(introduction...)

It is often stated that all problems with the small irrigator would vanish if he were simply given a regular, dependable, timely supply of water. Unfortunately the monsoon, principal source of water in the region under discussion, is neither entirely dependable nor always timely. Irrigation is conditioned by supply as well as by demand, and conflict between the two is to some extent unavoidable.

Major irrigation systems can be designed and operated with a formal supply system extending down to the individual farm, "formal" implying construction, operation and maintenance by government agency (e.g. departmentally). Alternatively, the formal system may stop some distance away from the individual farm, delivering to an area in which the farmers manage the distribution of water themselves. This may be as little as 30 or 40 ha, as much as 100 ha or more. However, due to the inefficiency of water distribution within that area, encountered in many such systems, the trend over the last two decades has been to reduce its size and to extend departmental activities to the design and, in some cases, the construction of the distribution system within it. Maintenance and operation of that system remain the responsibility of the group of farmers which it serves. For purposes of this discussion, the area served by the outlet from the formal supply system is referred to as the outlet command, and the distribution system within it is referred to as the tertiary system.

A number of factors have led to efforts in recent years to increase the role of cultivator groups in management of the lower end of the system. One is the major cost of staffing an organization capable of operating a formal system extending down to the individual farm, and the difficulty of recovering such cost from cultivators. Another is the conviction that the cultivators themselves are better capable of managing the system at that level. Finally, much faith is placed on the thesis that the tertiary system will be better maintained if the cultivators have full control of it. Group operation of the lower end of the system is not without its problems (discussed later) but in a smallholder situation, particularly in the absence of land consolidation, it is believed to be the practical solution. The remainder of this discussion is in that context.

There are two types of variable which must be considered in the design and operation of an irrigation system. The first is imposed by natural conditions, including variations in availability of water to the canal system, the occurrence and amount of rainfall on the project area, and variations in suitability of lands within the area for cultivation of particular types of crop. The second type of variable is inherent in the freedom of choice of the individual farmer, or group of farmers, regarding crops to be grown and hence the pattern of irrigation requirements. Such choices are often determined by changing market conditions for various crops. The site variables must be accommodated; the cultivator-related variables may be subject to some degree of regulation by project authority (Frederick 1993, Hoffman 1990).

The design of an irrigation system sets out to optimize the use of the water resource, taking into account the variable nature of the supply and demand as discussed above. Two basic issues are the degree of sophistication acceptable in the design and operation of the system, and the degree of freedom of choice in cropping pattern which is to be left to the cultivator.

A system capable of carrying optimization to the ultimate degree would be very sophisticated indeed, and probably inappropriate to the region under discussion. Limits in the practical level of technology in design and operation are imposed by several factors. These include cost in relation to productivity (cost effectiveness), financial resources available, the ability of the agency concerned to-construct, maintain, and operate the facilities including funding of operation and maintenance, and finally the prospective attitude of cultivators to the type of operation proposed. The latter can be a key factor. If an operation, although entirely logical from the system viewpoint, meets with cultivator disfavor it is likely to be subverted by interference with control structures, or construction of unofficial checks or outlets ("destructive self-help" as one commentator puts it). Vulnerability of structures to misoperation and susceptibility to theft of components may be a very real constraint on the degree of sophistication and on the practical limits of efficienct water use which can be aimed at in a particular project situation.

The question of freedom of choice of crop and consequently of timing of irrigation supply, versus project-wide standardization, is a fundamental one in the design of an irrigation system. Where soils and topography are uniform throughout a project area, and groundwater can be utilized by the individual cultivator who has special water needs, standardization of canal supply based upon the requirements of the most commonly grown crops is generally practiced in South Asia. On the other hand, where soils and topographic conditions vary to an extent compelling radically different classes of crop to be grown in different areas of a project, canal delivery tailored to these specific areas is desirable. However, where the variations in soil and topographic conditions commonly occur within small areas, as can be the case in some topographic situations, tailoring delivery to such areas becomes difficult.

Aside from situations in which site conditions impose variations in class of crop, there is the more general case of the cultivator who wishes to depart from the project norm and to grow specialty crops requiring special water scheduling, but is in an area where groundwater is not available. This brings up two questions: should the canal delivery system recognize such needs and should such diversification be encouraged? and what are the practical limits to meeting such demands?

Terms used in referring to the successive orders of the canal system differ regionally, causing some confusion in the literature. The following alternatives are in common use:

Irrigation Sub-System Conveyance System	Alternative Terms Primary canal and branches Secondary canal and branches	Main canal and branches Distributary canal and branches
	Sub-secondary	Minor (or tertiary)
Distribution system commonly serving	Tertiary (or watercourse)	Watercourse (or field channel)
30 to 40 ha	Quaternary (or field channel)
Within the individual holding	Farm Channel	Farm Channel

The capacity or area served by a particular order of conveyance canal varies widely with the size of the scheme. The area served by a tertiary (or watercourse), however, is more specifically defined, as discussed later. The terms employed herein are primary, secondary, sub-secondary, tertiary or watercourse, and field channel.

Designing for variable supply

Canal supply from a monsoon-fed source is inherently variable. A regulating storage, if such is available, can improve the situation considerably, but two basic questions remain. First, what proportion of the historic mean flow should the system be designed for, and second, how should it be operated in periods of less than the design flow. Greater system capacity means greater total utilization of water, but it also means more frequent operation at less than capacity, and more frequent deficiencies in supply to the cultivator. Both cause problems (Berkoff, 1987 and 1990).

Rainfall, in a monsoonal area, is also highly variable. Here again, a basic question is the proportion of the historic mean which should be taken into account in calculating irrigation needs. Where rainfall is substantial, but nevertheless insufficient to provide fully for paddy requirements, canal supply is complementary to rainfall. If too conservative a view of rainfall is taken, the necessary canal duty will be overestimated and the area which can be served by the available canal supply will be restricted. Conversely if rainfall projections are optimistic and canal duty is underestimated, periods in which rainfall deficiency cannot be entirely compensated by canal supply will occur too frequently. That some periods of deficiency in rainfall plus canal supply will occur is inevitable, particularly where rainfall in the catchment supplying the canal system and that in the irrigation area are subject to the same climatic conditions, hence when there is a rainfall deficiency in the project area there is also deficiency in the catchment. The relationship between the onset of monsoon rains in the supply catchment and in the project area is especially sensitive where canal supply is relied upon to provide a proportion of the water needs for puddling and transplanting of rice in the critical initial month of the monsoon. This may be a key factor determining the practicability of double-cropping of paddy versus single cropping, and the timing of planting of follow-on dry season crops.

The occurrence of deficiency in monsoon rainfall in rainfed paddy areas is a situation entirely familiar to cultivators. Up to a point such deficiencies can be accommodated. As previously noted, water consumption by paddy can be reduced by up to 20% by reverting from continuous flooding to intermittent flooding, without major loss of production, although at the expense of greater labor in weed control. Much paddy is traditionally grown in this fashion in rainfed areas. However, on introduction of canal supply, cultivators generally assume that problems of deficiency are in the past, and when that proves to be not entirely the case unofficial diversions and tampering with control structures, also headend versus tailend problems, must be anticipated. No solution to this problem has yet been found other than more intensive operational supervision, or hopefully more effective farmer-group management than has yet been generally achieved.

Where monsoon rainfall is inadequate for rainfed paddy, and other (dry-foot) crops are traditionally grown, there is a possibility that the advent of canal supply may bring about a radical change to monsoon-season paddy, resulting in substantial demand for water. However, assumptions that such a change will occur must be viewed with caution. For example, a clayloam soil believed to be suited to paddy under irrigation actually proved to be quite unsuitable for relatively obscure reasons (phosphorus deficiency under the conditions of wet-land paddy, and extremely sticky nature when wet). A design incorporating the use of the large "surplus" supply of water available in the monsoon season should begin with a careful analysis of agronomic factors which may influence such use. Either monsoon-season or dry-season irrigation may be found to be the determining factor in arriving at the capacity of the canal system. But in either case there will be periods in which supply or demand are considerably less than system capacity. The question of how the system is to be operated under such circumstances strongly influences its design, particularly the nature of the hydraulic controls. Consequently, planning of system operation must go hand-in-hand with, or precede, detailed engineering design.

Much ingenuity has been exercised in meeting periods of "scarcity" of supply, in both public and farmer-operated systems. Some measures taken include proportionate reduction in rate of supply or in frequency of supply throughout the service area or a selective sacrificial approach in which supply to a portion of the area is foregone for the whole season. Another measure is a system in which each secondary has a rotating schedule of weekly priorities (first, second, and third) which determines those secondaries which will have supply withheld in a particular week if supply to the system is in deficit. The latter system (traditional in north-western India) distributes adversity on a mathematically equitable basis and is easily understood by all concerned. However, it does not necessarily result in maximum production. An effort is being made to introduce other factors into the scheduling of priorities, but whether this more sophisticated (computer assisted) approach will be acceptable to cultivators is yet to be seen.

The most effective handling of "scarcity" is probably that observed in a village "tank" system, with a small reservoir and a service area of a few hundred hectares. During a period of deficiency in which the reservoir only partially fills, a village meeting may be convened, taking place on the slopes of the dam from which both the contents of the reservoir and the whole of the service area are visible. Questions of whether to reduce the rate of supply or to delete supply to a portion of the service area for the season, and whether to use the remaining storage to save the standing crop or to hold it over for the equally important purpose of pre-monsoon irrigation in the next year, are discussed, and decisions are made by consensus. A further level of ingenuity is displayed in some village schemes (e.g. the bethma system in Sri Lanka) in which cultivators with holdings in the lower end of the system are moved to the upper end during periods of scarcity, and share in cultivation of that area for the duration of the season, during which time irrigation supply is confined to that upper area.

However, expedients available to a small village scheme are not generally available to a major public system, where the cultivator is only aware of what goes on in the canal serving his area Generally, deficiency in supply in a major system results in illegal operation of control structures and illegal diversions. The net effect is that upstream irrigators secure their supply, and the deficiency is passed on to "tailenders".

The options available for distribution of water in times of reduced supply or demand are the following:

(a) To reduce the rate of flow in all channels;

(b) To reduce the rate of flow in the primary canal (still maintaining continuous flow in it) and to supply full flow in rotation to secondaries and their tertiaries;

(c) To reduce the rate of flow in both primary and secondary canals, and to supply full flow in rotation to tertiaries;

(d) Similar to (b) but instead of a fixed rotation to secondaries, priorities are rotated (the north-western Indian system discussed above);

(e) To delete supply to portions of the service area for the entire season;

(f) To supply on "limited demand".

The relative merits of these alternative systems is discussed subsequently.

Varying demand within the service area

As noted earlier, changing market conditions can bring about radical changes from the cropping pattern as conceived at the time of project formulation. The possibility of such changes should be taken into account in project design, with particular regard to determining desirable canal capacities.

Variations in cropping pattern and in irrigation requirements within the service area may be due to soil conditions, for instance the occurrence of valley-bottom lands with heavy soils and poor drainage in one portion of a project and uplands with very light shallow soils in another. A uniform project-wide cropping pattern would not be appropriate in such circumstances nor would a uniform schedule of water deliveries, with the cultivator being left to decide on its use, as crops appropriate to the lowland area are likely to require water at seasonally different times from crops in the upland area. Where the different soil situations occur in homogenous areas of substantial size, virtually as major sub-divisions of the project, independent branch canal systems with independent scheduling may be the logical solution.

A more difficult situation is encountered when substantially different soil types occur within a small area. An example is in gently rolling uplands dissected at frequent intervals by shallow valleys. Small streams meander between alluvial terraces in each valley bottom. The upland soils are free-draining and suited to a variety of non-paddy crops.

The valley-bottom lands are in heavy clay soils, poorly drained and slow to dry out after the monsoon season. They are suited to paddy only, but offer the possibility of double cropping of paddy. The differentiation of soil types is on a small scale, with typically one tertiary command (30 to 50 ha) in one soil type and the neighboring command frequently in the other. This is a common situation in the service areas of small (50 to 100 ha) village "tank" schemes. The area includes valley-slopes in irrigated upland crops, and in the valley bottom a narrow strip of paddy. The same supply canal which serves the upland crops also serves the paddy area, via downslope branches. Regulation of supply to upland or to lowland areas is not a problem in the closely-knit social structure of the small village. In the context of a larger public system, however, the same topographic situation, repeated throughout the project area, does pose a problem. To cater to the irrigation requirements of the two soil types (particularly differences in seasonal timing) is technically possible, but requires more intensive operational staffing than is normally provided and a more elaborate system of hydraulic controls.

The diversity in type of crop and in irrigation scheduling discussed above is imposed by soil conditions. A more general question is whether simply the preferences of the individual cultivator or group of cultivators should be accommodated in scheduling water deliveries. This is the question of the specialty crop. It may be vegetables, sugar-cane, bananas, fruit-crops, pond fish culture, etc. in an area otherwise devoted to regular field crops. With basic food-grains now reaching the stage of overproduction in some regions of S. Asia, there is much emphasis on diversification, particularly higher-value crops. This has become a matter of urgency in some areas.

The availability of good quality groundwater at depths accessible to the small cultivator may be the solution to the problem of irrigation of specialty crops within a canal system command. In areas of hard-rock substructure, however, the ground water yield may be small, although augmented by seepage from the irrigated area, and large-diameter dug-wells may be needed. Although expensive, they have the virtue of storage capacity and may function virtually as small farm-level short-term pondages, being filled in part by rotational supply from the canal system.

Larger lateral or terminal pondages are also attractive conceptually, insulating cultivators from the time constraints of rotational canal supply. However, in spite of considerable efforts to incorporate such pondages in some recently constructed schemes, sites which can be filled by gravity from the canal system and which can be drawn off by gravity have proved difficult to find. The addition of pumping could facilitate location of sites, but pondages are not expected to be a generally-applicable feature of irrigation in the region under discussion. Where small tank schemes already exist within an area about to come under canal supply they can furnish the desired storage. In some cases the function of the primary canal is in fact solely to supply existing tanks, and the scheme has no distribution system of its own. Alternatively, there may be a combination of new service areas supplied by the canal system and existing service areas served by tanks with supplemental canal supply.

Lateral and terminal pondages are, however, the exception rather than the rule, and in areas of hard-rock substructure the proportion of a command which can be irrigated from dug-wells, although important, is relatively small. This leaves the question of how deliveries from the canal system itself can best be adapted to serve areas of radically different soil types occurring throughout the project command, or the needs of diversification from the standard project-wide cropping pattern.

In this regard, it is noted that the situation in an irrigation project serving smallholders is very different from that in Western countries, where a project is likely to serve much larger holdings or larger Water Districts. In the latter case, supply of water in response to day-to-day "limited demand" is feasible. The request for water can be met provided it does not exceed the limits of conveyance capacity or the availability of water, taking into account concurrent requests from other irrigators. Technicalities of canal operation are not usually a constraint. Designing for such "limited demand" operation incidentally requires incorporation of a diversity factor in sizing of canals. As the probability of all the irrigators served by a lower-order canal requiring water at the same time is greater than in a higher-order canal with larger number of irrigators, in determining canal capacities a factor is applied to the peak water duty (flow per unit of area served) varying from unity for the main canal, and progressively increasing to as much as two or more for the lowest order canals. This procedure is not followed in those smallholder systems which are designed on the basis of a uniform cropping pattern. In this case the water duty is taken as constant throughout the system, subject only to conveyance losses which, in fact, marginally reduce the duty towards the lower end of the system. However, where a degree of "limited demand" capability is to be incorporated in a smallholder system the application of a diversity factor in determining canal capacities may be necessary.

Allocation of water and establishing water charges

Equity in allocation of water may or may not be a consideration in the design of smallholder irrigation. In some schemes, allocation in proportion to the area of holding is strictly followed, or at least it is aimed at. In others, each cultivator makes an application for water to irrigate a nominated area of a particular crop. Each application is viewed along with all other applications, and with such modifications as circumstances may indicate it is finally "sanctioned". The irrigation agency then endeavors to make the necessary water available to meet that sanction by incorporating the cultivator's holding in an appropriate rotational schedule. This is the "rigid shejpali" system practiced in central India. In sanctioning, restrictions on the proportion of certain crops, such as sugar-cane, are placed more for reasons of watertable control than for reasons of equity in water distribution.

Where water is allocated in proportion to area of holding it is in fact an entitlement, rather than an amount of water, which is allocated. In the original N.W. Indian system the cultivator was obliged to take his share of water when his turn came (the "warabandi" system). This presented no problem to the cultivator in that area, as the amount allocated was always less than he could use, i.e. water was always needed. The present concept is a broader one. Among the group of cultivators served by a tertiary (watercourse) each has a basic entitlement but he is not obliged to use it and he can transfer its use to other members of the group by informal adjustment of the rotational schedule.

Where water is distributed on the basis of uniform entitlement per unit of area of holding, water charges are based on a per-hectare per season basis. Where distribution is on the basis of sanctioned areas of particular crops water charges are usually based on area and type of crop. The latter charges taken into account the relative amount of water which the type of crop uses, and the economic returns from it.

Charging on the basis of actual volume of water used, usually regarded as the ideal system, is not generally practiced at the level of the individual cultivator or of the tertiary command in the South Asian region. This is due to the practical difficulty of maintaining recording equipment at that level. However, it is being employed in some cases where water is supplied to a cooperative, particularly for sugar-cane production. In tubewell systems, direct or indirect volumetric charging for water is commonly practiced.

Capacity of primary and secondary canals and size of irrigation area

In determining the capacity of the primary canal, two factors are relevant:

(a) The availability of water (rate of flow) at the primary canal head, modified by storage if any, month by month throughout the year

(b) The pattern of varying demand for irrigation corresponding to the proposed composite cropping pattern, throughout the year.

The two factors are related, in that the cropping pattern may be designed to make best use of the seasonally varying irrigation supply. Conversely the operation of the storage reservoir, if any, will be designed around the varying seasonal irrigation needs of the selected cropping pattern.

With regard to the second factor, in designing the system it is convenient to work with a hypothetical area of 100 ha under cultivation with the proposed crop mix and with 100% irrigated crop intensity (also hypothetical) in the season of maximum irrigation intensity. The irrigation demand for this area is estimated at weekly intervals and plotted throughout the year. The 100 ha area is then scaled up progressively until the plot of irrigation demand touches the monthly plot of flow available. The corresponding peak annual demand for that scaled-up area gives the desired nominal maximum design rate of supply at the primary canal head If the irrigation system were based on a design rate of flow smaller than that value, the available supply would only be partially utilized throughout the year. If on the other hand, the system were to be based on a design rate of flow greater than that value, there would be a deficit in part of the year.

As discussed above, cropping pattern and reservoir operation may be varied to optimize water use, for instance to maximize water use in the high-flow season, or to minimize demand in the low-flow season. Hence, the procedure described is part of a trial and error process in project design.

It is emphasized that the desirable maximum design rate of flow at primary canal head determined above is not necessarily the same as the canal capacity. Canal capacity may be made greater than the nominal maximum design rate of flow for a number of reasons, including provision for possible future changes in cropping pattern (particularly in reservoir-controlled systems) or to accommodate unsteady flow in the canal. It is, however, a key parameter in the design of the remainder of the system.

So far no reference has been made to size of irrigation area or to actual irrigation intensity. The hypothetical 100 ha irrigated at an assumed 100% intensity in the peak season has been scaled up until demand matches the available flow in the critical period. The area so derived by this scaling-up process is the net service area which could be supplied at 100% intensity during that period. However, the area actually served could be made considerably greater, for social or other reasons. In this case, the irrigation intensity would be correspondingly smaller. It is noted that reducing the irrigation intensity through increasing the size of the service area does not change the design rate of flow or canal capacity at canal head.- It does, however, affect the figure commonly referred to as the canal "duty", which is simply the design maximum flow per hectare of area served by the system. If the peak design irrigation intensity is reduced by 20% by increasing the size of the service area the canal duty is similarly reduced. Provided that the cropping pattern is uniform, the canal duty in the primary and secondary canal system is uniform throughout (adjusted for losses), and the design flow at any point on a canal is proportional to the area commanded by that point on the canal. An exception is noted when a secondary is designed to operate rotationally, half the time on and half the time off, even in the season of peak delivery. Such operation is the exception but it is practiced in some systems with long unlined secondaries, in order to minimize conveyance losses. In such circumstances, the canal duty has to be doubled if the same seasonal delivery of water per unit of area is to be achieved, as with a system based on continuous operation in the peak period.

The same canal duty, adjusted for conveyance losses, is also applied at the head of the tertiary, provided that the cropping pattern is uniform. However, what distinguishes a tertiary from the higher-order canals is the fact that the design flow remains constant throughout the tertiary command. For reasons discussed in the next section, it does not reduce as the area served reduces, toward the outer margin of the tertiary command.

Background

A tertiary channel or watercourse commonly serves 30 to 40 ha, in some cases up to 100 ha. In a major public scheme operation above the tertiary, including the intake to the tertiary, is usually the province of the Irrigation Department. Operation within the tertiary command is commonly (and preferably) carried out by the irrigators which it serves.

The design of the distribution system within the tertiary command is influenced by the size of holdings (the term holding is used in the sense of a contiguous area farmed as a unit), as well as by the type of crop being grown, particularly the extent of paddy. The following discussion presents the three cases of cultivation of non-paddy crops, paddy as the primary crop, and mixed-cropping.

Tertiary system design for non-paddy crops

Where holdings are relatively large (5 to 10 ha) the tertiary and its branches provide an outlet (usually a single outlet) to each holding. Within the holding, water distribution is by farm channel. Where holdings are small, typically less than one-half hectare, a further division of the canal hierarchy may be introduced. This is the quaternary or field channel. In this situation, the tertiary command is divided into sub-units of 4 or 5 ha, each provided with an outlet from the tertiary. Within the sub-unit, field channels carry the distribution down to the individual holding. There may be four or five holdings or as many as twenty or thirty in a typical sub-unit. It is noted that in the situation being discussed property boundaries are irregular. In the contrasting situation in which land consolidation has been carried out, or in new sub-division of lands for settlement purposes, boundaries may be near-rectangular and oriented with regard to land slope. In that case, channel layout within the tertiary command is simplified and the small holding may be served directly from the tertiary, without intervening field channels.

Hydraulically, tertiaries and field channels, are simply successive elements of a continuous branching system extending from the tertiary head down to the turn-out to each holding. Channel capacity, in most cases, is the same throughout. However, the field channel is closer to the farmer than the tertiary and may differ in the need for, or absence of, right-of-way procurement and in the extent of farmer participation in construction. Furthermore, tertiaries may be partly or wholly lined, but rarely are field channels.

As noted above, channel capacity is usually constant throughout the tertiary/field channel system. It does not reduce from tertiary to field channel. This reflects the important design philosophy that a key factor in obtaining satisfactory irrigation efficiency with non paddy crops is an adequate stream size at the field level. This desirable rate of flow at the point of delivery to the holding depends on the size of the holding, soil infiltration rates, ground slope, and other factors, but is generally in the range 30 to 60 liters/sec (about 1-2 ft³/sec). A rate that is too low will result in excessive seepage from field channels and poor field efficiency, while a rate that is too high will lead to water management problems for the farmer.

The tertiary is designed to supply water at this selected rate. In fact, this may be regarded as the definition of a tertiary. A corollary is that the flow in the tertiary should not be further divided enroute to the point of delivery to the holding, i.e. one farmer at a time should take the full flow of the tertiary/field channel system. This situation is in contrast to that in many older systems in which any number of cultivators may divert from the watercourse, each taking an indeterminate amount of flow, generally excessively small or nil at the outer perimeter of the tertiary command. The arrangement in which one farmer at a time takes water from the tertiary/field channel system, if complied with, assures adequate delivery stream size at all points. It also permits control of the respective amounts of water taken by each cultivator by fixing the duration of his irrigation within the rotational supply period.

The system described has much merit and is a key provision in World Bank assisted irrigation development in the Indian sub-continent over the last fifteen years. It has its origin much earlier in N.W. India. However, it is not universally practiced elsewhere in South Asia, and it has encountered operational problems in some areas in India, particularly where paddy is an important crop component.

A variant employed elsewhere involves supplying two or three holdings simultaneously from the tertiary, the latter being of correspondingly greater capacity and serving a greater area of command than with the single turn-out system. Supplying several holdings simultaneously in this manner poses the problem of how to ensure that each turnout takes its proper share of the flow the tertiary/field channel system. With appropriate control structures, this could be achieved, technically, but it would require a regulated outlet (not simply a fixed orifice) at each turnout, capable of delivering the desired flow under conditions of varying head in the channel As there are commonly thirty or forty or more holdings in a tertiary command, each with turnout, provision of such regulated outlets would be a substantial cost item and a major problem of operation and maintenance. In fact, where diversion to more than one holding at a time is practiced, the division of the flow between turnouts is usually on a judgement basis by the cultivators, without the use of formal regulating structures. The one-turnout-at-a-time system avoids this problem, the rate of flow being controlled at the intake to the tertiary only. The turnouts to each holding are operated simply on an on/off basis, each in turn taking the whole flow of the tertiary/field channel system.

While conceptually simple, the latter system does depend for its success on the cooperation, or restraint, of the cultivators. Out-of-turn diversions can defeat its operation, and Departmental supervision at the level of the tertiary command is virtually infeasible. Control must be by the cultivator group.

Two factors enter into determination of the size of the area commanded by a tertiary. One is the desirable rate of flow at the point of delivery to a holding (the farm stream). As previously discussed this should be in the range of some 30 to 60 liters/sec. The other factor is the canal duty (design maximum flow per unit of area served). The duty applies from primary canal head down to the outlet to the tertiary, subject only to adjustment for losses enroute and provided that the tertiaries are designed to run continuously during periods of peak demand and that water allocation is uniform throughout the project area. In such circumstances, the size of area commanded by a tertiary is obtained simply by dividing the water duty into the farm stream. Thus, a water duty of 1.2 liters/sec/ha and an adopted farm stream of 30 liters/sec would give a service area of 30/1.2 or 25 ha. In actual practice, the size of service area is likely to be influenced by topographic features and location of property boundaries, and the tertiary capacity may have to be adjusted accordingly, although still within a small range either side of the desirable value.

A common problem with tertiary channels is excessive seepage losses and poor service to tailenders. This can be encountered when the irrigation intensity is low, with correspondingly low water duty, and the area is in soils with a high infiltration rate. The low duty results in a relatively large area of tertiary command with correspondingly great length of tertiary/field channel in relation to its small stream size. In high infiltration rate soils this may result in excessive proportionate seepage losses. Adopting a farm stream at the higher end of the practical range for efficient water management can assist, also lining of the main stem of the tertiary. Other alternative courses are to design the tertiary system for rotational operation (50 % on, 50 % off) even in the season of peak supply, which doubles the duty and halves the size of tertiary command, or to design the system for supply to two or three holdings simultaneously rather than to one only. This doubles or triples the required capacity of the tertiary and reduces proportionate seepage losses. However, this course loses the operational advantages of single point delivery previously discussed.

Appropriate layout of the tertiary/field channel system is essential to preserving cultivator harmony. Although the channel network provides an important service, it nevertheless occupies a significant amount of agricultural land for which adequate compensation may or may not be paid, and it can greatly obstruct access to farm and field. Correct siting of the tertiary intake and of turnouts to individual holdings has considerable bearing on its ability to command distant fields. Provision of channel crossings on traditional village access routes and access over channels to farms are also items of vital interest to cultivators in the area Current policy of investment institutions is to favor maximum participation of cultivators in all phases of layout of the tertiary/field channel system, and if possible also in its construction. Such participation is essential if cultivators are expected to operate and maintain the system.

Although technically straightforward, tertiary system layout requires detailed topographic and cadastral (land ownership) data Most errors in layout and cultivator complaints stem from the use of inaccurate or insufficiently detailed contour plans. Verification of layouts by field survey, particularly levelling, is essential, as is flagging of alignments and turnouts prior to construction to ensure that cultivators are in agreement with their location.

The responsibility of the agency constructing the project is usually considered as ceasing at the boundary of the holding. Thereafter, the farmer takes care of water distribution. It is nevertheless emphasized that in irregular topography, or in slopes of 2-3 % or greater, water distribution within a holding as small as one or two hectares can pose problems which a cultivator cannot be expected to overcome without guidance. Erosion of down-slope farm channels resulting in loss of command of fields is a typical problem, requiring some form of control structure, however primitive. In such circumstances, advice should be provided regarding water distribution and associated land shaping. However, water distribution at this level is usually regarded as outside the scope (or beneath the professional dignity) of conventional irrigation engineering, nor is it covered by agricultural extension. Consequently, such assistance is conspicuously absent in most cases.

Tertiary system design for areas primarily under paddy

The previous discussion has been with regard to irrigation of non-paddy or "irrigated dry" crops. Where paddy is the principal irrigated crop, other crops being incidental, water distribution within the tertiary command can be very different from that described above. In the case of irrigated dry crops, a critical factor is field efficiency, including efficiency of distribution within the field by farm channel or furrow. Hence, there is an emphasis on adequate size of delivery stream. With paddy, however, the field is flooded and efficiency of conveyance or distribution within the field is not a factor. During the growing season a small continuous flow to the ponded field can be perfectly satisfactory. During cultivation and puddling, on the other hand, a substantial supply of water is required for a short period.

Water needs for wet-land paddy have been discussed earlier. Cultivation and water distribution practices vary widely. In many wet-tropic areas the traditional approach has been to supply large blocks (1000 ha or more) by field-to-field flow. Cultivation and puddling progress from upper to lower portions of the area, the whole process taking a month or more. Irrigation subsequent to transplanting is also field-to-field, requiring virtually no formal distribution (tertiary system) within this large area

In contrast, paddy cultivation in other areas is on the basis of the individual holding. A cultivator may have regular employment outside the farm and may wish to have cultivation and transplanting completed in as short a period as possible (two or three days), often by contractor, and largely independent of the activities of his neighbors. Further, during the growing period a cultivator may wish to withhold irrigation from his fields at intervals to optimize efficiency of fertilizer application, either coordinating this activity with his immediate neighbors or independently. This situation calls for a network of delivery channels capable of controlled supply to relatively small units of area This may also be necessary where irregular topography precludes full command of large blocks by purely field-to-field delivery.

However, tertiary channels within an area under paddy have their problems, and may be shortlived. The habit of encroaching on channels during puddling (in order to maximize planted area) often leads to collapse of the channel bank. The channel is lost, and the system reverts to fieldto-field distribution in large units. In fact, if paddy cultivators have a tertiary network imposed upon them, rather than requesting it, eventual destruction of the network is highly probable.

Nevertheless, with widespread adoption of High Yielding Varieties and associated increased fertilizer application, as well as increased double-cropping through the use of short-duration varieties, finer-tuning of paddy irrigation is increasingly desirable. A compromise employing a limited tertiary system delivering to sub-units of five to ten hectares, within which distribution is field-to-field, may be the solution. Whatever system is adopted, however, it must have the blessing of the cultivators concerned, and must be regarded as permanent. Where gradients permit, the use of buried-pipe tertiaries delivering to valved outlets serving each sub-unit could avoid much of the problem of open-channel distribution in paddy areas.

In a predominantly paddy area, determination of the maximum design flow in the primary canal, size of service area, and canal duty, follows the same procedure as previously described for irrigated-dry crops. However, estimation of water requirements for cultivation and puddling is likely to be contentious, being dependent on assumptions as to the amount and timing of early monsoon rainfall, the period over which land preparation and transplanting is likely to extend in the area as a whole (this may be four to six weeks), and the cultivation/puddling practices expected to be adopted by the cultivators. Duty at primary canal head, assuming 100% irrigation intensity in the peak season, is likely to be between 1.5 and 2.0 liters/sec/ha, with corresponding figures at tertiary intake (allowing for losses enroute) in the range 1.2 to 1.5 liters/sec/ha The possible adoption of peak season irrigation intensities of less than 100% for reason of social equity was discussed earlier in connection with irrigated dry crops. Such a course is unlikely in areas primarily devoted to paddy, where cultivators traditionally expect the entire area to be under that crop, for at least one season. The desirable rate of delivery to the farm (the delivery stream) for paddy during the stage of cultivation and puddling is likely to be higher than for irrigated dry crops, probably in the range of 40 to 80 liters/sec, assuming delivery to one turnout at a time.

The appropriate size of tertiary command is derived from the duty at tertiary outlet and the size of farm stream adopted, as discussed for irrigated-dry crops. Thus, with a duty of 1.2 liters/sec/ha and a delivery stream of 40 liters/sec the nominal size of tertiary command would be 40/1.2 or 33 ha. With a duty of 1.5 liters/sec/ha and a delivery stream of 80 liters/sec it would be approximately 50 ha.

In areas wholly under paddy, cultivators may choose to practice rotational delivery only during the period of puddling and transplanting and during the growing season to supply all sub-units simultaneously with a smaller continuous flow, just sufficient to maintain inundation. However, when limited water supply precludes continuous flooding, periodic (weekly) inundation may become necessary, with full-now rotational supply to each subunit in turn.

The above discussion assumes delivery to one sub-unit at a time in periods of rotational supply. As discussed in connection with irrigation of non-paddy crops, it is also possible to design the tertiary system for simultaneous supply of the full delivery stream to two or three sub-units rather than to only one, consequently doubling or tripling the required capacity of the tertiary channel and the area of tertiary command. The disadvantage is the need for installation and operation of regulating structures on the tertiary and the reduced level of control on the amount of flow taken at the sub-unit.

Tertiary system design for mixed cropping

Paddy and non-paddy crops may be irrigated at the same time, or sequentially, with paddy being irrigated during the monsoon followed by other crops in the dry season. Where both types of crop occur within the same tertiary command, which is the general situation, operational and design problems are encountered. The operational difficulties stem from differences in the amount and in the timing of irrigation needs of the two crops. The design problem is that nonpaddy crops require a closer tertiary/field channel network than does paddy. With paddy, formal irrigation distribution may stop at the boundary of the sub-unit, with field-to-field flow thereafter. Non-paddy crops, however, require formal distribution, by tertiary and field channel, right down to the border of the field.

While paddy can be irrigated with a tertiary/field channel system designed for non-paddy crops there is a strong possibility that the field channels (within the sub-unit) will be destroyed by encroachment during cultivation for paddy and will have to be reconstructed prior to the followon non-paddy crop. In &et, cultivator practice, in some areas where wheat follows paddy, is to each year construct temporary field channels after harvesting paddy, before planting wheat. This can be quite effective if the length of the temporary channels is small. If longer, their delivery efficiency can be poor, particularly to tailend cultivators.

An effective compromise is to carry the permanent tertiary system down to relatively small subunits (2 to 4 ha), and to leave it to cultivators to construct temporary field channels within the sub-units each season. The alignment of the temporary channels and the rights of tailenders to conveyance of water via such channels should, however, be officially established.

In mixed cropping, the peak rate of supply may occur in the early monsoon season during puddling for paddy and may greatly exceed the available rate of supply or the demand during the subsequent dry-season. Furthermore, the desirable size of delivery stream during puddling may be considerably greater than that required for the dry-season non paddy crop. In such circumstances, there is a case for reducing the rate of diversion to each tertiary in the dry season. This can be achieved, with some approximation as to sharing between outlets, simply by reducing the flow and the level in the parent secondary canal. Alternatively, gated outlets to tertiaries may be provided and adjusted seasonally, although this departs from the operationally preferred course of fixed outlets. The use of a second outlet (usually simply a pipe) for supplemental diversion in the paddy season is practiced in some areas. However, employing either adjustable or supplemental outlets invites their misuse in the dry season.

Layout of tertiary channels

The importance of involving cultivators in the layout and construction of tertiary channels has been referred to earlier. An issue which frequently arises in areas of irregular field boundaries or irregular topography is the extent to which tertiary alignment should follow boundaries or be routed more directly, crossing properties where necessary to do so. The latter course minimizes length and cost of channel. However, it maximizes interruption to access for cultivation by bisecting fields and for the smallholder may result in disproportionate loss of cultivated area. Furthermore, in many cases compensation is not paid for right-of-way for tertiaries, the channels being regarded as communal property. Following property boundaries, although inconvenient from the layout viewpoint, does come close to equitable contribution of land for tertiary construction. However, routing the tertiary along boundaries is not without problems, particularly where there is considerable difference in elevation between fields lying on either side of the boundary, as is commonly the case in areas already terraced for rainfed cropping.

Determination of size of tertiary command and its division into smaller sub-units have already been discussed. While this procedure is conceptually simple, some site situations can make implementation difficult and may require departure from the idealized design. A case in point is supply to the long narrow strip of command frequently encountered between a primary and a secondary canal in its upper reaches. A tertiary channel serving, for instance, 30 ha in such a location could be several kilometers in length, running generally along the top of the primary canal embankment. In some situations, this can be impractical for a number of site-specific reasons. One expedient is to use a number of "direct" outlets, each taking off from the main canal and serving an area of a few hectares. While solving the problem in one sense, this arrangement requires that the direct outlets be independently operated (rotated), otherwise they would run continuously with the primary canal, taking an excessive amount of water. There is no entirely satisfactory solution to this situation.

A further problem is encountered in irregular topography where a secondary canal runs down a narrow spur, with the area to be irrigated lying on the slopes on either side. Conventional layout would require relatively long tertiaries running parallel to and on either side of the secondary. However, the crest of the spur may not be sufficiently wide to accommodate all three channels if they are of regular trapezoidal section. One option is to omit the tertiaries and use direct outlets on the secondary, each serving an area of a few hectares. However, such outlets would require independent rotational operation, as the capacity (size of delivery stream) would be excessively small if scaled down commensurate with the small size of the area served so that they could be run continuously with the secondary. A second option has both secondary and parallel tertiaries. But in order to accommodate all three on the narrow crest of the spur, a rectangular composite flume section is used (concrete or brick) incorporating all three channels. This avoids the separate operation of outlets required in the first option. Choice between the two is a trade off between construction cost and operational simplicity.

A further example of conflict between technical and operational factors in tertiary system design involves the number of delivery points (turnouts) to a holding. To facilitate adherence to the rotational delivery schedule within the tertiary command, it is usual to observe the convention of a single turnout from the tertiary or field channel to each holding. Indeed this is very strictly enforced in some areas. However, there are site situations which strongly indicate otherwise. A case in point is a sandy soil with very high seepage losses in unlined channels. The tertiary is lined, but farm channels are not. A tertiary parallels the boundary of a holding of several hectares. The single turn-out rule would require a farm channel (unlined) to parallel the lined tertiary for several hundred meters, supplying down-slope branches. However, seepage losses in the farm channel would be high. The alternative would be to provide a second turnout from the lined tertiary at about half-way down the length of the boundary, thereby much reducing seepage losses when supplying the lower end of the holding. However, the existence of two turnouts to the holding, although intended to be operated as alternates, invites the possibility of their being operated together, doubling the rate of diversion to the holding and infringing upon the rotational- schedule. This again is a case of conflict between technical desirability and simplicity of operational control.

Finally, there is the question of possible use of multiple siphon tubes for delivery from tertiary/field channel to the holding. Such plastic siphon tubes (about 5 cm diameter) can be used in groups of five or more, functioning in effect as a portable turn-out or separately supplying individual furrows directly from the tertiary, thereby eliminating the farm channel. This can be attractive in certain situations and is widely practiced in Western irrigation. However, the presence of large numbers of siphon tubes in a smallholder situation would be likely to result in out-of-turn diversions and would write off any possibility of maintaining rotational distribution, unless the cultivators concerned were unusually capable of policing their system.

The above discussion of tertiary systems underlines the fact that their design is not always straightforward. Generalized layout criteria can be formulated, but judgement at the field level has to be exercised in their implementation. Experience in South Asian irrigation is that construction of tertiary/field channel systems is the controlling factor in completion of new projects and often lags years behind "creation of irrigation potential" (the construction of the main canal system). The efforts of international lending institutions to expedite such works by making reimbursement against primary and secondary canal construction conditional on completion of the associated tertiary systems have not entirely solved the problem. Training courses in tertiary distribution for irrigation engineers have undoubtedly assisted, but the subject remains a principal area of concern.