, and water bodies. The district

is bounded by the Talavera River to the east and the Ilog

Baliwag River to the west, and consists of an upper part, called

the Talavera River Irrigation System-Lower (TRISL), and a

lower part, called the Santo Domingo Area (SDA). Water is

supplied by Diversion Canal No. 1, which gets its water from

the Pantabangan reservoir, and the TRIS main canal, which

gets its water from the Talavera River through a run-off-the-

river diversion dam. The major direction of water flow is from

northeast to southwest, though locally, water flows in various

directions according to topography. The TRIS main canal first

supplies water to an irrigation system north of, and contig-

uous to, District I, called TRIS-Upper. The area is quite flat,

with elevations of around 20 m above sea level. Soils are

Vertisols, Entisols, and Inceptisols, and have typically silty

clay, silty clay loam, clay loam, and clay textures. The climate

is characterized by two pronounced seasons, dry from

November to April and wet for the rest of the year. The

water use and water productivity in a rice-based irrigation system

016/j.agwat.2007.05.006

Page 3

C

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

172

173

174

175

176

177178

179

180

181

182

183

184

185

a g r i c u l t u r a l w a t e r m a n a g e m e n t x x x ( 2 0 0 7 ) x x x – x x x 3

AGWAT 2454 1–9

E

average annual rainfall is about 1900 mm, of which 90% falls in

the wet season (Tabbal et al., 2002).

Our study was conducted in the 2000–2001 dry season. It

started with the first release of water in the main canals on 19

November 2000, and ended with the harvest of the last rice

crops on 18 May 2001. District I was subdivided into four

spatial units (Fig. 1): TRISL, SDA-A, SDA-B, and SDA-C. All

boundaries were roads, which were selected in such a way

that all surface water flowing in and out of the areas

underneath the roads could be measured. A downstream

part of District I in the southwest was excluded since this part

of the system was inaccessible by roads and no well-defined

boundaries were available across which water flows could be

measured. Six additional spatial units were obtained for the

water accounting analysis by combining contiguous units, so

that a total of 10 units were obtained: the four original ones,

plus SDA-AB, SDA-BC, SDA-ABC, TRISL + SDA-A, TRISL + SDA-

AB, and TRISL + SDA-ABC. Surface water (inflow and/or

outflow) flowed across the boundaries of all units.

2.2. Water accounting

The water accounting framework is basically an analysis of

the water balance, in which the outflows are classified

according to their use, or potential use, within or outside

that area (Table 1). The three-dimensional boundaries of our

study area were the horizontal outer boundaries of the 10

combined units, the top of the surface/vegetation, and the

bottom of the rootzone. Keeping in line with Loeve et al. (2004),

we focused our water accounting on surface water, and we

separately analyzed the net flow of water across the lower

boundary (rootzone) as the closing term of the water balance.

We computed all water flows as seasonal totals, from 19

November 2000 to 18 May 2001. The gross inflow was rainfall

plus all surface irrigation water. The net inflow was the gross

inflow minus the change in water stored at the surface (mainly

in the canals) and in water stored in the rootzone of the crops

UN

CO

RR

Table 1 – Water accounting and performance indicators in DisSakthivadivel (1999))

Indicator

Gross inflow Total amount of water flowing into the

surface irrigation

Net inflow Gross inflow less the change in storage

equals gross inflow in our study area

Committed water Part of outflow from the spatial units t

irrigation by the other spatial units

Uncommitted outflow Water which is neither depleted nor co

because of lack of storage or operation

Available water Net inflow less the committed water

Process flow That part of the water outflows that is

evapotranspiration is the only process

Depletion flow Water outflow of the spatial units that

percolation flows to unrecoverable sin

depletion flows are evapotranspiration

Process fraction (PF) Ratio of process depletion (rice evapot

(PFavailable), and to depleted water (PFde

Depleted fraction (DF) Ratio of total depletion (both rice and

Water productivity (WP) Ratio of rice grain production to gross

depletion (WPET)

Please cite this article in press as: Hafeez, M.M. et al., Scale effects on

(UPRIIS) in the Philippines, Agric. Water Manage. (2007), doi:10.1

TED

PR

OO

F

from the beginning to the end of the cropping season. Since

the canals were dry before the start of the season and after

harvest of the last crops, the change in surface water storage

was zero. We neglected the change in stored soil water in the

rootzone since our dry season crop followed straight after a

wet season crop and we assumed that the amount of soil water

was the same after a harvested wet season crop and after a

harvested dry season crop. Therefore, the net inflow was the

same as the gross inflow. All surface outflows were considered

‘‘committed’’ when they flowed into a neighboring spatial unit

or further downstream in the irrigated area of District I. All

water flowing out of District I was considered ‘‘uncommitted’’

since there was no immediate major water user downstream

of District I. The only outflow considered ‘‘depletion’’ was

evapotranspiration (ET) since no water percolated to irretrie-

vably deep or saline groundwater. Since the purpose of UPRIIS

is to irrigate rice, we considered only rice ET as ‘‘process

depletion,’’ and all nonrice ET as nonprocess depletion

(following Loeve et al., 2004). Nonprocess depletion is

evaporation from bare soil, build-up area, open water bodies,

and nonrice crops. The transpiration from nonrice crops is a

beneficial water use, of course, but no data on nonrice crops

were available for this study, so this particular use of water

could not be quantified.

2.3. Water balance

The water balance of each spatial unit was calculated as

dW ¼ I� O� ETðmmÞ (1)

where I is the net surface inflow by irrigation and rainfall,O the

surface outflows, and ET is the evapotranspiration of all rice

and nonrice surfaces. The term dW should be interpreted as

the net result of water percolating downward, capillary rise,

and groundwater pumping across the lower boundary (bottom

of the rootzone) of our study area. These components could

not be as accurately assessed as the surface water flows, and,

in the case of percolation, could not be readily classified as

trict I, UPRIIS (adapted from Molden (1997) and Molden and

Description

spatial units of the irrigation system, comprising rainfall and

over the time period within the growing season. Net inflow

hat is committed to downstream uses, in our study area for

mmitted and is available for use, but flows out of the spatial units

al measures

truly used by the intended user. In our study area, rice

flow

is no longer available for reuse, such as evapotranspiration or

ks such as deep or saline groundwater. In our study area, the only

ranspiration) to gross water inflow (PFgross), to available water

pleted)

nonrice ET) to gross inflow (DFgross) and to available water (DFavailable)

water inflow (WPgross), to available water (WPavailable), and to process

water use and water productivity in a rice-based irrigation system

016/j.agwat.2007.05.006

Page 4

E

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214

215

216

217

218

219

220

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243

244

245

246

247

248

249

250

251

252

253

254

255

256

257

258

259

260

261

262

263

264

265

266

267

268

269

270

271

272

273

274

275

276

277

278

279

280

281

282

283

284

285

286

287

288

289

290

291

292

293

294

295

296

297

298

299

300

a g r i c u l t u r a l w a t e r m a n a g e m e n t x x x ( 2 0 0 7 ) x x x – x x x4

AGWAT 2454 1–9

UN

CO

RR

‘‘committed’’ or ‘‘uncommitted.’’ Moreover, part (if not all) of

the water pumped from the groundwater is in fact reuse of

percolation water and is analyzed separately as internal water

reuse. When dW is positive, it means that water is added to the

groundwater or to the soil layers below the rootzone; when it is

negative, it means that water is extracted from the ground-

water or from the soil layers below the rootzone (Bouman

et al., 2007). It should be noted, however, that since dW is

calculated as the closing term of the water balance, it also

includes all errors in the measurement or computation of the

individual water balance components.

Linear regression was performed on water balance com-

ponents against surface area of the spatial scale units.

2.4. Internal water reuse

UPRIIS was designed to reuse surface water through check

dams in creeks and drainage ways. Farmers added to this

water reuse by constructing their own dams that have

subsequently been formalized by the irrigation system

management. There are now a total of 15 check dams in

District I, which are operated and maintained by either NIA or

by groups of farmers. We estimated water flows in inlets at 9

of the 15 check dams (see below). Farmers also informally

reuse water by pumping from shallow groundwater, creeks,

and drains. We surveyed all farmers in the area on pump

ownership and pump use, and counted the number of pumps

in each of our spatial units. We then selected 50 farmers

representative for the different types of pumps and pump

use, and monitored their pump operations during the

growing season. Each pump was calibrated, and pumped

water volumes from surface water and groundwater were

obtained by multiplying calibrated flow rates by recorded

durations of pumping. The pumped water volumes were

extrapolated to our spatial units using the total number of

pumps in each unit.

Groundwater pumping can mean the reuse of water

percolated down from rice fields and/or the use of ground-

water that originated from outside the area. We estimated the

total volume of water percolating down from rice fields by

multiplying the rice area by a mean percolation rate of

2 mm d�1 as reported for the TRIS and SDA areas by Lucero

(1984) and Sattar (1992). This calculation is a conservative

estimate of total percolation flows through the lower

boundaries of our areas since it does not include water

percolating from waterways and nonrice fields.

We also estimated the change in groundwater storage from

the start to the end of the growing season by measuring

groundwater depths in 50 observation wells throughout

District I. The differences in depth were multiplied by a

soil-type specific storage coefficient that was estimated as the

difference in water content between saturation and field

capacity derived from data reported by Ramos (1986). Changes

in groundwater storage for each spatial unit were obtained by

overlaying our spatial units with the soil map (containing the

calculated storage coefficients) and with Thiessen polygons

around the observation wells (containing the measured

changes in groundwater depth). The calculated change in

groundwater storage indicates whether net groundwater

recharge or depletion takes place.

Please cite this article in press as: Hafeez, M.M. et al., Scale effects on

(UPRIIS) in the Philippines, Agric. Water Manage. (2007), doi:10.1

CTE

D P

RO

OF

2.5. Water flow measurements

Surface water flows were tracked twice a day by measuring

water depths and establishing rating curves to estimate flow

volumes. At the inlets of six of the internal check dams, we

installed staff gauges and at three others we installed

automatic water level recorders. We could not measure water

flows at the remaining six dams because of inaccessibility

and/or semi-stagnant water conditions. Surface inflows and

outflows across the boundaries of our spatial units were

measured in all drains, creeks, channels, and culverts (a total

of 158 points) underneath the roads that formed the

boundaries. For most open waterways, we measured water

depth with installed staff gauges, while, for some smaller

waterways, we used V-notch weirs or Parshall flumes. For

culverts, we used standard flow equations and measured

water depths up- and downstream where possible, or we

established our own rating curves again. Rating curves were

established by plotting measured water depths against flow

volumes obtained with current meters. Calibrations were

made until each rating curve (for each staff gauge location)

had an R2 of 0.95 or more. Further details on the measurements

are given by Hafeez (2003). An accuracy analysis using the

electronic water level recorders showed that daily flow rates

obtained from the twice a day readings of the staff gauge were

on average 13% higher than the daily flow rates obtained from

5 min interval readings of water depth with the electronic

recorders.

Rainfall was measured twice a day with eight rain gauges

installed throughout District I, and spatially extrapolated

using Thiessen polygons. Total rainfall for each of our spatial

units was obtained by overlaying with the Thiessen polygons.

We used six MODIS and three Landsat optical satellite

images and the surface energy balance algorithm for land

(SEBAL; Bastiaanssen, 1995) to estimate actual ET. SEBAL is a

thermodynamically based model, which partitions between

sensible heat flux and latent heat of vaporization flux. All

meteorological data for calibration of SEBAL, such as air

temperature, skin water temperature, soil temperature, air

humidity, wind speed, and solar radiation, were collected

hourly on the day of satellite overpass from two weather

stations in District I. The SEBAL-derived ET values were

extrapolated in time using measured (Class A) pan evapora-

tion and calculated Penman–Monteith reference ET using the

daily weather station data. The ET was divided over rice

(process outflow) and other land covers (nonprocess outflow)

based on a supervised classification checked by groundtruth-

ing using a Landsat image from 31 March (the classification

accuracy was estimated as 90%; Hafeez, 2003). The ‘‘other land

covers’’ included mainly built-up areas, open water bodies,

nonrice crops, and bare soil. The final output was a pixel-based

map of seasonal ET in District I, which was then used to obtain

spatially aggregated ET values for all our spatial units. More

details of the approach are given by Hafeez et al. (2002).

2.6. Water performance indicators

For each spatial unit, we calculated a number of water

accounting indicators and water productivity (WP) following

the procedures presented by Molden (1997), as summarized in

water use and water productivity in a rice-based irrigation system

016/j.agwat.2007.05.006

Page 5

CTE

D P

RO

OF

301

302

303

304

305

306

307

308

309

310

311

312

313

314

315

316

317

318

319

320

321

322

323

324

325

326

327

328

329

330

331

332

333

334

335

336

337

338

339

340

341

342

343

344

345

346

347

348

349

350

351

352

353

354

355

Ta

ble

2–

Ma

inch

ara

cteri

stic

sa

nd

wa

ter

acc

ou

nti

ng

com

po

nen

tsfo

r1

0sp

ati

al

sca

les

inD

istr

ict

I,U

PR

IIS

SD

A-A

SD

A-B

SD

A-C

SD

A-A

BS

DA

-BC

SD

A-A

BC

TR

ISL

TR

ISL

+S

DA

-AT

RIS

L+

SD

A-A

BT

RIS

L+

SD

A-A

BC

To

tal

are

a(h

a)

1513

2240

3011

3753

5251

6764

11,2

39

12,7

52

14,9

92

18,0

03

Ric

ea

rea

(ha

)1177

1709

1972

2886

3681

4858

8,7

13

9,8

90

11,5

99

13,5

71

Up

lan

dcr

op

(ha

)86

242

415

328

657

743

886

972

1,2

14

1,6

29

Rest

(ha

)250

289

624

539

913

1163

1,6

40

1,8

90

2,1

79

2,8

03

Ric

ey

ield

(th

a�

1)

4.7

35.5

84.8

35.1

65.2

05.0

56.0

95.4

15.4

75.3

1

Fa

rmers

(nu

mb

er)

751

901

1051

1652

1952

2703

7,2

07

7,9

58

8,8

59

9,9

10

Pu

mp

s(n

um

ber)

109

107

419

216

526

635

519

628

735

1,1

54

Wa

ter

flo

ws

(all

in10

6m

3)

Irri

ga

tio

nin

flo

w110

129

112

226

127

228

355

355

358

358

Ra

inin

flo

w3

59

814

17

33

37

41

50

Co

mm

itte

do

utfl

ow

95

109

98

195

102

198

231

245

239

250

Un

com

mit

ted

ou

tflo

w0

00

00

049

49

49

49

Av

ail

ab

lew

ate

r18

25

22

39

39

47

157

147

160

159

Ric

eE

Td

ep

leti

on

812

13

20

25

33

57

65

77

90

Oth

er

ET

dep

leti

on

23

65

10

11

11

13

16

22

Ba

lan

ce8

10

214

43

40

20

18

�3

Inte

rna

lw

ate

rfl

ow

s(a

llin

10

6m

3)

Ric

efi

eld

per

cola

tio

n4

67

11

13

18

32

36

42

49

Reu

seb

ych

eck

da

ms

07

29

736

36

54

54

61

90

Pu

mp

ing

surf

ace

wa

ter

00

00

00

11

11

Pu

mp

ing

gro

un

dw

ate

r2

29

310

12

14

16

17

26

Gro

un

dw

ate

rch

an

ge

�1

�2

�1

�3

�4

�4

�3

�3

�5

�7

a g r i c u l t u r a l w a t e r m a n a g e m e n t x x x ( 2 0 0 7 ) x x x – x x x 5

AGWAT 2454 1–9

UN

CO

RR

E

Table 1. Rice yield to calculate WP was obtained from the NIA,

which kept track of the yield of each farmer in the area. For

each spatial unit, total rice production was obtained by

summing the yields of the individual farmers. Linear regres-

sion was performed on the performance indicators against

surface area of the spatial scale units.

3. Results

General characteristics and results of the water balancing and

water accounting are given for each spatial unit in Table 2. The

sizes of the spatial units varied from 1500 ha to about

18,000 ha. Some 10,000 ha of District I, mainly at the down-

stream end in the southwest, were excluded from the analysis.

Rice covered about 75% of the surface area in most of our

spatial units. In SDA-C, the rice area covered only 65% of the

total area, probably because less water was available here,

which prompted farmers to grow less water-demanding crops.

Rice yields were highest in the upstream area TRIS and lowest

in SDA-A, with an absolute difference of 1.36 t ha�1.

3.1. Water accounting and balance

Because of little rainfall, irrigation comprised 88–97% of all

surface water inflows. The irrigation water inflow generally

increased with increasing spatial scale within the whole SDA

(Fig. 2). However, when TRISL was merged with any of the

units of SDA, no more irrigation water was supplied between

scales of 11,000 ha (TRISL) and 18,000 ha. Total water outflows

and net water inflows followed the same trend (Fig. 2). These

trends indicate that large amounts of surface water flowed

overland through the system without being depleted. Out of all

surface water outflows, only 49 � 106 m3 was uncommitted as

it flowed directly into the Talavera River from TRISL. All other

outflows were committed and flowed either into another

spatial unit or into the downstream irrigated area of District I.

Units that had a relatively large irrigation water inflow also

had a relatively large surface water outflow. The result is that

the net surface inflow increased quite linearly with spatial

scale up to the 11,000 ha of TRISL.

Per unit rice area, total net applied surface water (all

surface inflows minus surface outflows) decreased linearly

with increasing scale within the SDA from 1500 mm at 1500 ha

to 1000 mm at 6800 ha (Fig. 3). There was a break in the linear

relation when TRISL was added, but, over all units combined,

applied surface water decreased linearly again from 1200 mm

at 11,000 ha to 800 mm at 18,000 ha. Overall, applied surface

water decreased by 30 mm for every 1000 ha. Out of the

applied amounts of surface water, only 213–295 mm was

rainfall in the different spatial units.

The volume of rice and nonrice ET increased linearly with

spatial scale, indicating uniform evaporation conditions

within District I. Per unit area, the average rice ET was

665 mm for the whole season and 3.7 mm d�1. The nonrice ET

was 503 mm for the whole season and 2.8 mm d�1.

The water balance term (net surface inflows minus surface

outflows and all ET) was relatively small, being 1–10% of total

surface inflows at different scales. The term was positive for

all spatial units except for the combination of all units, for

Please cite this article in press as: Hafeez, M.M. et al., Scale effects on water use and water productivity in a rice-based irrigation system

(UPRIIS) in the Philippines, Agric. Water Manage. (2007), doi:10.1016/j.agwat.2007.05.006

Page 6

E

OO

F

356

357

358

359

360

361

362

363

364

365

366

367

368

369

370

371

372

373

374

375

376

377

378

379

380

Fig. 2 – Volume of irrigation water inflow (^), surface water

outflow (^), and net inflow (irrigation plus rainfall minus

surface outflow; T) vs. spatial scale. The lines are linear

regressions.

Fig. 3 – Depth (volume divided by rice surface area) of

available surface water (^) and rain water (^) vs. spatial

scale. The lines are linear regressions.

Fig. 4 – Process fraction of net surface inflow (^), available

water (^), and depleted water (T) vs. spatial scale. The

lines are linear regressions.

a g r i c u l t u r a l w a t e r m a n a g e m e n t x x x ( 2 0 0 7 ) x x x – x x x6

AGWAT 2454 1–9

RRwhich it was close to zero. These positive values suggest that

water percolated down and recharged groundwater or flowed

as subsurface water into neighboring units. There was no

relation between the water balance term and scale within SDA

UN

CO

Table 3 – Process fraction (PF), depleted fraction (DF), and wate

SDA-A SDA-B SDA-C SDA-AB SDA-B

PF (%)

PFgross 7 9 11 8 18

PFavailable 45 47 61 50 65

PFdepleted 81 78 68 79 72

DF (%)

DFgross 9 11 16 11 24

DFavailable 56 60 89 63 89

WP (kg grain m�3 water)

WPgross 0.05 0.07 0.08 0.06 0.14

WPavailable 0.32 0.39 0.43 0.38 0.50

WPriceET 0.70 0.82 0.72 0.76 0.77

See Table 1 for explanation of abbreviations.

Please cite this article in press as: Hafeez, M.M. et al., Scale effects on

(UPRIIS) in the Philippines, Agric. Water Manage. (2007), doi:10.1

CTE

D P

R

alone. However, the water balance term decreased consis-

tently with scale starting with TRISL and sequentially adding

units.

3.2. Water performance indicators

The performance indicators are given in Table 3. The process

fraction of all surface water input (irrigation and rainfall) was

relatively low but gradually increased with spatial scale up to

22% for all scale units combined (Fig. 4). The process fraction of

available water was much higher. For SDA, it linearly

increased with spatial scale up to 70% for all SDA units

combined. In TRISL, though the spatial scale was larger, it

dropped down to 36%, partly because of the relatively large

uncommitted surface water outflows from TRISL (Table 2).

However, when SDA units were subsequently added, the

process fraction increased again to a maximum of 57%. The

process fraction of depleted water was relatively independent

of scale because the fraction of the surface area covered by rice

was similar for all spatial units (Table 2).

The depleted fractions of all surface water input and of

available water increased with spatial scale and followed the

same trends as the process fractions (Fig. 5). Again, with the

r productivity (WP) for 10 spatial scales in District I, UPRIIS

C SDA-ABC TRISL TRISL +SDA-A

TRISL +SDA-AB

TRISL +SDA-ABC

13 15 17 19 22

70 36 45 48 57

74 84 84 83 80

18 18 20 23 28

95 43 53 58 71

0.10 0.14 0.14 0.16 0.18

0.52 0.34 0.36 0.40 0.45

0.75 0.92 0.82 0.82 0.80

water use and water productivity in a rice-based irrigation system

016/j.agwat.2007.05.006

Page 7

C

OO

F

381

382

383

384

385

386

387

388

389

390

391

392

393

394

395

396

397

398

399

400

401

402

403

404

405

406

407

408

409

410

411

412

413

414

415

416

417

418

419

420

421

422

Fig. 5 – Depletion fraction of net surface inflow (^) and

available water (^) vs. spatial scale. The lines are linear

regressions.

Fig. 7 – Volume of reuse of surface water by check dams (^)

and volume of pumped water (^) vs. spatial scale. The

lines are linear regressions.

a g r i c u l t u r a l w a t e r m a n a g e m e n t x x x ( 2 0 0 7 ) x x x – x x x 7

AGWAT 2454 1–9

RR

Elarge uncommitted outflow of water, the depleted fraction of

available water in TRISL was lower than in SDA despite its

larger surface area.

Water productivity with respect to gross inflow and

available water increased with spatial scale and followed

the same trend as the process fractions (Fig. 6). Water

productivity with respect to ET of rice was less scale

dependent and varied between 0.7 kg grain m�3 and

0.9 kg grain m�3 water.

3.3. Water reuse

Data on water reuse are given in Table 2. The reuse of surface

water through check dams was well distributed throughout

the area and increased linearly with 4.6 � 106 m3 per 1000 ha

(Fig. 7). At the highest aggregation level, the reuse of surface

water was 22% of all applied surface water and 57% of all

available surface water. A large number of farmers used

pumps for complete or supplemental irrigation. On average,

12% of the farmers owned a pump, though more farmers used

a pump because of shared use and rental arrangements (Moya

et al., 2002). The highest pump density occurred in SDA-C with

212 pumps 1000�1 ha and 40% of the farmers owning a pump,

UN

CO 423

424

425

426

427

428

429

430

431

432

433

434

435

436

437

Fig. 6 – Water productivity per unit net surface inflow (^),

per unit available water (^), and per unit depleted water

(T) vs. spatial scale. The lines are linear regressions.

Please cite this article in press as: Hafeez, M.M. et al., Scale effects on

(UPRIIS) in the Philippines, Agric. Water Manage. (2007), doi:10.1

TED

PR

confirming the relative lack of surface water suggested by the

relatively low percentage rice area compared with the other

units. The pump density in the other units was 60–90

1000 ha�1 and pump ownership ranged from 7% to 40%.

Pumping from surface water was negligible and nearly all

water was pumped from the shallow groundwater. The total

(re)use of water through pumping increased by 1.3 � 106 m3

per 1000 ha (Fig. 7). At the highest aggregation level, the water

(re)use by pumping was 7% of the applied surface water and

17% of the available water.

3.4. Percolation and groundwater recharge

The estimated amount of percolating water from rice fields

was about one to three times the amount of water pumped

from the groundwater across the spatial units (Table 2).

Despite this percolation flow, which can be interpreted as

recharging the groundwater, groundwater tables in our

observation wells decreased from an average depth of 2.3 m

at the start to 3.4 m (standard error 0.2 m) at the end of the

growing season. Though the estimated changes in stored

groundwater were small (Table 2), the negative values suggest

that groundwater leaked out of the irrigated area.

4. Conclusions and discussion

Our results support the hypothesis that water use becomes

more efficient with increasing scale because of water reuse:

the amount of net surface water input decreased and the

process fraction, depleted fraction, water productivity, and

amount of water reuse increased with increasing spatial scale.

We calculated that in the whole of our study area, 22% of the

applied surface water was reused by internal check dams and

7% through pumping from shallow groundwater. The 22%

reuse through check dams is probably even an underestima-

tion as 6 of the 15 check dams could not be monitored. Because

of the large amount of water reuse, the overall water use in the

area is quite efficient as evidenced by fairly good performance

indicators (Table 3) and low values of uncommitted outflow

(drainage water that is not used downstream). Most of the

water use and water productivity in a rice-based irrigation system

016/j.agwat.2007.05.006

Page 8

E

438

439

440

441

442

443

444

445

446

447

448

449

450

451

452

453

454

455

456

457

458

459

460

461

462

463

464

465

466

467

468

469

470

471

472

473

474

475

476

477

478

479

480

481

482

483

484

485

486

487

488

489

490

491

492

493

494

495

496

497

498

499

500

501

502

503

504

505

506

507

508

509

510

511

512

513

514

515

516

517

518

519

520

521

522

523

524

525

526

527

528

529

530

531

532

533

534

535

536

537

538

539

540

541

542

543

544

545

546

547

548

549

550

551

552

553

554

555

556

557

a g r i c u l t u r a l w a t e r m a n a g e m e n t x x x ( 2 0 0 7 ) x x x – x x x8

AGWAT 2454 1–9

UN

CO

RR

water applied to District I is used within the district, and only

49 � 106 m3 is lost as uncommitted water and could potentially

be saved or used for rice production in downstream units

within the district (such as SDA-C, which is relatively water

short).

A closer analysis of the water performance indicators can

point to possibilities to further increase the general efficiency

of water use. The overall process fraction of depleted water (ET

from rice over total ET from the whole area) was 80%, which is

close to the 75% area covered by rice. In a comparable-sized

area of 28,500 ha (first main canal command area) in the ZIS in

China (see Introduction), the process fraction of depleted

water was 27% for a rice area of 19% (Loeve et al., 2004). In our

area, there is little scope to further increase the process

fraction of depleted water by changing land use. About 9% of

our area was covered with upland crops, and their evapo-

transpiration resulted in agricultural produce that can be

considered as process fraction as well (being a beneficial use).

The remaining 14% land area is used mainly by built-up area,

roads, and open water bodies. The evaporation from canals

could theoretically be reduced by introducing piped convey-

ance systems, but it is doubtful whether the value of water in

UPRIIS is high enough to justify such investments.

The overall depletion fraction of available water (all ET from

the whole area over the available surface water) was 71%, which

is similar to the 67% reported for the same scale size in ZIS. Like

the process fraction of depleted water, the fraction of depleted

water is quite high and possibilities to increase it seem limited.

The depleted fraction can be increased by reducing the amount

of seepage and percolation water and/or increasing the internal

reuse of water. The amount of seepage and percolation water

was estimated at 49� 106 m3 and accounted for about 12.5% of

all surface water inflows. Seepage and percolation can be

reduced by the adoption of water-saving irrigation technologies

such as alternate wetting and drying (AWD). In field experi-

ments in the Philippines and China, Belder et al. (2004) and

Cabangon et al. (2004) reported water-savings of 15–30% by

adoption of AWD with no yield loss, with nearly all of the

savings coming from reduced seepage and percolation flows. If

all farmers in our area would adopt AWD, and if all water saved

would be used as process ET by rice, the depleted fraction would

increase from 71% to 76–80%. However, we should realize that

part of the seepageand percolation flowsis alreadybeing reused

by farmers and that the increase in depleted fraction would

probably be less. If all seepage and percolation water could be

recaptured by pumping, the depleted fraction would reach

100%. The NIA actively encourages farmers to use pumps for

water reuse,but any further increase inpumping will dependon

the economics of pump water use (Moya et al., 2002). It is

unlikely that all seepage and percolation water can be

recaptured as there will always be leakage losses of ground-

water to outside the area as evidenced by the drop in

groundwater table during the growing season (Table 2).

The overall water productivity with respect to available

water (WPavailable) was 0.45 kg grain m�3 water, which com-

pares well with the average of 0.4 kg grain m�3 water for rice at

the field level (Tuong et al., 2005). However, the maximum

WPavailable reported for rice by Tuong et al. is 1.2 kg grain m�3

water, suggesting that there is considerable scope for

improvement in our study area. The WPavailable can be

Please cite this article in press as: Hafeez, M.M. et al., Scale effects on

(UPRIIS) in the Philippines, Agric. Water Manage. (2007), doi:10.1

CTE

D P

RO

OF

increased by reducing the nonbeneficial outflows from rice

fields (evaporation from the ponded water layer, seepage, and

percolation) and/or by increasing the rice yield (Bouman,

2007). In our area, there may be great scope to reduce

nonbeneficial outflows by shortening the total operation time

of the irrigation system. Though it takes only about 110–120

days to complete a cycle of a rice crop, the irrigation season in

2000–2001 lasted 182 days. After the first water release in the

canals on 19 November 2000, it took from 1 December to 7

February 2001, to soak the land, from 23 December to 22

February to prepare the fields (puddling), and from 1 January to

22 February to transplant the crops. Since each farmer had an

individual seedbed in a corner of a main field, all fields were

kept flooded from the start of land soaking and the establish-

ment of seedbeds onwards. Thus, from 1 December 2000, to

halfway between 1 January and 22 February 2001, hardly any

crop was growing in the fields and most water was lost by

nonproductive evaporation, seepage, and percolation flows.

During this period, the seepage and percolation flows were

mostly lost to the whole system as little reuse by pumping

occurred at that time. The duration of land soaking to crop

establishment can be shortened by synchronized planting,

using community seedbeds (so that main fields do not need to

be kept flooded), or using direct seeding (Tabbal et al., 2002). If

the total duration from land soaking to crop establishment

could be reduced by 30 days, the total amount of water applied

to the rice fields would be reduced by about 17% and WPavailable

would increase from 0.45 kg grain m�3 to 0.55 kg grain m�3

water. After crop establishment, evaporation, seepage, and

percolation flows can be reduced further by the adoption of

AWD (Belder et al., 2004; Bouman et al., 2007), though potential

gains in reductions in seepage and percolation flows are offset

by reuse of these flows. There is also the possibility to increase

WPavailable by increasing yield. The average yield of 5.3 t ha�1 in

our area was a bit higher than the UPRIIS-average of 4.5 t ha�1

realized in 1990–2000 (Bouman et al., 2002), but still consider-

ably lower than yields of up to 8.4 t ha�1 obtained in field

experiments in the area (Tabbal et al., 2002; Belder et al., 2004).

The nitrogen (N) fertilizer application rate in these experi-

ments was as high as 180 kg N ha�1, and since average

fertilizer N use in Central Luzon where UPRIIS is located is

only about 100 kg ha�1 (Bouman et al., 2002), there may be

scope to increase yields through increased fertilizer N

application. Other yield-increasing measures include

improved pest and disease control and reduced post-harvest

losses. If yields could increase to 8 t ha�1, WPavailable would

increase to 0.68 kg grain m�3 water. If the water-saving

measures discussed above would be combined with a yield

increase, WPavailable could increase to 0.83 kg grain m�3 water.

The overall water productivity with respect to evapotran-

spiration (WPET) was 0.8 kg grain m�3 water, which was close

to the average of 1.1 kg grain m�3 water for rice at the field

level as reported by Zwart and Bastiaanssen (2004). The only

option to increase WPET is again crop protection measures

such as pest and disease control and reduced post-harvest

losses (Bouman, 2007).

The results of our study are influenced by the sizes and

locations of the spatial units we could establish based on the

existing road network and accessibility. There was a relatively

large amount of uncommitted outflow from TRIS whereas no

water use and water productivity in a rice-based irrigation system

016/j.agwat.2007.05.006

Page 9

C

558

559

560

561

562

563

564

565

566

567

568

569

570

571

572

573

574

575

576

577

578

579

580

581

582

583

584

585

586

587

588

589

590

591

592

593

594

595

596

597

598

599

600

601

602

603

604

605

606

607

608

609

610

611

612

613

614

615

616

617

618

619

620

621

622

623

624

625

626

627

628

629

630

631

632

633

634

635

636

637

638

639

640

641

642

643

644

645

646

647

648

649

650

651

652

653

654

655

656

657

658

659

660

661

662

663

664

665

666

667

668

669

670

671

672

673

674

675

676

677

678

679

680

681

682

683

684

a g r i c u l t u r a l w a t e r m a n a g e m e n t x x x ( 2 0 0 7 ) x x x – x x x 9

AGWAT 2454 1–9

UN

CO

RR

E

uncommitted water flowed out of any of the other spatial

units. This caused a ‘‘break’’ in some of the linear relation-

ships between water accounting and performance indicators

with spatial scale when we lumped spatial units in our

analysis. Although the slope of the relationships between

water accounting and performance indicators with spatial

scale will be different with another layout of spatial units, the

trends we have found will be the same. A hydrological model

study is needed next to quantify the options to improve the

efficiency and productivity of water use as discussed above,

and to disentangle spatial tradeoffs in water accounting and

water performance indicators.

Acknowledgments

The following IRRI staff assisted in data collection, data

processing, and overall logistical arrangements: Domeng

Tabbal, Ruben Lampayan, Lizzel Llorca, and Lucio Caramihan.

We thank the NIA and their staff who contributed to this study

and generously shared their data with us.

r e f e r e n c e s

Bastiaanssen, W.G.M., 1995. Regionalization of surface fluxdensities and moisture indicators in composite terrain: aremote sensing approach under clear skies inMediterranean climates. Report 109. Agricultural ResearchDepartment, Wageningen, The Netherlands.

Belder, P., Bouman, B.A.M., Spiertz, J.H.J., Cabangon, R., LuGuoan, Quilang, E.J.P., Li Yuanhua, Tuong, T.P., 2004. Effectof water-saving irrigation on rice yield and water use intypical lowland conditions in Asia. Agric. Water Manage. 65,193–210.

Bouman, B.A.M., 2007. A conceptual framework for theimprovement of crop water productivity at different spatialscales. Agric. Syst. 93, 43–60.

Bouman, B.A.M., Tuong, T.P., 2001. Field water management tosave water and increase its productivity in irrigated rice.Agric. Water Manage. 49, 11–30.

Bouman, B.A.M., Castaneda, A., Bhuiyan, S.I., 2002. Nitrate andpesticide contamination of groundwater under rice-basedcropping systems: evidence from the Philippines. Agric.Ecosyst. Environ. 92, 185–199.

Bouman, B.A.M., Humphreys, E., Barker, R., Tuong, T.P., 2006.Rice water. Adv. Agron. 92, 187–237.

Bouman, B.A.M., Feng, L., Tuong, T.P., Lu, G., Wang, H., Feng, Y.,2007. Exploring options to grow rice under water-shortconditions in northern China using a modelling approach.II. Quantifying yield, water balance components, and waterproductivity. Agric. Water Manage. 88, 23–33.

Cabangon, R.J., Tuong, T.P., Castillo, E.G., Bao, L.X., Lu, G., Wang,G., Cui, Y., Bouman, B.A.M., Li, Y., Chen, C., Wang, J., 2004.Effect of irrigation method and N-fertilizer management onrice yield, water productivity and nutrient-use efficienciesin typical lowland rice conditions in China. Paddy WaterEnviron. 2, 195–206.

Choudhury, B.U., Bouman, B.A.M., Singh, A.K., 2007. Yield andwater productivity of rice-wheat on raised beds at NewDelhi, India. Field Crop. Res. 100, 229–239.

Dawe, D., 2005. Increasing water productivity in rice-basedsystems in Asia—past trends, current problems, and futureprospects. Plant Prod. Sci. 8, 221–230.

Please cite this article in press as: Hafeez, M.M. et al., Scale effects on

(UPRIIS) in the Philippines, Agric. Water Manage. (2007), doi:10.1

TED

PR

OO

F

Dong, B., Molden, D., Loeve, R., Li, Y., Chen, C., Wang, J., 2004.Farm level practices and water productivity in Zangheirrigation system. Paddy Water Environ. 2, 217–226.

Hafeez, M.M., 2003. Water Accounting and Productivity atDifferent Scales in a Rice Irrigation System: A RemoteSensing Approach. Ecology and Development Series No. 8.University of Bonn, Germany.

Hafeez, M.M., Chemin, Y., Van De Giesen, N., Bouman, B.A.M.,2002. Field evapotranspiration estimation in central Luzon,Philippines, using different sensors: Landsat 7 ETM+, TerraMODIS and ASTER. In: Proceedings of the Joint Conferenceon ‘‘Geospatial Theory, Processing, and Applications’’organized by the International Society of Photogrammetryand Remote Sensing (ISPRS)/Canadian Institute ofGeomatics (CIG) Conference, Ottawa, Canada, July 8–12, pp.142–149.

Loeve, R., Dong, B., Molden, D., Li, Y., Chen, C., Wang, J., 2004.Issues of scale in water productivity in the Zhangheirrigation system: implications for irrigation in the basincontext. Paddy Water Environ. 2, 227–236.

Lucero, L.C., 1984. Improved project water management inirrigated rice. Ph.D. Thesis. Utah State University, Logan,UT, USA.

Maclean, J.L., Dawe, D., Hardy, B., Hettel, G.P. (Eds.), 2002. RiceAlmanac. International Rice Research Institute, Los Banos,Philippines.

Molden, D., 1997. Accounting for water use and productivity.SWIM Paper. International Water Management Institute,Colombo, Sri Lanka.

Molden, D., Sakthivadivel, R., 1999. Water accounting to assessuse and productivity of water. Water Res. Dev. 15, 55–71.

Moya, P.F., Dawe, D., Valencia, S., 2002. The effects ofpumping on water use and profitability in dry-seasonrice: a case study in UPRIIS, Philippines. In: Bouman,B.A.M., Hengsdijk, H., Hardy, B., Bindraban, P.S., Tuong,T.P., Ladha, J.K. (Eds.), Proceedings of the InternationalWorkshop on Water-wise Rice Production, InternationalRice Research Institute, Los Banos, Philippines, April 8–11,pp. 333–345.

Ramos, E.P., 1986. Evaluation of soil parameters for irrigationplanning and design for upland crops. M.Sc. Thesis. CentralLuzon State University, Munoz, Nueva Ecija, Philippines.

Rijsberman, F.R., 2006. Water scarcity: fact or fiction? Agric.Water Manage. 80, 5–22.

Sattar, M.A., 1992. Water management and technology adoptionfor direct seeded rice in an irrigation system. Ph.D. Thesis.Central Luzon State University, Munoz, Nueva Ecija,Philippines.

Sharma, P.K., Bhushan Lav, Ladha, J.K., Naresh, R.K., Gupta,R.K., Balasubramanian, B.V., Bouman, B.A.M., 2002. Crop-water relations in rice-wheat cropping under differenttillage systems and water-management practices in amarginally sodic, medium-textured soil. In: Bouman,B.A.M., Hengsdijk, H., Hardy, B., Bindraban, P.S., Tuong,T.P., Ladha, J.K. (Eds.), Proceedings of the InternationalWorkshop on Water-wise Rice Production, InternationalRice Research Institute, Los Banos, Philippines, April 8–11,pp. 223–235.

Tabbal, D.F., Bouman, B.A.M., Bhuiyan, S.I., Sibayan, E.B., Sattar,M.A., 2002. On-farm strategies for reducing water input inirrigated rice: case studies in the Philippines. Agric. WaterManage. 56, 93–112.

Tuong, T.P., Bouman, B.A.M., Mortimer, M., 2005. More rice, lesswater—integrated approaches for increasing waterproductivity in irrigated rice-based systems in Asia. PlantProd. Sci. 8, 231–241.

Zwart, S.J., Bastiaanssen, W.G.M., 2004. Review of measuredcrop water productivity values for irrigated wheat, rice,cotton and maize. Agric. Water Manage. 69, 115–133.

water use and water productivity in a rice-based irrigation system

016/j.agwat.2007.05.006