|Design and Operation of Smallholder Irrigation in South Asia (WB, 1995, 134 p.)|
|Chapter 4 - Water supply and demand|
The region under discussion is monsoonal, with river-flows characterized by highly variable seasonal and annual discharge. Storage reservoirs can be provided in some cases where topographic and ecological considerations permit. In general storage capacity can provide partial regulation only, and the project must accommodate to such limited control of flow, or to completely unregulated flows in the absence of any reservoir.
The highly variable nature of monsoon precipitation makes for considerable difficulty in both yield and flood hydrology. The onset of the monsoon, upon which so many agricultural operations depend, may vary by several weeks from year to year, and gaps of weeks duration may occur within the monsoon period. Much of the monsoon precipitation is in the form of discrete, local rainstorms, often violent, rather than the popularly conceived uniform countrywide downpour. This pattern results in wide random variations in seasonal rainfall between adjacent areas (as much as 50% difference in a particular year, between locations as little as 25 km apart) and makes for considerable difficulty in a statistical approach to estimation of water yield, particularly for small catchments. The problem is aggravated by the limited number of rainfall and river-flow recording stations. While international agencies commonly call for at least five years of actual stream-flow records as a basis for the design of small projects, (much longer for major projects) in remote areas there are commonly none and extrapolation from similar catchments must be resorted to. In these circumstances expansion of the network of rainfall recording and stream gauging stations is a priority item. It is noted, however, that maintenance of calibration of stream gauging stations is no small task in rivers subject to heavy siltation and frequent changes of channel during flood-flows.
The impact of the widely varying pattern of monsoon precipitation on the life of the small cultivator is illustrated by two situations. In one, the monsoon had begun propitiously and then failed, and paddy stood wilting in the fields. It was ploughed in, an unusual event, and when the rains returned was replanted with yellowing spindly seedlings remaining from seed-beds. The monsoon then became violent, flooding and destroying the replanted crop. Cultivators in the area, in the path of monsoon storms moving from the Indian ocean to the Himalaya, commonly borrow ostensibly for purchase of fertilizer but actually for "pujas", religious ceremonies to placate the deity held to be responsible for such outrageous events.
In the other case, the young maize crop, newly sprouted from the red lateritic soil, stood wilting under the backdrop of heavy grey monsoon clouds, but it did not rain. And nearby, the Door of the village reservoir was cracked and dry. The monsoon had failed for two successive years. The next monsoon rains were nine months away.
The seasonal variations in monsoon rainfall can, of course, be studied statistically, and this must be done in project design, but the realities of the situation for the cultivator and his family must also be kept in view.
Given the large variability in water supply, the immediate problem is to take into consideration the uncertainty of water supply into the design of the project. To design for an assured level of supply would avoid certain operational problems, but would grossly underutilize the water available.
Much of the debate over the design and operation of surface irrigation systems centers around the question of how to handle the non-assured component of supply. One approach to limiting the variability of supply to be accommodated is to design the system for the "75% probable" year (or other degree of probability). Then statistically in three years out of four, the amount of water available equals or exceeds the amount for which the system is designed; only in the fourth year is there a deficit. A calendar of twelve months each which is "75% probable" may also be constructed, becoming the "design years". While this is a useful concept for purposes of establishing system capacity, it still leaves the question of how to operate the system in the deficit years, or months. This will be discussed in the next chapter.
If the system is to have storage, a question influencing design and operation is how the storage will be utilized, whether for seasonal regulation within a twelve-month period, or over-yearly. In the first case water stored in the wet season is used in the following dry season, possibly with some carry-over for pre-monsoonal irrigation (particularly puddling and transplanting of paddy) in the following year. In the second case, applicable only to major reservoirs, the storage cycle may extend over several years, partially evening out years with good and bad water supply.
Once the amount of water to be taken as seasonally available for design purposes is determined, the key question is then the area to be supplied. This involves consideration of cropping pattern, water requirements of individual crops, land availability, and the socioeconomic question of intensity of irrigation. The latter is the contentious item. Should the project be confined to an area all of which can be fully irrigated with the available water (intensive irrigation)? Or should the benefits of irrigation be spread more widely, supplying less than the full irrigation requirements to a larger area (extensive irrigation)? In the second case each cultivator can irrigate only part of his holding, or optionally he can supply all of it with less than the "optimum" quantity of water. The alternatives are described by the irrigated crop intensity (irrigation intensity). This is the percentage of the holding which is to be supplied with irrigation in a particular season, or annually if all seasons are totalled. The question of whether the figure is based upon application of the full "optimum" amount of water, or less than that (a common practice), is usually left unanswered. In some respects a more useful index of intensity of irrigation is simply the depth of water to be supplied, seasonally or annually, calculated as if applied uniformly over the whole area of the holding. Use of this index avoids the question of what water requirements to assume in calculating irrigation intensities.
The relative merits of intensive vs. extensive irrigation system design are much debated. The intensive approach leads to a smaller area to be served by canals (the "command") and lower canal cost, also lower total cost of land development. The extensive approach is often imposed by social pressures. In fact some states decree an upper limit on the design irrigation intensity, on the grounds that any higher intensity would unfairly benefit those within the command at the expense of those excluded from it. The pressure to expand the area served may continue through the life of the project, with petitions to extend the canal system to peripheral areas, or to introduce or permit pumping from canals to higher areas not served by the original system. Extensive irrigation has certain advantages. By limiting the supply of water to less than apparent need, it imposes an incentive for prudent use of water. It may also permit on-farm rotation of irrigated crop benefiting productivity in light soils. Of particular importance, it encourages development of supplemental groundwater, where wells are technically possible. This in turn may benefit watertable control.
Extensive irrigation may well increase productivity per unit of water supplied. However, it may introduce operational problems, particularly in large projects. In a small system that is village owned and operated, decisions on water-management, including the use of stored water, are likely to be made by consensus of the cultivators. In a large public system the cultivator is aware only of the canal which serves him. He is not aware of project-wide supply problems, the "grand design" of the system. If he receives less water than his apparent needs, he may endeavor to take it by whatever means are available. The subject of operation of supply systems in situations of water deficiency is discussed later. For present purposes, it is sufficient to underline the fact that supply of sufficient water to irrigate the whole command, in at least one season, is not automatically a design feature. It is a question to be decided in each case.
Estimation of crop water needs, a basic factor in irrigation design, is by no means as straightforward as might be assumed. Actual water consumption (evapotranspiration), is influenced by climatic factors, including air temperature, humidity, radiation, cloud cover, and wind, and by the nature of the plant itself including its stage of growth. It is also influenced by the amount of moisture in the soil at the time (soil moisture tension). In the face of this number of factors, values for many of which are frequently not known, simplified approximate methods of estimation are commonly used. These employ a limited number of parameters, for instance air temperature and number of daylight hours only, or the measured evaporation from an open pan, as the basis for estimation. Alternatively, approximate estimations of values of climatic factors for which actual measured values are not available are inserted in more general formulae. "Plant factors", the water-consuming characteristics of each particular type of plant at each stage of growth, are based on field observations for which generally-accepted tabular data are available. There is, of course, a more direct method of water-use estimation, which measures water abstraction from a lysimeter containing soil and the growing plant. However, the difficulties of using the Iysimeter have limited its application to basic research.
Values of consumptive use obtained by the various methods of estimation vary widely. A comparison between actual measured water use and estimates made by eighteen different methods was given in the 19 74 report on Irrigation Water Requirements by the Irrigation and Drainage Division of the American Society of Civil Engineers. The investigation was related to alfalfa and grass crops, grown at ten stations in varying climate situations. The two most commonly used methods of estimation, Penman and Modified Blaney Criddle, gave results ranging from 14% low to 30% high (Penman), and 46% low to 35% high (Modified Blaney Criddle), compared with actual measurements. A.S.C.E has issued a further comprehensive report on the same subject (Jensen 1990).
A widely used reference for the estimation of crop water requirements is the Irrigation and Drainage Paper No. 24 (Revision of 1977) of the Food and Agriculture Organization of the U.N. (Doorenbos 1977). This covers the Penman, Blaney Criddle, Radiation, and Pan-evaporation methods of estimation and extends their applicability by calculating coefficients based on climatic factors not otherwise included in the estimation (particularly for the latter three methods, Penman is already comprehensive). However, estimates prepared by the four methods still differ substantially.
The estimates of consumptive use discussed above refer to "optimum" conditions, i.e. with unrestricted availability of water at plant roots or virtually zero soil moisture tension. These are the basic E to values. The customary use of the word "optimum" in this situation is misleading, in that such moisture conditions while possibly optimizing vegetative growth may not result in optimum economic use of water.
The effect of restricting the availability of soil moisture on plant growth is an important issue with respect to two questions. First, can less than "optimum" amounts of irrigation be used without significantly reducing crop yields, and second, how do the fluctuations in soil moisture tension between conventional periodic irrigations affect yields (Jensen 1990, Hillel 1987).
Research relevant to these two questions continues, but work to date indicates that any reduction in transpiration imposed by soil moisture stress automatically reduces the rate of vegetative growth in an approximately linear fashion, and as a corollary, cycling the soil moisture in the root zone from field capacity down to near wilt point, a basic feature of conventional irrigation practice, inevitably adversely affects yields.
However, the above conclusions must be treated with caution, in view of the results of extensive field station trials, which indicate that crop yields can be highly responsive to irrigation at critical stages of plant development, but that with-holding irrigation between such stages for periods of a month or more (with inevitable stress) has little effect on yields. This is notably true for certain crops and less so for others. Moreover, cycling of soil moisture in the root zone is an unavoidable feature of all irrigation systems (other than trickle or sprinkler), and the question of period between irrigations, which affects the range in soil moisture tension, has considerable implications on system design. More data is needed on the relationship between range of soil moisture tension between irrigations and crop yields.
Added to the level of uncertainty regarding crop water use is field efficiency, a factor involving considerable approximation. Consumptive use refers to water use at the plant. Field efficiency is the ratio between the amount of water consumptively used by the crop and the amount applied at the outlet to the field. Factors contributing to field inefficiency are percolation beneath the reach of the plant root system, evaporation from areas not occupied by the crop, seepage from distribution furrows, spillage from the end of the field, and non-uniformity in distribution of water on the field (i.e. some areas receiving more than sufficient and some less). Some elements contributing to field inefficiency are not, in fact, a loss to the project. Seepage below the root zone may fill a necessary leaching function (unless this is provided seasonally by monsoon rains) or may be recovered by groundwater development. Spill from the end of the field may be used elsewhere in the system. However, these elements contribute to the amount of water which must be applied at the field boundary.
Values of field efficiency are simply judgement figures. They may vary from an upper limit of some 80% to a more generally applicable range of 70-75%, and be much lower in less inadequately managed systems. One procedure which largely avoids the need for separate estimation of field efficiency is to base the estimation of crop water needs on field station data on irrigation requirements at the field boundary (which includes field inefficiency). Such data usually gives crop production under a range of seasonal water applications and irrigation schedules, in particular relating time of watering to stage of plant growth.
Thus, estimation of crop water requirements by conventional formulae inevitably involves considerable approximation. Estimates using different, but well accepted, formulae are likely to differ by 25% or more. Calculation of basic Eto figures for consumptive use under "optimum" soil moisture conditions is a necessary step, as a point of reference. However, for actual project design the use of agricultural field station data is preferable, if such data is available. If it is necessary to extrapolate, the ratio of Eto values for that station and for the project area can be used.
Because of the differences likely to be obtained in consumptive use estimates using different but reputable approaches, it is most desirable that agreement be reached in this respect between the agencies concerned with formulation and appraisal of a particular project. It is preferable to avoid a situation in which a government agency, or a consultant, carries out detailed designs and prepares cost estimates for a project, only to find at appraisal that the prospective financing organization disagrees with the basic assumptions regarding water requirements.
In a monsoonal environment rainfall can provide a major part of crop water requirements in the wet season, and a much lesser part, or none at all, in the dry season. However, not all rainfall can be utilized by the crop. During periods of heavy precipitation much is lost from the field by run-off and during very light showers most rain is intercepted by leaves and reevaporated without ever reaching the ground. Bunding of fields provides temporary pondage of heavy rain, although where crops other than paddy are being grown impondment has to be limited. On the other hand, where paddy is being grown, the bunded plot is likely to have standing water prior to the rainstorm, which limits its capacity for further storage. The soil moisture situation prior to a rainstorm also influences the extent of retention of rainfall, for instance pre-monsoon or early monsoon rain on dry soil may be fully retained, while later in the season it would not be.
Procedures for estimation of the "effectiveness" of rainfall are set out in the paper previously referred to (Doorenbos 1977). However, operational factors make it desirable to view each project separately. Also to be considered are the operational implications of unusual deficiencies in rainfall at particular times, for instance late arrival of the monsoon or rainless periods in mid monsoon. Hold-over storage may be included in the design of the project operation as insurance against delayed rains. Aside from the amount of storage to be reserved for this purpose, irrigation distribution system capacity may be determined by its function during such times of rainfall deficiency. Simulation ("paper operation") of the system, under various historic or postulated rainfall conditions, is the only satisfactory means of testing the system under these circumstances.
Rice is the most important single crop in the region under discussion. It is the only food crop which can be grown under conditions of continuous inundation of the root-zone, a feature which makes it uniquely suited to wet-tropic monsoonal cultivation. However yields are also responsive to sunshine, and are inhibited by cloud cover. Hence, highest yields are obtained in lower rainfall areas under irrigation as in the Punjab.
Rice is conventionally grown under conditions of inundation, when it is referred to as paddy (the term is also used for the bunded plot in which rice is grown) or as wet-land or low-land rice. It can also be grown without inundation, soil-moisture being held at near field capacity, in which case it is generally referred to as upland rice. It is basically the same plant in either case, although preferred varieties for the two situations may differ. Between the two limits, of continuous inundation on the one hand and upland cultivation on the other, lies a wide range of conditions under which rice can be successfully grown and which have a considerable bearing on water requirements.
For wet-land paddy, water is required for cultivation and puddling, and to compensate for seepage and to meet evapotranspiration. Cultivation (initial plowing) may be carried out in dry conditions, but in view of the limited capacity of the draft animals employed prior softening of the soil either by irrigation or by pre-monsoon showers is desirable. Subsequent puddling serves to convert the soil into a fine saturated slurry suitable for transplanting. It also provides weed control, and reduces seepage rate.
With regard to estimation of water requirements for cultivation and puddling, there are two widely different approaches. In traditional wet-tropic areas cultivation and puddling of a plot may extend over a period of a month or more. Emphasis is laid on the merits of allowing time for rotting of the ploughed-in stubble of the previous year's crop, under saturated conditions, before completion of puddling and transplanting the new crop in order to conserve nutrients. In contrast, there are extensive areas where water requirements and time are critical, where cultivation, puddling, and transplanting of an individual plot all occur within a period of twenty-four hours. The difference in water requirements between the two procedures is, of course, substantial (300 to 400 mm compared with 150 mm). It is noted, however, that even where puddling and transplanting in each plot is carried out in short order the operation is likely to be in progress in a large command over a period of several weeks due to limitations in availability of labor, draft animals, and cultivation equipment.
The capacity required of main and distributary canals during puddling in an area predominantly under paddy in the wet season is influenced both by the amount of water used per unit of area (the procedure employed on the individual plot), by the amount of time during which this operation is in progress in the command as a whole, and by the contribution of rainfall during the period. It should be noted that a plot puddled and transplanted early is likely to have little assistance from rainfall during the process, while a plot prepared later may benefit from already being saturated from prior rains. Consequently, averaging water requirements over the whole command is not entirely appropriate to determine the rate of supply required (the "water duty") for an individual sub-area. In this regard, the practice in some small village schemes is to make cultivation, puddling, and transplanting a communal or social event, with the whole population of the village concentrating its labors in one local area at a time, with virtually the entire flow of the main canal temporarily directed into that area. The water duty required at the tertiary or minor canal level in this situation is much higher than at the project-wide level.
Seepage is likely to be a substantial part of water requirements for the standing paddy crop. Rate of seepage is influenced not only by the character of the soil and the extent of puddling (collectively determining its permeability), but also by external factors including topography and watertable depth. On terraced slopes, seepage in upper paddies is likely to be entirely controlled by soil conditions. In the lower paddies, however, it will be influenced by seepage from up-slope areas and may be negative, presenting a drainage problem. In large areas of near-flat terrain, soil conditions will be the controlling factor early in the monsoon, but the watertable is likely to rise to the surface and limit seepage later in the season. An extreme case is provided by a near-flat area of highly permeable sand in a riverine delta. Heavy monsoon rains rapidly raise the watertable by several feet to the surface and the rate of infiltration becomes virtually nil. A late season crop of paddy is successfully cultivated.
The problem is how to estimate seepage rates for the purpose of project design. Generalized figures based on soil texture may be a useful guide for low permeability soils, but not for more pervious material, where external factors may control. The results of standard field tests (ring infiltrometer) can be entirely misleading for the latter reason. Seepage rates determined from a bunded plot several square meters in area are more relevant, although not necessarily reflecting the effects of repeated puddling, nor of seasonal rise in watertable. Better still are observations from a plot which has been under rainfed paddy cultivation for some time, in the same area, if such is available. It is noted in this connection that much nominally rainfed paddy in fact benefits from run-off (small drainage or surface flow) from adjacent uncultivated slopes. It is "semiirrigated". This fact has a bearing on the relevance of published statistical data comparing irrigated and "un-irrigated" yields (also on comparative projections of "with project" and "without project" crop production).
The above discussion refers to paddy cultivation by transplanting, which is the most common method in the South Asian area. However, paddy may also be direct seeded. This is the usual practice in Western countries, but is also being adopted in some areas of South Asia due to rising costs of labor for puddling and transplanting. Direct seeding also reduces water requirements in the initial stages of the crop, compared with the process of puddling and transplanting.
While the traditional procedure with wet-land paddy is to keep the crop continuously flooded, this is not essential. Much research has been devoted to the question of by how much crop yields are reduced if paddies are drained at intervals, and how much water can be saved by doing so. The question is particularly relevant to the rotational supply of water to paddy, which may be operationally convenient in some situations. Periodic withdrawal of water does, in fact, have some advantages. It is desirable at the time of fertilizer application (to avoid loss of nutrients with Bow from the end of the field) and it promotes oxygenation of the root zone. Published data indicates that water use can be reduced by 15 to 20% compared with continuous flooding, without significant reduction in crop yield.
However, considerably less water is regularly being used by some cultivators (e.g. Nepalese Terai), with flooding of paddies at fortnightly intervals only, on relatively high infiltration rate soils. With local varietal selection, surprisingly good yields are being obtained with this practice, which is intermediate between wet-land and upland cultivation. It must be acknowledged, however, that the conventional procedure of puddling and transplanting followed by near continuous inundation exercises very effective weed-control. Any departure from that procedure may be at the price of other means of control, although the extent of this problem varies from severe to very modest, from one area to another. The subject of "sub-optimal" irrigation of paddy warrants further investigation.