|
I.W. Makin and G.A. Cornish, HR Wallingford Ltd., Wallingford, UK |
SUMMARYIrrigation water management generally has not received adequate attention from system operators but, as competition for water resources continues to intensify, irrigation departments in many countries are looking for methods to improve water use efficiency. Improved water management is being included in the objectives of many rehabilitation projects, with computer-based irrigation scheduling viewed as a promising tool.
This paper examines the operational environment in which main system scheduling models are likely to be used. It describes the origins and functions of the INCA software (Irrigation Network Control and Analysis), an irrigation water management package developed by HR Wallingford, and draws lessons from field applications of this software.
INCA has been implemented on schemes in seven countries since 1991. The experience gained in implementing water management procedures in irrigation systems offers valuable lessons regarding the level of complexity that can be sustained in agency managed smallholder irrigation schemes. The paper draws conclusions regarding the broader management framework required before computer-based scheduling tools such as INCA can bring about sustainable improvements in scheme performance.
In an extensive analysis of irrigation canal management, Chambers (1988) identified a number of blind spots in the thinking of those charged with the operation of canal irrigation systems, describing main system management as the 'central gap'. He emphasized the need for appropriate and reliable methods of calculating target water releases and their timing (scheduling) coupled with operational practices to achieve those targets (delivery), if the performance of systems is to see sustained improvement. Other authors including Bottrall (1981), Sagardoy et al. (1986) and Plusquellec et al. (1990) have drawn similar conclusions regarding the importance of improved main system scheduling in bringing about improved productivity of irrigated agriculture.
Abernethy (1985), reviewing six years' research into system operations on the Kaudulla scheme in Sri Lanka, identified the following issues as key factors influencing the performance of the scheme:
· inadequate utilization of wet season rainfall, leading to restricted cropping opportunities in the following dry season;· lack of feedback about actual field water status, leading to inappropriate responses resulting in ineffective water releases, or crop yield losses;
· poor operational practices, with only limited knowledge of actual water distribution achievements and little guidance for field staff;
· little use of data collected in decision making.
He suggested that a micro computer with database and analysis software could perform a central role in enabling system managers to assess water requirements and determine supply schedules, ultimately leading to more equitable and reliable water distribution.
Irrigation practitioners, researchers and academics have developed and applied computer tools to the tasks of planning, scheduling and monitoring releases as a means of improving the operational performance of schemes. However, although numerous computer models are described in the literature (Lenselink and Jurriens, 1992) their widespread use in the operational management of irrigation schemes in the developing world is yet to be seen. Recent studies examining the factors contributing to this apparently slow uptake of computer-based water management tools (Murray-Rust and Cornish, 1993; Oskam, 1994; FAO, 1994) have concluded that even technically appropriate and robust models will remain under-utilized when the wider issues of management, such as performance incentives and communications between system managers and water users, are not properly addressed.
This paper describes the operational environment in which main system scheduling models are commonly applied. It goes on to draw lessons from field applications of the INCA model (Irrigation Network Control and Analysis), an irrigation water management package developed by HR Wallingford. The software has its origins in the Kaudulla study reviewed by Abernethy. It aims to address the need to improve the operation of water distribution systems, both by reducing the quantity of water consumed by irrigation and improving the reliability and equity of deliveries to water users. INCA is designed for application by irrigation managers responsible for water distribution in medium to large-scale schemes.
OPERATIONAL ENVIRONMENT AND THE ROLE OF INCA
Where water deliveries are through open canal networks, with levels and discharges regulated manually, operators face a daunting task in quantifying demands at the field level. They must account for losses in the tertiary system and main canal network, adjust for contributions from rainfall and allow for rotational and/or physical capacity constraints of the system, to derive a workable schedule of releases. When and if such a schedule is determined, field staff must then take action to ensure that scheduled volumes are actually delivered, striving to control and stabilize a hydraulic network where farmer activities do not always complement their actions.
Effective decision support tools to assist managers to schedule and monitor water releases at the main system level must be able to operate in an environment where much of the operations data will not be reliably available or completely accurate. Knowledge of areas planted, the crop mix and dates of planting will, almost always, be imprecise. Values for field application efficiencies, seepage and percolation losses and conveyance losses at all levels in the canal network will be estimates, occasionally supported by limited field validation.
In the light of this operational environment, INCA provides a flexible database structure holding the information necessary for the calculation of water requirement, the planning of water schedules and monitoring of system performance. The software does not define a frequency and density of data collection that must be attained before the model can be applied. Rather, the model may be applied with a minimum of data available and subsequently extended as the scope and intensity of data collection is increased.
Used as a main system planning and operational tool the model is designed to schedule water to discrete units of the scheme, referred to as Management Units (MU). Within each MU, responsibility for water allocation and distribution lies in the hands of another agency or organization, typically water user organizations or individual farmers. Each MU may be sub-divided into 'fields' - a notional division which can be used to represent actual field plots, different villages or administrative units such as water user associations, or different soil types. The 'field' is the unit at which information about cropping activities and actual soil moisture conditions is input to the model.
INCA calculates crop water requirements at 'field' level on the basis of reported cropping activity and monitored climatological conditions. Field requirements are routed through the irrigation network according to operational rules specified by the system management. The required discharge at each control location is evaluated against the minimum and maximum permitted flows at that structure. Warnings are issued wherever the required flow lies outside structure limits and flows are automatically re-scheduled to comply with these system constraints.
The resulting operational schedules form the targets for system operators. The database holds records of structure operations, observed water levels and flow rates passing through the system computed using individual calibration relationships. By linking target setting and monitoring, INCA enables the evaluation of scheme performance following each irrigation release. In this way, any discrepancies between intended and actual operations can be quickly identified and remedial actions initiated.
MODEL FUNCTIONS
The calculation of crop water requirement and the definition of target water releases at system control structures involves the following stages.
Determination of field water requirement
By default, the field water requirement is calculated from long-term mean daily evapotranspiration and crop coefficients for each crop development stage. Indenting is provided as an alternative method allowing water requirement to be based on water orders generated by water user organizations, department officials or farmers. In either case, demand is assessed on a daily basis and the total volume to be delivered is calculated.
Scheduling to meet field demand
INCA adopts two distinct approaches to scheduling which simulate supply regimes typical of paddy rice irrigation and the more irregular irrigation demands of upland crops.
Paddy rice scheduling
For paddy rice, water is scheduled during every rotational period. Where the crop water requirement is based on evapotranspirative demand, the depth supplied is the crop water requirement, plus seepage and percolation losses and land preparation requirements over the schedule period, less the depth expected to be supplied from rainfall.
A correction model, based on feedback of actual field water conditions in the previous rotation and the difference between actual effective and expected rainfall, is used to modify theoretical water requirements in the following period (see below: Correcting the field irrigation requirement and soil moisture balance).
Upland crop scheduling
For non-rice crops, irrigation is not assumed to be required in each period. The model maintains a daily soil moisture balance to simulate field crop behaviour for each field/crop combination. The balance is initialized at a user specified moisture content on the date of planting. The model follows the approach of Doorenbos and Pruitt (1977) regarding total and readily available moisture content. It is not intended to simulate the soil moisture status of a specific farmer's plot, but rather, to be representative of the soil moisture status typical of the field/crop combination in the MU. Again, field data collected during rotations are used to update the model simulations.
Correcting the field irrigation requirement and soil moisture balance
Irrigation scheduling seeks to maintain a favourable soil moisture status at the field level by anticipating future demand and ensuring that water is supplied to meet the demand as it arises. Because neither demand prediction nor operation of the delivery system can be accomplished with complete accuracy, updating and correction of model predictions is included to ensure that the model represents actual field conditions.
INCA incorporates separate correction models for paddy rice and upland crops. For paddy rice, the correction model uses expected and actual effective rainfall, combined with observations of field wetness, to adjust the calculated field irrigation requirement. For upland crops, corrections are made to the soil moisture balance based on either field wetness assessments or, more accurately, soil moisture measurements.
Main canal flow routing
Calculations are based on the calculated volume required to pass each control structure during the schedule period. Scheduled flows are routed through the network at time steps of one hour to model rotational operations. In addition to the corrected irrigation demand, computed at the 'field' level, the routed flows also include:
Inputs and outputs on the network. These represent flows into or out of the canal system in addition to the main supply and demands. Common examples include inflow of cross drainage and abstractions for domestic or industrial use. Each input or output is defined by a daily hydrograph.
Conveyance loss in the main canal system. Losses are defined as a percentage of either the design discharge or the total scheduled flow between two scheduling locations.
Compensation flows. Where a minimum flow must be maintained, the flow required to meet that compensation is added to the scheduled requirements. Compensation flows are defined by daily hydrographs.
Summation of required volumes gives a 'period demand volume' at each scheduling location, and dividing by the remaining time for which the structure is open during the rotation, yields the required discharge. This is compared against the upper and lower flow limits of the structure and, if necessary, the flow is re-scheduled according to the following rules.
Required flow < structure minimum flow limit: Minimum flow limits are used to prevent a canal being operated wastefully to supply small discharges relative to design capacity. Where the required discharge is less than the minimum authorized discharge, the structure and all downstream structures are scheduled a zero discharge for that current time step (one hour).
Required flow > structure maximum flow limit: Where required flows exceed the maximum authorized discharge the advised discharge is reduced to the maximum for that structure. The advised discharges at all downstream structures are proportionally reduced. Similarly, the advised discharges at structures upstream of the constraining structure are reduced to reflect the reduced demand.
The model functions described here have been developed on the basis of field experience in a range of systems. The resulting software provides a robust and flexible management information system, well suited to supporting the planning, implementation and monitoring phases of system management. However, the following section stresses the broader issues that must be addressed to achieve sustainable improvements in performance through the use of computer management aids.
EXPERIENCES WITH IMPLEMENTING COMPUTER ASSISTED WATER MANAGEMENT
INCA has been implemented as a component of water management improvement projects on schemes in Sri Lanka, Thailand, Bangladesh, the Philippines, Jamaica, China and Turkey. The principal characteristics of six of these schemes are summarized in Table 1. From these projects and from ongoing research it has been found that the following issues must be addressed if sustainable improvements are to be achieved.
TABLE 1 - Characteristics of schemes where INCA has been applied
|
Location |
Nominal command (ha) |
Density of control structures1 ha/structure |
Supply source |
Major crops |
Operational pattern |
|
|
Normal |
Water shortage |
|||||
|
Sri Lanka |
2 050 |
60 |
Tank |
Rice |
Intermittent (6-day intervals) |
Head-end farmers served |
|
The Philippines |
3 725 |
230 |
Run-of-river |
Rice |
Continuous flow |
Head-end farmers served |
|
Turkey |
17 500 |
550 |
Regulated river |
Wheat/sugar beet |
Limited on-demand |
Overall supply reduced |
|
Thailand |
21 000 |
250 (all)2 |
Regulated river |
Rice/sugar cane |
Continuous flow |
Within canal rotation |
|
China |
22 000 |
450 |
Pumped from river |
Wheat/maize/cotton |
Rotation below secondaries |
Area limited |
|
Bangladesh |
110 000 |
820 |
Pumped from river |
Rice |
Continuous - main Rotation - sec. |
Within canal rotation |
Notes:1. Density of control structures refers to head and cross regulators and division structures in the main system. Tertiary offtakes are excluded.2. 'All structures' includes weirs.
Appropriate management institutions
Decision support software is designed to improve the way managers make choices between alternatives. In many cases, this is anathema to bureaucratic or administratively oriented agencies where staff are more accustomed to following a single course of action. Deviations from this course may not be tolerated by superiors, and many lower and middle level managers are afraid to deviate from previous practices. Creating a proper management context that ensures a clear and lasting commitment from the agency's hierarchy is therefore a prerequisite for effective intervention to improve water management (Rey et al., 1993).
Adopting computer models requires that a management framework exists, or can be established, that recognizes the need to set objectives, develop implementation plans and monitor the consequence of management choices (Murray-Rust and Cornish, 1993). Without an organization able to respond to the output of the model, any implementation is likely to result in failure.
An institutional factor, that will become increasingly important, is the need to be able to recognize and reward managers who achieve higher levels of performance. There is little motivation to undergo training and adopt new work practices if the rewards are identical to those obtained following traditional processes (Murray-Rust and Cornish, 1993).
System control structures and distribution monitoring
Monitoring water distribution is central to improved operations. In many cases it is necessary to calibrate regulating and gauging structures before quantitative control of water distribution can be implemented. Furthermore, on many schemes, existing data collection procedures are weak. Changing the way water requirements are estimated will not result in improved delivery unless repairs to control gates and water level gauges are carried out at an early stage of implementing changes. Strengthening the procedures to collect and verify data and ensuring field staff are present at control sites, and are sufficiently well trained to execute the control functions, are essential components of improving system management.
Training of staff
Comprehensive software, such as INCA, may at first be daunting for new users. Training should not be compressed into too short a period of time, users require time to learn the software's capabilities progressively. Users need to be introduced to additional capabilities when those functions are actually required rather than when it is convenient in the training programme. We have found a combination of formal training and extended on-the-job training programmes for users to be the best approach to ensure that new procedures are fully understood and adopted.
The training of field staff, to collect field data and set discharges at control structures, and of supervisory staff, who will run the software and oversee the application of output, is of paramount importance. Parallel to providing training in the practical procedures required to monitor water levels and structure settings, use calibration tables or assess field water status, training is essential to sensitize engineers and technicians to the potential for improving water management and the importance of accurate and timely field procedures. Where such training is lacking, or where those trained are unresponsive, changes brought about during the implementation phase are likely to be short lived.
Communications with water users
In part, the impetus for improving water management derives from the trend for turnover of tertiary distribution systems to water users. However, to have an impact on overall project performance provision must be made for the introduction of effective dialogue between the scheme management and water users. This may take the form of regular meetings between field staff and water user groups, at which farmers can report on the acceptability of deliveries at the turnouts and on the status of crops within the command area. More importantly field staff and farmers can discuss any problems foreseen for future deliveries. For such dialogue to be effective and sustainable it must be transparent, enabling farmers to see that, where practical, action is taken in response to their observations.
Time scale
Installation of software and training of staff in its operation is only one component of what must often be a larger intervention, focusing on improved main system management. Whilst staff, with training, may become proficient in the operation of the software within six weeks of installation, other actions required, before the software can play a useful role, may take a number of seasons to implement. Apart from the time required for installation of gauges and structure calibration there is also the need to establish reliable data collection routines and develop dialogue between the agency and farmers. Frequently the need to change entrenched patterns of thinking regarding water management, both on the part of agency staff and water users, is essential. A period of two or three years should be allowed before a water management improvement programme can be effectively evaluated.
CONCLUSIONS
Experience of implementing INCA in a number of countries has led to the following conclusions about the practical aspects of irrigation water scheduling on medium to large-scale smallholder irrigation schemes. The primary conclusion is that, irrespective of the capabilities of management software, improvements in irrigation system performance cannot be achieved solely by providing tools to assist in decision making.
It is essential that the managers of irrigation systems recognize the need for improved water management and are committed to changes in working practices. Support for achieving improved system performance must extend throughout the agency, which in turn requires agreement on the definition of performance and the means of measuring it.
An appropriate management framework must exist before advanced decision support software, such as INCA can be effective. The framework must deal with recurrent processes, including setting objectives, implementing plans, monitoring the impacts of management actions and evaluating the best course of action. Weak data collection programmes must be strengthened. Without a realization of the value of accurate and timely information it is unlikely that computer assisted water management will be successful.
Implementation of new technology to irrigation schemes must be carried out carefully. New decision support technology should replicate the management activities existing prior to interventions. This assists project staff to better understand the function of the software during training.
Training requires time. A single intensive course in the operation of the software will not be effective. New users require time to become familiar with the operation of the software. Once familiar with the day-to-day use of the software, users are able to focus on its application. One or more follow-up periods of training may be required before it can be said that a sustainable transfer of technology has been achieved.
Where the software has been implemented initial results have been encouraging. In Sri Lanka water releases over the wet season have been cut by 22% without detriment to the equity or adequacy of water supply. Water savings in the wet season have carried over into the following dry season to provide a more assured supply to the full command area. In Thailand the water saving was less marked, but a saving of 9% has been shown when comparing wet season releases with the mean releases over the seven seasons before implementation. At the Ganges-Kobadak scheme in Bangladesh the introduction of systematic water management based on the INCA software led to a 19% reduction in volume pumped at the headworks, with consequent savings in pumping costs and parallel improvement in the equity of water distribution, (findings based on comparison of performance for kharif II 1992 and kharif II 1993).
These improvements in system performance are not brought about simply by more accurate determination of water requirements, but are the logical consequence of additional attention being directed to systematic monitoring of system operations, and the existence of mechanisms that are responsive to observed distribution patterns. INCA software can play an important role in enabling these tasks to be performed, when the other requirements identified in this paper are addressed.
REFERENCES
Abernethy, C.L. 1985. Irrigation water management at Kaudulla, Sri Lanka: a review of measurements made 1978 to 1983 and a proposal for new management methods. Hydraulics Research, Wallingford. Report OD70.
Bottrall, A. 1981. Improving Canal Management - The Role of Evaluation and Action Research. Water Supply and Management 5: 67-79.
Chambers, R. 1998. Managing Canal Irrigation - Practical analysis from South East Asia. Cambridge University Press.
Doorenbos, J. and Pruitt, W.O. 1977. Crop water requirements. FAO Irrigation and Drainage Paper 24. FAO, Rome.
FAO. 1994. Irrigation Water Delivery Models. Proceedings of the FAO Expert Consultation, Rome, Italy, 4-7 Oct. 1993. Water Report 1. Rome.
Lenselink, K.J. and Jurriens, J. 1992. Inventory of Irrigation Software for Microcomputers. ILRI. Wageningen.
Murray-Rust, D.H. and Cornish, G.A. 1993. Improving irrigation management using decision support systems, A framework and review of IIMI's Experience. Irrigation Management Improvement Project, Final Report. IIMI and HR Wallingford.
Oskam, R. 1994. The relevance of Computerised Decision Support for Water Management. Unpublished M.Sc. Thesis, University of Twente, Netherlands.
Plusquellec, H.L., McPhail, K. and Polti, C. 1990. Review of irrigation systems performance with respect to initial objectives. Irrigation and Drainage Systems 4: 313 - 327.
Rey J., Murray-Rust D.H. and Sally, H. 1993. Improving irrigation management using decision support systems. A framework and review of IIMI's experience. Irrigation Management Improvement Project, Final Report. IIMI and HR Wallingford.
Sagardoy, J.A., Bottrall, A. and Uittenbogaard, G.O. 1986. Organization, operation and maintenance of irrigation schemes. FAO Irrigation and Drainage Paper 40. FAO, Rome
Van der Krogt, W.N.M. 1993. OMIS: A Package for Irrigation System Management. Paper presented at FAO Expert Consultation on Irrigation Delivery Models. FAO Rome, 4-7 October 1993.
