FUTURE  PERSPECTIVES ON MODERNIZATION

Modernizing irrigation operations:
spatially differentiated resource allocations

D. Renault & I.W. Makin

Irrigation Specialist, and Head of Design and Operations Programme, respectively, IWMI

Abstract

Modernization of irrigation implies interventions in different components of system management. This paper focuses on operations and proposes a methodology for an improved assessment of irrigation canal behaviour and the environment in which operations take place. An underlying assumption is that irrigation systems are generally heterogeneous and therefore the allocation of operational resources should be matched to the spatial distribution of management requirements.

A descriptive model of irrigation systems is presented by defining three domains. First, the cause, frequency of occurrence and magnitude of perturbations to the flow regime are considered as the perturbation domain. Second, the behaviour of the physical system when subject to perturbation is considered as the sensitivity domain. Third, the impact of system operations on agricultural yields is examined in the vulnerability domain, which enables the development of the specifications of a required water service.

Combining the vulnerability and sensitivity domains makes the definition of the precision with which systems must be operated possible. The inclusion of the perturbation domain allows for the specification of the required mode of operation to be implemented to achieve the required water service, including specification of the required frequency of intervention. The whole provides scope for the definition of the demand for operation at a spatially de-aggregated level.

Introduction

Irrigation modernization is increasingly recognized as a fundamental transformation in the management of water resources within agricultural areas. Such transformations may include improved structures, physical or institutional or both; rules and water rights; water delivery services; accountability mechanisms and incentives. In this paper we address how modernization provides an opportunity to redefine and update operational procedures within irrigation schemes. By incorporating broader perspectives and paying attention in particular to the spatial distribution of significant variables, this paper defines new approaches to the allocation of operation resources. The critical step, i.e. the means by which to obtain and manage the resources, is not addressed here.

The objective of this paper is an improved methodology for evaluation of the resource demands for effective canal operations to enable more cost-effective operational management. The basis of the proposed approach is whether or not operational requirements are homogeneously distributed throughout the entire scheme. If not, we would argue, operations require different responses in different sections of the scheme.

Demand for operational resources consists primarily in answering the following questions:

• What mode of operation?
• What degree of precision?
• With what frequency of checking and intervention?
• What monitoring system is required?

The need to re-evaluate and update approaches to operations is given impetus by the tremendous changes that have occurred in the irrigation sector over the last few decades. These result from increasing competition for both water and financial resources and also growing concern over the environment and health impact of irrigation. Water management is no longer narrowly focused but must embrace a broad perspective including water quality, conjunctive management, multiple uses of irrigation waters, a watershed perspective, new water rights and priorities for distribution. A traditional quantitative and rather uniform management system for irrigation schemes is no longer sufficient to address current issues. Furthermore, these trends will continue and system operators will have to develop more cost-effective operational plans to satisfy the increasingly influential users-payers.

As opportunities to develop new areas are increasingly restricted, many existing irrigation schemes are, or will in the near future be, undergoing major changes, either physical or institutional, or both. It is necessary to scrutinize the basic irrigation activities, operation and maintenance, in order to ensure that the new systems are economically sustainable.

Canal operations in technical literature

Canal operation and flow control techniques are well documented, particularly for system design analysis - Zimbelman D.D. (1987), Paudyal and Loof (1988), Plusquellec H. (1988), Plusquellec H. et al (1994), RIC (1997). However, there are few published studies on how managers should operate existing systems, evaluate the operational requirements or allocate resources and effort to optimize system performance. In many schemes, a mixture of rule of thumb and local experience is the basis for operational decision-making. There is no standardized base for retention of operational experience and, due to senior staff rotation, there is a risk of permanent loss of knowledge if such information is not formally recorded in an understandable form.

Any renewed approach to canal operations must bridge the gap between on-site management and official plans for operation and maintenance, and other such operational guides. These guide manuals are increasingly required by authorities and funding agencies at the completion of structural works (new projects or rehabilitation). However, unless it is recognized that the operational framework cannot be fully planned at design stage and that finetuning over some years of practice is a fundamental requirement (Uittenbogaard and Kuiper, 1993), it is proposed that an adaptive or learning process is preferable to strictly prescriptive approaches (Handbook, 1990; Skogerboe and Merkley, 1996).

Operations in the irrigation process

Operations are the manipulation of physical structures in the irrigation system to implement management decisions about water allocation, schedules of delivery and distribution. Operations are also the routine actions taken to minimize the impact of perturbations by maintaining steady or quasi-steady state water profiles in the system and to prevent overtopping at peak discharges.

Operations are routinely required to implement distribution decisions and, as a consequence, the terms are sometimes confused, even though they are fundamentally different. To clarify the thinking, technical irrigation management implies three levels of decisions - allocation, scheduling and distribution - and one level of implementation - operations.

Operations and types of irrigation systems

It is self-evident that irrigation systems are not identical in regard to their operational requirements. Some are highly automated and, although this requires larger investments in construction, they often need fewer human and financial resources for day-to-day operation. Other systems are manually controlled and require full and intensive operations during irrigation. We can classify the irrigation systems as:

• Fully operated systems, where all structures (intake-outlets-cross-regulators) require regular and routine operation during irrigation, (setting on-off, setting and monitoring);
• Non-operated systems, generally operating on the proportional distribution principle, common in India and Pakistan. Fixed dividing structures ensure an equitable distribution of water. No operations are required to adjust the on-going flow within the structured system (Shanan, 1992).
• Minimal operation systems, such as those equipped with modules and combined with automatic or fixed regulators. Interventions are generally limited to on-off operations and flow regulation is achieved by control modules or baffles.

This classification of control systems is essentially valid only for intermediate level canals such as distributaries. Main canal systems and field canals are generally fully regulated. We can therefore conclude that whatever control technique in the intermediate distribution system, major irrigation systems include (at least) portions with gates that must be operated.

The basic assumption of heterogeneity and the spatial analysis

Technical manuals for irrigation operations, in general, implicitly assume homogeneity: first, homogeneity in the requirements for operation and therefore homogeneity in the distribution of operational efforts. In many cases this assumption simply does not hold true. Rather, the basic assumption in operating an irrigation system should be that the scheme is heterogeneous, unless it can be clearly shown to be homogeneous.

There is limited literature on the heterogeneity in irrigation. One very noteworthy approach, proposed by Ng Poh-Hok (1987), for the design of an irrigation system uses the concepts of irrigation form and irrigation context. Poh-Hok proposed that these must match in order to be successful. He considered the assumption of heterogeneity, a generic term regrouping variability, uncertainty, diversity and complexity, before presenting a conceptual model of irrigation as a consistent aggregation of elementary homogeneous units. These elementary units were defined as a socio-geographic unit, homogeneous in form and context.

Steiner and Walter (1993) considered the spatial variability of all factors influencing irrigation management, such as the physical characteristics of the context, the quality of infrastructure, etc. These authors later on focused exclusively on the level of spatial variability of climate and simulated the consequences of different allocation schedules.

Consideration of heterogeneity also underpinned the methodology developed for water management on a large scale for the Bhakra system in Haryana, India, by Schakel and Bastiaansen (1997). Irrigation management throughout an area of 1.2 million ha was de-aggregated considering 67 homogeneous geo-hydrological units.

The assumption of heterogeneity in the physical characteristics, the context and therefore of demands for operation is fully recognized here. It is proposed that this assumption is valid not only for large-scale systems but also for smaller ones, of say, one thousand hectares. Therefore the analysis of demand for operational resources should start with a spatial analysis leading to partitioning systems into elementary units with homogeneous characteristics, for convenience's sake called subsystems.

An important consideration is the link between heterogeneity and equity. It is clear that the justification of the widespread application of the assumption of homogeneity is partly related to the goal of achieving equity within a system. This goal should not be ignored in any new approach to operations. Without care, the introduction of the heterogeneity concept may conflict with equity: for example, considering the value of crop per area could lead to reinforcing existing inequity by providing better service to already well-served users.

Methodology

Open-surface canals are subject to modification of flow characteristics (discharge-water depth) resulting from scheduled and unscheduled events. In the usual operational mode the management objective is to maintain steady state conditions when such events occur. The methodology developed here aims to characterize the frequency and magnitude of perturbation events likely to occur in a subsystem. The frequency of change in the distribution pattern defines the perturbation domain. By characterizing the physical properties of the irrigation structures and evaluating the behaviour of canal systems when operated or affected by perturbations, the sensitivity domain is defined. Finally, the analysis of the impact of operation on agricultural yields, on the environment and on the watershed makes it possible to define the vulnerability domain.

Analysis of the vulnerability domain makes the definition of the required water service possible. Considering the required water service performance and combining this with sensitivity analysis of the infrastructure enables the specification of the precision of water depth control required. The mode of canal operation required is defined by the combination of the vulnerability and perturbation domains. Finally the perturbation domain determines the required minimum frequency of system observation and regulation.

The approach can be viewed as a series of overlays of spatially distributed variables, illustrated in Figure1 (appended). Although defined by technical considerations, the process must be sociologically acceptable and also fit the defined objectives of the irrigation scheme.

Opportunities and constraints for water service

The vulnerability domain

Vulnerability is a generic term employed here to describe opportunities and constraints or impact of operation at different scales of space and time. Vulnerability of an irrigated area can be seen as the propensity to be positively or negatively affected by irrigation operations. For instance, a highly vulnerable area would be a unit in which impact and side effects of low-quality operations are high (sensitive crops, areas without drainage facilities). Inversely, low vulnerability areas are those in which impacts and consequences of low-quality operation are either temporally or permanently dampened (paddy fields can stand interruption of water supply for short periods). Vulnerability goes beyond the confines of water for crops and includes consideration of larger-scale water management. Finally vulnerability leads to the estimation of requirements for water service, as both are proportionate.

Some of the wider aspects of water management that define the vulnerability domain are:

• Water quality. Modern agricultural methods and scarcity of fresh water result in irrigation having to deal with water loaded with chemicals (pesticides, nutrients) and other pollutants. Acknowledgement of the importance of water quality is one of the main challenges for current irrigated agriculture, with implications for both surface and groundwater. Many shallow aquifers are important for domestic supply. These often receive some recharge from dry season percolation from irrigated areas representing simultaneously a gift - an additional source - and a threat - pollution. In situations of this kind, managers will have to consider both uses and arrive at an effective compromise.
• Recycling of irrigation water. Drainage flows from irrigated areas can be important assets in water management. Losses in one place become inputs for other areas. The presence of such recycling can substantially ease the upstream management problem by allowing less precision in distribution, knowing that any surplus will not be lost. Return-flow systems - drainage and surplus irrigation are channelled back to the irrigation network itself - represent an opportunity for managers to store positive perturbations, for example to harvest rainfall.
• Water harvesting and conjunctive management. Water harvesting during rainfall is an important opportunity for water management, and specific operation procedures may be designed to maximize harvesting while preventing overtopping. Conjunctive use of water (surface, groundwater and rainfall) can provide additional flexibility to farmers. Groundwater is frequently used to compensate for rigidity or low performance in the surface water delivery system. Areas lacking access to additional supplies from groundwater deserve greater attention than areas where pumping facilities can compensate for inadequate or unreliable deliveries.
• Soil and water salinity and water logging. The rise in soil and water salinity, the increase in waterlogged areas, are environmental hazards of great importance in arid regions. They represent a severe threat to irrigation schemes. It is clear that operation of irrigation systems must take into consideration the spatial distribution of these hazards in order to provide a selective and adapted service of water. In practice, solutions are relatively site-specific and generic guidelines are difficult to derive. But, as a principle, partitioning of the irrigated area should distinguish areas where freshwater has to be provided and areas in which excessive percolation should be avoided to prevent saline groundwater from rising.
• Multiple uses of water. In many irrigation schemes, water is used not only for crops but also for many other purposes including domestic water supply, environmental uses, fisheries, perennial vegetation and hydropower. Rules for multipurpose systems operations are complex because of potential conflicts in setting targets for the different uses and also, on occasion, by the lack of suitable accounting procedures. Multiple use of water will be increasingly integrated in management concerns, whether these uses are considered at design time or not. A first step in the management of multiple uses is to define consistent water and productivity accounting procedures, as proposed by Molden (1997).
• Water rights, equity and priorities in distribution. Water distribution priorities may be based on rights and established uses. However, in systems experiencing water shortages these priorities should define a policy to share limited water among shareholders. Priorities may be defined on the value of crops (high/low), soil water-holding capacity, etc. As the mission of irrigated agriculture changes from subsistence to more highly productive agriculture, it may be necessary sometimes to revise previous policies. It may become necessary to avoid penalizing highly sensitive or high-value crops in case of shortage. In many cases, distribution policies should be rethought and, where appropriate, changed to enable new operational strategies.
• Health impact. Despite the positive effects of irrigation on the economy and income of farmers, there is no doubt that, in some circumstances, it also has a negative impact on the health of communities through vector-borne diseases. The maintenance of water in canals for long periods can affect the reproductive cycle of disease vectors. The link between system operations and community health can be strong. The recommendations of health experts are converging on a desire for more variability in canal flow regimes to, for example, reduce mosquito breeding (Hunter et al, 1993). However, there is a clear conflict between the requirements of the health sector for fluctuations in water depths, and the irrigation management objective of stable water profiles. New techniques of operation may be required where mosquito breeding is related to irrigation practices.
• The position within the system. The impact of operations on the command area is evidently greater for structures located toward the head of the canal system. Therefore location is included in the analysis of vulnerability.

Water service and irrigation performance indicators

The spatial characteristics of the vulnerability domain can be converted into specific water service targets and measured with water supply performance indicators (Bos M.G., 1988). Adequacy, efficiency, dependability, timeliness and equity are the common indicators of performance (Molden and Gates, 1990). Flexibility of access to water and reliability of deliveries are important criteria of performance that should be considered.

Performance indicators for operation can be derived from the vulnerability domain considering both water deliveries for irrigated crops and water management in a broad perspective. In the analysis presented here, only the primary indicators are considered, namely adequacy, efficiency and timeliness. Performance targets are expressed as tolerances with respect to the target discharge rate as shown in equation 1.

(1)

Equation 1 shows that discharge at a given location should be maintained within the two limits, i.e. target discharge -z % or +y%.

A similar relationship can be developed considering the time of delivery, equation 2.

(2)

in which

The perturbation domain

Free surface irrigation systems are hydraulically complex. In general, system operations are reduced to controlling water levels at cross-regulators in an attempt to maintain stable water levels at off-take structures. However, steady water level profiles seldom occur in irrigation systems due to variations at the upstream boundaries of the system (perturbations of intake flow rate) and also the effects of operational interventions themselves. Hence operation is a never-ending challenge as adjustments are made to bring the system to the intended steady conditions in spite of the perturbations.

A perturbation at a given location is defined as a change to the on-going discharge. Such change arises from two sources, first, planned changes in the delivery, and second, unexpected or transient changes. Unexpected or transient perturbations are more difficult to manage precisely because they are unexpected and effective control depends on early detection (degree of information).

Management of unexpected perturbations

When a perturbation occurs in a canal, the effects travel both up and downstream from the location at which the perturbation is created. However, the main impact is noticed downstream. For analysis, the perturbation domain is divided into two parts: generation, and propagation, also expressed as the active and reactive processes.

The active process can be analysed in three constituent parts: the causes of perturbations such as return flows, illicit operation of structures, and drift in the setting of regulators; the frequency of occurrence; and the magnitude of the perturbations experienced:

• Causes of perturbations are to a large extent determined by the network properties of the system (source of supply; hydraulic layout; interconnections with other networks such as drainage, unregulated return flows, etc). These properties are described more fully by Renault and Goddyaladda (1999). However, a second source of perturbations is the operation of the irrigation regulation system itself. Off-take and regulator operations generate transient conditions in the network which may translate upstream from the branch channels if the submerged flow occurs at the division from the main. In such cases the sensitivity of off-takes is the major determinant of the propagation of the transient (Renault and Hemakumara, 1997). Perturbations may also be generated at the off-take due to deliberate or accidental modification of the flow section, either because of changes in gate setting or of trapped debris. Perturbations are also generated by unscheduled operation of structures for unauthorized withdrawals, flow rejection or over-tapping. Table 1 (appended) summarizes the major components and properties causing perturbations.
• Position in the network, to some extent, controls the frequency of occurrence of transients and partially explains the well-known head vs. tail issue in irrigation. In this analysis, the occurrence and magnitude of perturbations occurring at any given point depend on the number and behaviour of upstream structures (cross-structures and off-takes). Generally, the more numerous and the greater the sensitivity of upstream cross-regulators, then the greater the magnitude and frequency of perturbations. Inversely, for systems with sensitive off-take structures, perturbations generated in the head reaches will be attenuated by upstream off-takes and the lower off-takes will see smaller transients.
• Water service considerations: perturbations are expected whenever a change in the distribution takes place. Therefore, the distribution policy (on demand, supply-based, free access) is a key determinant of the frequency of perturbations. The greater the flexibility of the delivery service provided, the higher the frequency of changes in discharges in the canal system. Proper consideration of the impact of service flexibility is essential to identify the specific operation modes and structure characteristics required for acceptable performance.

The sensitivity domain

Sensitivity describes the ratio of output to input of a particular process. In the context of irrigation, sensitivity analysis describes the behaviour of structures during the propagation of transient conditions (the reactive process). The behaviour of delivery structures, such as off-takes and outlets, in response to water level perturbations in the parent channel is the delivery sensitivity, described by the ratio of the relative off-take discharge (dq/q) to the change in upstream water level (?H_US), equation 3.

(3)

All irrigation structures (off-takes, regulators, canal reaches) have a distinct sensitivity. A comprehensive analysis of the sensitivity of irrigation off-takes leads to the identification of several indicators defining delivery and conveyance impact, including up and downstream translation of transients, and water level changes due to hydraulic conditions and adjustment of structures (Renault and Hemakumara, 1997). The relative sensitivity of regulator and off-take combinations has been studied in depth (Albinson, 1986). The rationale for sensitivity analysis is that more sensitive structure groups must be monitored and operated with greater care than less sensitive groups.

An important consideration for canal operations is the sensitivity of structures and their impact on the propagation or attenuation of transient flows that enter the canal system. In the absence of operational interventions the evolution of perturbations through the subsystem defines a decay curve integrating the conveyance sensitivity of the reaches and associated regulators and off-takes. Systems with sensitive structures tend to attenuate the transient flows by diverting surplus through off-takes, less sensitive structures propagate the perturbation downstream (Renault, 1999a).

Converting water service objectives to operational targets

A study of the domains discussed above enables the specification of requirements for operational interventions in a specific subsystem. By converting tolerance for discharge variations to a tolerance on water depth, the frequency and precision of control interventions can be specified. The link between operation and irrigation performance is established through generic dependency below:

The first relation indicates that the required precision of structure operations is the product of the tolerance on delivery and the sensitivity of the structure. The second relation defines the mode and the frequency with which the system should be operated in view of the type, frequency and magnitude of perturbations the system is subject to.

Control of water levels along the canal is the result of the combined effects of the hydraulic properties of the canal section, regulator characteristics and periodic operational manipulation of cross-regulator structures. The precision with which target water levels are controlled at cross-regulators (?H) is an indicator of operational performance directly influenced by management. Conversely, the extension of influence of cross-regulators, the backwater curve, is controlled by the physical characteristics of the reach and discharge rate.

In an analysis of the demand for operations, the determination of the precision of control can be assessed quantitatively. Given a target of water service, defined by tolerance factors (equation 1), and considering the delivery sensitivity (equation 3) the required precision of operations can be determined as:

(4)

in which

The required operational precision is proportional to the specified tolerance and inversely proportional to the delivery sensitivity. Therefore, an off-take of low sensitivity (0.5 m^-1) would require a precision equal to twice the tolerance in discharge expressed in relative terms. Thus if the tolerance on adequacy or efficiency is set as _10%, then the subsystem may be operated with a precision of _20cm. Equation 4 is valid for a single structure; however, similar relationships can be determined at system level linking system sensitivity indicators, the required precision of control, and operational performance (Renault , 1999b).

In general, evaluation of the requirements for operational inputs requires a qualitative approach with the goal of clearly identifying the significant properties strongly influencing potential operational strategies in each subsystem. These properties may include, for example, opportunities for recycling losses or the vulnerability within the system. Ultimately, these properties can be combined to classify the demand for operation as low, medium or high demand.

Case study of the Kirindi Oya irrigation settlement project in Sri Lanka

The proposed methodology is illustrated using the Kirindi Oya Irrigation and Settlement project, one of the largest agricultural development programmes in Sri Lanka. The system was completed in 1987.

Scheme summary

Kirindi Oya has two different command areas, which can be subdivided into four subsystems:

• The Old Ellegala command area that existed prior to the development of the new system. The Ellegala zone is in a flat alluvial area covering about 4 000 ha. Water is delivered to the area from an interconnected system of five old tanks, a diversion structure (anicut) across the Kirindi Oya and also from the new project. The area has priority in water allocations. It can be subdivided into two subsystems, the Left Bank Old, 2 850 ha, and the Right Bank Old, 1 150 ha.
• The new command area was completed in 1987. This command is in slightly undulating topography located upstream and on both sides of the alluvial plain of the Ellegala system. The new command is served by two subsystems:
  • the Right Bank New canal, delivering to five newly developed subcommands (Tracts 1, 2, 5, 6 and 7), totalling 3 300 ha and a reservoir (Bandagiriya) and the associated command area at the tail of the main canal. The Bandagiriya command receives scheduled deliveries from the Kirindi Oya area and surplus flows from the tail of the main canal.
  • the Left Bank New canal, whose command area is fully developed in tracts 1 and 2, and partly developed in tract 3 with a command area totalling 1 835 ha.
• Climate. The climate is classified as tropical humid with two seasons: a wet season between October and February-March (maha) and a dry season from April to September (yala). The average seasonal rainfall is 750 mm in maha and 240 mm in yala. Annual evapotranspiration is approximately 2000 mm.
• Water resources and water management. Water for agriculture is derived from direct rainfall on paddy fields and releases from the system reservoirs. The main reservoir (Lunugamwehera, 200 Mcm) was developed to extend irrigation to the new command areas and to secure irrigation for a double crop in the old Ellagala command. An important characteristic is the cascade of tanks in the system that enables capture of run-off, overflows and drainage from upstream areas. These captured flows are collected and stored in downstream reservoirs and used for irrigation at a later stage of the season. The Kirindi Oya project is the last water user before the river discharges in the ocean and therefore water savings in this scheme are true savings as defined by Seckler (1996).
• Crops. High-yielding varieties of rice are grown over the entire project area during the maha season. However, during the yala season, only the Ellagala area is routinely cultivated for rice; the new command is only cultivated when water availability is high. Some other field crops are grown on uplands in the new area during yala.

Improving system performance

Due to a perceived mismatch between available resources and potential uses of water, the entire extent of the new command has not been fully developed. Even though development is not complete, cropping intensity in the irrigated areas has not reached the expected levels but has remained at about 178 percent (increased from 140 percent) in the Ellagala area and only 108 percent in the new commands. Current operational strategies are largely based on overflow practices, which result in large water losses from the command areas where recycling is not feasible.

Schemes in coastal areas, such as Kirindi Oya, should seek to maximize effective water use, as water not used is lost to the sea. It can be shown that irrigation intensity at the project can be raised to 200 percent in both new and old areas provided a global efficiency of 43 percent is obtained (Renault, 1997). To achieve this level of efficiency, operational resources must be allocated effectively. Such allocations of resources depend on accurate assessments of the required levels of operational control. The analysis of operational requirements at the project addresses two aspects: the water service required at the command area, and the management of the operation of reservoirs. In addition, specific operational procedures should be evaluated to improve the management of rainfall, aiming to harvest and store as much rainfall in reservoirs and paddy fields as possible.

An analysis of the demand for operation at the Kirindi Oya project is presented here, based on the framework proposed above, examining in turn the vulnerability, perturbation and sensitivity domains of the system.

Vulnerability domain. Water is quite abundant and annual average resources (local rainfall plus reservoir inflows) are sufficient to sustain two crops a year provided the system is operated effectively (Renault, 1997). The maha rainfall is reasonably dependable, however the yala rainfall is less so. There are no major salinity or waterlogging problems in the area.

The existence of the cascade system with several tanks makes it possible for the scheme to be very efficient in harvesting rainfall. During the maha season, the cascade tanks should be operated at the lowest level possible to maximize storage capacity. This requires the direct supply from the main reservoir, and drainage return flows from new command areas should be restricted.

Single bank or contour canals are common in Sri Lanka. One characteristic is the potential to capture run-off from lateral watersheds. The Right Bank Old canal is a contour canal and this opportunity could be combined with the storage capacity of three intermediate reservoirs during rainy periods. Some parts of the Left Bank New canal are also of the single-bank type.

Water management. The potential to recycle drainage or spilled water from command areas is one criterion that divides the entire scheme into two categories. All tracts on the Left Bank New canal and tracts 1 and 2 of the Right Bank New canal drain to tanks supplying the old area. Conversely tracts 5, 6 and 7 of the Right Bank New canal drain to a lagoon and ultimately to the sea, resulting in large water losses. Drainage flows from that canal subsystem largely return straight to the main river channel and on to the ocean, with little opportunity to recycle the losses.

The Left Bank Old command area is characterized by a widespread interconnection between drainage and irrigation networks due to the flat topography. It is almost impossible to define precise command areas for small outlets (Mallet, 1996) or to specify the hydraulic characteristics of channels or structures. Surplus flows at one point become inputs elsewhere and therefore, in the terminology developed by Renault and Godaliyadda (1999), this unit is classified as a return-flow system. The Left Bank Old subsystem must therefore be managed as a single unit, considering several entry points to the network such as tank outlets and canal inlets, a number of drainage outlets to the river and the ocean. To increase the efficiency of water use, all drainage outlets should be monitored to avoid excessive losses. An effective feedback control system is essential to enable proper control of the inlets of the subsystem.

There is no conjunctive use in the area, pumping from the river, drainage and irrigation canals are restricted to small-scale gardening enterprises.

Multiple use of water is important in the project area. However, there are no major conflicts between irrigation and the other uses of water, such as domestic supply, bathing, homesteads, gardens and perennial vegetation, environmental uses (wetlands, wildlife habitat), tourism (lagoons and national parks) and fisheries. Irrigation is the major user of water, representing more than 90 percent of water use in the basin. Availability of water for irrigation ensures availability for other uses. There are no specific health-related issues.

Agriculture. Paddy cultivation is relatively less vulnerable to variations in water supply than other field crops, due to the buffer effect of the flooded paddy field. As the area is mainly cultivated for paddy rice in both seasons, the area can be considered as homogeneous and of low vulnerability. This is an important characteristic for tracts where recycling is not feasible, as it may allow implementation of strategies to reduce overflows. Special consideration may be required for the new areas where some farmers are cultivating other field crops that will be more vulnerable to water shortages. Soils in the Ellagala area are heavier than in the new command areas. Although this has some implications for water allocation and drainage flows (percolation rates are estimated at 3mm/d and 6mm/d respectively, IIMI, 1994), it has little impact on operational strategy, or on system efficiency, as the dominant criterion is the ability to recycle water.

Water rights and equity. In theory, all farmers at the project have equal water rights. In practice, farmers of the old areas have established a powerful position and are able to impose allocations of water in their favour. Records of cropping intensity show that the Ellagala area has averaged 178 percent whilst the new areas have achieved an average of only 108 percent (Renault, 1997). The Ellagala area also obtains irrigation supplies in advance of the new areas, contrary to an effective water savings policy. Under these conditions, any attempt to improve water management must secure 200 percent irrigation intensity to farmers in the Ellagala subsystem before attempting to implement any changes of supply to the new area.

Environment. The area surrounding the project has several facets of environmental and wildlife importance: the entire area is a recognized wetland sanctuary of importance to migrating birds; the Bundala National Park is to the south-west of the scheme. The lagoons in the park are partly supplied by water draining from the right bank new canal area. Fortunately there are no conflicts between improved irrigation management and existing environmental concerns: improved water management in irrigated areas will extend the period of water in tanks and will reduce fresh water inflows to the lagoons, which are felt as a hazard at present.

Water service and performance indicators. In subsystems with no opportunities for the recycling of excess flows, the tolerance on deliveries (equation 1) must be minimized and a feedback link between drainage flows and inlet settings should be established. In areas where recycling is possible, the delivery tolerance can be less strict. However, a feedback loop control is required to maximize potential storage in downstream tanks.

Perturbation domain. Analysis of the perturbation domain focused on the occurrence and magnitude of external and internally generated perturbations. The upstream boundary conditions of each subsystem are homogeneous; all systems are supplied from the main reservoir or tanks and are regulated by manually operated gates. The anicut supplying the Ellagala area is now supplied, indirectly, from the main reservoir.

Lateral flows. The Right Bank New canal is a double-bank canal and therefore not greatly influenced by rainfall. The Left Bank Old canal is also a double-bank channel. Parts of the Left Bank New and the entire Right Bank Old canals are single-bank and therefore susceptible to be affected by large perturbations during periods of rainfall.

Position in the system. Field observations have confirmed that the Left Bank New and Right Bank New canals are subject to an increasing range of water level fluctuations between head and tail locations. The range variation of water level at selected off-takes during maha 1993 in the Left Bank New canal increased from 65 mm at head to 90 mm in the middle reaches to 110 mm at the tail. Observations on the Right Bank New canal show a similar trend; records for six seasons indicate average increases from 75 mm at head to 120mm at the tail.

Users. Discipline varies between the systems. In the old system, there are few problems of discipline, probably as a result of the relatively reliable water supply. In the new system, farmers must contend with shortages of water to the extent that some local people have not been able to establish themselves as farmers and have had to seek other employment. Even those who have been able to establish themselves as farmers have less influence in decision-making regarding allocations of water. As a result, unauthorized operations of gates and harmful interventions at cross-structures do occur. System managers have coped with these problems by issuing more water than theoretically required to the main canal. The lack of discipline may be a serious constraint to increased precision in operations aimed at improved efficiency. The strategy should be to achieve highly reliable supplies in all areas.

The Left Bank New canal illustrates the impact of unreliability of supplies. Although built to the same design and at the same time as the Right Bank New canal, its structures are in poor condition compared to the latter's. Many gates are broken or missing at cross-regulators after only twelve years of operation. One cause may be the relatively high delivery sensitivity along this canal, causing farmers to make unilateral interventions when supplies are inadequate.

Operational procedures. To improve economy of water use in command areas with no opportunity to recycle drainage flows, managers will have to adopt more effective procedures than the existing overflow method of management. Two alternative procedures might be considered, first a strategy of progressive reduction of deliveries, second, the introduction of rotational delivery. Progressive adjustment to reduce downstream drainage discharges would impose permanent and progressive modifications of inflows (deliveries). This option would require precise operation and methods to fine-tune deliveries so as to minimize inflows while avoiding the drying-up of downstream field units. Ultimately this method would result in a minimum steady state discharge. Rotational operations, either an on/off schedule or with alternating high and low discharges, will result in frequent fluctuations in canal discharges, requiring greater supervision of the whole system.

The sensitivity domain

• Offtake sensitivity. The sensitivity of off-takes clearly distinguishes different subsystems at the Kirindi Oya project. Along the Right Bank New canal, the off-takes are relatively insensitive (average S = 0.46) and homogeneous (CV = 0.1). However, along the Left Bank New and Right Bank Old canals, the off-takes are sensitive (average = 2 and 2.2 respectively) and fairly inhomogeneous (CV = 0.4 and 0.96 respectively). This means that the same level of precision in water depth will generate discharge deviations four times greater in the latter two canals than in the Right Bank New canal.
• Regulators and reaches. Three situations can be distinguished with regard to the regulation of water levels in the main channels:
  • not regulated: for example, the Right Bank Old canal is effectively not regulated, as the density of regulators is very low and most existing structures are no longer functional;
  • poorly regulated: the Left Bank New canal has adequate provision of regulators but the existing condition is poor, with gates missing or inoperable; and
  • well regulated: the Right Bank New canal is well equipped for regulation and regulators are in good shape.

Spatial variation of operational demands. The operational requirements to achieve specified levels of water delivery service and acceptable levels of water use economy at the project are analysed for daily operation of water releases and regulation of the canal systems. Requirements for improved scheme operations related to the scheduling and tank management-rainfall harvesting tasks are not addressed here.

Considering four classes of operational requirements, varying from low demand to very high demand (D1, D2, D3 and D4), five subsystems were identified. An evaluation of the characteristics of the demand for operations in each is summarized in Table 2. Although the ranking used here may be subject to discussion, the identification of significant operational features of each subsystem allows for a spatially differentiated allocation of management resources, Figure 2.

The next step would be to determine what allocation of resources would be required to match the demand. It seems clear that the number of operators required will vary from area to area to match the operational demand in order to improve the overall efficiency of the system.

Conclusions

The case study of Kirindi Oya illustrates the existence of heterogeneity of requirements for operational resources, even within a medium-sized, mono-cropped irrigation system. The analysis is based on an overlay process considering three operational domains: vulnerability, sensitivity and perturbation.

System managers can address heterogeneity of operational demands through two different strategies. They may accept the reality of spatially variable operational requirements and allocate resources accordingly. Alternatively, the effects of spatial variability can be minimized by interventions in the physical system. In either case it is expected that the improved evaluation of the spatial variability of demands for operation will be useful in the design of:

• more cost-effective strategies and procedures for operation, leading to a better use of available means; and
• priorities for the rehabilitation or modernization of physical infrastructure.

References

Albinson, B. 1986. Designing and operating guidelines for structured irrigation networks. Fourth draft. Irrigation II Division, South Asia Projects Department

Bakker et al. 1998. Multiple use of water in irrigated areas: a case study from Sri Lanka. IWMI-SWIM Research Publication

Bos, M. G., Murray-Rust, D. H., Merrey, D. J., Johnson, H. G. & W.B. Snellen. 1994. Methodologies for assessing performance of irrigation and drainage management. Irrigation and drainage systems, 7(4):231-261

Handbook 1990. Louis Berger Intnl. Inc. and Water & Power Consultancy Services (India) Ltd. 1990. Handbook on irrigation system operation practices. Water Resources Management and Training Project. Irrigation Management and Training Program. Technical Report No33

Hunter J.M., Rey L., Chu K.Y., Adekolu-John E.O. & K.E. Mot. 1993. Parasitic diseases in water resources development. The need for inter-sectoral negotiation. WHO, Geneva

Mallet Thibault. 1996. Geographic information system for water management in a cascade system, Kirindi Oya, Sri Lanka. Internal report

Molden, D.J. & T.K. Gates. 1990. Performance measures for evaluation of irrigation-water-delivery systems. 804-823, in Journal of irrigation and drainage engineering. Vol116, No6. Nov/Dec 1990

Molden D.J. 1997. Accounting for water use and productivity. SWIM Paper 1. International Irrigation Management Institute, Colombo

Ng Poh-Kok. 1987, Irrigation design: a conceptual framework, pp 61-78, in Proceedings of the Asian regional symposium on irrigation design for management. 16-18 Feb 1987, Kandy, Sri Lanka

Paudyal G.N. & R. Loof. 1988. Improvement of irrigation system operation. Research Report 211: Agricultural, Land and Water Development Programme, Division of Water Resources and Engineering, Asian Institute of Technology, Bangkok

Plusquellec Hervé. 1988. Improving the operation of canal irrigation systems. An audio-visual presentation. World Bank

Plusquellec H.L., C.M. Burt & H.W. Wolter. 1994. Modern water control in irrigation. Concepts, issues, and applications. World Bank Technical Paper 246. Irrigation and Drainage Series. World Bank

Renault Daniel. 1997. Technical background for KOISP 200. Internal IWMI note

Renault, Daniel & H.M. Hemakumara. 1997. Irrigation off-takes sensitivity analysis. Fourth international ITIS network meeting on modern techniques for manual operation of irrigation canals, Marrakesh, 25-27 Apr 1997. Proceedings. D. Renault (Ed) IMMI. Pp.74-84

Renault D. & Godaliyadda C.G.A. 1999. Typology for irrigation systems operation. IWMI Research Report 29

Renault, Daniel. 1999a. Aggregated hydraulic sensitivity indicators for irrigation system behaviour. Agricultural water management (forthcoming)

Renault, Daniel. 1999 b. Sensitivity, control and performance of irrigation systems. Journal of irrigation and drainage engineering. Vol125 No3

RIC. 1997. Regulation of canals: state of the art of research and applications. Proceedings of the international workshop held in Apr 1997, Marrakesh, Morroco

Shanan, Leslie. 1992. Planning and management of irrigation systems in developing countries. Agricultural Water Management (Oct 1992) 22(1+2). The Netherlands: Elsevier

Schakel J.K. & Bastiaansen W. 1997. Regional water and salt balances obtained from GIS and hydrological models. in ITIS Newsletter Nov 1997 Vol4, No1. IIMI.

Skogerboe, Gaylord V. & Gary P. Merkley. 1996. Irrigation maintenance and operations learning process. Water Resources Publications, LLC:USA

Steiner, Roy A. & Michael F. Walter. 1993. The effect of allocation schedules on the performance of irrigation systems with different levels of spatial diversity and temporal variability. In Agricultural water management, 23: 213-224. Elsevier Science Publishers BV

Uittenbogaard, G.O. & N.R. Kuiper. Improved operation strategies in India. Paper presented at the Asian regional symposium on maintenance and operation of irrigation/drainage schemes for improved performance held in Beijing, People's Republic of China, 24-27 May 1993. HR Wallingford

Zimbelman, Darell D. 1987. Planning, operation, rehabilitation and automation of irrigation water delivery systems. Proceedings of the symposium sponsored by the Irrigation and Drainage Division of ASCE, Portland, Oregon, 28-30 Jul 1987. Zimbelman D.D. (Ed). ASCE. 381pp

Table 1. Components and properties significant for unexpected perturbation generation
(adapted from Renault and Goddyaladda, 1998)

Table 2. Evaluation of the demand for operation per subsystem in KOISP

1. The agricultural and environmental aspects do not partition the scheme (mono crop minor concerns)
2. The tolerance for time is irrelevant here as deliveries are continuous
FBC: feedback control -