Você está na página 1de 10

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/291974266

Simulation of soil moisture movement under rice field using Hydrus-2D

Article · April 2013

CITATIONS READS

0 78

3 authors, including:

Mahesh Chand Singh A K Jain


Punjabi University, Patiala Punjab Agricultural University
22 PUBLICATIONS   11 CITATIONS    17 PUBLICATIONS   24 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Protected cultivation-A Review View project

IRRI IFAD PROJECT ON RESOURCE CONSERVATION TECHNOLOGIES View project

All content following this page was uploaded by Mahesh Chand Singh on 13 July 2016.

The user has requested enhancement of the downloaded file.


Crop Res. 45 (1, 2 & 3) : 45-53 (2013)
With twenty-one figures
Printed in India
Simulation of soil moisture movement under rice field using
Hydrus-2D
MAHESH CHAND SINGH*, ANCHAL KUMAR JAIN AND SUNIL GARG
Department of Soil and Water Engineering
Punjab Agricultural University Ludhiana-141 004 (Punjab), India
*(e-mail : mahesh25pau@yahoo.co.in)

(Received : April 2012)

ABSTRACT

Soil moisture is a key variable in controlling the exchange of water between


land surface and atmosphere through evaporation and plant transpiration. A field study
was conducted at Punjab Agricultural University, Ludhiana (India), in the year 2009, to
simulate the soil moisture movement under rice field using numerical model Hydrus-
2D. The treatments included two dates of transplanting viz., June 5 (D1) and June 20
(D2), two varieties inbred PAU-201 (V1) and hybrid RH-257 (V2) of 120 and 90 days durations
(from transplanting to harvest) and two irrigation regimes viz., intermittent irrigation at
2-days drainage period (I1) and irrigation based on soil water suction (SWS) of 16 k Pa
(I2). A total of eight treatments were replicated thrice in 24 plots of each having size 10
x 4 m. The saturated hydraulic conductivity (Ks varied from 0.36 to 1.48 cm/day with an
average value of 0.92 cm/day) and the fitting parameter (n varied from 1.58 to 2.76 with
an average value of 2.17) were found to be very sensitive during calibration. The results
from the experiment revealed that the soil moisture content was more in upper layer of
the soil profile as compared to the lower layers due to effect of puddling. The application
of Hydrus-2D confirmed that the simulated depth wise soil moisture content values
were in good agreement with observed data. The model performance was evaluated by
using parameters, namely, root mean square error, absolute percentage error, correlation
coefficient and model efficiency. During validation, the average absolute error varied
from 2.19 to 13.21%, root mean square error varied from 0.006 to 0.032 cm, correlation
coefficient varied from 0.773 to 0.996 and the average model efficiency was 98.6%. The
use of Hydrus-2D model can be successfully adopted for simulating soil moisture profiles
under rice crop.

Key words : Hydrus-2D, simulation, soil moisture

INTRODUCTION Tuong and Bouman, 2003). The amount of


water consumed to produce rice is much
Rice is one of the most important staple higher than for other cereal crops (Bhuiyan,
food crops in the world. In Asia, more than two 1992), as the water use efficiency of rice is
billion people are getting 60-70% of their quite low. Rice is generally grown under puddled
energy requirement from rice. In India, the conditions mainly to reduce percolation losses
average productivity of rice is only 2-3.5 t/ha and control of weeds. However, the puddling
(Ladha et al., 2000). In India, the production of process besides consuming a substantial
rice was 99.18 mt (Anonymous, 2009a) and the amount of irrigation water results in sub-
current production of rice is 87.56 mt surface compaction (Kukal and Aggarwal,
(Anonymous, 2010). To sustain present food 2003).
self-sufficiency and to meet future food With the dominance of rice-wheat
requirements, India has to increase its rice system in the state of Punjab, the demand of
productivity by 3% per annum (Thiyagarajan the water has increased manifold. The cropping
and Selvaraju, 2001). intensity increased from 161% in 1980 to 189%
Rice is an important target for in 2008 (Anonymous, 2009b). The area under
irrigation water use reductions, because of its rice crop in Punjab is about 26 lakh ha. In the
relatively large water requirements compared central part of the state, the water table is
with other crops (Li, 2001; Wang et al., 2002, declining at the rate of about 0.75 m/year due
46 Singh, Jain and Garg

to over exploitation of ground water as the Research Farm of the Department of Soils, PAU,
surface water supply is not adequate. This Ludhiana (30°56′ N latitude and 75°52′ E
declining water table has caused increased cost longitude) during the kharif season of 2008-
of production and energy use vis-à-vis threat to 09. The soil at the site was sandy loam. The
sustainability of agricultural production. bulk density of the soil was 1.6825 g/cm3, field
The globally growing demand for water capacity was 20.68% and saturated hydraulic
has ushered the need for its efficient and conductivity was 0.92 cm/day. Neutron probe
judicious utilization particularly in agriculture was used for determination of depth-wise soil
sector being a single largest consumer of water. moisture content, the access tubes were
The majority of irrigation projects in India installed at different depths for measurement
perform at a very low efficiency, which calls for soil moisture content. The climate of Ludhiana
application of efficient water management is semi-arid sub tropical with hot dry summers
technologies for meeting the increasing water and cold winters. The mean annual rainfall is
demands. A variety of analytical and numerical about 680 mm. The average rainfall received
models have been developed in the past decades during the crop period was 6.5 mm/day. The
to predict water process between the soil treatments included two dates of transplanting
surface and the groundwater (Goncalves et al., viz., June 5 (D1) and June 20 (D2), two varieties
2005). Hydrus-2D is one such numerical model inbred PAU-201 (V1) and hybrid RH-257 (V2) of
used to simulate soil moisture movement with 120 and 90 days durations (from transplanting
or without crop under different field conditions. to harvest) and two irrigation regimes viz.,
In this study, Hydrus-2D has been adopted to intermittent irrigation at 2-days drainage
simulate the soil moisture profiles in a rice period (I1) and irrigation based on soil water
field. suction (SWS) of 16 k Pa (I2). A total of eight
treatments were replicated thrice in 24 plots
MATERIALS AND METHODS of each having size 10 x 4 m. Number of rows
transplanted in each plot was 20 and number
General Description of Study Area of hills per row 68. Number of plants per hill
was two. Row to row spacing was 20 cm and
This experiment was conducted at the hill to hill 15 cm. The layout of the
4 m 

               
               
10 D1V1I1  D1V2I1  D1V2I2  D1V1I2  D2V1I1  D2V2I1  D2V2I2  D2V1I2 
m        

1 IRRIGATION WATER CHANNEL


m

               
               
D2V2I1  D2V1I1  D2V2I2  D2V1I2  D1V2I1  D1V1I1  D1V2I2  D1V1I2 
         
     

IRRIGATION WATER CHANNEL

               
               
D1V2I1  D1V1I1  D1V1I2  D1V2I2  D2V2I1  D2V1I1  D2V1I2 D2V2I2 
     

Fig. 1. Layout of field experiment.


Simulation of soil moisture movement under rice field using Hydrus-2D 47

experimental field is shown in Fig. 1. The [LT-1] given by


amount of irrigation water applied at each
irrigation from transplanting till maturity was K (h, x, z)=Ks (x, z) Kr ( h, x, z)
monitored with Parshal flume. Differential
irrigations were started 15 days after Where, Ks is saturated hydraulic conductivity,
transplanting. (LT -1 ] and K r is the relative hydraulic
The daily meteorological data were conductivity.
collected from the records of the Meteorological
Observatory located at the experimental area. Initial and Boundary Conditions
The variables collected were temperature,
relative humidity, vapour pressure, wind speed, The simulations were carried out with
sunshine hours and rainfall on daily basis. The atmospheric boundary conditions (BC) at the
data collected were utilized for computation of surface of the soil for the time dependant data
potential evapotranspiration, potential of precipitation (cm/day), potential evaporation
evaporation and potential crop transpiration. (cm/day), potential transpiration (cm/day),
irrigation rate (cm/day) and minimum allowed
Model Description pressure head (cm). Left and right boundaries
of the flow domain were given no flux of water.
The Hydrus-2D program numerically A free drainage was applied at bottom boundary
solves the Richards equation for saturated and (z=-180 cm) of the flow domain. The irrigation
unsaturated water flow. The flow equation was implemented by applying variable flux BC
incorporates a sink term to account for water to an area of cross-section of 15000 cm2 at a
uptake by plant roots. Hydrus-2D can handle flow depth of 15 cm below soil surface. The initial
domains delineated by irregular boundaries. The water content in the flow domain was set equal
flow region itself may be composed of non- to field capacity.
uniform soils having an arbitrary degree of local
anisotropy. Flow and transport can occur in the Criteria for Model Evaluation
vertical plane, the horizontal plane and in a three
dimensional region. The water flow part of the To check the performance of Hydrus-2D
model considers prescribed head and flux model, four performance indicators, namely, root
boundaries, boundaries controlled by atmospheric mean square error (RMSE in %), absolute
conditions, free drainage boundary conditions, percentage error (APE), correlation coefficient
as well as a simplified representation of nodal and model efficiency (EF) were used, which
drains using results of electric analog compared the observed and simulated values
experiments. of soil moisture content. RMSE, APE, correlation
coefficient and model efficiency (EF) were
Governing Flow Equation calculated using equations 1, 2, 3 and 4.

Governing flow equation in Hydrus-2D Root mean square error


is given by the following modified form of the
Richards’ equation (Simunek et al., 2006) : The model was evaluated by using root
mean square error (RMSE) which quantifies
dθ d ⎡ ⎛ A dh ⎞ ⎤
= ⎢ K ⎜⎜ K ij ⎟⎟ + K izA ⎥ − S the differences between observed and
dt dx i ⎣ ⎝ dx i ⎠ ⎦ simulated data.
Where θ=Volumetric water content [L3 L-3]
∑(Pi − Oi )
1
RMSE =
2
h=Pressure head [L] ...(1)
S=Sink term [T-1] (The flow equation N
incorporates a sink term to account Where, N is the number of observations, Pi and
for water uptake by plant roots) Oi are the ith simulated and observed values,
xi=Spatial coordinates (i=1, 2) [L] respectively.
t=Time [T]
Absolute percentage error
S=Components of a dimensionless
anisotropy tensor KA Pi − O i
A PE = × 100 ...(2)
K=Unsaturated hydraulic conductivity Oi
48 Singh, Jain and Garg

Where, Pi and Oi are the i th simulated and the input parameters required to use Hydrus-
observed values, respectively. 2D were determined by detailed field
experimentation, however, a few were taken
Correlation coefficient from published literature matching to our soil
and crop conditions. In order to obtain the
The correlation coefficient, R2 varied initial values of model parameters, the
from 0 to 1 and describes the degree of calibration of Hydrus-2D was conducted on
association between observed and predicted computer through simulation runs using input
values. data. It was done against two parameters
{
Correlation R 2 = N∑ i =1 i
N
(Oi }
− O )( P − P )
2

...(3)
saturated hydraulic conductivity [K s varied
from 0.312 to 0.408 in upper layer (0-15 cm)
coefficient (
∑i =1 i
O − O )2
∑ i=1 (Pi − P )
N 2
and 1.03 to 0.93 in lower layers (15-180 cm)]
and fitting parameter [n varied from 1.26 to
Model efficiency 1.89 in upper layer (0-15 cm) and 1.89 to 3.6 in
lower layers (15-180 cm)].
The model efficiency (EF), the measure
of the deviation between model predictions and Model Validation for Treatment-D2V1I1
measurements relative to the scattering of the
observed data was calculated by the formula The depth-wise observed and simulated
(Nash and Sutcliffe, 1970), given below : values of soil moisture content are shown in
Figs. 2, 3, 4, 5, 6 and 7. From Figs. 2 to 7 and
⎡n 2⎤
⎢∑ (Pi − Oi ) ⎥ Table 1, it is clear that on the 43rd (17 July)
EF= 1 − ⎣ i =n1 ⎦
...(4) day after crop transplantation, the average

( )
2⎤
⎢ ∑ Oi − O ⎥ absolute error between simulated and observed
⎣ i =1 ⎦
values of soil moisture content was 4.98%, the
RESULTS AND DISCUSSION average RMSE value was 0.014 cm, average
correlation coefficient was 0.982 and model
Simulated Soil Moisture Content efficiency was 99.4%. At 62nd (5 Aug.) day after
the crop transplantation, the average absolute
Hydrus-2D was calibrated to predict the error between simulated and observed values
soil moisture content in the rice crop with the of moisture content was 8.56%, the average
help of observed soil moisture content. Most of RMSE value was 0.02 cm, average correlation
Table 1. Summary of results of statistical performance parameters for validation

Time (days) 17 Jul. 5 Aug. 11 Aug. 25 Aug. 9 Sep. 18 Sep.

Treatment-D2V1I1
Absolute percentage error 4.98 8.56 5.03 3.68 2.19 3.81
RMSE (cm) 0.014 0.02 0.014 0.01 0.006 0.011
Correlation coefficient (R2) 0.982 0.92 0.938 0.979 0.996 0.983
Model efficiency 99.4 98.9 99.4 99.7 99.9 99.6
Time (days) 5 Aug. 7 Aug. 25 Aug. 9 Sep. 18 Sep.
Treatment-D2V1I2
Absolute percentage error 6.84 6.89 5.3 5.18 5.9
RMSE (cm) 0.016 0.019 0.012 0.014 0.016
Correlation coefficient (R2) 0.945 0.88 0.967 0.973 0.964
Model efficiency 96.0 94.3 97.1 97.0 96.2
Time (days) 17 Jul. 5 Aug. 11 Aug. 26 Aug. 9 Sep. 18 Sep.
Treatment-D2V2I1
Absolute percentage error 3.15 7.24 13.21 3.92 6.58 2.71
RMSE (cm) 0.007 0.022 0.032 0.01 0.016 0.007
Correlation coefficient (R2) 0.995 0.916 0.773 0.984 0.97 0.981
Model efficiency 99.9 98.6 96.6 99.7 99.2 99.9
Time (days) 5 Aug. 7 Aug. 11 Aug.
Treatment-D2V2I2
Absolute percentage error 4.46 6.65 4.18
RMSE (cm) 0.013 0.017 0.01
Correlation coefficient (R2) 0.981 0.956 0.962
Model efficiency 99.6 99.1 99.7
Simulation of soil moisture movement under rice field using Hydrus-2D 49

Fig. 5. Trend between simulated and observed


Fig. 2. Trend between simulated and observed values (25 Aug.).
values (17 July).

Fig. 3. Trend between simulated and observed


values (5 Aug.). Fig. 6. Trend between simulated and observed
values (9 Sept.).

Fig. 4. Trend between simulated and observed


values (11 Aug.). Fig. 7. Trend between simulated and observed
values (18 Sept.).
coefficient was 0.92 and model efficiency was
98.9%. moisture content was 3.68%, the average
On 68th (11 Aug.) day after the crop RMSE value was 0.01 cm, average correlation
transplantation, the average absolute error coefficient was 0.979 and model efficiency was
between simulated and observed values of 99.7%. On 97th (9 Sept.) day after the crop
moisture content was 5.03%, the average transplantation, the average absolute error
RMSE value was 0.014 cm, average correlation between simulated and observed values of
coefficient was 0.938 and model efficiency was moisture content was 2.19%, the average
99.4%. On 82nd (25 Aug.) day after the crop RMSE value was 0.006 cm, average correlation
transplantation, the average absolute error coefficient was 0.996 and model efficiency was
between simulated and observed values of 99.9%. On 106th (18 Sept.) day after the crop
50 Singh, Jain and Garg

transplantation, the average absolute error


between simulated and observed values of
moisture content was 3.81%, the average
RMSE value was 0.011 cm, average correlation
coefficient was 0.983 and model efficiency was
99.6%.

Model Validation for Treatment-D2V1I2

The depth-wise observed and simulated


values of soil moisture content are shown in
Figs. 8, 9, 10, 11 and 12. From Figs 8 to 12 and
Table 1, it is clear that on the 62nd (5 Aug.)
day after crop transplantation, the average Fig. 10. Trend between simulated and observed
absolute error between simulated and observed values (25 Aug.).
values of soil moisture content was 6.84%, the
average RMSE value was 0.016 cm, average
correlation coefficient was 0.945 and model
efficiency was 96.0%. On 64th (7 Aug.) day after
the crop transplantation, the average absolute
error between simulated and observed values
of moisture content was 6.89%, the average
RMSE value was 0.019 cm, average correlation
coefficient was 0.88 and model efficiency was
94.3%.

Fig. 11. Trend between simulated and observed


values (9 Sept.).

Fig. 8. Trend between simulated and observed


values (5 Aug.).

Fig. 12. Trend between simulated and observed


values (18 Sept.).

On 82nd (25 Aug.) day after the crop


transplantation, the average absolute error
between simulated and observed values of
moisture content was 5.30%, the average
RMSE value was 0.012 cm, average correlation
coefficient was 0.967 and model efficiency was
97.1%. On 97th (9 Sept.) day after the crop
Fig. 9. Trend between simulated and observed transplantation, the average absolute error
values (7 Aug.). between simulated and observed values of
Simulation of soil moisture movement under rice field using Hydrus-2D 51

moisture content was 5.18%, the average


RMSE value was 0.014 cm, average correlation
coefficient was 0.973 and model efficiency was
97.0%. On 106th (18th Sept.) day after the crop
transplantation, the average absolute error
between simulated and observed values of
moisture content was 5.90%, the average
RMSE value was 0.016 cm, average correlation
coefficient was 0.964 and model efficiency was
96.2%.

Model Validation for Treatment-D2V2I1


Fig. 14. Trend between simulated and observed
The depth-wise observed and simulated values (5 Aug.).
values of soil moisture content are shown in
Figs. 13, 14, 15, 16, 17 and 18. From Figs. 13 to
18 and Table 1, it is clear that on the 28th (17
July) day after crop transplantation, the
average absolute error between simulated and
observed values of soil moisture content was
3.15%, the average RMSE value was 0.007 cm,
average correlation coefficient was 0.995 and
model efficiency was 99.9%. At 47th (5 Aug.)
day after the crop transplantation, the average
absolute error between simulated and observed
values of moisture content was 7.24%, the
average RMSE value was 0.022 cm, average
correlation coefficient was 0.916 and model
Fig. 15. Trend between simulated and observed
efficiency was 98.6%.
values (11 Aug.).
On 53rd (11 Aug.) day after the crop
transplantation, the average absolute error
between simulated and observed values of
moisture content was 13.21%, the average
RMSE value was 0.032 cm, average correlation
coefficient was 0.773 and model efficiency was
96.6%. On 68th (26 Aug.) day after the crop
transplantation, the average absolute error
between simulated and observed values of

Fig. 16. Trend between simulated and observed


values (26 Aug.).

moisture content was 3.92%, the average


RMSE value was 0.01 cm, average correlation
coefficient was 0.984 and model efficiency was
99.7%.
On 82nd (9 Sept.) day after the crop
transplantation, the average absolute error
Fig. 13. Trend between simulated and observed between simulated and observed values of
values (17 July). moisture content was 6.58%, the average
52 Singh, Jain and Garg

Fig. 19. Trend between simulated and observed


values (5 Aug.).
Fig. 17. Trend between simulated and observed
values (9 Sept.).

Fig. 20. Trend between simulated and observed


values (7 Aug.).

Fig. 18. Trend between simulated and observed


values (18 Sept.).

RMSE value was 0.016 cm, average correlation


coefficient was 0.97 and model efficiency was
99.2%. On 90th (18 Sept.) day after the crop
transplantation, the average absolute error
between simulated and observed values of
moisture content was 2.71%, the average
RMSE value was 0.007 cm, average correlation
coefficient was 0.981 and model efficiency was Fig. 21. Trend between simulated and observed
99.9%. values (11 Aug.).
the crop transplantation, the average absolute
Model Validation for Treatment-D2V2I2 error between simulated and observed values
of moisture content was 6.65%, the average
The depth-wise observed and simulated RMSE value was 0.017 cm, average correlation
values of soil moisture content are shown in coefficient was 0.956 and model efficiency was
Figs. 19, 20 and 21. From Figs. 19 to 21 and 99.1%.
Table 1, it is clear that on the 47th (5 Aug.) On 53rd (11 Aug.) day after the crop
day after crop transplantation, the average transplantation, the average absolute error
absolute error between simulated and observed between simulated and observed values of
values of soil moisture content was 4.46%, the moisture content was 4.18%, the average
average RMSE value was 0.013 cm, average RMSE value was 0.01 cm, average correlation
correlation coefficient was 0.981 and model coefficient was 0.962 and model efficiency was
efficiency was 99.6%. On 49th (7 Aug.) day after 99.7%.
Simulation of soil moisture movement under rice field using Hydrus-2D 53

CONCLUSIONS relation to crop production : case study on


rice. Outlook Agric. 21 : 293-99.
The study revealed that the soil Goncalves, M. C., Simunek, J., Romos, T. B.,
moisture content was more in upper layer of Martin, J. C., Neves M. J. and Pires, F. P.
(2005). Using Hydrus to simulate water and
the soil profile as compared to the lower layers
solute transports in soil lysimeters. Proc.
due to effect of puddling. The saturated Workshop on Hydrus Application. Utrecht
hydraulic conductivity (Ks varied from 0.36 to University, Netherland. pp. 38-41.
1.48 cm/day with an average value of 0.92 cm/ Kukal, S. S. and Aggarwal, G. C. (2003) Puddling
day) and the fitting parameter (n varied from depth and intensity effects in rice-wheat
1.58 to 2.76 with an average value of 2.17) were system on a sandy loam soil (I).
found to be very sensitive during calibration. Development of sub-surface compaction.
The simulated values of depth-wise soil Soil Till Res. 72 : 1-8.
moisture content were in good agreement with Ladha, J. K., Fischer, K. S., Hossain, M., Hobbs,
P. R. and Hardy, B. (eds.) (2000). Improving
the corresponding observed values. The
the productivity and sustainability of rice-
average absolute error varied from 2.19 to wheat systems of the Indo-Gangetic Plains
13.21%, RMSE varied from 0.006 to 0.032 cm, : a synthesis of NARS-IRRI partnership
correlation coefficient varied from 0.773 to research. IRRI Discussion Paper Series
0.996 and the average model efficiency was No. 40. pp. 31. Rice Research Institute,
98.6% for all the treatments. It was observed Los Banos, Philippines.
that the simulated values were in good Li, Y. (2001). Research and practice of water saving
agreement with observed data. Thus, Hydrus- irrigation for rice in China. In : Water-
2D model can be successfully adopted for Saving Irrigation for Rice, Barker, R., Loeve,
R., Li, Y. and Tuong, T. P. (eds.). Proc
simulating soil moisture profiles under rice
International Workshop, Wuhan, China,
crop. March 23-25. International Water
Management Institute, Colombo, Sri
ACKNOWLEDGEMENTS Lanka. pp. 135-44.
Nash, J. E. and Sutcliffe, J. V. (1970). River flow
The numerical model Hydrus-2D was forecasting through conceptual models
provided by the Department of Soil and Water part I-A discussion of principles. J.
Engineering, COAE & T, PAU, Ludhiana. The Hydrology 10 : 282-90.
authors appreciate the assistance from their Simunek, J., Sejna, S. and van Genuchten, M. Th.
(2006). The Hydrus-2D model for
colleagues and thankful to them, specifically,
simulating water, heat and solute
S. K. Jalota, Senior Soil Physicist, Department transport in two-dimensional variably
of Soil Science, PAU, Ludhiana. Additionally, saturated media, version 1.0. U. S. Salinity
the authors are very thankful to Rita Dahiya, Laboratory, USDA, Riverside, CA, USA.
Department of Soil Science, CCS Haryana Thiyagarajan, T. M. and Selvaraju, R. (2001). Water
Agricultural University, Hisar, Haryana and saving in rice cultivation in India. In : Proc.
Jiri Simunek, Professor and Hydrologist, International Workshop on Water Saving
Department of Environmental Sciences, Rice Production Systems. Nanjing
University of California Reverside, Reverside, University, China. pp. 15-45.
Tuong, T. P. and Bouman, B. A. M. (2003). Rice
CA.
production in water scarce environments.
In : Water Productivity in Agriculture : Limits
REFERENCES and Opportunities for Improvement, Kijne, J.
W., Barker, R. and Molden, D. (eds.). CABI
Anonymous (2009a). Review of the Economy. Publishing, UK. pp. 53-67.
Economic Advisory Council to the Prime Wang, H., Bouman, B. A. M., Zhao, D., Wang, C.
Minister, Hall-E, Vigyan Bhawan, Maulana and Moya, P. F. (2002). Aerobic rice in
Azad Road, New Delhi. northern China–Opportunities and
Anonymous (2009b). Punjab Agricultural Hand Book. challenges. In : Proc International Workshop
Punjab Agricultural University, Ludhiana. on Water-Wise Rice Production, April 8-11,
Anonymous (2010). Review of the Economy. Bouman, B. A. M., Hengsdijk, H., Hardy,
Economic Advisory Council to the Prime B., Bindraban, P. S., Tuong, T. P. and Ladha,
Minister, Hall-E, Vigyan Bhawan, Maulana J. K. (eds.). International Rice Research
Azad Road, New Delhi. Institute, Los Banos, Philippines. pp. 143-
Bhuiyan, S. I. (1992). Water management in 54.

View publication stats

Você também pode gostar