PART II: CHANGING APPROACHES TO THE DESIGN OF IRRIGATION PROJECTS

The time for grand vision and flowery rhetoric has passed. The challenges ahead require sharper focus, real commitment, and concrete action.

VII. IRRIGATION DESIGN CONCEPTS IN SELECTED COUNTRIES

This chapter presents the salient features of the design, management and performance of irrigation systems in key countries. Obviously it is not an exhaustive presentation. The objective is to highlight the large differences regarding designs and management of systems caused by the climatic differences and the economic, political and social relations in the different countries. The second part of the chapter discusses the problems with the transfer of technology from one region to another one with different social environment.

A. Traditional irrigation systems

Traditional irrigation is rooted far back in history. Although traditional irrigation schemes now represent a small percentage of the 265 million hectares under irrigation worldwide, they still play an important role in most developing countries such as Nepal, Indonesia, Morocco, Peru or the Philippines. Century-old schemes in Spain have been the models for the development of irrigation in the New World and have attracted the attention of engineers from colonial powers in the 19^th century at the onset of irrigation development in the Indus basin and elsewhere. Traditional schemes are relatively small in scale, from a few tens to a few hundreds of hectares. However, some traditional schemes reach a few thousand hectares, such as the Maujis Chautra project (10 000 ha) in Nepal and the Khanabad project covering more than 30 000 ha in Afghanistan. These schemes have been built and maintained by local communities with little or no government support. Local customs regarding water allocation and distribution in these systems have evolved over time and are well adapted to local and ecological conditions. Although the term "local customs" may be interpreted as the opposite of scientific, the rules of water distribution can in fact be very sophisticated. Their complexity increases with the degree of water scarcity. By contrast, the infrastructure for water allocation is rather simple and all the users can understand their operation. The most frequently found water control structure in traditional irrigation is the flow divider to allocate water proportionally to fixed water allocation ratios related either to water rights or to the irrigated areas. The most famous traditional irrigation scheme is the 16 000 ha Valencia project in Spain, known for its oldest water tribunal. Complex operational procedures of this project prescribe different rules for three levels of water availability.

The cohesion and social bounds among members of local communities are the main reasons for the success and sustainability of traditional schemes which have been under existence for hundred of years. These social bounds do not usually exist in rural areas where settlers of various ethnic groups move shortly after the development of irrigation systems. Extending the finding of social research studies on traditional irrigation systems to the large-scale systems built in the last decades should be done with great caution. Self-enforcement of the rules is much weaker in these projects. For example, stealing water from the distribution canals in the Indus basin is common practice since water is perceived by the farmers as belonging to the government. By contrast, rules are strictly enforced once the water is diverted to the lower level managed by the users.

B. Country experience

1. India

The large irrigation systems built in northern India in the 1800s were designed for drought protection to avoid famine. The objective was to distribute irrigation water to the maximum number of farmers. The design capacity of canals was as low as 0.25 litre/second/hectare or three times less than the irrigation requirements of intensive irrigation.

Agro-climatic and socioeconomic conditions in India vary widely depending on geographic location, and irrigation systems have evolved reflecting this diversity.

There are several models of distribution of water below field outlets in surface irrigation systems: i) the warabundi system of Northwest India (Punjab, Haryana, Rajasthan and Uttar Pradesh), ii) the shejpali and block systems (Maharashtra and Gujarat) and iii) the localized system used in the southern states.

Under the warabundi system the available water is allocated to all farmers within a block, irrespective of their crops and location of their holdings, under a rigid weekly scheduling. The share of water is proportional to the holding area in the outlet command and allocated in terms of time interval as a fraction of the total hours of the week. Advocates of warabundi claim that this system is highly equitable. This would be right if the seepage losses in the field channels of the block were negligible. Seepage losses of unlined channels may represent 40 percent of the diverted flows.

Under the shejpali system, the government enters into some sort of agreement with the farmers for supplying water to them. The farmers file applications and the government issues permits for the supply of water and the two constitute the agreement. The water is distributed according to a predetermined date in each rotation. A preliminary programme is drawn every year depending on the availability of water. Farmers submit applications for supply of water indicating the crops they wish to grow and the areas under them. Water is then apportioned on the basis of the crops and the overall demand. A schedule fixing the turns of supplying farmers in the sanctioned areas is prepared for each rotation. The irrigation interval depends on the rate of water consumption by the crops. The schedule is notified in advance and every farmer of the command area has prior information about his turn of supply. The system is called "rigid shejpali" if the duration of supplying water in the various fields along with the date is also recorded on the permits issued to the farmers for sanctioned areas. Application of the shejpali system is based on a high intensity of adjustable gates and their frequent resetting. The objective of matching supply with demand is rarely met because of the difficulty of operating manually adjustable gates.

Under the block system, a long-term arrangement for supply of water is done particularly for perennial crops, but irrigation from season to season proceeds through shejpali. The blocks are sanctioned for six to twelve years. (Mandavia)

The advantages and disadvantages of the different designs used in India have been the subject of several research studies, which led to intense debate. There is, however, wide recognition that, overall, the performance of surface irrigation in India needs considerable improvement. Bhavanishankar states: "The reliability and predict-ability of water supplies is not assured in most of the schemes. Conflicts are common in most of the systems, leading to vandalism and disruption of the physical facilities and degradation of the system. The present method of delivering the water as per the demand of the powerful group among the farmers is often arbitrary and wasteful with considerable inequity in distribution."

The unreliability and/or rigidity of water distribution from the surface irrigation systems in India as well as the under-sizing of the canals to deliver water for intensive irrigation have contributed to the uncontrolled development of groundwater during the last decades.

2. Pakistan

Of the 16.2 million hectares irrigated in Pakistan in the early 1990s, about 93 percent are under the command of the Indus River Irrigation System. This system, the largest in the world, encompasses three storage reservoirs, 19 barrages or head works, 12 link canals, 43 command areas and 107 000 watercourses, each one serving an area of about 250 to 700 acres. Water to the watercourse is diverted from the distributary and minor canals through ungated structures known as mogha. The design is similar to the one used in northwest India. With the exception of the link and main canals, the system was designed to operate at or near full supply. The mogha is designed to allow for a constant discharge. Within the watercourse command, farmers receive water proportionally to their landholding. The entire discharge to the watercourse is given to one farmer for a specified period on a seven-day rotation.

The canal system was designed as a run-of-the-river project to maximize the cropped area, with minimum water consumption, and simple operation and administration. Canals were intended to provide equitable distribution, with no interference by the canal establishment^[15].

Extensive performance studies by IWMI and others have demonstrated that water distribution, contrary to the stated objectives, is not equitable. The greatest inequity is between watercourses. Some head mogha draw two or three times their allocated shares while tail mogha may only receive half or less. The main causes of inequity are the opposition of the farmers to the abolition of the privileged water allocations granted during the colonial period, the tampering of the mogha structures and installation of illegal outlets, as well as changes in water profile due to siltation^[16] and lack of maintenance. The overall efficiency of canal systems serving individual command areas is below 40 percent. Losses contribute to groundwater recharge.

Groundwater use has been a main factor in the intensification of irrigated agriculture in Pakistan during the last two decades. Groundwater not only supplies additional water but provides the flexibility to match water supplies with the crop demands. The originally expected cropping intensity has increased on average from 120 to over 160 percent in some areas of Punjab. The overexploitation of groundwater is discussed in Chapter 9.

Irrigation systems were initially developed without provision for drainage. Irrigation without drainage in an environment like the Indus basin inevitably leads to the rising of water tables and salinity. About 30 percent of the Indus command area is currently waterlogged and about 14 percent severely or moderately salt-affected. (World Bank 1994)

The basic design concept in northwest India and in the Indus basin was to provide equitable distribution of water with minimum interference and low-cost operation by limited staff and means of communication. Distributary and minor canals were operated in on/off rotation from continuously running main systems. Ungated outlets discharge water from these canals into the watercourses from where farmers get their water shares under the warabundi system. This system was expected to be effective and equitable but it was not related to crop water requirements. It is up to the farmers to arrange their cropping pattern and watering to suit the delivery of water at a fixed flow and predetermined time. For the reasons given above, there is great inequity in actual withdrawals between head and tail watercourses.

Groundwater development has obviously contributed to the reliability and flexibility of water allocation in the Indus basin. A valid question is whether the development of groundwater has improved the equity of distribution. Few research studies have been made on the performance of conjunctive use of water in the Indus basin. It is assumed that the equity between the mid and upper sections of command areas has improved. However inequity may have increased for the lower sections because of the poor quality of groundwater and the lower density of wells.

Use of untreated wastewater is a usual practice around most cities in Pakistan - and many other countries. Wastewater is valued by the farmers not only because of its nutrient content, but also for its reliability of supply, which makes cultivation of vegetables, the most common crop in Punjab, possible.

Adoption of modern design approaches was attempted in the North-West Frontier Province, Pakistan, which benefits from relatively abundant water resources. The higher capacity of the canals, without enough consideration of variable flow conditions and risk of siltation, has resulted in the adoption of a control "by refusal". This ad-hoc strategy consists in operating the main and secondary canals at or near full supply and letting the farmers and operators release the excess water directly from the canals to the drainage systems.

3. Egypt: The Nile Valley system

Irrigation was practised throughout the Nile Valley from the earliest times. Until the mid 19^th century, this was realized by natural inundation from flood waters. The system has been converted from flood to perennial irrigation following the construction of the Aswan dam and delta barrages. Of the 3.3 million hectares irrigated in Egypt, nearly 95 percent are supplied from the Nile irrigation system. Barrages on the Nile divert water to the main canals. Main canals supply branch and sub-branch canals, which provide water to private farm watercourses, called meskas. Flows released in the main canals are based on crop water requirements and expected distribution and farm losses. The branch canals are operated on rotation based on the requirements of the dominant crop. During a typical 12-day rotation, branch canals receive water during 4 days and are off during 8 days. A unique feature of the Nile system is that most branch canals and meskas are below ground level. Farmers used to lift water from the meskas through animal-driven pumps. Irrigation was mostly practised during daytime. The relatively low flow of individual pumps ensured a high level of equity of water allocation between head- and tail-enders and avoided over-watering of the cultivated lands, by contrast with gravity systems. The situation changed dramatically with rapid replacement of traditional pumps by individually owned diesel pumps or electric pumps since the 1970s, creating large inequities of water extraction along meskas and social inequities between tail- and head-enders. Tail farmers responded by looking for other sources of water, mainly by pumping from the drain system. The Nile system is similar in its architecture and operation by rotation to the Indus system. However it differs in four key aspects:

• the releases from the Nile to the canal system are based on irrigation requirements;

• the watercourses are below ground level, forcing the farmers to pump;

• the farmers are free to irrigate at any time when their branch canal is "on"; and

• the design capacity of the canals is about three times higher.

The Ministry of Irrigation is now implementing a modernization irrigation project to reduce the inequity and the re-use of poor-quality water in the Nile delta following the adoption of diesel pumps. The objective of the modernization is to create night storage in the secondary canals and to replace individual lift pumps by a common lift pump serving a raised meska. This plan is based on the adoption of a rotational distribution of water by the farmers organized for this purpose in water associations at the meska-pump level. This is a unique case of modernization in which farmers have to accept the conversion from a free-demand (when their canals are "on") to a rotational system requiring coordination and discipline.

4. Sudan: The Gezira project

The Gezira project lies between the Blue and White Nile rivers south of Khartoum. The Sennar diversion dam built in 1925 and the multipurpose Roseires dam completed in 1966 regulate the flow of the Blue Nile. The Gezira scheme was designed in the 1920s after prolonged experiments had been carried out at pilot scale. It was designed with the main objective of producing cotton, a single cash crop. Other crops are grown to provide food for the farmers and to help in the maintenance of soil fertility. Cotton, wheat, groundnut and sorghum are now cultivated in a four-crop rotation including fallow. The farmers do not own the land. The scheme is divided between 102 000 tenants with an average of about 8 hectares.

The irrigation system was laid out to suit the size of tenancy and crop rotation. The flat and featureless topography was favourable to the adoption of a regular gridiron layout. The basic unit is a group of four adjacent fields of 90 feddans. One crop is grown on each strip following the four-crop rotation system. Each block is divided into 18 tenant fields of 2.2 hectares each.

The irrigation system comprises twin main canals running from head works at the Sennar dam with a combined capacity of 354 m³/s, a network of 2 300 kilometres of branch and main canals, and about 1 500 minor canals with a total length of over 8 000 kilometres. All canals are divided into reaches by cross regulators which are the control points for the off-taking canals.

The minor, branch and main canals are designed as regime conveyance channels. The minor canals are also designed for storing water flowing continuously from the main canals at night.

Operation of the scheme is centrally controlled: the management is divided between the Ministry of Irrigation (MOI), which is responsible for the irrigation network, and the Sudan Gezira Board (SGB), which is responsible for agricultural operation and for determining the irrigation water requirements. The water orders (or indents) are passed to the MOI engineers, summed out throughout the system up to the head works at the Sennar dam.

Water flows from the main to the minor canals are controlled by movable weirs, which provide accurate and easy water measurements, but have the serious disadvantage to be highly sensitive to upstream variations of water level.

The Gezira scheme is not a sophisticated one by present-day standards. It was designed before the development of modern technologies of canal water control. The design, however, took the best advantage of some favourable and unique features of Gezira: the flat topography and the adopted tenancy system, i.e. the absence of constraints imposed by small, fragmented, field plots found in many developing countries. The adoption of the night storage system resolved the issue of night irrigation found in many schemes. It provides a remarkable solution to the complex problem of adjusting water releases at the head works and at critical points of the system to the demand without excessive losses. A negative characteristic of the minor canals, which was probably overlooked, is their ability to trap the silt released into the system.

For about forty years, the Gezira scheme was operated satisfactorily on the basis of the original design and operational concept. The management of the Gezira scheme ran into problems in the early 1970s shortly after the scheme reached its present extension. The steady deterioration of trade in Sudan led to shortages of financial resources. Funds became insufficient to finance the high recurrent operation and maintenance costs and to replace machinery and equipment. For lack of financial resources, MOI was not able to cope with the removal of silt and clearance of weed. The situation was worsened by the breakdown of the telephone system, which was a vital tool for communication between SGB and MOI for the water indent process. All these factors resulted in inadequate use of the system. The degree of siltation in some minor canals is such that precious little water reaches the tail blocks and some areas are out of production. The tenants lost confidence in the untimely operation of the system and, to some extent, took over the management of the minor canals. The original night storage system gave way to continuous irrigation water delivery to the fields. By adopting unattended continuous irrigation, the tenants have reduced irrigation labour costs. They also appreciate the flexibility of the new de-facto on-demand system since they took control of the opening of the field outlets. The departure from the originally planned method of irrigation has given rise to a new management and water application. The intention of MOI is to re-establish the night storage system, which was based on a strict discipline of water scheduling. An informal management offset the decline in the performance of the official system. However, the unique design of the system played a major role in maintaining irrigation service in the 1980s and in the adoption of a new informal management system. The minor canals playing the role of terminal reservoirs are the key features of that transformation. It is now demonstrated that water delivery in the Gezira scheme can be based on either rigid or highly flexible scheduling, as long as the indenting ensures adequate refilling of the minor canals. In other words, the design was able to adjust to a major departure from the original management thanks to the flexibility in operation provided by the design of the minor canals. The main drawbacks of this unique feature are its silt-trapping efficiency and high health hazards during manual weed clearance.

The suggestion made by a foreign consultant to narrow the minor canals to reduce weed and silt clearance would not totally solve the problem of siltation and weed infestation, although it would eliminate the buffer storage in the minor canals, a critical feature of the design of the Gezira scheme. It would also considerably increase the complexity of operation.

5. China

China has very detailed standards and regulations for the design of large water structures. Typical structures are found in most provinces. Irrigation projects belong to the category of manually operated gated systems. Some gates built in concrete in the 1960s are very difficult to handle. Many systems in China are operated during periods of about 20 days each totalling about 80-100 days per year, generally at or close to full supply. Water is released after a long consultation between the local authorities and farmer groups. The design is very basic but management relies on the active participation of users or local communities at all levels. This strongly advocated approach for improving irrigation performance has rarely succeeded in other countries.

A characteristic feature of the configuration of irrigation systems in mid and southern China is the number of large, medium and small reservoirs, which form an integral part of the systems. Large reservoirs created by the construction of large dams are connected to medium reservoirs and to hundreds of thousands of village reservoirs and ponds. For example, the Pi-Shi-hang Irrigation districts, which serve 680 000 hectares in Henan province, consist of a network of canals connected to five large reservoirs, 24 medium-sized reservoirs with a total active capacity of 420 million m³, 113 small reservoirs with an active capacity of 205 million m³ and 210 000 storage ponds. The reliability and flexibility of water delivery of these systems are very high.

China has developed a water-saving technique for irrigating paddy fields consisting in alternating wetting with shallow water and drying periods. This method is now applied on about 3.5 million hectares out of the 21.3 million hectares of irrigated paddy fields in China. It not only saves a considerable volume of water but also leads to higher yields of rice. Application of this method requires a high level of management at both on-farm and off-farm levels.

A singular feature of irrigation in China is found in its management as a result of a series of reforms that took place in the water management sector during the last decades. The authority for owning and managing irrigation projects is determined according to investment. Large irrigation districts are usually managed by organizations at various levels such as prefecture, county, township and village. In the 1980s, the government launched the production contract responsibility systems in the rural areas to support individual initiatives. The government also changed irrigation management from centralized control to contract management in order to facilitate decentralization. The contracting organization may be a company, a farmers' group, a joint household or an individual. The contractors have the right to operate and manage the irrigation facilities and should take full responsibility for profits and losses. As a result of the contract management, the management organization is optimized; especially, the workers' income is closely related to their performances of the contracted targets. China is now experimenting with several models of contract management, particularly in Shandong province. China is clearly a country where management improvement has been a substitute for a very basic water control infrastructure.

6. North African countries

Development of modern irrigation in the three North African countries (Tunisia, Algeria and Morocco) started in the late 1930s, more than fifty years later than in South Asia and Egypt, accelerating only after World War II to reach peak development after Independence. This late development was possibly the reason for the fundamentally different approaches to the planning of irrigation projects in that region. According to verbal sources, the low level of education of the rural population stimulated the colonial government agencies to design irrigation projects which met the dual objective of being operated with minimum intervention of operators and simple operational procedures and being flexible enough to meet the irrigation requirements.

An intensive research programme with the support of the private industry led to the development of hydro-mechanical equipment to automatically control upstream or downstream water levels and water flows as well as fixed static structures such as flow limiters and long-crested weirs, known also as duckbill weirs. These weirs provide nearly stable water levels in canals. The concept of canals operated by downstream and upstream control and the combination of these techniques were refined over the years. The use of these innovative designs became standard practice for irrigation projects in these three countries and in most Mediterranean countries. It was later extended to other regions but generally on a project, case-by-case basis.

The canals are designed to be able to answer irrigation requirements during peak demand for the cropping pattern adopted at design stage. However, the specific design capacity is multiplied by a factor of up to 50 percent from the main system to the tertiary canals to provide some flexibility in order to accommodate variations in demand and deviations from the project cropping pattern. For example, the design capacities of the Doukkala project in Morocco increases from 0.65 l/s/ha for the main canal to 1 l/s/ha at the tertiary level.

Most of the main canals are concrete-lined and the secondary and tertiary canals consist of prefabricated canaletti (flumes) using the most advanced pre-stressing concrete techniques. The concrete lining of main canals is about 30 percent thicker than the lining of canals of similar sizes in other countries.

As a result of the high standards of design and construction, and the small variations in water levels, the life of the irrigation systems is remarkably better in these three North Africa countries than in some other regions. Some projects built in the 1950s are still under operation. The first rehabilitation projects in Morocco were related to undersized projects built before World War II, which became incompatible with the intensification of irrigated agriculture.

Another feature of irrigation in Morocco is the systematic consolidation of irrigable lands before the installation of the infrastructure. Before project, irrigable lands are highly fragmented and boundaries of individual plots are randomly organized. The model adopted by the Moroccan administration in the 1960s, after testing different models, is based on the same principle as the model used in Gezira in Sudan. The objective was to facilitate the adoption of modern irrigation scheduling and mechanized farming practices in a context of smallholdings. Geometric blocks of 30 hectares were divided into four to six crop strips of equal width and the farm holdings were arranged with boundaries parallel to the other direction. Permanent quaternary canals were associated with a farm road and farm ditch. The number of farm plots was reduced about five times in some projects.

To be successful this model requires the strict discipline of the farmers in respecting the government-imposed cropping patterns and joint organization of agricultural works within each crop strip. The farmers progressively deviated from the imposed cropping. How-ever, the most serious deviation from the original plan was in on-farm water management. The quaternary canals owned collectively by the farmers of a block were not maintained, and land levelling badly degraded, eliminating the possibility to adopt furrow irrigation or border irrigation. The farmers came back to the century-old inefficient irrigation method of small basins.

Although there are some variations between regions, water distribution in Morocco is largely decided by the irrigation agencies (ORMVA). The basic principle of water distribution is that each farmer receives a predetermined volume of water per irrigation turn. ORMVA decide on the implementation of the irrigation turn, its duration, and the volumes per hectare for the various crops, depending on the availability of water in the storage reservoirs. Farmers can decide whether to take water during a turn or to reduce the duration. They sign a note of acceptance, which specifies the date, time, duration discharge and total volume delivered which will be used for assessing the water charges. Although the system has the capacity to be operated on prearranged demand and to provide the flexibility required to meet the farmers' needs, it is essentially a centralized system. This mode of operation was justified when rain-fed farmers were converted into irrigators a few decades ago. It does not respond to the needs of modern agriculture in Morocco.

Irrigation in the North African countries is not performing at the expected level, although the level of technology of the delivery system is of the highest standards. The main reason may be found in the poor on-farm use of water, which is related to the outdated delivery procedure and land consolidation model.

7. Iran

Iran is an interesting example of a country without national design standards. Two basically different approaches to irrigation planning are found in that country. In the Khuzistan province in the south, old design standards of the Bureau of Reclamation have been used for the design of the Dez multipurpose project and adopted for all irrigation projects in that region. In the northern provinces, the most frequent design standards are those introduced for the design of the Isfahan and Guilan projects by a French consulting firm with long experience in North Africa. The two design standards used in the northern and southern parts of Iran belong to the category of fully gated systems. The design objective in both cases is to distribute water according to requests of individuals or group of farmers with flexible scheduling. However, they differ by the control function. All gates in the south are manually operated whereas the northern systems benefit from a high degree of hydraulic automation, which simplifies their operation. Box 7 provides a detailed discussion of the Guilan project, which is a unique success story of transfer of technology from an arid region to the paddy systems with humid climate along the Caspian Sea.

The Guilan project contradicts the paradigm that a design consisting in reticulated fully gated canals is not suitable for irrigation projects in the humid tropics. It also contradicts the well-accepted paradigm that the irrigation technology of arid and semi-arid countries is not suitable for humid tropics.

8. Malaysia

Malaysia is another example of a country without national design standards. Foreign consultants have introduced three different design standards that reflect their own experience, as illustrated by the examples of the Muda, Kemubu and Kriang-Kerian schemes.

The Muda scheme: This 98 000-ha scheme, designed by a British firm, accounts for 40 percent of the national rice production and is critical to the rice policy of Malaysia. The main infrastructure is comprised of two storage reservoirs connected by a tunnel, a diversion dam further downstream, and two main canals. At the time of construction, a remote monitoring system was installed to provide the operating engineers with real time information on reservoirs and canal water levels and on rainfall in the catchment area between the storage and diversion dams to predict the unregulated flow. Cross regulators on the main canals are equipped with overshot motorized gates. Furthermore, pumping facilities and tidal gates were installed to recapture the drained water in the lower part of the scheme. The combination of these devices with remote monitoring has contributed to the efficient operation of the main system. Service to rice growers was irregular, however, because of the difficulty of managing the delivery system equipped with manual gates. To achieve the better control over volumes of water and timing required for new techniques of direct seeding, the farmers install their own pumps to lift water from public canals and drains.

The Kemubu scheme: This low-lift pumping scheme, designed by French consultants, adopted downstream control for the main canal and the pumping station and upstream control for the secondary system equipped with long-crested weirs and modular distributors^[17]. As in the Muda scheme, the operational problem is the difficulty of controlling flows in the minor system and meeting the requirements of increasingly diversified cropping. Different control structures were later adopted for an extension of the scheme, consisting of adjustable flow-dividing structures.

The Kriang-Kerian scheme: That scheme was developed using the old standards of the Bureau of Reclamation, mainly the use of constant head orifice. These are discussed in the next section. This project is now under modernization through low-cost automation.

9. Indonesia

Irrigation in Indonesia, particularly on Bali and Java islands, has been practised for the cultivation of rice since ancient times. The old and non-technical systems represent a large part of the six million hectares currently irrigated. Design of the systems built during the colonial period and soon after Independence was rustic. Individual control structures were improved over time, but not enough consideration was given to the functioning of the entire system.

• To improve measurement and control of flows diverted from one parent canal to the next-level canal, an adjustable weir gate, known as Rominj gate, was developed in the 1950s. This gate is a precise measuring device but has the disadvantage to be sensitive to the variations of water level in the parent canal, which are frequent in run-of-the-river projects.

• Indonesia design standards were improved in the 1980s by foreign consultants. One of the proposed innovations was to replace the flashboards of check structures by conventional sliding gates. The main reason for this change was that flashboards were risky and difficult to handle.

The result of these two independent local improvements is a combination of hydraulic structures, the worst solution for the operation of a canal system. The sensitivity and hydraulic stability of structures are discussed in several books and design manuals (Horst, Ankum).

The Rominj gate was again unsuccessfully introduced in the design of the Mae Khlong project in Thailand.

10. United States of America

Application of modern technologies in water projects in the United States attracted the attention of many foreign visitors, as is the case in other developed countries (drip irrigation and wastewater use in Israel, Canal de Provence in France). The most comprehensive application of automation through central supervisory control is found in the control system of the Central Arizona project, which delivers water to urban, agricultural and industrial water users in central and southern Arizona. This system includes a large number of in-line pumping plants. The first effort to develop devices for local automatic control of canal systems in the United States dates back to the mid 1950s and was faced with the problem of instability in case of large flow changes.

Less well known is that many of the canal irrigation systems in the United States are far from having been modernized. Almost all control on irrigation canals is upstream control. Some systems still operate on rigid rotation schedules. In California, it is unusual to operate on pure demand. Water delivery to users is usually arranged. The average advance time for request is 26 hours for the 60 irrigation districts surveyed by Burt. Almost none of these districts have downstream control. Farmers, however, enjoy flexibility in flow rates. The flexibility in delivery can be offered because of excellent communications, high mobility of staff, high density of turnouts and judicious use of proper equipment such as weirs, regulating reservoirs and recirculation of excess water through interceptors and numerous applications of remote monitoring through SCADA. There is almost always measurement of flow rates at all turnouts.

In the Grand Valley district in Colorado, water is delivered on demand with a crude upstream control and few regulators. The system is operated at high flows with a large proportion of flow back to the river.

Burt has identified some aspects of the social and legal environment in the United States that have a bearing on the success and failure of irrigation projects, such as:

• The projects benefit from water rights and have the ability to enforce water rights and rules;

• Projects have excellent legislation for the formation of water user associations. Most of these associations are operated as businesses with professional management staff responsible to the elected boards of directors;

• Most consultants in modernization are private local consultants who must live with the results of their work; news on bad projects travel fast;

• Good living conditions (health, education) in rural areas; and

• There is excellent infrastructure for spare parts and new equipment.

This environment is not found in many developing countries, where modernization of irrigation projects is therefore more difficult to undertake and sustain.

C. Use of Old U.S. Bureau of reclamation standards in developing countries

Design standards for the projects supported by the Bureau of Reclamation in the Western States are the most detailed standards that can be found worldwide. They have been widely disseminated through technical assistance and consulting firms to a number of developing countries. In some of these countries, such as Thailand, the Philippines, Mexico and Turkey, U.S. Bureau standards have become de facto national standards for a few decades. In countries without national standards, they were used for specific projects, such as the Kriang project in Malaysia or the San Lorenzo project in Peru.

The basic design consists of a reticulated network of canals equipped with manually operated structures. Cross regulators are equipped with one or more radial or flat gates which are hand-operated or motorized. In some cases, a small lateral weir section is provided for emergency purposes, not for normal operation. Typical off-takes are equipped with constant-head orifice gates designed to measure and control flows. That infrastructure is in theory compatible with different methods of water distribution: prearranged, rotational or centralized.

A large number of these projects have a low hydraulic, agronomic and economic performance, as demonstrated by a number of recent studies (FAO). The U.S. Bureau standards were acceptable for the specific conditions of some Western States: short rainy season, relatively large farms, good road network and communications, and highly dedicated and trained operating staff. Good quality of construction was also a condition of success. All these conditions were generally not present in the countries which adopted these standards. An exception is the arid north-western region of Mexico (states of Sonora and Sinaloa). As in the U.S. irrigation districts, water is distributed in Mexico on a prearranged basis whereas centralized distribution is the rule in East Asian countries.

Projects supported by other federal or state agencies in the United States are not necessarily applying the Bureau standards. A number of irrigation systems and large conveyance systems in the United States have been upgraded either at design stage or later through rehabilitation and improvement programmes. Automated data collection and control has become an integral part of most large water delivery systems and is becoming more prevalent on smaller projects as well. From its beginning with simple gate controllers, canal automation has evolved to include large supervisory control systems that oversee entire projects. The California Aqueduct and the Central Arizona project are operated under a central remote system. The Salt River project, which was under remote monitoring, has been upgraded to remote control. As noted earlier, small-scale canal modernization projects are now widespread throughout the western United States. For example the gates of the composite check structures of the Dolores project have been automated. (Composite structures consist of a combination of one or two automated gates and long-crested weirs.) Several terminal reservoirs were built in the Coachella project, still operated for gravity application, to allow the farmers to convert from surface irrigation to low volume application methods. The Sevier River water user association in Central Utah has installed low-cost solar-powered automatic gates and SCADA. This user association has adopted the principles of gradual upgrading and retrofitting of existing infrastructure that are generally adopted for the modernization of irrigation systems in the western United States. None of these upgrading/modernization tools were applied in developing countries to improve the performance of gated projects. The 175-km-long main canal serving the 90 000 ha Phitsanulok project in Thailand is a case in point (Box 8).

In 1991, the Bureau of Reclamation issued a manual on canal system automation. The preface of this manual points out that the earliest designs of canals were based on the maximum flow conditions (known as full supply). This design does not provide the requisite flexibility to operate a canal efficiently. The first canal automation was crude but it was immediately successful. Advances in the operation of canals through the use of automation have paralleled the development of the electronics industry.

This statement is of major relevance to the canal systems in developing countries designed for maximum flow conditions; they simply cannot work as designed. Unfortunately, the signal sent by the Bureau of Reclamation on the limitations of old designs was ignored by many design engineers. Institutional reforms alone, including participation irrigation management, will not change this situation.

Technical and managerial deficiencies in irrigation projects

Photo 1 Dominican Republic: A tampered gated check structure

Viet Nam. Dau Tieng Project

Photo 2 Canals designed for operation at full supply cannot deliver water to the lower level because of a lack of check structures

Photo 3 Farmers correct th design deficiencies by constructing bamboo weirs

Proportional division and structured design

Photo 4 Pakistan: SWABI Project in the North West Frontier Province. Use of flow dividers, a more transparent device for proportional flow strategy than the conventional outlets used in the Indus River Basin

Photo 5 Argentina: An adjustable flow divider providing some flexibility in apportioning flow in proportional systems

Photo 6 Nepal: Sunsari-Morang Project. An example of transfer of the "structured design" in a humid tropic area resulting in some conflicts between farmers on when to close the irrigation season

Hydraulic automation: examples of long-crested weir type cross-regulators

Photo 7 Iran: Guilan Project. A typical long crested weir

Photo 8 Iran: Guilan Project. A double long-crested weir on the 100 m³/s capacity main canal

Photo 9 Malaysia: Kemubu Project. A composite cross-regulator consisting of a gate and a weir

Photo 10 India: Majalgaon Project. A double weir on a distributary canal

Photo 11 Iran: Guilan Project. A cross regulator consisting of an automatic constant upstream level gate associated with a conventional sliding gate

Photo 12 France: A cross regulator consisting of two automatic constant level gates

Photo 13 Philippines: An example of low cost modification of a gated structure in a weir check structure

Land consolidation in Japan

Photo 14 Farm layout before land consolidation

Photo 15 Farm layout after land consolidation

Conjunctive use

Photo 16 Pakistan: Conjunctive use of ground and surface water using the same canal system

Remote monitoring and remote control

Photo 17 U.S.A. Coachella Valley Water District. SCADA

Photo 18 Spain: Cabral Project, SCADA

Photo 19 Morocco: Haouz Project. Remote automatic control (dynamic regulation)

The active role of water users' associations in the modernization of irrigation projects

Photo 20 Mexico: Rio Fuerte Project. First stage: Modernization of maintenance equipment and office technology

D. Cross-country transfer of technology

Transferring any technology from one environment to another should be approached with caution. Great caution is particularly needed in the irrigation sector where the site conditions are specific and the success of the transferred technology depends on physical, social and economic factors. Horst points out in his recent book that "many donors stipulated that foreign consultants were to be involved in the planning, design and supervision of construction. These consultants came from different parts of the world with different irrigation technologies and traditions. Each of them was educated and experienced in one of the distinct irrigation schools. Owing to the weak position of the national irrigation departments in terms of experience in planning and design and the dominant role of the donor agencies, the consultants were able to decide on the technology to be adopted, that is to sell or impose their own technology. In other words, [it was] the country of origin of the consultants [that] determined the type of technology, and not the compatibility with the local physical and socioeconomic environment".

Horst's observation applies to a number of developing countries which do not have well-established practices of irrigation engineering. In countries with large development of irrigation, the state officials have often entrenched engineering practices. Foreign consultants may face strong resistance from local engineers, sometimes justified, to any proposed change in design of irrigation projects. Irrigation departments in India and Pakistan have rigidly adhered to their design standards for decades. It is only recently that some innovative departments have agreed to adopt new standards. Examples include a few projects in India (such as the Majalgaon project in the state of Maharashtra and the major Narmada project in Gujarat), and the high-level Pehur project and SWABI- SCARP in the North-West Frontier Province in Pakistan.

A number of lessons can be learnt from the examples of failed transfer of experience.

1. India: Transfer of rotational distribution from northwest India to the southern states

Despite the undoubted achievement of irrigation development in India and the success of the Green Revolution in that country, performance of irrigation projects, particularly in the southern states, was far below potential. A strong group of local engineers supported by financing agencies promoted the idea to transfer, with some adjustments, the design package of the northwest states to the southern states, which they considered as the best system in India. This idea materialized in the World Bank-supported National Water Management Project (NWMP) in 1985. The main objective of the project was to improve agricultural productivity through the provision of a more reliable, predictable and equitable irrigation service. The most important element in scheme improvement was the preparation of an operational plan. On the basis of water availability, system characteristics and agricultural options, the plan was expected to define how the system would be operated with respect to the timing and quantities of water deliveries. The project concept was to convert the demand-type systems into supply systems. To ensure equity in water distribution, the "structured design", an adaptation of the rigid rotational warabundi delivery combined with ungated canal technology, was developed for the project. The structuring level is the point downstream of which the canal system is ungated. This system was not tested in a pilot project in paddy-growing areas where field-to-field irrigation was traditionally practised. The system does not have the flexibility to adjust to the important variations in rainfall and soil conditions which prevail in southern India. In the warabundi states, soils are rather uniform and rainfall does not contribute much to the total water. More important, groundwater, which is a reliable and flexible supply, accounts for a large portion of the water resource, a critical difference for the farmers of southern India, where groundwater resources are not as rich and widely spread as in the alluvial northern plains.

The project was rated unsatisfactory at completion. During a seminar on modernization of irrigation systems in 1998, IWMI reported the failure of the concept of the structured design and equitable supply technology in the Bhadra scheme, which was part of the NWMP project: "The provision of reliable and equitable supply was not achieved as expected at appraisal. The changes in cropping patterns and agricultural calendars stipulated in the operational plans had not been followed; the advancing of the kharif season could not be implemented" (Sakthivadivel). Lessons from this project were taken into consideration in the formulation of a new generation of projects in India. The thrust is now on improving productivity through system improvement linked to turnover of management of the systems to user associations.

2. Transfer of rotational irrigation from India to Thailand and Nepal

Thailand. An attempt was made in the Nong Wai project in Northeast Thailand in the early 1980s to introduce the warabundi delivery system to irrigation projects in Southeast Asia. This ill-designed transfer of a rigid delivery system got dismal results because of its incompatibility with rice irrigation and the local culture of Thai farmers.

Nepal. Because of the difficulties in operating the Stage-I rehabilitation of the Sunsari-Morang project (58 000 hectares), which was designed as a fully gated, manually operated system, a structured design was adopted for the rehabilitation of Stage II: downstream of the off-takes of the sub-secondary canals, the system is ungated. Tertiary canals are supplied through concrete flow dividers and the watercourses through adjustable proportional modules, as used in northern India. The operation of the Stage II subsystem has been considerably simplified compared to Stage I. However, it has lost flexibility in meeting the variations in demand due to factors such as local variations in rainfall and excessively long staggering of rice cultivation. Towards the end of the growing season, some farmers still request irrigation water while others are ready to harvest. The lack of a drainage system and of operational flexibility in the structured design imposes severe operational constraints, which affect productivity. The duty limitations of the main system (0.7 l/s/ha) and the low flows of the Kosi River during the dry season make this run-of-the-river project very dependent still on monsoon rainfall.

This example from Nepal challenges the opinion shared by some irrigation professionals that farmers prefer proportional distribution. They certainly prefer equity over anarchy but they can also understand that equity and higher productivity can be achieved through improved water control and alternative operational procedures.

If the water user associations in Sunsari-Morang in Nepal were organized before the physical improvements, farmers would have been in a position to formulate their preferences for not freezing the infrastructure into an inflexible distribution system.

3. Transfer of technology to user-managed systems in Indonesia: A case of farmers' rejection of inappropriate technology

Agro-socio-religious associations in Bali, called subaks, have developed a water division technology throughout the ages. The subak water division technology consisted of institutional arrangements backed by temple priests and of weirs dividing flows into negotiated shares. These centuries-old subak systems were first disrupted by the arrival of the Green Revolution in the 1970s, which changed the agricultural practices required for cultivation of shorter maturing rice varieties and more dramatically by the government plan to "modernize" the subak systems with the assistance of a financing agency and foreign consultants. The "modernization" project attempted to introduce the technology and water distribution procedures which were the norms in the government-built and -managed irrigation projects in Java. The project consisted in the replacement of the dividing weirs existing at each bifurcation by adjustable structures to be set and reset on the basis of frequent calculations of the crop water requirements^[18]. Gates were installed to regulate and measure flows. In general the subak members did not accept this technology. They handled the gates to accommodate the division of water according to their perceptions of water allocation and removed the gates to restore the former fixed-proportion division structures. Towards the end of the project, the state officials reluctantly accepted the subak technology and built new or improved existing proportional division structures.

Curiously neither the change of irrigation technology nor its social implications have been discussed in any of the design reports or even in the completion reports. The technology practised on Java was transferred to Bali with no concern for the opinions and perceptions of the subak members (Horst). This pure engineering approach is unthinkable today, given the emphasis now placed by financing institutions on a participatory approach in the design of irrigation projects.

E. Conclusions

Different strategies of water delivery and water control have been used for the development of canal irrigation schemes throughout the world. Obsolete designs can be found in both developed and developing countries. Many irrigation projects in the United States, Western European countries and Australia built decades ago are inefficient in terms of water and energy use and are in urgent need of modernization. Advanced concepts and electronic-based sophisticated technology have been used for nearly three decades and have been introduced in a number of developing countries such as India (Majalgaon, Narmada), Egypt (the Nile telemetry system), Morocco and Jordan (dynamic regulation). Hydraulic regulation has been used in most Mediterranean countries for about fifty years. Modernization of irrigation, as defined earlier, is not an issue limited to developed countries, as has been stated during some workshops.

The concepts used for the development of irrigation by colonial powers since the mid 1800s in India, Egypt and Sudan were well adapted to the conditions and to the objectives of irrigation in the past. Irrigation was extensive and the water resources were not regulated by large storage reservoirs. Conditions have changed with the intensification of irrigation due to the pressure on land related to the escalating demography and to the construction of regulating reservoirs. Relaxation of the discipline of the users required for an adequate operation of ungated systems in the Indo-Gangetic plain is often mentioned as the cause of poor performance. The diversion of larger volumes of water due to the construction of large dams has exacerbated the problem of siltation, particularly in the smaller canals, which in turn has contributed to the inequity of distribution since the flows diverted through the farm outlets are influenced by the upstream water level. The farmers responded to the economic changes by tapping additional water resources to overcome the limitations of the existing systems, which were undersized for intensive irrigation. Farmers captured more water from the canals by illegal means (Indus basin), replaced animal-driven pumps by motor pumps (Nile valley) and installed a dense system of shallow wells or deep tube wells. Groundwater accounts now for a large proportion of the water used for irrigation in the alluvial plains, which is a normal evolution of irrigation development. The unique feature of the Gezira scheme in Sudan, consisting of night storage in minor canals, makes possible to shift from a rigid delivery to an on-demand system, an advantage greatly appreciated by the farmers of that scheme in the absence of groundwater or any other water resource, including drainage water in the area.

The design standards adopted in many developed and developing countries after the mid 1900s to deliver water according to crop demand were conceptually more advanced. However, most of them failed to meet that objective because of the deficiencies of the water control technology and complexity of the operational procedures. Managing an irrigation system equipped with manually operated gates at each branching point is a very complex task. In many cases, the systems were designed to be operated at full capacity without consideration for operation at less than full supply. Even with the best vigilance of the operators, operation of these systems is usually inefficient and/or costly in developed countries. Computer-assisted calculations of irrigation targets based on assessments of crop water requirements were developed for use in some countries where centralized scheduling is practised, such as Indonesia. The demands imposed on agency and irrigators to collect innumerable data, to calibrate devices and to control flows often prove beyond their capabilities and interests. Some information which is frequently lacking or inaccurate is data about seepage and percolation, return flows and spatially variable rainfall. Where the inability to take such factors into account renders irrigation targets unacceptably inaccurate, water is distributed based on qualitative judgments by field staff or through interference by farmers. This makes the water delivery system uncertain.

The use of technology with adjustable structures, which has been the norm during the three decades of intensive development of irrigation in developing countries from 1960 to 1990, has badly affected the performance of irrigated agriculture in many countries. It is now impeding the transfer of management to user associations.

With hindsight, the outcome appears to have been inevitable, raising questions about the realism of the foreign consultants' plans and the Bank support to them. These experiences give the impression of donors and technical assistance teams using the (East Asia) region as testing ground to try out new designs, with encouragement from agency headquarters, but without a realistic assessment of local management capacities of the incentives for irrigators (Rice). This paper argues that even the best qualified managers and operators would not be able to manage these systems to the highest standards over long periods without the assistance of modern communication systems and/or remote monitoring. The issue is in the deficiencies of the design that imposes very complex methods of operation, not in the organizational weaknesses of the irrigation agencies.

The farmers served by these low-performing manually controlled systems have reacted in different ways to be able to adopt modern cultivation practices and diversified cropping patterns: tampering with control structures, pumping from canals, drains, borrow pits and, more recently, tapping groundwater resources which provide the flexibility and reliability needed for modern irrigation at farm level. These responses from the farmers are inevitable and irrigation agencies are generally passive since they can do very little to stop them. However, it is not a proper use of limited water resources. It is an unacceptable situation with regard to the increasing competition for water and environmental considerations. In some countries, farmers use untreated water which is rich in nutrients and constitutes a reliable resource for yearly intensive cultivation of high-value crops in suburban areas (for example in Punjab, Pakistan).

The development of hydraulic automation in North African countries in the 1950s helped to a large extent the operation of canal irrigation systems by reducing the number of structures requiring readjustments and the frequency of resetting control structures. Automatic downstream control eliminates the need for complex calculations of water releases. It is therefore puzzling that these innovative design standards have not been adopted in other countries. The reasons for the slow adoption of these or any modern techniques are both administrative and behavioural:

• Lack of economic pressure on irrigation agencies;

• Lack of contractual motivation for consultants to introduce new concepts;

• Resistance to change by irrigation managers, engineers and others; risk aversion and adherence to outdated designs;

• Lack of operational experience and service motivation by planners and irrigation departments;

• Lack of sufficient training at all levels, from the university to the field;

• Lack of evidence of the superiority of modern systems in terms of agricultural productivity;

• Failure of some pilot projects for technology transfer (Sidorejo in Indonesia, Cupatizio in Mexico)

• Use of economic tools during the preparation of projects focusing on cost comparisons of different equipment and overlooking the potential benefits to be expected from modernization.

Some transfer-of-technology projects have been unquestionable success stories, such as the Guilan project in Iran or the Muda project in Malaysia. Regrettably these projects have not attracted the attention of international research organizations.

Many transfer-of-technology experiments have failed because of inadequate attention to all the key factors that determine the selection of an appropriate irrigation strategy. For example, the transfer of a rigid method of water allocation from arid zones to rice projects is doomed to fail if the farmers have not been involved in the process which affects their cropping patterns and practices and if they do not perceive some improvement in the quality of service. Replacing fixed water-dividing structures in traditional run-of-the-river schemes without a regulation of the water resources is also doomed to be rejected by the farmers, as was the subak system in Bali.

Suitable water control technology is not enough, however, to achieve high agricultural productivity. It is assumed that the mediocre productivity of irrigated agriculture in Morocco, particularly for cereal crops, is mostly related to the centralized method of irrigation scheduling, with little participation of the irrigators, lack of maintenance of on-farm works, and constraints imposed by the land consolidation model, possibly in combination with deficient use of non-water inputs. It would be useful to carry out an in-depth analysis of the performance of selected irrigation projects in that country to check the validity of the above assumption.

A last category of projects is those with faulty design, such as wrong selection and combination of control structures which amplify the fluctuations of hydraulic conditions in irrigation canals and those using unrealistically complex procedures to determine irrigation releases.

VIII. THE FORCES OF CHANGE

Irrigation has well served in the past in supporting the increase in food production, but it must evolve to adjust to the new economic environment (Gardner).

In the above chapter, it was noted that the farmers are responding to the changes affecting irrigation in their own environment by looking for more reliable and flexible water supply. Other fundamental and potentially far-reaching changes are challenging some of the basic premises supporting the use of irrigation, as least as traditionally practised. This chapter systematically explores these changes and their effects on the future of irrigation.

The key forces that are going to influence the role and performance of irrigation over the next decades are:

• Population growth, with an even faster growth of the urban population and the continuous prevalence of rural poverty in several regions;

• Competition over water supplies between agricultural and other uses (municipal, industrial, recreational, energy generation, and environmental uses) and the rising cost of developing new resources;

• Globalization of the economy resulting from international and regional agreements (GAAT, NAFTA, European Union) and rapid advance of the information and communication technologies;

• General public awareness that the environment should be protected;

• Diminishing government implications due to changes in institutional policies; and

• Climatic changes such as a higher recurrence of drought years.

These changes have and will continue to have considerable consequences for irrigated agriculture. The demand for food from irrigated lands will increase, as demonstrated by all models of food-supply predictions. In high-income developing countries, changes in diet patterns from the increased urban population will shift demand for staple food towards processed food obtained predominantly from irrigated fruits and vegetables. For example in Taiwan, rice consumption has declined from 130 to 65 kilograms per capita over the last three decades. In some countries, the governments will gradually eliminate the protection of commodity prices to comply with international agreements. As a result of the globalization of the economy, irrigated agriculture will be driven by market forces and will have to compete on both international and national markets.

The forces of change have and will continue to have serious impact on water supply to irrigated agriculture. Given the increasing competition for water, supply to irrigation is going to decline in some countries. The excessive cost of developing new water resources and the reduction of subsidies to public irrigation systems will contribute to this decline.

Irrigated agriculture is gradually becoming more accountable for the environmental degradation, in particular the degradation of water quality through contamination by salt and agrochemicals. Stronger regulations are required, together with mechanisms to arbitrate conflicts between environmental and irrigation interests.

Response from farmers

In some countries, subsistence irrigation will continue to cover a large share of irrigated lands and provide staple food to poor farmers. However, the general trend towards the modernization and efficiency of irrigated agriculture will apply to small-farmer communities operating now at subsistence level. The objectives will be yield increases, reduction of water consumption and energy costs, and crop diversification.

In some countries irrigated agriculture will predominantly become an economic activity driven by market forces rather than a way of living supported by government subsidies. Increased economic efficiency will be a condition for farmer survival, implying continuous improvement in technology, agricultural practices, farm management and marketing. Savings on water, labour and energy costs will increasingly become major considerations. Confronted with increasing costs of water supply, combined with reduced protection on agricultural prices, farmers will have to react by producing more with less water. They will shift to higher-value crops and crops consuming less water and they will adopt water conservation strategies to reduce water losses.

Farm structures will shift towards well-operated and well-financed units with strong integration in domestic and international marketing and processing industries. This trend took place in OECD countries during the last decades, where the people directly involved in farming activities now represents 3 to 5 percent of the total population. Small farms persist, however, in countries like Japan and Taiwan through weekend farming by aging farmers and highly subsidized agriculture. Irrigated agriculture is in the shrinking phase in some countries, as a result of urbanization and other factors. In Taiwan, irrigated lands decrease from 560 000 to 380 000 hectares over the last three decades.

Some irrigation projects with excessively high operating costs because of high lift pumping or adoption of extensive irrigation with high maintenance costs, which are not sustainable without government subsidies, will have to be abandoned unless governments keep a policy of subsidizing irrigated agriculture.

Response from technology

Few industries produce the same product in the same way they did fifty years ago and irrigation should not be an exception. However, irrigation technologies have slowly evolved for centuries and age-old practices can still be observed in many rural areas. Significant changes took place in the late 1800s with the construction of large reservoirs providing regulated water to the users and the possibility to balance water supply and demand. A second wave in advance in irrigation technology was the development of more efficient application methods at farm level, including surface methods and pressurized systems. However, application of most of these techniques is still limited in most developing countries. The most striking change in irrigation in the large irrigated areas in Asia during the last two decades has been the phenomenal development of groundwater, which is discussed in Chapter 9.

Because of farm sizes, markets and other factors, farmers in developed countries have readily adopted advanced irrigation technology. This adoption has had a major effect on productivity. Labour, energy and costs of water have considerably influenced the adoption of technologies and the use of water for irrigation in the United States. Since 1968, the total irrigated acreage has increased by 37 percent, but the average water application has decreased by about 15 percent from 6 300 to 5 400 m³ per hectare. Surface irrigation methods that were used on 90 percent of irrigated lands are still the most common methods of irrigation but are now used on only 55 percent of the lands. Sprinkler irrigation is now used on 41 percent of the lands. However, the traditional sprinkler methods, hand moved and solid set, are rapidly loosing ground in favour of centre pivot and linear move, which are now used on one quarter of irrigated lands in the United States.

Labour requirements for irrigation systems vary greatly. Automated systems, such as automated micro-irrigation and centre pivot, have relatively low labour requirements. The recent success of the low-energy precision application labelled LEPA is due to its combined low labour, low energy and water-saving advantages.

The slow adoption of new irrigation technology in developing countries is a perplexing issue. While there have been changes in irrigation technology in the United States, Australia and Western European countries for example, little of this development has affected irrigation in many developing countries. Some of the constraints are obviously the unavailability of capital, low costs of water and energy and pricing policies that fail to provide incentives to conserve water, and the absence or limitations of high-value crop markets and marketing facilities.

However, a main reason for the slow transfer is that the focus of attention in irrigation technology and research in developing countries has occurred at farm level, and not at the level of operation of the main and conveyance systems. Farmers will not invest in water-saving technologies if the service of water is not reliable and if the incentives for saving on water, energy and labour are not strong enough. Bottrall observed in 1979 that "it is only if the main water distribution system is well operated that many other important management objectives can be satisfactorily realized, and it is only then that high returns can be obtained from agricultural extension advice and the increased application of other complementary inputs".

Response from agricultural research

The agricultural research centres deserve credit for their contribution to the achievements of the Green Revolution. The Irrigated Rice Research Institute estimates that their new rice varieties have increased water productivity threefold through increased yield and reduced crop duration. IRRI and CIMMYT are optimistic that they can develop high-yielding drought-tolerant varieties. However, development of reliable irrigation is crucial to realizing the benefits of high-yielding modern varieties. Growing crops under a mild water deficit requires a high degree of water control and farmers' confidence in the irrigation system (Molden).

Advances in genomics and genetics will certainly contribute to the challenge of food and water production but would have to be associated with improvement in water delivery.

Response from the governments

As a consequence of diminishing implication of governments in irrigation management, stakeholders and the private sector will take an increasing share of responsibilities at all stages of irrigated agriculture. The governments will support the establishment of new policies, legislation and institutions. The decisions about water allocation and planning will be gradually made at the river-basin level through a consensual process among users.

The governments will continue shifting from supporting the construction and rehabilitation of large irrigation infrastructure towards establishing new policies to support private-sector participation, water conservation and environmental protection.

The forces of change discussed in this section will force the irrigation engineers to design more efficient and responsive irrigation systems.

IX. THE EXPLOSIVE EXPLOITATION OF GROUNDWATER RESOURCES

The contribution of irrigated agriculture to food and fibre production has continued to increase despite the lower level of investments for developing new irrigable areas and the focus on rehabilitation of existing schemes. One of the reasons is the exceptional increase in groundwater development in recent decades. Declining extraction costs due to advances in technology and in many instances government subsidies for power and pump installation have encouraged private investment in tube wells. Groundwater in India now supplies more than 50 percent of the irrigated area. Due to higher yields in groundwater-irrigated areas, groundwater is central to a significantly higher proportion of the total irrigated output.

The significance of groundwater in the Indian economy is due to the fact that agricultural yields are generally higher - by one third to one half - in areas irrigated by groundwater than in areas irrigated from other sources. Groundwater offers greater control over the supply of water than do other sources of water. As a result groundwater irrigation encourages complementary investments in fertilizers, pesticides and high-yielding varieties, leading to higher yields (World Bank 1998). It is the reliability of groundwater that allows farmers to take the risk of investing in fertilizers, which substantially increase their crop productivity (Ahmad).

In Pakistan, groundwater development through private tube wells has grown exponentially, especially in Punjab. According to a 1991 survey, about 46 billion m³ of groundwater are used for irrigation in the Indus basin, 85 percent of which comes from private tube wells. However, salinity continues to present a threat to the sustainability of agriculture because of the recycling of large quantities of poor-quality groundwater from the top of underlying aquifers.

Groundwater exploitation for irrigation is not limited to arid or semi-arid countries. The explosive use of diesel pumps in the Chao Phraya and Mae Khlong river basins in Thailand has responded to the increase demand for dry-season cultivation of high-value crops and the unreliable supply from the large gravity irrigation systems. Commenting on the changes that have affected the Phitsanulok project in Thailand, Manuddin pointed out the advantages of groundwater over canal water: "With an average of one well for 5 hectares, virtually all the farmers now have access to groundwater. The development of groundwater has given farmers a high level of control over their crop calendar. They do not have to wait for the availability of canal water, and they can plant their crops at the time that seems the best according to their own situation. The benefits that the changes have brought to farmers include increased quantity of water, increased reliability of water and freedom for the families to choose their own crop strategies."

The rapid development of groundwater has recently been observed even in some projects that were designed to meet the full requirements needed for intensive irrigation and to provide reliable service. About 9 000 private deep wells have been installed in the 40-year-old Tadla project in Morocco during the last five years, i.e. nearly one well for 10 hectares. The main reason for this recent farmer initiative might be the higher frequency of dry years that affects annual water allocation and the constraints imposed by the current water allocation strategy.

Overexploitation and an associated decline in water quality have been occurring in many parts of the developing world, particularly in the arid and semi-arid regions. Water tables are falling at an alarming rate - often 1 to 3 metres per year. These regions include some of the world's main grain production areas such as the Punjab in India and the North China Plain. About two thirds of farmlands in North China are facing serious problems of groundwater exploitation (Shah).

Groundwater has played a critical role in food production by agriculture over recent decades. However, groundwater is a major emerging problem in many parts of the world. Some areas have reached the point where overexploitation is posing a major threat to the environment, health and food security. The potential of groundwater development for irrigation may be reached soon. Seckler pointed out that "many of the most populous countries of the world have literally been having a free ride over the past two or three decades by depleting their groundwater resources" and he concluded that "the results could be catastrophic for these countries and, given their importance, for the world". The explosion of groundwater irrigation in some countries is a farmers' response to the lack of flexibility and, in the worst cases, to the unreliability of the canal irrigation systems. "Water recycling and the conjunctive use of groundwater are rarely considered in the original design of irrigation projects. They mostly happen as a desperate response from farmers who are unable to obtain their share of irrigation water from the canal or from system managers as a way to rectify problems of management capacity and shortcomings of the original design" (Bhuiyan).

These brief considerations on groundwater use in irrigation lead to the conclusion that there is an urgent need to improve the quality of service from water surface systems and for a well-thought conjunctive use of both surface and groundwater. The present passive attitude is no longer acceptable.

X. THE PLANNING PROCESS: A GLOBAL GAME PLAN

Definition of modern design

Several definitions of modern design have been proposed. The following definition was adopted during an FAO seminar on modernization held in Bangkok in 1996: "Modernization is a process of improving resource (labour, water, economic and/or environmental) utilization by upgrading (as opposed to merely rehabilitating) the hardware and software in irrigation projects, while maintaining or improving water delivery service to farms."

Another definition was proposed during another FAO-supported workshop on the valorization of irrigation water in the large-scale irrigation schemes of North Africa in 1999: "Modernization is a process of rehabilitation of irrigation systems during which substantial modifications of the concept and design are made to take into consideration the changes in techniques and technology and to adapt the irrigation systems to the future requirements of operation and maintenance. Delivery of water should be as flexible as possible, with demand irrigation being the ideal solution."

These two definitions slightly diverge on the technical aspects. The first one focuses on the shift from supply to service-oriented and the second one on future requirements of operation and maintenance. The first one suggests the use of advanced concepts of hydraulic engineering, the second of new techniques and technology, which may include modern equipment for remote control. The main difference is that institutional and organizational changes, including more active farmer participation, are attached to the first definition. The principles attached to the 1996 definition have been elaborated in a number of publications. These principles are summarized in the next section.

Principles of modern design

The overriding principle of modern irrigation is that irrigation is a service to farmers which should be as convenient and efficient as possible. Farmers ultimately have to generate the benefits which keep the system functioning. Modern irrigation schemes can be conceived to consist of several subsystems or levels with clearly defined interface, where water is measured and controlled.

• Each level is as financially autonomous and hydraulically independent as possible.

• Each level is technically able to provide reliable and timely water delivery to the next lower level. At each level there are the proper types, number and configurations of gated turnouts, measuring devices, communications systems and other means to control flow rates and/or water levels as desired.

• Each level is responsive to the needs of its clients. Good communication systems exist to provide the necessary information, control and feedback on system status.

• Each level of delivery has confidence, based on enforceable rights, in the reliability, timeliness and equity of the water which will be supplied from the next higher level. Effective mechanisms for conflict resolution are in place.

• The hydraulic design of the water delivery system is created with a well-defined operational plan in mind. The operational plan is established with a clear understanding of the needs of the end users.

• The hydraulic design is robust, in the sense that it will function despite changing dimensions, siltation, and communication breakdowns. Automatic devices are used where appropriate to stabilize water levels in unsteady flow conditions.

• Motivated and trained operators are present at all levels of the system. They are not necessarily the farmers themselves but preferably hired staff. Instructions for individual operators are well understood and easy to implement.

• Maintenance is the obligation of each level. Maintenance plans are defined during design and are adequately funded and implemented.

• There is a clear recognition of the importance and requirements of agriculture and of the existing farming systems. Engineers do not dictate terms of water delivery; rather agricultural and social requirements are understood at all levels and in all stages of the design and operation process.

A modern design is the result of a thought process that selects the configuration and physical components in light of a well-defined and realistic operational plan that is based on the service concept. A modern design is not defined by specific hardware components and control logic, but use of advanced concepts of hydraulic engineering, irrigation, agronomy and social science should be made to arrive at the most simple and workable solution.

The most important issue and the one that often receives superficial attention during project preparation and appraisal is the ability of the system to achieve a specific level of operational performance at all levels. A precondition for high performance is that the design must reflect the objectives and requirements of the future operation. As long as the design of irrigation systems is understood as a classical engineering task of designing structures, essential operational questions will not be addressed.

A proper operational plan is one that combines the various perspectives and helps reconcile conflicting expectations between the users, the project manager, the field operators and the country's policy objectives. The two preliminary steps in the planning of irrigation projects are the definition of objectives and the associated decision about water deliveries.

Project objective

A preliminary step in the planning of an irrigation project or its rehabilitation is the definition of project objectives that depend on the country policy for irrigated agriculture. The objective could be to maximize the value of production by unit of water or unit of land. Project designs vary whether the project objective is to develop export-oriented commercial farming or to support rural population, alleviate poverty in rural areas and limit rural migration to urban centres. The original design of a project may no longer be compatible with the forces of change, which are affecting irrigated agriculture in many countries.

Water delivery

The second step in the planning of an irrigation project is the decision about water delivery. This can be described as the frequency, rate and duration of water deliveries at all levels of an irrigation system. The various systems of water delivery are described in the literature (FAO, Clemmens): rotation, arranged schedule, limited rate demand and centralized scheduling. Most traditional delivery systems have no or little flexibility built into them. They do not attempt to match water deliveries to crop needs. The stated objective is to obtain equity through simplicity of design, although poor design, maintenance and operational problems may prevent from achieving the objective. Modern irrigation projects are designed with the stated objective to deliver water according to crop requirements. At a minimum, the frequency and sometimes the duration of irrigation should be adjustable.

As noted earlier, a water delivery schedule does not necessarily imply a specific design or technology. A fairly rigid schedule of water delivery to the farm turnouts may use modern irrigation hardware and computerized decision support systems to make the water delivery reliable and equitable. A primary advantage of using modern design and equipment is that the operators can choose the flexibility offered at various levels of the system. For example, the different methods of water distribution in modern irrigation systems in Morocco are variations of centralized scheduling (with the choice given farmers on duration and timing), although the systems have the capability to be operated on an arranged schedule. The reverse is not possible. A project designed for rigid rotation through simple non-adjustable structures or for proportional distribution cannot be operated for flexible water distribution.

The most important step in the design of an irrigation system is water control defined by three strongly related elements: the configuration of the distribution system, the control strategy and the hydraulic equipment.

Configuration

The configuration of an irrigation system is obviously determined by the relation of land and water resources, the topography and economic considerations. The design should incorporate as much as possible features that facilitate operation and provide flexible irrigation service, such as buffer reservoirs and on-farm reservoirs and use of low-pressure pipes instead of canals for tertiary distribution. Buffer reservoirs are widely used in mid and south China, in what is known as the "melon on the vine" system. Buffer reservoirs can be located alongside main systems or at the interface between two levels of management, such as between main and secondary canals. The incorporation of reservoirs reduces to some extent the need for sophisticated water control methods of the main systems.

Water control strategy

The designer has the choice between many control strategies for the operation of the system:

• Upstream, downstream control or controlled volume

• Local versus remote monitoring and control

• Proportional versus adjustable control

The designer has also the choice between various configurations for the automation of the canal systems:

• Distributed control in which control is achieved through independent automatic units;

• Centralized control in which control is achieved through a master station; and

• Supervisory control combining distributed automation under master supervisory control^[19]. This configuration is known as SCADA (supervisory control and data acquisition).

These three systems are depicted in Figure 2 a, b and c showing the same canal system equipped with different types of control structures. The advantages and disadvantages of these different configurations are discussed in technical publications. Under distributed control, the system manager is not in a position to supervise or control the entire canal system. Centralized automatic control makes possible the use of highly efficient control logics but the operation depends on the reliability of a communication system. Under supervisory control, the central station makes decisions on the lower-level strategy based on the data received from the local controllers, also known as remote terminal units or programmable logics controllers. These local controllers make changes to the control devices according to the target instructions received from the master station, such as maintaining a target flow rate or water level. This system is less susceptible to communication system failure. The centralized and supervisory control methods can involve varying levels of participation of the master station personnel in making decisions from manual to computer-directed control, which uses specially developed computer programs using data from the entire canal systems and modelling studies. Computer-directed control is applicable to the most complex systems involving a number of canals, reservoirs, pumping and/or power stations.

Figure 2 Alternative configurations of canal automated systems

A. Distributed Automation.

B. Centralized Control.

C. Supervisory Control.

The selection of water control strategies can have very different effects on day-to-day operations. Certain control strategies eliminate the need for advance scheduling of water deliveries, the need to know exactly the flows at various sections of the canals, the determination of lag time and the estimation of seepage and operational losses. This is the main advantage of the basic downstream control compared to the upstream supply-oriented control methods, which require elaborate and complex predictions of irrigation requirements. However, downstream control is not necessarily associated with demand delivery. It is essentially a control strategy often used to greatly simplify the operation of very long canal systems, which is very complex under upstream control. Demands from the next lower level of secondary canals can be either under rigid rotation or flexible.

A single strategy is rarely used for an entire irrigation system. Most projects combine two or more control strategies. A main canal can be under downstream control and the distribution system under upstream control; or the main canal can be under upstream control and the distribution under proportional control (structured design in Nepal). Many examples of combination of control strategies could be provided to illustrate the wide number of solutions. The Narmada system is designed for remote central control and the tertiary system for proportional control using the structured design standards. The King Abdullah canal in Jordan is operated by remote control under dynamic regulation although rigid rotation is used for the distribution of water from the pressurized pipe systems. The Coachella project in California is operated under upstream control with the assistance of a remote monitoring system. A large buffer reservoir at the end of the canal absorbs the daily differences between orders and deliveries. This reservoir supplies open pipelines which receive rigid deliveries during 24-hour periods. Farmers have built numerous reservoirs on their farms to increase the flexibility of irrigation required for the mix of on-farm irrigation systems most suitable to the variety of crops.

The selection of a control strategy has a major impact on several aspects of the future performance of an irrigation system: ease of operation, ability to provide a high quality of customer service, and general efficiency. Projects designed for proportional division of water through rigid and passive structures are the easiest to design, construct and operate, but provide the least flexible service to the users. Manually operated gated systems are the most complex to operate to provide both quality of service and efficiency. Modern design makes possible to improve service and efficiency but requires more design skills and higher quality of construction and installation.

Figure 3 Complexity of different control strategies at design, construction and operation stages

Some guiding principles for selecting a control strategy and equipment

No control strategy and no equipment is ideal for all situations found in irrigation projects. Many physical and institutional factors have to be taken into consideration by the planners and designers. However, there are some general principles, which are summarized here:

• The delivery service should be as much as possible user-oriented. Reliability and equity of water delivery are the basic features of irrigation service. However, providing some form of flexibility in duration, flow and interval of irrigation should be considered during the planning stage.

• Different control strategies could be combined within an irrigation system. The degree of flexibility could differ from one level to another level of a canal system.

• A key objective should be the ease of operation - not necessarily the simplicity of design or installation. This principle is widely applied by other industries.

• Automatic downstream control is well suited for long canals because it eliminates the need for advanced scheduling and a number of estimations. Use of downstream control for large canals does not mean on-demand delivery at farm level. Efforts to convert existing systems to downstream control through the use of control algorithms have generally failed. However, adoption of downstream control with the use of automatic float-gates has been successful in new projects.

• Manually operated gated systems are the most complex irrigation systems to operate with high efficiency and reliability. Using simple hydraulic principles and equipment could make operation simpler.

• Hydraulic automation requires a minimum of skills and training, compared to electronic-based automatic controllers.

• Proportional-division systems are the easiest to operate and design. Basically these systems provide no flexibility in water delivery. They cannot respond to agricultural changes. However, some improvements in the design of proportional dividers make it possible to easily adjust the sharing of incoming flow.

In summary, the selection of a control strategy is not limited to a simple choice between gated and ungated systems, as implied in the oversimplification of the 1994 World Bank irrigation review. There are a number of options available for each level of a canal system and they can be combined to define the most desirable global solution in order to provide ease of operation and a higher level of service. These options are summarized in Figure 4.

Figure 4 Options for ease of operation and a higher level of service

Control equipment: The last step is the selection of control equipment that fits with the selected control strategy. A number of publications provide detailed description of the equipment available to control flows in canal systems. The 1993 ICID publication on Automation of canal irrigation systems presents the salient features and fields of application of the various types of equipment which may be used, including:

• Passive regulators: long-crested weirs, flow dividers, level controllers;

• Conventional gates: leaf gates, drop-leaf and flap gates;

• Automatic controllers: electro-mechanical and electronic controllers;

• Self-operating gates for automatic level control: float-operated gates;

• Instrumentation: position, level and flow sensors;

• Means of communications: radio and cable methods; and

• Equipment for remote monitoring and control master stations.

A number of physical, social and institutional factors, which are discussed in Chapter 11, should be considered in the selection of control strategy and equipment. Questions such as the possibility of crop diversification or conversion to crops with higher irrigation requirements, the risk of silting of canals operated under variable flows, the capabilities of the field staff to operate and maintain electronic equipment, the acceptance of the operating rules by the farmers and their understanding of how the structures function should all be considered. The answers to some of these questions are beyond the scope of responsibilities of a design engineer. However, it is his responsibility to select control structures that are robust, easy to operate and interact with the other structures in the vicinity to minimize the fluctuation of hydraulic conditions.

One of the most important points is the right selection of the combination of check structures and turnouts. Clustered structures react differently to fluctuations of upstream level and flows depending on their characteristics. The sensitivity of these structures and the hydraulic flexibility of the different combinations of overshot and undershot gates are discussed in detail in several reference books (Ankum and Horst). Figure 5 shows the sensitivity of overshot and undershot hydraulic structures (weirs and orifices) and the effect of a twofold increase of the head on the flow rates.

The World Bank review of the Indonesian irrigation sub-sector (1990) rightly observed that "very often the solution adopted has been sluice-gated controls along the parent canal combined with Rominj gated off-takes (overshot gates). Unfortunately this is the worst of all combinations from the hydraulic viewpoint since it is extremely unstable. Small deviations have a proportionally great effect". The negative effect on the operational stability of a system was overlooked when selecting the Rominj gate, which has excellent metering potential when considered in isolation. Figure 2 shows the best and the worst combinations of hydraulic devices to minimize the fluctuation in water levels and diverted flows.

The constant-head orifice gate, found in many schemes throughout the world, is a flow-control and -measuring structure which is particularly difficult to operate. Very few field operators and gate-keepers are familiar with the functions of the two gates. In practice, these devices are rarely used for water measurement despite their good capability in laboratory conditions.

Figure 5 Flow rate fluctuations through weir and orifice control structures

Note: When the relative head doubles, the relative flow rate increases by 180 percent over a weir and only by 40 percent through an orifice

Detailed design and construction drawings are the last steps in the design of irrigation projects. The design process moves from art to conventional structural engineering, which is the domain of civil and mechanical engineers. The emphasis is on the dimensioning of the structures and reinforcements and the mechanical and structural integrity analyses. Manuals and guidelines are widely used during this phase.

The configuration of an irrigation system and the selection of control strategies are the design steps that require the most imagination. Although there are economic considerations to select among various options, there are no design manuals that can be applied as in structural engineering. The selection of the control equipment requires up-to-date skills to keep up with the developing technology. The principles of hydraulic regulation, including passive and reactive regulators, although developed half a century ago, have not widely been used outside the Mediterranean countries, because of a lack of awareness of these techniques and in many cases resistance and aversion to innovation. Computer-assisted operation and telecommunications entered the irrigation sector about two decades ago and have been used to improve the performance of old systems under manual operation.

Figure 6 Combination of check and turnout structures

Worst combination

Best combination

Modernization of existing schemes

Modernization of irrigation schemes raises major challenges for designers and policymakers. How much of the existing infrastructure can be saved? Which level of investments can be supported by the users? How much can be achieved by substituting hydraulic infra-structure by management inputs? The first step in this modernization is an in-depth diagnosis of the present performance of the system. The objective of this diagnosis is to identify the changes that have taken place since its original design, and the deficiencies in design and management. The diagnosis should determine the best approaches to solving the problems and if changes in water deliveries and system control strategy are desirable or necessary. The rapid appraisal process presented in Chapter 13 proves to be a very successful diagnosis tool.

After the diagnosis of an existing scheme is complete, a master plan needs to be developed. The master plan needs to define short-term and long-term improvements. A list of priorities must be developed based on realistic financing availability. Of major importance is the choice of the configuration of the automation system between distributed, centralized and supervisory control.

Changes or improvements in control strategy are difficult to test in real conditions without disrupting the operation of canals. The development of digital computers and advances in numerical methods in recent years has gradually helped to solve this problem. Flow simulation models in recent years have made possible to develop and test various control strategies (Mutua). Two approaches have been adopted to make use of these new tools.

Simulation of canal response for different scenarios

This approach has been used in recent years by IIMI and other researchers to simulate different operational scenarios for improving the operation of complex canal systems, for example the Chasma Right Bank canal in Pakistan, affected by high silt content, and the Gal Oya canal in Sri Lanka, which has a large number of gated cross regulators. The complex operational rules derived from these studies have not been fully adopted by the agencies responsible for the operation of these canal systems. A simulation study of the Pyramid Hill No 1 canal in Victoria State in Australia concludes: "Operational scenarios to improve manual operation yielded only marginal improvements in the operational performance, thus reinforcing the view that significant gains in the quality of operation cannot be attained under manual operation."

The studies show that canal operation can be considerably improved by adopting the SCADA model because of the advantages of real-time information on water levels and flow rates and the possibility to respond to fluctuations more accurately and timely.

Centralized automatic control systems

A number of problems have been reported with centralized automatic control making use of control logics. Although a number of different control algorithms and automatic devices have been used, the desired results have not always been attained. Balogum states that some projects have failed and even led to unstable control because of the failure to take into account the dynamic properties of canal systems. Some specialists do not recommend to use these models for real-time control but to use them for defining operational procedures. Several efforts were made in Alberta, Canada, to achieve downstream control; but to date none has been satisfactory due to control algorithm limitations (Ring). As stated earlier, this method should be used for the most complex systems because of the complexity of design and equipment and the skill level required which is not always available in either developed or developing countries.

An emerging approach to the modernization of existing systems is the basic low-cost, incremental SCADA model. This model is commonly adopted by the irrigation districts in the United States for the modernization of old manually gated systems. Based on financing availability, four or five sites are equipped every year. For example, the proposed SCADA system for the Yuma irrigation district serving about 4 500 hectares provides for the automation of 20 sites, including the head of main canal, laterals, waste way and supply wells, which have been given five levels of priority. The total cost estimate is about US$350 000.

As discussed earlier, some developing countries have adopted or developed modern design standards, but very few have converted from conventional to modern design standards for the modernization of existing schemes. One remarkable exception is the modernization of the irrigation system in the Nile River delta, which is at the cutting edge of the technology by attempting to solve the problem of night storage within the secondary canals, a solution imposed by the constraints on land availability for the creation of farm reservoirs. Another example is the modernization of irrigation in the Jordan Valley, where the originally manually operated main canal has been converted to dynamic regulation and the canal distribution to pipelines. In Asia, modern designs are being used for the construction of the High-level Pehur canal in Pakistan, for the Narmada main canal and for the GAB project in Turkey. After a long period of failed attempts to install automated systems, Taiwan has now successfully placed most of its large irrigation districts under basic SCADA control. Malaysia has commissioned a study on "Modernization of irrigation water management systems in granary areas of Peninsular Malaysia". The 20-year-old remote monitoring system of the MUDA project has been recently converted to a SCADA system.

XI. PARAMETERS INFLUENCING PLANNING AND DESIGN OF IRRIGATION PROJECTS

The design of a project configuration and the selection of an irrigation strategy and of control equipment to meet that strategy depend on a number of physical, social, managerial and economic considerations. Not all of these are of equal importance in different projects, but all should be considered during the planning stage.

Water resources

Water resources and their variability are the critical elements in the determination of the irrigable areas. Studies on the balance of water supply and demand form the main part of the conventional feasibility studies of irrigation projects. Simulation model programming techniques have made it possible to examine various alternative solutions for different cropping patterns and seasonal or multi-year storage considerations.

The less reliable the water supply, the less feasible it is to adopt a water control strategy to meet precise crop irrigation requirements. There is little need for precise flow and water-level control. It is for this very reason that most traditional run-of-the-river projects have a proportional control strategy. The main objective is the equitable distribution of diverted natural flows that respond rapidly to the local variations of rainfall. The adoption of another strategy is doomed to fail. A case in point is the modernization of the subak projects in Bali, Indonesia. Farmers or groups of farmers can, however, improve the dependability of available water by constructing farm reservoirs, as was done in China before the construction of large reservoirs, or by tapping groundwater.

Groundwater resources

The development of groundwater in surface irrigation projects is a relatively recent phenomenon. The original objective may have been the mobilization of additional water resources, particularly in projects where canal systems were designed for extensive irrigation. However, the prevailing incentive for farmers to develop groundwater may now be the flexibility and reliability of that resource. Groundwater now accounts for about 40 percent or more of the total irrigation resources available in some originally water-surface projects. The contribution of groundwater has changed the overall game plan. Irrigation strategies adopted in the past to support a drought protection policy should be reassessed. In some cases, use of groundwater can be limited to drought years, if the canal capacities and surface resources can meet the requirements in normal years (United States). In other cases, such as in the Indus basin in Pakistan, conjunctive use is needed year round to satisfy the intensification of irrigation. The poor quality of irrigation service in the middle and lower sections of these systems frequently requires the use of groundwater for precise irrigation for high-value crops, and even for pre-germination rice-seeding techniques.

The two positive and obvious effects of groundwater exploitation that have led to its rapid development are that it is an easy means to get access to a huge extra resource and, when developed privately, it provides the ideal flexible water delivery service. At the same time, these developments have had negative side effects that have important consequences for the future planning of irrigation development:

• The development and use of groundwater as an additional source of irrigation water has often served to indemnify the irrigation agencies responsible for canals systems from bad or poor water deliveries, effectively taking away the incentives and need to improve their systems and service;

• The exploitation of groundwater has often escaped the regulation measures of allocation and scheduling, thus facilitating its overexploitation. This situation requires an integrated approach in the planning and design of future irrigation developments.

How the integrated management of both ground and surface water resources can best be integrated in the design of the delivery and conveyance system still remains a challenge. The recent increased attention paid to integrated management of surface and ground water at the river basin level has also brought about new concepts of irrigation management. The aquifer can be regarded as a reservoir that can be refilled with surface water. Some imaginative solutions could also be developed to improve the performance of the old-fashioned surface systems designed for protective irrigation. These systems could deliver multiple services ranging from a basic proportional delivery to a highly flexible demand delivery, for example if operated under the refusal operation mode. In surface projects designed for meeting full crop requirements, groundwater could be saved for dry years if the farmers are satisfied with the quality of service and particularly the flexibility offered by the surface system.

The strategy to be adopted for conjunctive use of surface and ground water strongly depends on the quality of groundwater. If ground-water is brackish or saline, one way is to alternate its application with application of surface water of better quality; another way is to blend good-quality water with brackish water in order to extend the water supply, gaining quantity at the expense of quality (Hillel). The choice depends on the tolerance of crops to brackish water, the degree of salinity of groundwater and the type of soil. The cyclic strategy allows the soil to be flushed from time to time.

Hillel points out the great importance of the frequency of irrigation in salinity management. If irrigation is applied frequently, the concentration of salts in the soil solution is maintained at a level close to that of the applied water, and the progressive build up of salinity is prevented, which points to the need for modern irrigation at farm level.

Silt load

The problems and challenges associated with silt-laden water in irrigation are frequently poorly understood, and underestimated, by irrigation specialists. There is an inherent conflict between flexible delivery operation and maintenance costs of schemes with a high sediment load. Flexible delivery results in unsteady flow conditions and occasionally low flow velocities, thus increasing the risk of siltation of canals. Unstable channels put enormous strains on maintenance and undermine the operation of canals. The problem of silt management was underestimated in designing the Chasma Right Bank canal in Pakistan, possibly because of optimistic assumptions on the silt trap effect of the upstream reservoir. As indicated earlier, considerable studies using computer simulation models were later carried out by IWMI to determine how to manage silt and operation of this canal at less than full supply. Techniques and management procedures to reduce substantially the silt load should be developed.

Rainfall

The variability and intensity of rainfall require flexibility in the operation of irrigation systems to achieve overall efficiency. The system should be able to respond quickly to a sudden fall in demand of irrigation water. Operation of irrigation schemes in arid regions is usually easier because smaller variations in demand require fewer provisions to regulate unsteady flows. In humid areas, the system should be able to satisfy the total evapotranspiration requirements in case of dry spells during the wet season.

It is still normal practice to use the concept of excess probability of rainfall in the calculations of crop requirements, as recommended in FAO Paper No 24. A more realistic method would be to use the actual data rainfall for each 10-day period of the entire period. The conventional method has the disadvantage of smoothing out the deficits of water during the drought periods. Irrigation departments in China use the more precise method of actual rainfall for each period. The capacity of computer modelling is no longer an obstacle to the wide adoption of that method.

The underestimation of water deficits due to the use of rainfall probability for the calculations of canal capacity contributes to the discontent of farmers in humid areas, as they try desperately to save their crops during dry spells.

Soil conditions

Differences in soil conditions influence on-farm irrigation requirements. Crop evapotranspiration is equal in well-managed fields on sandy and clay soils. There are major differences between these soils, however, with regard to optimum frequency flow rate and duration. Sandy and clay soils have different holding capacities and water intake rates. Coarse soils have a low water-holding capacity and a high intake rate. Fine soils have a high water-holding capacity and a low intake rate. Sandy soils must be irrigated frequently. No single irrigation schedule (rate, duration and interval) is optimal for all soil types. Recognition of the importance of customizing water deliveries to different soil types is a main reason why modern irrigation schemes provide as much flexibility in water delivery as possible, rather than forcing the users to adapt to a rotation with a specific flow rate, duration and frequency.

Crop diversification

There are inherent differences among crops in relation to needs for water at particular growth stages, root depths, optimal frequency of irrigation, and drought resistance. For example, grain crops are very sensitive to water stress during the critical periods of pollination and milking of grains. Vegetables are particularly sensitive to water stress because of shallow root zones. Water supply must be very reliable to meet the quality requirements of high-value markets.

Rice cultivation has particular water requirements: high flow rates are required during land preparation. Low design capacity of canals increases the time of land preparation and imposes its staggering over too long a period to benefit from the optimal use of rainfall. During the growth period, most farmers prefer the traditional method of continuous supply of water, which enables them to maintain the desired level in the paddy fields to reduce weed problems. Attempts to introduce rotational irrigation in rice schemes have failed in many countries, such as Thailand and the southern states of India, because of farmer resistance. By contrast, water-saving irrigation (WSI) techniques are practised in many projects in Southern China where distribution of water from terminal reservoirs is fairly reliable. WSI techniques involve maintaining a very thin water layer in the field and alternate wetting and drying. Bhuiyan observes that "WSI techniques as those applied in China, however, require a high degree of management control and infrastructure at both the farm and system levels. For much of developing Asia, management capacity to implement such strategy does not exist. More supervision and labour are required. Adoption may also be hampered by farmers' concern about not having access to water when they need it because of lack of reliability in the system water supply performance".

The very high capacity of irrigation canals for paddy irrigation in Western Africa, over 3 l/s/ha in some cases, where high yields are obtained, contrasts with the limited flows used for the design of projects in South Asia - for instance, the Sunsari-Morang project in Nepal, where main canals were designed on the basis of Indian standards. At the end of the growing season, conflicts arise between farmers who still require irrigation water and those who are ready to harvest. There is no management solution to this situation in a structure design scheme in the absence of drains. Undersized capacity of canals imposes the staggering of cultivation over too long a period. It is not only a question of management capacity as argued by Bhuyian.

Existing infrastructure

To a large extent, the layout, original design criteria and standards used for an irrigation project limit the options for its rehabilitation and modernization. The slope of the main canals, if too steep (over 15/20 cm per kilometre), determines whether or not operation can be converted from upstream to downstream control. The relative design capacity also is a major constraint, unless remodelling of the canals is found to be economically viable. In extensive irrigation projects with the objective of spreading water thinly, the design capacity decreases from upstream to downstream since only seepage losses are considered. In responsive irrigation projects, the design capacity increases when moving downstream to accommodate the need for flexibility.

Considerable imagination is required of the designers to modernize the existing infrastructure of irrigation projects since in most cases there are severe constraints. Construction of collectors and buffer reservoirs, conversion to pressurized systems and modifications of cross regulators are some of the tools available for the modernization of the configuration and control technology.

Land tenure and consolidation

Modern principles of irrigation scheduling and water application methods require specific arrangements of farm boundaries. Ideally plots have to be arranged within a geometric grid, with the proper choice for their orientation and slope for the application of surface irrigation methods. Since the 1960s, the policy of the Ministry of Agriculture in Morocco has been to systematically consolidate irrigable lands before the construction of any modern irrigation system. Design of the distribution system layout and of the blocks within which consolidated plots will be arranged is the first step in the design process. Ideally each plot should have a direct outlet from/to the irrigation and drainage system.

Basically, two land consolidation models have been tested and later adopted in a few developing countries. These two models reflect two different ideologies of irrigated agriculture: planned economy or liberalism. Under the former model, the irrigation blocks are divided into equal crop strips, crossing the farm boundaries, for-semi-collective farming activities. Governments largely impose the crop-ping pattern. This model is used, for example, in the Gezira project in Sudan, in most of the modern systems in Morocco and in some smallholder systems in Zimbabwe. Irrigation in principle should be organized by crop and not by farm. Under the latter model, the farm plots are arranged so that each farm has individual access to the tertiary canal. Distribution of water is organized by turns of individual farms, not by crop strips.

Figure 7 Land consolidation model in an interventionist agricultural economy

Note: Each crop has an individual turnout

Figure 8 A land consolidation model in a liberal agricultural economy

Note: Each farm has one or more individual turnout

Management and technical capability: The field reality

The operational performance of irrigation systems is influenced by the capacity of the management agency to apply the operational rules defined by the designer. Central scheduling used in most Asian countries is apparently simple since there is no need for farmer input. However, increasingly complicated operational procedures have been developed to determine the irrigation requirements as accurately as possible. Voluminous operation manuals have been compiled in which lengthy stepwise procedures are given to arrive at operational schedules. Horst observed that "these procedures require an enormous amount of data collection and processing. Shortage of staff, in combination with little contact or feedback from the field and in-sufficient or unreliable water measurements because of malfunctioning structures, often results in a situation in which the administrative activities remain largely paper exercises with little relevance outside the office".

The difficulty of managing manually operated, fully gated irrigation systems has already been discussed. If the strategy of water delivery is to closely follow the irrigation requirements, frequent adjustments of control structures are necessary. Horst again cautions about that design: "When combined with hydraulically unstable canal systems with structures cumbersome to operate, the often poorly trained field staff are confronted with an operational task which is effectively impossible."

Friendly user design is a concern that seems not to have permeated the irrigation industry, as it has the car or computer industry. In the irrigation sector, designers have rarely operated the systems they have designed. They have not confronted the operational reality at field level, the poorly trained operators, the poor road communications, and the harsh and changing climatic conditions.

Economics and maintenance costs

The IPTRID issue paper entitled "Realizing the value of irrigation system maintenance" provides innovative thinking on the consequences of neglect of maintenance and the issue of low investment cost and high maintenance costs versus higher capital cost and less maintenance: "Conventional economics, using a high discount rate for future costs and benefits, fails to show the importance of maintenance in sustaining the life of a system and the livelihood of farmers. Since costs and benefits occurring in the future are highly discounted, little benefit is apparently to be derived from extending the life of a new project beyond 10-15 years. The result of this thinking is that a project with a low initial cost, which deteriorates quickly and is dependent for continued survival on timely and properly funded maintenance, is preferred to one that is constructed to need less maintenance because it appears cheaper. Yet for a farmer and also for a nation it is important that a scheme endures."

Designs for low maintenance giving easier and less costly management, reduced maintenance, lower service fees and greater effective life need to be given more importance during project preparation. In the future, systems will be operated and maintained by farmers. Engineers and planners need robust information on the links between design, maintenance spending, performance, whole life cost and sustainability.

The above considerations are particularly applicable to the lining of irrigation canals, which often deteriorates very quickly because of the poor quality of construction of rigid canals.

Institutional setup

Management of a relatively large system is generally divided between various units. The locations of the interfaces between these levels have a great influence on the way the system is operated and its hydraulic performance. In many cases, the irrigable area is divided in geographic areas from upstream to downstream and includes both the main and the distribution systems (Thailand). That approach has many disadvantages since there is not a single unit responsible for the main system, which forms a continuous hydraulic unit. If management of the main system has to be divided between units, the interface should be located at hydraulic "breakdowns" such as reservoirs. This principle is widely used in South China, where the provincial, counties and village water bureaus share the responsibilities of management.

Modern irrigation systems consist of several levels, with clearly defined interfaces. Each level is responsible to provide any agreed service to the next lower level. The trend is to transfer the management of large sections of irrigation systems to large user associations, such as in Turkey and Mexico. Precise, yet user-friendly, control of flows and measurement of volumes at the interfaces between management levels are needed.

Operational capability of irrigation agencies and user associations

It has been stated earlier that the performance of irrigation systems is influenced by the capacity of management agencies and of user associations, if any, to apply the operational rules defined by the designer. This statement can also be turned around: the performance of an irrigation system is influenced by the capacity of the designer to design systems that conform to the operational capacity of the management agency. Both statements have some value.

The designer should identify the current restrictions in the institutional setup before incorporating concrete changes in the procedures and rules of water allocation, scheduling and delivery. In some cases, designers introduce changes in their designs that have too far-reaching consequences on the institutional and operational setup. These changes may not conform to the expectations of the agency or of the water associations. The rejection of the conversion of the traditional systems in Bali from proportional division to adjustable water allocation, which was implemented without consulting the users, was discussed earlier.

Changes in the managerial skills of an irrigation agency could be part of a planned modernization programme, however. For example in the State of Victoria, Australia, the modernization programme was accompanied by a change in the workforce. The job of the newly created positions of planners was more complex than that of the original water-masters and commanded higher pay. On the other hand, the job of field operators required less skill since it did not include the planning and customer contact roles of the original water-masters. A new pay and career structure was designed for the operation staff. The new planners were recruited using a competitive selection process. Improved productivity of the workforce was a major benefit gained from the introduction of centralized communication and planning. Most of the user associations in Mexico have turned down the staff from the irrigation agency at the time of the turnover and recruited younger staff with better academic skills.

Pricing and water allocation strategy

To improve the economic efficiency of water use, donor agencies have encouraged borrowing countries to adopt reforms in institutions and policy. These reforms include the establishment of water rights and trade of these rights, and the pricing of water on a volumetric basis. The design of irrigation projects should take these reforms into consideration. A rigid system with fixed distribution structures is not compatible with water trading, since physical modifications of dividing structures may be needed for each transaction. In adjustable systems, measuring devices are needed at each bifurcation at the level water trading is expected. For example, if water trading is encouraged between user associations along a branch canal, measuring devices are needed at the outlets serving each association.

Capacity of the construction industry

Failure to meet construction standards is a common factor in some projects being unable to fulfil their specific function. This is particularly true for the modern irrigation systems making extensive use of hydraulic regulation devices, such as level-control automatic gates, modular distributors and long-crested weirs. This modern design is very sensitive to problems with improper installation. Baffle distributors provide constant flow rates within very strict fluctuations of the upstream water levels and do not tolerate downstream-submerged conditions. Such design requires perfect vertical settings of all the control structures. The lack of awareness of the requirements of high-quality construction standards in some countries has contributed to the failure of some pilot projects. It may justify the resistance of some specialists to adopt or recommend that technology.

Projects making use of conventional sliding gates are much less affected by inaccurate installation of the control equipment. However, they are more difficult to operate and have to be associated with measuring devices and/or local controllers.

XII. THE CHALLENGE OF MODERNIZATION: OPTIONS AND PROCEDURES

Irrigation project design and management are very complex and each project has different constraints. Designers and institutional reformers must have a very comprehensive understanding of these constraints and options in order to make the proper choices for modernization. The modernization phase of formulating an appropriate operational plan and water control strategy, in which deliberate choices are made on the water allocation and scheduling principles to be adopted, is a crucial phase that requires special multidisciplinary skills. Apart from a sound and up-to-date knowledge of hydraulic control concepts and technology, it also requires management and institutional skills. This section summarizes some of the points discussed earlier and proposes some guiding principles for selecting control strategy and control equipment.

Infrastructure versus management inputs

It has been argued that a certain amount of substitution may exist between staffing requirements and hydraulic control, depending on the type of service that must be provided with a given hydraulic infrastructure. Mutua (2001) recognizes that the ability to substitute hardware for management inputs can only take place within certain limits. As mentioned earlier, the type of infrastructure may have inherent physical limitations in the way it can be operated.

The trade-off between infrastructure investment and management to meet a certain level of service is based on the assumption that that level has been defined as part of a comprehensive strategic plan for the agency. It is founded on the principle that a training programme will be in place to support the strategies designed to achieve the strategic goals and guarantee that the staff will acquire the appropriate skills to operate and maintain the types of infrastructure chosen for flow control (Malano).

Step-by-step versus full-fledged modernization

Some modernization projects aim at "perfect" post-project operations that require huge behavioural changes from both agency personnel and water associations. Other specialists argue that the modernization of existing schemes can and in practice should be more often a phased, step-by-step process, in which better operational management (both in efficiency as in service rendered) is achieved in a series of concrete steps of incremental water control and changes in water scheduling. A step-by-step approach might often prove the only feasible option in difficult contexts where the options to come on the desirable operational plan and water control strategy are limited by an institutional resistance to change by the irrigation agency and/or institutional reforms where newly created water user associations and federations have to take up operational tasks for which they have had little experience. The dilemma of irrigation development will always remain to strike the right balance between the operational capacity of the prevailing management institutions and the technological and hydraulic configuration of the system. A proper and successful irrigation modernization programme will thus strive to reach a coordinated and congruent development of both the institutional capacity and the technology. In order to achieve this, it is imperative that the first phase of defining an agreed-upon operational plan and water control strategy be conducted with utmost care and consideration. (Personal communication from G. van Halsema)

Experts from the Bureau of Reclamation (Stringam) who stated that canal control systems should be designed with operations and operators in mind also share these views. Operating personnel must be comfortable with any changes. This objective is attained by tailoring the control system to the needs and understanding of operators and by following an evolutionary installation procedure. Great care should be taken to match equipment and control software to the needs, understanding and management practices of the project operators. Designers will be further ahead if they begin with a simple system. To support these views, these experts from the Bureau of Reclamation note that "managers and governing boards of user associations at many irrigation projects hesitate to adopt high-tech control systems because of bad experience and sometimes lack of understanding. In some cases, unfavourable experiences have resulted because designers have not been sensitive to the needs of understanding and experience of canal operating personnel. Control systems were designed and installed without adequate instructions to the operators, and the systems did not perform well because operators did not understand how to use them. On these projects, managers may view automatic controls as a waste of time and money, and they will be hesitant to allow any future automation development".

This paper does not challenge the views that the modernization of existing systems should be an evolutionary process. This is particularly valid if modernization plans are aiming at the adoption of technologies that have been developed during the last three decades in parallel with the development of electronics, such as high-tech sensors, telecommunications, supervisory control, computers and flow simulation models and automated controllers. However, some improvements can be achieved with simple one-time modifications of the hydraulic infrastructure, such as the construction of long-crested weir sections at gated cross regulators, flow limiters, compensation reservoirs and interceptor canals, the conversion of surface systems to pressurized pipe systems, and possibly the installation of modular distributors, if the hydraulic conditions of installation and quality of construction are appropriate. All these techniques have been used in a number of countries and quickly understood by the operators and farmers. The rate of implementation of these technical improvements is often limited by financial resources, not by the management capability of the operators. Wherever some of these new structures, such as long-crested weirs, have been installed, farmers have expressed their appreciation by requesting additional structures. Obviously the irrigation projects built on wrong design, such as those using the wrong combination of overflow and underflow structures, should be modified as fast as possible.

An important conclusion from this discussion is the crucial importance to have global strategic planning for modernization from the beginning, whether the modernization is incremental or not.

The role of water users' associations in the modernization process

In general, only large well-established water user associations have the staff and resources to conceptualize modernization plans. The role of small associations is inevitably limited to the definition of the type of service they need from the water provider, either a government agency or a higher level of farmer association. Generally the members of small associations do not have a concept of modernization and are primarily concerned about just having reliable and equitable service. Farmers who are struggling with their present-day problems cannot have a vision of what is needed to support their future needs. The interest and (partial) funding for modernization will have to come first from government organizations, which will have to stimulate the interest of large associations. Even in the United States, although the irrigation districts have been managed by user associations for decades, modernization is still a relatively new concept. It may take considerable time and efforts to have water user associations in developing countries very active in modernization conceptually and financially. The creation of a national federation of water user associations, as in Colombia and Mexico, may accelerate the process.

Financing of rehabilitation programmes: A few examples

In practice, the financing of rehabilitation programmes may be the determinant factor in the rate of implementation. The irrigation districts in Alberta, Canada, have benefited since 1969 from cost-sharing rehabilitation programmes which have allowed that province to develop a world-class water distribution system.

As a rule, irrigation districts in the western United States have to use their own financial resources to modernize their irrigation systems. In some cases, however, modernization is financed by other organizations, which have direct interest in water saving programmes. For example, the Metropolitan Water District of Southern California currently finances a huge modernization of the Imperial Irrigation district. Modernization of the Grand Junction district is financed by fish interest groups. No doubt the irrigation districts in the United States are very cautious about the pace of modernization and the adoption of sophisticated and costly equipment. In most countries, modernization works are financed by the government, with a very modest contribution by the farmers. This is the case of the Mula irrigation scheme in Spain. The century-old canal infrastructure was replaced by pressurized systems. The innovative management initiative associated with the physical modernization is the creation of an individual water account.

A modernization by substitution of the 5 500-ha Lugan Sur project in the province of Mendoza, Argentina, is under consideration. Even the main canal will be replaced because of the relocation of the intake on the Mendoza River. Although the project is designed to boost high-quality wine export, there is no clear decision on the contribution of the users to this project, which costs about US$3 000 per hectare, excluding on-farm water-saving techniques.

There are very few cases in developing countries where the farmers have been involved in the planning, design and financing of modernization of their irrigation infrastructure. The business-oriented water user associations in Mexico have progressively matured since their creation about ten years ago. Their first efforts for improving performance were concentrated on acquisition of office equipment and special light maintenance equipment. Efforts to improve the delivery infrastructure were slow until 1997 when an agreement on a 50/50 cost-sharing programme was signed between the National Water Commission (CNA) and ANUR, the national association of irrigation water users. Modernization activities consist in the substitution of canals by low-pressure pipelines at secondary and farm levels, construction of long-crested weirs in secondary canals, subsurface drainage, agro-meteorological equipment, broad-crested weirs and electronic flow meters. Efforts to install automatic controls in canals have failed for a number of reasons, including the use of tailor equipment.

Specific design issues

Two specific design issues that are neglected during the design of irrigation projects are discussed here because of their important effect on the performance of irrigation projects.

Hierarchy of canals

The layout of the canal network should be designed so as to be integrated with not only the roads and drainage system but also multilevel management, whether from an agency or user associations.

Figure 9 An irrigation project with a well-established hierarchy of canals

Note: The setting-up of a multi-tiered organization of water user associations, combined or not with higher level management by an irrigation agency, will naturally be derived from the layout of the canals

In many existing schemes in South Asia, the hierarchy of canals is not adequate for the creation of multi-tiered user associations. As much as 40 percent of the irrigable area in some projects is served through direct outlets from the branch distributary and even the main canals. It is unlikely that a water user association consisting of user groups with individual points of supply would be very active since the groups have little common interest. The construction of a new canal parallel to the parent canal and serving all the direct outlets may be a technical solution to correct the deficiencies of the original layout in relation to the social and managerial aspects. However, the farmers served by direct outlets may strongly object to that solution which would deprive them of their privileged situation.

Figure 10 A typical irrigation system with a loose hierarchy of canals and a high proportion of direct outlets

Note: The figure shows several options for the creation of water user associations above the 50-150 ha block level

If management of the system is divided between units, the interface should be located preferably at hydraulic breakdowns, such as reservoirs. This principle is widely applied in mid and southern China, where most systems consist of a large reservoir connected to many medium-sized and small reservoirs supplying thousands of village and farm reservoirs.

Figure 11 Typical configuration of an irrigation project in mid and southern China (known as "melon on the vine")

Note: Medium-sized and small reservoirs are natural interfaces between different levels of management: province, counties and villages

Control of seepage losses

Seepage losses from canal irrigation systems may reduce substantially the volumes of water to be delivered at farm level. The losses from the field channels in the large alluvial plains of Asia are reported to account for about 40 percent of the volumes of water supplied to these channels.

Lining of irrigation canals is a very costly exercise and may increase the total cost of a project by 30 to 40 percent compared to the cost of an unlined system. The decision whether or not to line irrigation canals on the basis of economic considerations should therefore be a very important part of the design process. It is puzzling that not much thought has been given by designers to the robustness of the conventional rigid lining techniques routinely adopted during design. Most disturbing is the doubt about the actual amount of water saved by canal lining despite considerable investments for canal lining and the ambitious rehabilitation programmes in many countries. The reduction of seepage losses is generally assumed to be constant for the expected life of a project to have a chance of achieving a favourable economic return, an assumption generally unrealistic.

Projects with extensively cracked concrete lining abound throughout the world. The effectiveness of these linings to reduce seepage losses is very low. Even lining with still good appearance may not be an effective seepage barrier because the small cracks or deficient joints form water flow paths to the entire canal prism, which considerably reduce the effectiveness of the lining. Saturation of the soils after commissioning of a canal causes some settlement of the soil mass resulting in wide gaps underneath the slabs. The rigid slabs eventually rest on a limited number of supports resulting in further cracks. The gaps provide opportunities for seepage loss under most of the lining.

The use of rigid materials (cast-in-situ concrete, pre-cast concrete panels or bricks) is still the most common technique of canal lining in developing countries. The causes of the ineffectiveness of rigid canal lining are linked to the poor quality of construction, particularly the compaction of the sub-base and the placing of concrete, and to the inadequate operation of the canals with too rapid drawdown and frequent dewatering. The adoption of design standards of irrigation agencies from some developed countries recommending thin concrete lining, which is adequate in case of high standards of construction using modern canal lining machinery and of strict respect of operation rules, is also the cause of the problem. Designers should consider the difficulty of canal lining in humid conditions, using rustic methods of concrete placing. Increasing the thickness of canal lining by about 30 percent may be a realistic but costly solution.

Another option is to adopt the use of flexible geo-membranes protected with a rigid lining or with loose materials. This technique was used first in Bureau of Reclamation projects and in some projects in the Middle East with very difficult soils or gypsum soils. If geo-membranes consist of stable long-lasting material and are well installed, seepage rates from canals are negligible. Despite the accumulated evidence of the short life of the conventional lining techniques in many projects, adoption of geo-membranes in irrigation canal lining is still very slow. The Fordwah-Sadiqia project in Pakistan and the Tarim project in western China are two remarkable exceptions. The former included a very intensive research component on estimation of canal seepage losses before and after lining and an evaluation of different construction techniques. The latter, involving the lining of over 400 kilometres of canals, with the use of over 5 million m² of geo-membranes, is the largest construction work ever using that technique.

A managerial option to reduce seepage losses is to operate the canals by turns, and to close them during the low season. That management approach has obvious disadvantages for the irrigators.

The economic considerations on the issue of capital investments versus maintenance costs and the short life of low capital investments discussed above are particularly valid for canal lining. Before deciding on a costly lining programme, it is important to use realistic values for the reduction of seepage rates, life of lining and maintenance costs. A comparative evaluation of earlier lining programmes in the same region should be carefully evaluated through intensive measurement of seepage losses of unlined and lined canals several years after construction.

XIII. A PROCESS FOR REVISING DESIGN PROCEDURES AND STANDARDS

Assessing the need for changes: The use of internal indicators

Very few developing countries have adopted the full spectrum of modern irrigation concepts and standards. In a few cases, the design makes use of the most advanced technology for water control but the water distribution strategy lacks the flexibility required for a service-oriented delivery (for example in Jordan and Morocco). In other cases, the technology is inadequate to satisfy the stated objective of modern irrigation. This is the case of the fully gated, manually operated systems, which were designed to meet the irrigation requirements but are far too complicated or too costly to operate, especially during the wet season and in the humid tropics. Many examples are found throughout the world. More frequently, neither the technology nor the strategy meets the definition of modern design. That category includes the projects with faulty designs, and operational procedures designed for the convenience of the operators, not of the users.

Most countries would have to assess the needs for changing their approaches to irrigation development. It would be useful to develop a kind of checklist to assist irrigation agencies in this diagnosis with questions falling under the two standard categories:

• Are we doing the right thing?

• Are we doing the thing right?

Questions to be addressed could include:

• Realism of original assumptions for the design of existing schemes

• Changes in water demand

• Layout of the farms: direct access of each farm to irrigation and drainage farm

• Reliability and flexibility of the service provided

• Is irrigation scheduling imposed by the agency or based on arranged demand?

• Days for advance request

• Can water be shut without notice? Possibility of flow rate changes

• Possible options of on-farm water application for the farmers

• Hierarchy of canals

• Compatibility of the layout of the canal system with the organization of multi-tiered user associations with well-defined hydraulic boundaries

• Water measurement at each level of management

• Degree of independence of operation at each level (reservoirs or other means)

• Frequency of changes in settings of devices required to meet set targets such as water levels or flows

• Sensitivity of the control structures used for cross regulators and off-takes

• Simplicity of operation of the control devices

• Interaction between control structures and hydraulic stability

• Feasibility of volumetric water pricing (robustness of volumetric devices, stability of flow conditions)

• Feasibility of water trading at different levels

• Quality assurance of construction

• Capability of maintenance of sophisticated equipment (telemetry, remote control and monitoring)

All these questions were considered in the 1996 World Bank research study on performance of 16 irrigation projects. These 16 projects had some elements of modernization, and represented a variety of design concepts, climates, crops and water supply conditions. That study broke new grounds in project evaluation by developing a new framework for assessing systematically the internal process of irrigation projects. The research includes two features: a rapid appraisal process, and the development of a number of internal indicators. The internal indices provide a systematic rating of hard-ware, management and service throughout the entire system. Internal indicators examine the mechanism of water control and allocation at all levels of the project. The complete picture enables the evaluator to visualize where changes are needed, and what impact the changes would have at various levels. That approach was developed to answer the question: what specific actions need to be taken so that benefits can justify the investment in rehabilitation and modernization?^[20] None of the external indicators which compare inputs and outputs of irrigation projects provides specific information on what needs to be corrected to improve performance. They do not provide insight regarding the working of the internal mechanisms within an irrigation project, be it management, social or hardware.

The internal indicators provide a new evaluation tool. They can serve as a valuable training and diagnostic tool for modifying the internal hardware and operation of irrigation projects (Burt).

The rapid appraisal process developed for that research differs substantially from the walk-through techniques with farmer representatives that aim at the identification of the worst spots such as poorly located or badly damaged farm outlets.

Development of new design procedures

In the preparation of new design procedures and standards, an important distinction is to be made between the design of new projects and the rehabilitation and modernization of existing ones. The modernization of existing projects is a complex compromising process between the objective of making use of existing facilities and minimizing the costs on one side and the objective of improving performance and changing the quality of service on the other. The solution adopted will depend on the potential productivity of the project and on the government policy on the development of water resources. Development of a modernization plan requires an assessment of the service currently provided by the system, the impact of the current service on farmer irrigation management and the potential benefit that would result from improved service. Successful modernization requires the adoption of a service attitude, as well as comprehensive strategies for design and operation of water distribution facilities at all levels of a project.

A number of technical documents have been published during the last decade to stimulate and promote the development and application of modernization to irrigation projects, be they new projects under design or projects under operation. Three documents that are intended to serve as guidelines to owners, designers, operators and users are of particular interest^[21]:

• The ICID publication on automation of irrigation systems, which deals with the concepts and the various automatic control logics and describes the different types of equipment, including passive regulators, hydro-mechanical equipment, local automatic controllers and hardware for remote monitoring and control.

• The U.S. Bureau of Reclamation Manual on canal system automation also deals with the control fundamentals, methods and algorithms. The automatic control of irrigation systems in the United States is based on the development of the electronic industry, which includes the use of electronic components, usually consisting of microprocessor-type equipment (microcomputers and software, keyboards, digital displays, master stations and control centres) and communication equipment.

• The publication of the IHE in the Netherlands on flow control in irrigation systems discusses in detail the various delivery strategies and canal control systems, canal capacities, and other hydraulic structures such as stilling basins, drops and chutes.

Revision of design standards

Most irrigation agencies should proceed with a diagnosis of the concepts and design standards of existing projects and of their current standards to assess whether they could meet the requirements of present and future operation and maintenance given the forces of change that will affect the irrigated sector. To meet the conditions of the future, water delivery from irrigation projects should be more flexible and reliable. Operation rules should be transparent and understood by the users. Volumetric measurement of water is a prerequisite to the establishment of water rights, water trading and water charges. Potential benefits should be attractive enough to the users to be able to recover at least the sustainable costs of irrigation, which include operation, maintenance and replacement costs. These objectives cannot be met without modern management of irrigation projects, including automation. Automation has the potential to yield agricultural benefits as well as environmental benefits: increased delivery flexibility and dependability will benefit farmers and canal operators while better management will reduce water waste and the deterioration of quality of water flowing to the drains and aquifers. Of greater priority are:

• The countries using water control strategy and equipment that are well known for their difficulty of operation in field conditions and their inefficiency, such as the manually adjustable gated systems and those designed for operating at full supply. The manual from the Bureau of Reclamation referred to above states that the operation of irrigation systems under conventional design is costly and inefficient. Even under continuous vigilance by operating personnel, the latter cannot eliminate undesirable variations in water levels, which adversely affect the water delivery rate and schedule.

• The countries using designs that are faulty from an engineering viewpoint, such as the wrong combination of control structures which amplify the variations in supply.

Strategies for dissemination

Revision of FAO guidelines: The 1996 FAO guidelines for the preparation of irrigation projects should be revised to provide better guidance on the questions of water delivery strategy, selection of control strategy and control equipment. Some chapters of this document should be incorporated in these guidelines.

Building up national capacity on irrigation modernization through workshops, seminars and training programmes: These programmes could use extensively the tools and findings of the World Bank research study (FAO Water Report No 19) referred to above. This innovative approach to modernization training has been successfully tested in Thailand and provides quick changes in the thinking of trained irrigation engineers. The complete picture provided by the evaluation of internal and external indicators enables the participants to visualize the deficiencies of the present systems and the type of changes that are needed.

Training of water users' associations: Simple training materials should be prepared for dissemination to the boards of directors of user associations and their technical management staff. The difficulty of operating a project 24 hours a day seven days a week under manual control should be highlighted. The objective of modernization should be clearly stated in terms of cost saving, better use of labour force, more efficient and flexible service to the users and overall increased productivity. The possibility of incremental modernization should be emphasized. The active participation of the users in the design and their provision of a consistent level of funding are two ingredients for the sustainability of the modernization of existing systems.

XIV. CONCLUSION: THE NEED FOR COUNTRY IRRIGATION POLICIES

The need to improve the performance of large-scale irrigation is compelling. Easy, low-cost solutions to mobilize additional resources and improve the quality of service, such as the development of groundwater resources underlying canal irrigation systems, have nearly reached their limits in many arid and semi-arid areas. Furthermore, these solutions are not sustainable when they lead to overexploitation of the groundwater resources. There is no choice but to modernize existing systems to provide the level of service required for reaching the level of productivity needed to meet the food production goals of the 2020s.

Few irrigation systems in developing countries and even in developed countries meet all the characteristics of service-oriented modern irrigation. The financial resources required are considerable. However, the main constraints to modernization may be the still widespread denial of the importance of technology in irrigation performance, the strong resistance to change of the irrigation community and the limited expertise available to promote and apply modern concepts and design of irrigation systems. These obstacles would not be overcome without recognition of the importance of technology in the performance of irrigation projects by all international and national organizations involved in the development of water resources (GWP, IWMI, IRRI, IFPRI, ICID, FAO and development banks) and the internalization of modern design concepts in water resource development/irrigation policies of donor agencies.

The new CGIAR Challenge Programme on Water in Agriculture and the dialogue consisting of the dialogue on water, food and the environment and of a comprehensive assessment of water management in agriculture, which would coordinate the efforts of all stakeholders, including farmers organizations, may be an opportunity to achieve the above two objectives.

The water resource development policy papers prepared by the World Bank and other financing institutions which do not address the specific issues of irrigated agriculture should be supplemented by a long-overdue irrigation policy. Although it is beyond the scope of this paper to provide an exhaustive list of the points to be addressed, the following issues relevant to the performance and sustainability of irrigation projects should be examined:

Objectives of irrigation: Donors should avoid supporting irrigation policies that have conflicting objectives. The objective of sustainability may not be compatible with the policy of benefiting the maximum number of farmers through extensive irrigation. If this were the case, the level of subsidies should be clearly established.

Economic evaluation: The method for the economic evaluation of irrigation projects, which have largely contributed to under-design and to the notorious over-optimism of evaluation of irrigation projects, should be reconsidered. This document proposes to use the same economic method as the one used for other infrastructure or service projects. As for the water supply where projects are evaluated on the value of the water sales, independently of the final use of water, irrigation projects should be evaluated on the sales of water sold for irrigation purposes, at a price acceptable to the users. This approach, revolutionary in the econometrics of rural development projects, would have several advantages. It would break the vicious tradition of designers from consulting firms or governments of under-designing in order to reduce costs and overestimating the irrigation benefits, including the cropping intensities, yields of crops and projected prices of agricultural commodities.

The present use of the economic rate of return for the evaluation of irrigation projects is in favour of low capital and high maintenance cost projects, an unrealistic approach in most countries. Further attention should be given to the life of the infrastructure and sustainability of irrigation projects. It was bad service to borrowing countries to finance the lining of irrigation canals that have little or no effect on seepage losses less than ten years after installation.

Revision of design concepts: The strong link between the irrigation technology and the performance of irrigation projects and the need for the revision of design standards and operational procedures to respond to the forces of change should be recognized. There is no other infrastructure sector that has given so little attention to the needs of the final users of water in terms of flexibility and reliability. Many systems have been sized for the average irrigation requirements of a predetermined cropping pattern without any provisions for change in crops, deviations from average rainfall, the reluctance of farmers to irrigate at night, and special requirements such as land preparation of paddy fields. No other infrastructure sector has given so little attention to the ease of management by the field operators.

Relation between physical improvements and institutional reforms: The role of the user associations in the modernization process of irrigation projects should be recognized. Any strategy for improving performance of the irrigation sector should consider the relationship that exists between the design and functions of user associations and the plans for a better level of service. Physical and organizational improvements are not isolated actions but part of a well-planned process.

Relation between policy reforms and technology: Finally, the irrigation policies should emphasize that policy reforms cannot be implemented without an appropriate physical environment. Public investments may be required to improve irrigation systems to provide better control and measurement of water delivery before volumetric pricing, the establishment of water rights and trade of these rights can be implemented. On the other hand, public policies may have to be implemented while modernizing large-scale irrigation systems, to encourage improvement in farm-level water management and to enhance agricultural productivity.

In many projects, the farmers have pointed out the need for a new approach by looking for other short-term sources of water, which are not necessarily environmentally sound. It is a challenge to the irrigation profession to respond to their signals.