Content uploaded by N H Rao
Author content
All content in this area was uploaded by N H Rao on Feb 06, 2014
Content may be subject to copyright.
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 87, NO. 5, 10 SEPTEMBER 2004 628
GIS-based decision support system for real
time water demand estimation in canal
irrigation systems
N. H. Rao1,*, Sheena M. Brownee2 and P. B. S. Sarma2
1National Academy of Agricultural Research Management, Rajendranagar, Hyderabad 500 030, India
2Water Technology Centre, Indian Agricultural Research Institute, New Delhi 110 012, India
In canal irrigation systems in India, water supplies
reach the fields through a network of main canals,
branch canals (secondary canals) and distributaries
(tertiary canals). The distributary is the basic unit of
irrigation management in large canal systems, as it is
the last point of control in main irrigation systems
operation. This study presents a scheme for the deve-
lopment of a Geographic Information Systems (GIS)-
based decision support system (DSS) for real time wa-
ter demand estimation in distributaries. The DSS dy-
namically links a field irrigation demand prediction
model for the area irrigated by a distributary with a
GIS of the canal network. The system allows interac-
tive selection of distributaries and on-line real time
estimation of water demands in each distributary over
the entire network. For real time estimates, the model
is used with current season information on weather,
weather forecasts and distributary level information
on crops and soils. Since the distributary is the unit of
operation, the DSS integrates well with the actual
process of decision-making by the operators of canal
irrigation systems in India. The availability of such a
quantitative decision-support tool for irrigation systems
operation can have a powerful impact on the overall
water management strategy to be adopted in an irri-
gation project area, particularly in the event of a
shortfall in water supplies. The development of the
overall scheme and procedures is illustrated with data
from a case study area in India.
THE creation of a number of large irrigation systems in
India contributed significantly to gains in food produc-
tion provided by the Green Revolution. It also improved
food security in the region and reduced the dependence
of agriculture on the vagaries of the monsoons. However,
the increasing costs of creation and maintenance of these
systems, and doubts about the long-term sustainability of
the soil and water resources in their command areas, have
led to much criticism and concern. This relates mainly to
the low operational efficiencies of the large systems (rang-
ing between 30 and 40%), and consequent water losses in
transmission and low crop yields1–5. There is agreement
that substantial benefits can be derived even from rela-
tively small increases in operating efficiency6.
The operation of large canal irrigation systems is a
complex task. In the major irrigation systems of India, water
is delivered over a large area (10,000 to a million hec-
tares or more in India) with spatially variable soils, crops
and weather conditions. The irrigation supplies reach the
fields through a hierarchical network of main canals,
branch canals (secondary canals) and distributaries (terti-
ary canals). The distributary is usually the last point of
control for main irrigation system management as down-
stream of this level, irrigation is either field-to-field or
under the direct control of the farmers. The irrigation
supplies into each distributary are decided based on the
estimated water demands of the crops in the area irrigated
by it, after accounting for field-application losses. The
demands depend on soil, weather and crop conditions in
the irrigated area. Further, the total areas irrigated by
different distributaries also vary. The irrigation demand
estimation for each distributary is therefore independent
of other distributaries. The individual distributary-level
water demands are aggregated to assess irrigation supply
requirements at higher levels (branch canals and main
canals) of the irrigation system after accounting for
transmission losses7. The operational efficiencies depend
on the extent to which the irrigation supplies match the
demands at each hierarchical level of the network. Thus,
estimating periodically, and in real time, the water de-
mands of individual distributaries of the canal network
are critical for improving the overall operational efficien-
cies of large irrigation systems.
Most earlier studies in real time irrigation system manag-
ment to improve operational efficiencies8,9 have focused
on water releases at the reservoir level, after aggregating
the irrigation demands at different hierarchical levels to
this level. The temporal variability of reservoir inflows is
the primary source of uncertainty in these studies. They
do not include the variability of distributed demands in
real time at the field, distributary and other levels of the
water distribution system. When the problem of assessing
irrigation requirements for real time operation of irriga-
*For correspondence. (e-mail: nhrao@naarm.ernet.in)
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 87, NO. 5, 10 SEPTEMBER 2004 629
tion systems at the distributary level was addressed10,
the spatial variations in soil and crop were not explicitly
considered. Some studies11,12 addressed the problem of
real time operation of irrigation systems at three stages,
reservoir releases, transfer to field level after accounting
for channel losses and water allocation to crops. Here
too, the uncertainty was limited to variations in inflows
and the irrigation requirements of crops were considered
to be constant in different periods. In nearly all these
studies, the focus is on coping with uncertainty caused by
variable water supplies.
To increase operational efficiencies, variable irrigation
supplies need to be matched in real time with the variable
irrigation requirements over space. Spatial data manage-
ment tools like Geographic Information Systems (GIS)
can effectively include spatial variability of soil, crop,
water supply and environment in dealing with the com-
plex problems of water resources management13. Earlier
studies in large-canal irrigation systems management that
involved use of GIS tools, dealt with agricultural per-
formance evaluation by combining them with remote sensing
and hydrologic models14,15. These studies were essen-
tially diagnostic in nature and aimed at evaluating the
uniformity and sustainability of irrigation as reflected by
the crop yields. While such studies can be useful in irri-
gation planning and policy, they do not address the oper-
ational problems faced by irrigation managers in real
time.
A basic operational problem faced by irrigation man-
agers is the estimation of irrigation requirements at the
level of each distributary at the beginning of every irriga-
tion cycle. The difficulty lies in obtaining quick, systematic
and realistic estimates of the demand in real time for dif-
ferent distributaries in the canal network in the presence
of spatial variations in weather, soil and crop in the areas
irrigated by them. This study develops a scheme for pro-
viding a GIS-based tool for irrigation system managers to
assist them in making such estimates. It is shown that the
features of GIS for storing, manipulating and analysing
spatial data related to soil, crop and weather can be used
to (i) provide an effective information system for the pro-
ject area that is interactive and representative of the hier-
archy of irrigation system operation, and (ii) obtain real
time, systematic and quick estimates of irrigation demands
in the distributaries taking-off from different canals/branch
canals.
For (ii), irrigation demand prediction models are inte-
grated with current season information on weather, weather
forecasts and local information on crop and soil in a GIS
environment. Since the distributary is the basic unit of
decision making in the operation of the canal system, it
allows quick estimation of the spatial variations in irriga-
tion requirements in different distributaries that form the
canal network, by interactively selecting them in the GIS.
The procedures are illustrated by applying them to a case-
study area.
The case study area and problem
The case-study area forms a part of the Sone irrigation
project in Bihar, India. The Sone project is a river diver-
sion scheme built on the river Sone. The river is a tribu-
tary of the Ganga. The project irrigates about 400,000 ha
during the monsoon (kharif season) and 175,000 ha dur-
ing the rabi (winter season). The area receives about
1100 mm of rain, over 80% of which occurs over the
monsoon season (June to September). Soils are alluvial
and vary from light to heavy-textured clays. Rice is the
main crop grown in the area in the kharif (monsoon) sea-
son. The irrigation system has been operative16 since
1871. The canal network consists of main canals and sev-
eral branch canals (Patna canal, Arrah canal, Behea branch
canal, Dumraon branch canal, Buxar canal, Chausa branch
canal, Garachoubey branch canal). Each branch canal has
a network of several distributaries (tertiary canals) and
minor distributaries or minors. Data from the irrigation
distributaries of the Patna canal network of the Sone irri-
gation system are used for the case study.
The Patna canal (Figure 1) extends over 70 km and
irrigates nearly 190,000 ha through a network of 40 dis-
tributaries. The study area is bounded by the river Sone
Figure 1. Command area of Patna canal system.
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 87, NO. 5, 10 SEPTEMBER 2004 630
on the west, river Punpun on the east and the river Ganga
in the north. The area irrigated by each distributary ranges
from < 200 to > 25,000 ha. The distributaries operate on
a two-week cycle (10 days on and 4 days off). Rice is the
sole irrigated crop during the rainy season, grown under
standing-water conditions. Rice requires about 150 mm
depth of standing water in the field during transplanting
of seedlings. The designed discharge capacities of the dis-
tributaries are not adequate to meet the irrigation re-
quirement of all the fields for transplanting the entire
area irrigated by them in one irrigation cycle. For this
reason, transplanting of the rice fields under each dis-
tributary is usually ‘staggered’ (distributed) over 3 or 4
irrigation cycles (6–8 weeks during mid or late June to
August). In effect, this would mean that within the total
area irrigated by an individual distributary, the rice crops
in different fields would be at different stages of deve-
lopment depending on the date (irrigation cycle) of trans-
planting. This leads to variable soil water and standing
water conditions in different rice fields irrigated by a
distributary during any irrigation cycle.
The irrigation managers in the area are required to pre-
pare a ‘water indent’ (release requirement) for each dis-
tributary before the beginning of each irrigation cycle,
based on the anticipated water requirements of the rice
crop in its area to the end of the cycle, after accounting
for channel losses through seepage. The indents for indi-
vidual distributaries are aggregated to prepare a water
indent for releases into the Patna canal of which they form
the water distribution network. Precise estimation of irri-
gation requirements of the crops for each irrigation cycle
at the distributary level is therefore critical for main sys-
tem management in the project area.
The irrigation requirements for an irrigation cycle of a
distributary depend on the standing water depth in the
rice fields at the beginning of the cycle and the antici-
pated evapotranspiration and percolation losses to the end
of the cycle. These would vary for different fields depend-
ing on the dates of transplanting. They would also depend
on the forecast information on rainfall. A systematic pro-
cedure for assessing the crop water status and irrigation
release requirements in different distributaries in real
time, which includes the variations in transplanting dates
and other spatial variations within its total area, can assist
irrigation managers in making more realistic irrigation
demand estimates. Incorporating the procedure in a GIS
environment will permit interactive selection of distribu-
taries on a computer screen to estimate their release re-
quirements. This will provide a powerful decision support
tool for the main irrigation system management in the
area.
Development of decision support system
The decision support system (DSS) has two components.
The first component is essentially a spatial information
system of the canal network with the distributary as the
basic unit of information. The second component will
enable irrigation managers to decide on the ‘water in-
dents’ for each distributary, by estimating in advance the
irrigation releases required at the head of the distributary
for each biweekly cycle of its operation. The basis of the
DSS is that if a soil water balance model is linked dyna-
mically to the GIS of the canal system with the distribu-
tary as the basic unit, the irrigation releases for any
distributary can be estimated on-line by simple interac-
tive selection of the distributary in the GIS. The selection
will automatically identify (from the attribute table in
GIS) the relevant input data files for the soil water bal-
ance model (rainfall, soil and crop data files) for the selec-
ted distributary. Since rice is the only irrigated crop in
the area in the rainy season, the crop transplanted on dif-
ferent dates is treated as an independent crop and the soil
water balance model is run separately for each date of
transplanting. The releases for each distributary will de-
pend on water requirements of crops (rice transplanted on
different dates) estimated by the model, their areas and
conveyance efficiencies. The soil water balance model is
thus run at daily time steps in two stages: (i) with current
season data of daily weather up to the starting date of the
irrigation cycle, and (ii) with forecast data of weather to
the end of the irrigation cycle. (For model development
and demonstration, historical data are used as a perfect
forecast.)
This two-stage process is repeated for each irrigation
cycle of the distributary. In this way, the GIS-based frame-
work facilitates interactive selection of the distributary,
and in preparing a water indent for the distributary for the
next irrigation cycle.
GIS of study area
The canal network map of the Patna canal system with 40
distributaries was digitized using ARC/INFO GIS soft-
ware. The distributary is treated as the basic unit of the
GIS. Relevant design data (name, reach of main canal,
chainage, length, design discharge, cultivable command
area, nearest raingauge station, dominant soil type, etc.)
of each distributary were added to the arc attribute table
created in the GIS (Table 1). The corresponding data for
any distributary are automatically identified, when it is
selected interactively in the GIS.
The soil water balance model
A daily time-step soil water balance model for rice, deve-
loped in earlier studies7,17,18, was adopted for use in the
study area. The equations describing the processes in the
model are sufficiently general, and the model and pro-
grammes have been tested and used over a variety of en-
vironmental conditions. The parameters of the model are
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 87, NO. 5, 10 SEPTEMBER 2004 631
also directly derived from information of soil textural
properties. The major modifications for this study are
to the input and output routines of the model to accept
input from the GIS and variable transplanting dates for
rice in the area. The output routines provide information
on the depths of standing water at the end of each day,
and predict these depths over the next irrigation cycle
of two weeks after including information of weather
forecasts.
Traditionally, rice is transplanted and grown under con-
tinuously flooded conditions with about 5 to 10 cm or
more depth of standing water throughout the season. This
practice has been considered desirable not only for ade-
quately meeting the water needs of rice, but also for an
efficient supply of nutrients to the crop and effective
weed control in the field. Such practices and others like
puddling of soils for reducing percolation rates, compli-
cate the soil water balance computations. The water balance
of a puddled rice field (Figure 2) is given by:
FRt = FRt–1 + Rt–1 + IRRt–1 – ETt–1 – DPt–1 – Qt–1. (1)
In eq. (1), FRt is water depth in the field at the beginning
of day t, Rt–1 and IRRt–1 are rainfall and irrigation applied
respectively, at the beginning of day (t–1), and ETt–1,
DPt–1 and Qt–1 are evapotranspiration, percolation and
surface runoff during day (t–1). These components are
expressed in depth units (mm) and the time period con-
sidered is one day. Capillary rise from groundwater is
ignored on the basis that the plough layer remains under
saturated conditions for a considerable length of the crop
growth period and is at a higher moisture potential than
the capillary fringe at deeper depths19.
Table 1.
Attribute data of distributaries of the Patna canal system
Distributary
identification no.
Name Design
discharge (cusecs)
Irrigable
area (ha) Raingauge*
identification no.
Soil*
identification no.
1 Manora 139 4695 2 1
2 Teldiha 60 2029 2 1
3 RPC1 24 817 2 1
4 Tejpura 24 800 2 1
5 Tejpura feeder 17 575 2 1
6 Chanda 191 6465 2 1
7 RPC2 18 605 2 1
8 LPC1 0 0 2 1
9 Tuturkhi 113 3838 2 1
10 Mali 387 13118 2 1
11 Ancha feeder 37 1265 2 1
12 Kochasa 588 19904 2 1
13 RPC3 15 505 2 1
14 Amra 210 7106 2 1
15 Imamganj 167 5671 2 1
16 RPC4 5 180 2 1
17 LPC2 8 273 2 1
18 RPC5 8 286 2 1
19 LPC3 8 260 2 1
20 Rampur Chauram 83 2816 2 1
21 Aiyara 144 4887 2 1
22 Murka 272 9216 2 1
23 RPC6 0 0 1 1
24 LPC4 0 0 1 1
25 Paliganj 360 12197 1 1
26 RPC7 0 0 1 1
27 Dorwa 131 4436 1 1
28 RPC8 0 0 1 1
29 Dhana Minor 10 319 1 1
30 Jwarpur 0 0 1 1
31 Maner 784 26547 1 1
32 Adampur 97 3270 1 1
33 Rewa 115 3904 1 1
34 Tangrila 31 1050 1 1
35 Khajuri 44 1485 1 1
36 Fatehpur 317 10741 1 1
37 Manjhauli 69 2337 1 1
38 Kurkuri 88 2962 1 1
39 Danapur 126 4263 1 1
40 Patna 26 876 1 1
*Data from two raingauges are used. Raingauge 1 is at the tail end of the Patna canal and Rain
gauge 2 is near
its headworks. Data for only one soil type are used, as the soils in the case study area do not vary signifi-
cantly, particularly after puddling the rice fields. However, the DSS software is suffi
ciently general to deal
with such variations.
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 87, NO. 5, 10 SEPTEMBER 2004 632
Rainfall in excess of bund height leaves the system as
surface run-off (Q).
Q = R + FR – BH, (2)
where BH is bund height in mm. Further, according to the
recommended agronomic practices, the rice field is
drained once during the season, at the peak tillering stage
(about 45 days after transplanting for medium duration
varieties of rice) of the crop and maintained in the drained
condition for 3 or 4 days to prepare the field for applica-
tion of fertilizers. The specific days of drainage period
are 45 to 47 days after transplanting. Standing water depth
at the beginning of three days and rainfall which occurs
during these days, are treated as run-off.
Deep percolation is the vertical downward movement
of water below the crop root zone. Percolated water is not
available for use by the crop. Percolation is governed by
the hydraulic conductivity of the soil profile and the
depth of standing water on the field. The percolation rate
of puddled rice fields is affected by a variety of factors
like soil structure, texture, bulk density, mineralogy, organic
matter content and concentration of salt in soil solution20.
The percolation rate is further influenced by water re-
gimes in and around the field. Increased depths of ponded
water increase percolation due to the larger hydraulic
gradient21. Further, because of puddling, the soil layer
below the root zone (approximately 30 cm from the
soil surface) gets compacted. As a result, the saturated
hydraulic conductivity of this layer is reduced when com-
pared to that of unpuddled fields. The reduction in satu-
rated hydraulic conductivity caused by puddling varies
with the texture of the soil. The reduction is four to
five times for clay loam soil, eight to ten times for
sandy loam soil22–24 and ten to fourteen times for sandy
clay loam soil25,26. Accordingly, in the present study, the
saturated hydraulic conductivity values are selected
based on texture and reduced by a suitable factor to take
care of the puddling effect. The daily percolation rate out
of the root zone (30
cm) layer is computed by Darcy’s
law as:
DP = – ks dh/dz,
(3)
where DP is
percolation out of the root zone in mm/day,
ks
is saturated hydraulic conductivity in mm/day (after
accounting for puddling effects) and dh/dz is head gradi-
ent. DP is calculated using eq. (3), so long as FR >
0. If
the standing water disappears in any irrigation cycle, DP
is given by the difference in water depth between the soil
at saturation and field capacity. When soil moisture falls
below field capacity, DP
is assumed to be zero. This is a
common assumption in field scale soil water balance models.
Evaporation from the soil and water surface and tran-
spiration from the plant leaves are combined and treated
together as evapotranspiration (ET). ET
for rice depends
mostly on climatic conditions20. Maintaining ET
at the
potential rate, i.e. the rate whic
h is not hindered by water
shortage, is essential for high yields of rice27
, because
yields will decline with decreasing rate of ET. In the pre-
sent study, actual ET is assumed to equal potential ET
, if
soil moisture content is above or equal to field capac
ity.
During the periods, when the soil moisture falls below
field capacity (this can happen when the irrigation sup-
plies are inadequate in any irrigation cycle to meet the
potential ET re
quirements over its entire duration and when
water supplies are cut off before the crop reaches maturity),
actual ET is assumed to decrease linearly with soil mois-
ture content between field capacity and permanent wilt
ing
point. Crop coefficients used in the estima
tion of actual
ET are adopted from FAO28. The following relat
ionship
was used to estimate daily mean evapotranspiration:
PET = Kc * ETo,
(4)
where ETo is crop reference evapotranspiration in mm, Kc
is crop coefficient and PET is potential evapotrans-
piration. ETo
values were estimated using the modified
Penman method28.
The model runs from transplanting date up to the
harvest date. An irrigation depth of 150
mm is applied
at the time of transplanting to both saturate the total
root depth and allow for about 10
cm of standing water
on the field. The high irrigation requirement at trans-
planting and the limited channel capacity lead to stag-
ger
ing of the transplanting activities over the command
area of the distributary. A staggering period of two weeks
is consid
ered to coincide with the irrigation cycles. In
each
cycle, 25% of the command area is transplanted.
These reflect the general conditions in the study area. The
irrigation requirements in subsequent cycles are estima-
ted using the soil water balance model by scheduling irri-
gations of 50 mm depth on any day within the irriga
tion
cycle, when the depth of standing water falls to 10
mm
or less. Irrigation is cut-off two weeks before harvest.
Figure 2. Water balance components of rice field.
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 87, NO. 5, 10 SEPTEMBER 2004 633
The dates of irrigation scheduling during the growing
season and the cut-off date for irrigation before harvest
will vary with the transplanting date of the crop. Thus
rice transplanted on different dates needs to be treated as
different crops for running the soil water balance model.
Since there are four transplanting dates, the soil water
balance model is run four times in each cycle. The irri-
gation requirements for the areas of the crops trans-
planted on the four different dates are cumulated to arrive
at the irrigation indent for the next irrigation cycle. The
input data required for running the model for each dis-
tributary are summarized in Table 2.
Dynamic linkage between model and GIS
The GIS of the Patna canal system and the rice water
balance model were dynamically linked for real time appli-
cation in any season (Figure 3). This linkage allows:
(i) Selection of the distributary of interest on screen to
identify the corresponding weather station and soil data
files.
(ii) Running the rice field water balance model for each
transplanting date in real time up to current date in any
year, after entering the current date in response to screen
queries.
(iii) Preparing a report of the current water status in rice
fields in the command area of the distributary transplan-
ted on different dates.
(iv) Preparing a water indent for the irrigation require-
ments at the head of the distributary for the next irriga-
tion cycle, after accounting for weather forecasts and
conveyance losses.
(v) Proceed to next distributary.
Steps (i) to (v) are carried out sequentially and on-line
within the GIS environment. The user need not at any
stage come out of the GIS environment. For steps (i) to
(iv), the complete sequence is run for each distributary
with actual rainfall data up to the current date and with
the forecast data of daily rainfall for the next 14 days of
the irrigation cycle. At the end of this cycle, which is also
the beginning of the next cycle, the actual rainfall data
for this period would be available. Before the irrigation
indents are prepared for the next cycle, the actual rainfall
data of the previous cycle are used to assess the water
status at the beginning of the cycle, and the entire se-
quence is repeated. For this reason, the model needs to be
run twice for any irrigation cycle – first with forecast
rainfall for the current cycle and then with the actual
rainfall in this cycle, when the irrigation cycle advances
to the next.
Results and discussion
Individual distributaries can be selected by users from the
GIS and reports of water status in fields and indents for
water for the next irrigation cycle on any given date can
be prepared quickly on-line and in real time. The results for
one cycle are presented in Figure 4. Once the distributary
is selected, the soil water balance model runs at daily
time steps with the soil, rainfall and crops data up to the
starting day of the irrigation cycle for which the water
indent for the distributary is to be prepared. For the two-
week period following this date, the model uses the fore-
cast rainfall data to calculate the daily water balance to
the end of the irrigation cycle. To illustrate the method,
daily historical data for the period of the irrigation cycle
are used as a perfect forecast. The report shown in Figure
4 is prepared for such a perfect forecast. The report also
lists information on the soil water and crop conditions at
the beginning of the irrigation cycle. The entire process
(after the distributary is selected in GIS) is automated
and made user-interactive within the GIS. The results of
running the model for one distributary with historical
rainfall data of one season, assuming perfect rainfall fore-
casts for all the irrigation cycles of 14 days beginning
July 1 are presented in Figure 5.
The demands vary significantly over the different cy-
cles, and in the initial irrigation cycles they exceed the
capacity discharge of the distributary. In such periods,
Table 2.
Input data for soil water balance model
Identification number of distributary (selection in GIS)
Corresponding identification numbers of raingauge stations and soil types (from GIS attribute table)
Daily rainfall for each raingauge station (mm)
Forecast rainfall
Initial soil moisture and soil moisture content at saturation for each soil type (mm/cm)
Reduction factor for hydraulic conductivity because of hard pan (for each soil type)
Transplanting date (day and month)
Number of transplanting dates
Crop duration (days)
Maximum root depth (cm)
Days to attaining maximum root depth
Bund height (mm)
Date of start of drainage period and duration
Days to cut-off date of irrigation
Daily reference evapotranspiration (mm)
Daily crop coefficients
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 87, NO. 5, 10 SEPTEMBER 2004 634
there may be a need to prioritize water allocations among
the crops in the area irrigated by the distributary. Since
the report produced for each irrigation cycle (Figure 4)
also includes information on the soil moisture, standing
water on fields and crop development conditions, irriga-
tion managers can use this report, together with agro-
nomic knowledge of crop responses to stress, to support
decisions on prioritizing water allocations among crops.
In the later periods (Figure 5), the requirements are far
less than the capacity discharge, even in periods with low
rainfall. This is because the crops transplanted in the early
irrigation cycles reach the maturity stage in these cycles
and do not need irrigation. Only the area under the late
transplanted crops will need irrigation, leading to a
reduction in water demand for the distributary. In the ab-
sence of a formal decision support framework to assist
irrigation managers, they tend to be conservative and
indent in excess of the requirement even in the later
irrigation cycles. If the DSS is used to prepare the water
release indents, water can be conserved in the reservoir in
the later irrigation cycles and used for the winter season
crop, which is planted after harvest of rice.
The results presented in this study were based on his-
torical data, and perfect forecasts of rainfall at the begin-
Figure 3. Dynamic user-GIS-model linkages in decision support system.
Figure 4. GIS-model output for selected distributary.
in real time
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 87, NO. 5, 10 SEPTEMBER 2004 635
ning of each irrigation cycle were used to demonstrate
the feasibility of developing a GIS-based framework for
real time water demand estimation in irrigation projects.
However, field applications in real time would need to use
actual medium range weather forecasts (5–7 days ahead),
which are now being made available by the IMD on re-
quest for any location in the country. The use of these
forecasts would not alter the model development and appli-
cation presented above in any way. The only change is to
the input data for the irrigation cycle, which would
change to the forecast data for the cycle instead of the
historical data for the period which was used for the per-
fect forecast case. The framework presented here is ide-
ally suited for the effective use of medium range weather
forecasts, as the system status is clearly defined (Figure
5) at the beginning of each irrigation/forecast cycle.
However, it is interesting to note29 that in situations
where the available water storage capacity of the soil is
high, irrigation depths are fixed and irrigation decisions
are based on continuous monitoring of soil and crop con-
ditions (as with the soil water balance model); the deci-
sions themselves are not likely to be significantly influ-
enced by errors in medium range weather forecasts. What
is relevant and important is the monitoring of the soil
water balance with respect to the current season weather
data up to the time when the forecast is used. Since the
situation described here meets the above conditions (and
there is provision for an additional storage of water up to
the bund height), errors in forecast may not significantly
influence the irrigation indents for each cycle.
Conclusion
The study developed a GIS-based decision support sys-
tem for water demand estimation in canal irrigation systems.
The Patna canal network of the Sone Irrigation Project in
India was used as a case study. The main problem faced
by water managers is estimating demand at the head of
the distributaries of the canal network in advance for each
irrigation cycle. This is mainly because of spatial variations
in weather and rice crop transplanting dates. It was shown
that real time water demands for any distributary can be
estimated by linking dynamically the GIS of the canal
system with a soil water balance model and current sea-
son data of weather, weather forecasts, and crop and soil
conditions. The system managers can obtain the required
information by simply selecting the distributary in the
GIS. The DSS also allows quick estimation of the varia-
tions in irrigation requirement in different distributaries
that form the canal network and comparisons with the
available channel capacities and actual supplies. Such
visualizations, when combined with strong agronomic
knowledge and judgment, can have a powerful impact on
the overall water management strategy to be adopted in
the command area of the irrigation project. Though the
various procedures have been developed for the case study
area selected, they are sufficiently general to be adopted
to other canal irrigation networks.
1. Water Technology Centre, Resources and plan for efficient water
management – A case study of Mahi right bank canal command
area, Gujarat, Res. Bull. No. 42, IARI, New Delhi, 1983.
2. Water Technology Centre, Development of guidelines for sustain-
able water management in irrigation projects including conjunctive
use of canal water and groundwater. In Contributions to Water Sci-
ence and Technology, IARI, New Delhi, 1998, pp. 11–60.
3. FAO, Improved irrigation system performance for sustainable
agriculture. Proceedings of the Regional Workshop on Improved
Irrigation System Performance for Sustainable Agriculture, Bang-
kok, Rome, 1991.
4. Irrigation, command area development and flood control. Annual
Plan 2000–01, Planning Commission, Govt. of India, Chapter 6.2,
pp. 312–323.
5. Wood, S., Sebastian, K. and Scherr, S., Pilot analysis of global
ecosystems: agroecosystems, a joint study by International Food
Policy Research Institute and World Resources Institute. Washing-
ton, D.C., USA, 2000, p. 94.
6. Maidment, V. R. and Chow, V. T., Stochastic state variable dyna-
mic programming for reservoir systems analysis. Water Resour.
Res., 1981, 17, 1578–1584.
7. Hajilal, M. S., Rao, N. H. and Sarma, P. B. S., Planning intrasea-
sonal irrigation requirements in large projects. Agric. Water Man-
age., 1998, 37, 163–182.
8. Darianne, A. B. and Hughes, T. C., Application of crop yield func-
tions in reservoir operation. Water Resour. Bull., 1991, 27, 649–
656.
9. Vedula, S. and Majumdar, P. P., Optimal reservoir operation for
irrigation of multiple crops. Water Resour. Res., 1992, 28, 1–9.
10. Sriramany, S. and Murty, V. V. N., A real time water allocation
model for large systems. Irrig. Drain. Syst., 1996, 10, 109–129.
11. Hajilal, M. S., Rao, N. H. and Sarma, P. B. S., Real time water
management in storage based irrigation systems. Central Board of
Irrigation and Power, New Delhi, Publ. No. 256, 1997, p. 370.
12. Hajilal, M. S., Rao, N. H. and Sarma, P. B. S., Real time irrigation
reservoir operation. Agric. Water Manage., 1998, 38, 103–122.
13. Tsihrintzis, V. A., Hamid, R. and Fuentes, H. R., Use of geo-
graphic information systems (GIS) in water resources: a review.
Water Resour. Manage., 1996, 10, 251–257.
Figure 5. Irrigation indents (required releases) for Paliganj distribu-
tary in different irrigation cycles.
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 87, NO. 5, 10 SEPTEMBER 2004 636
14. Bastiaanssen, W. G. M., Molden, D. J., Thiruvengadachari, S.,
Smit, A. A. M. F. R., Mutuwatte, L. and Jayasinghe, G., Remote
sensing and hydrologic models for performance assessment in
Sirsa irrigation circle, India. Research Report 27, International
Water Management Institute, Sri Lanka, 1999, p. 29.
15. Sakthivadivel, R., Thiruvengadachari, S., Amarasinghe, U., Bas-
tiaanssen, W. G. M. and Molden, D. J., Performance evaluation of
the Bhakra Irrigation System, India, using remote sensing and GIS
techniques. Research Report 28, International Water Management
Institute, Sri Lanka, 1999, p. 22.
16. Svendsen, M. and Sinha, B., The evolution of large irrigation sys-
tems: a brief look at the Sone Command, 1850–1992. In Strategic
Change in Indian Irrigation (eds Svendsen, M. and Gulati, A.),
Indian Council of Agricultural Research, New Delhi, India and In-
ternational Food Policy Research Institute, Washington, D.C.,
USA, 1994, pp. 13–36.
17. Chowdary, V. M., Rao, N. H. and Sarma, P. B. S., GIS based deci-
sion support system for groundwater assessment in large irrigation
project areas. Agric. Water Manage., 2003, in press.
18. Chowdary, V. M., Rao, N. H. and Sarma, P. B. S., A coupled soil
water and nitrogen balance model for flooded rice fields. Agric.
Ecosyst. Environ., (accepted) 2003.
19. Odhiambo, L. O. and Murthy, V. V. N., Modeling water balance
components in relation to field layout in lowland paddy fields. 1.
Model application. Agric. Water Manage., 1996, 30, 185–199.
20. Wickham, T. H. and Singh, V. P., Water movement in wet soils. In
Soils and Rice, International Rice Research Institute, Loss Banos,
Philippines, 1978, pp. 37–57.
21. Ferguson, J. A., Effect of flooding depth on rice yield and water
balance. Arkansas Farm Res., 1970, 19, 4.
22. Varade, S. B. and Ghildyal, B. P., Mechanical impedance and
growth of paddy in artificially compacted lateritic sandy loam soil.
J. Indian Soc. Soil Sci., 1967, 15, 157–162.
23. Naphade, J. D. and Ghildayal, B. P., Effect of puddling on physi-
cal properties of rice soil. Indian J. Agric. Sci., 1971, 41, 1065–
1067.
24. Sharma, P. K. and de Datta, S. K., Puddling influence on soil, rice
development and yield. Soil Sci. Soc. Am. J., 1985, 49, 1451–1457.
25. Ghildayal, B. P. and Satyanarayana, T., Effect of compaction on
physical properties of four soils of India. J. Indian Soc. Soil Sci.,
1965, 13, 149–155.
26. Pande, H. K., Water management practices and rice cultivation in
India. In Symposium on Water Management in Rice Field, Tropi-
cal Agriculture Research Centre, Ministry of Agriculture and For-
estry, Japan, Technical Report P13, 26–29 August 1975.
27. IRRI, Annual Report for 1986, International Rice Research Insti-
tute. Los Banos, Philippines, 1987.
28. Doorenbos, J. and Pruitt, W. O., Crop water requirements. Irriga-
tion and Drainage Paper No. 24, FAO, Rome, Italy, 1977.
29. Hajilal, Rao, N. H. and Sarma, P. B. S., Can medium range
weather forecasts influence irrigation scheduling? Curr. Sci., 1994,
66, 60–63.
ACKNOWLEDGEMENTS. We thank the Department of Science and
Technology, New Delhi for funding and the authorities of the Sone
Command Area Development Authority for providing data for the case
study area.
Received 7 November 2003; revised accepted 26 May 2004
