RSS | Register/註冊 | Log in/登入
Site search:
Home>FFTC Document Database>Extension Bulletins>Agricultural Water Management Systems in Taiwan: Current Status and Policy Directions
facebook分享

MingDaw Su, Professor
Dept. of BioEnvironmental Systems Engineering
National Taiwan University, Taipei, Taiwan, 10617
sumd@ntu.edu.tw

Yi Fong Ho, Technical Specialist
Council of Agriculture, Executive Yuan, Taiwan
No.37, Nanhai Road, Taipei, Taiwan 10014
yifong@mail.coa.gov.tw

ABSTRACT

Water demands increase rapidly in most countries from rapid population growth and booming economic and industrial activities. Water and land are vital natural resources for food production. Agricultural sector usually uses the largest water supply share, especially in the rice production countries like Taiwan, Japan and Korea. The agricultural water sector faces an acute pressure for improving its water use efficiency to produce more crops from fewer drops. There is also a request in Taiwan for water rights reallocation among sectors to shift part of agricultural water rights to the domestic and industrial water use sectors for water supply deficits mitigation. This paper reports the experiences of using Geographic Information System (GIS) and Spatial Decision Support System (SDSS) to implement an agricultural water management system in Taiwan. The irrigation demand planning system based on spatial database and map-based user interfaces can better use the limited water supply especially under water supply deficit situations such as in a drought year. Some strategic water resources management policies currently studied in Taiwan were also reported including: fallow strategy for saving more water, a risk-based water demand planning scheme for reallocation among different water use sectors.

Keywords: Decision support system, GIS, Management, Irrigation demand planning


INTRODUCTION

Most countries in the world are facing increasingly severe water supply deficit problems due to rapid population and growths in socio-economic activities. Agriculture usually accounts for the biggest portion of the total water consumption among all water use sectors. The agricultural sector uses more than 70% of the total water supply in Taiwan (as shown in Fig. 1), even under that situation of continuously decreasing irrigated areas. As it consumes more than 60% of the total water supplies, irrigation water is frequently considered as a potential water reallocation source for municipal or industrial sectors during drought for water deficit relief.

Under the impact of climate change, the spatial and temporal fluctuations in regional precipitation distribution become larger in recent years. The new water supply developments become more difficult and less economic efficient due to environmental concerns. The pressure of water supply deficit is getting more and more severe. There raises a keen request to improve irrigation water use efficiency, because its high water demand and low GDP share. The estimation of regional irrigation water demand becomes more and more important in such circumstances.

REGIONAL IRRIGATION DEMAND PLANNING

Enormous amounts of knowledge can be extracted from the literature regarding the estimation of on-farm evapotranspiration (ET) for crops. Most of this research concentrates on finding the water requirements for different crops under certain field conditions related to soil, climate, and the groundwater table. Less work has been done in the area of regional irrigation water demand estimation. For large-scale irrigation planning and management practices, regional irrigation water demand is usually estimated from multiplying the irrigated area by the irrigation demand per unit area. There is actually a huge amount of data involved in estimating regional irrigation water demand, such as soil, crop, climate, water distribution infrastructures, and management practices. Most of these related data are spatially distributed. Since these spatial complexities are very difficult to handle, they are usually combined into aggregated forms. Traditionally, irrigation water demand estimations have had to assume a particular soil type for each crop due to a lack of data (Knox et al. 1996). Average soil characteristics were used regardless of any non-uniform field soil distribution. Because there were no suitable tools for handling the spatial complexities of soil, crops, and climate in the past, it was thought that the small-scale physical processes described by the sophistical ET equations were point measures that could not be scaled-up (Morton 1983).

Because of the ignorance of the spatial distribution, the impacts of the spatial variations of related parameters on the regional irrigation water demand are not clearly identified. Decisions based on these traditional "lump estimation" models may lead to inadequate water system planning. Therefore, a spatial approach is particularly appropriate (Knox and Weatherhead 1999). The Geographic Information System (GIS) is the most efficient tool for spatial data management and utilization that can allow us to improve our understanding of the spatial variance. A GIS-based approach to demand modeling allows us to consider local variations in cropping, soils, and climate, and so facilitates the production of irrigation demand maps (Weatherhead and Knox 1999).

Over the last decade, GIS-based irrigation water mapping techniques have been the subject of widespread discussion. A regional irrigation water demand model using a vector-based data model is introduced in this paper. The model framework is shown in Fig. 1. This GIS-based irrigation water demand model is not only for mapping the irrigation demands; it also provides the ability to accurately capture spatial variations and quickly reflect changes in irrigation water demand in response to changes in cropping patterns. The relationship between irrigation water demands and cropping patterns is crucial to large-scale irrigation planning and for regional water resource allocation during severe droughts. The purpose of regional irrigation water-demand planning is to set up suitable cropping patterns and estimate irrigation water demand for agricultural development (Wen, et al. 2004).

A SCENARIO-BASED DECISION SUPPORT MODEL

Decision making during regional irrigation water-demand planning typically involves many stakeholders and numerous competing objectives, such as maximum pro?ts for farmers, maximum production for the agricultural sector, and minimum water consumption. Weighing these objectives is dif?cult for decision makers especially, in some cases, these objectives are in con?ict. Developing an explicit and structured decision making process is dif?cult due to the multiple implicit objectives and factors affecting cropping patterns. Therefore, in practice, decision making typically depends on the experience of water managers rather than mathematical decision models (Fig. 2).

Despite the non-structured nature of the decision making process, researchers commonly employ structured approaches such as system optimization to determine optimal cropping patterns for regional irrigation planning. Linear programming models have been extensively applied to establish optimal cropping patterns that maximize net return or crop yield (Raman et al. 1992; Prajamwong et al. 1997; Singh et al. 2001). However, these optimization techniques only work in the speci?cally prede?ned situations. Once signi?cant changes occur (e.g., changes in land use, climate, or irrigation practices), decision makers cannot rely on these prede?ned "optimal" cropping patterns, but rather need tools to reformulate the problems, establish the models, and validate the analytical results. As a result of these in?exible or inef?cient processes, decision makers require additional alternatives for evaluation, including tools for scenario building and making comparisons among scenarios (Buras, 2000).

Therefore, rather than determining optimal cropping patterns, a scenario-based planning framework (as shown in Fig. 3 and Fig. 4) that allows decision makers to generate and compare planning scenarios is introduces (Wen, et al. 2007). For instance, it is possible to evaluate the effects on water-demand reduction when the irrigated area is reduced or if some paddy ?elds are shifted to other crops that require less amounts of water. The increased ?exibility for evaluating alternatives improves the effectiveness of decision making in regional irrigation demand planning. This scenario-based planning process using a GIS platform to aid water managers in spatial decision making during irrigation demand planning.

The concept of scenario planning is not new to systems evaluation. From business strategic planning to environmental analysis, the use of scenario planning is common in the planning community (Klein 1994; Deaton and Winebrake 1999; Martin et al. 2003). For example, in municipal water-demand forecasting, different scenarios may be developed based on population growth, economic development, or pricing policies that have certain probabilities of occurring in the future (Levite et al. 2003). This approach differs from conventional planning approaches, such as prediction or optimization. Rather than providing a single optimal solution, scenario-based analysis allows decision makers to establish scenarios based on speci?c criteria to determine the complexity of systems, generate experience and knowledge from different combinations of possible future events, and ?nally, produceeffective and robust strategies or decisions.

CHIA-NAN IRRIGATION ASSOCIATION, TAIWAN

The command area of the Chia-Nan Irrigation Association of Taiwan, which covers three river basins, with two major reservoirs and several river diversions, is used as demonstration for this regional irrigation demand planning model. The whole irrigated area, of approximately 78,000 hectares, is divided into seven management subdivisions. Each subdivision consists of several workstations, and each workstation is further divided into several irrigation groups. An irrigation group, made up of 3_5 rotational units with an area of approximately 50 hectares, is the basic irrigation management unit for this irrigation association. The term "rotational unit" comes from the cropping pattern. Paddy rice is the traditional cash crop in Taiwan. A cropping pattern with paddy rice and other crops like soybean, corn and sorghum, rotating each year between these rotational units within an irrigation group, is adopted in the study area because of the insufficient water supply for all-paddy cropping.

Demand planning during a drought in the ChiaNan irrigation command area is used as an example demonstrating the proposed framework for spatial scenario analysis. Three scenarios for fallow planning in the drought year to cut down the irrigation demands are shown in Fig. 5. Scenario A-1 is to leave area with distance from water sources longer than 35 km (areas shown in black) as fallow. Scenario A-2 is to leave area with percolation rate larger than 17 mm/day as fallow. Scenario A-3 is to leave area of two paddy crops. The detailed spatial databases build in the model and the GIS analytical capability make this scenario set up possible.

In addition to decision scenarios based on above mentioned criteria of soil percolation loss, cropping patterns and distance to water supply source, the optimistic (low temperature/high rain), pessimistic (high temperature/low rain), and most likely (medium temperature and rain) meteorological conditions for irrigation are also generated to examine the impact of stochastic characteristics of meteorological conditions to irrigation water demands. Based on simulations under different conditions of climate uncertainty, decision makers can identify possible ranges of water demand variations. The simulation results are shown as in Table 1.

The total fallow area is 21,132 ha in Scenario A-1, and 72.76_74.91million m3 of water are saved from paddy fallow. Conversely, in Scenario A-3, 12,330 ha are left fallow, accounting for about half of the fallow area in Scenario A-1; however, roughly the same amount of water is saved. Ef?ciency of demand reduction for Scenario A-3 is highest. Notably, the amount of fallow land is not proportional to the amount of irrigation demand reduction. The main reason for these differences is the percentage of fallow land that was originally planned for rice paddies. The paddy area is approximately 60% of total fallow area in Scenario A-1 and 100% of that in Scenario A-3. Leaving more areas designated as paddy crops fallow could result in higher ef?ciency in irrigation water demand reduction.

Other than the water supply problem, some fallow policy is required to modulate paddy production due to the globalization issue as Taiwan joined the WTO in 2001. The irrigation demand reduction per unit fallow area is not uniform due to many factors such as spatial aggregation, distance from the water source, soil texture and percolation rate. Thousands simulation were run using the proposed model for irrigation demand estimation under different percentages of fallow. The results are shown in Fig. 6. From the figure, the percentage of irrigation demand reduction is not proportionally varied with the percentage of fallow area. For example, a 50% fallow of the total area may result in an irrigation demand reduction between 40%-55%.

There are many feasible solutions for saving the same amount of irrigation demand with different spatial combination of fallow areas. For example, if the irrigation manager is asked to cut down 8% of the total regional irrigation demand due to water supply deficit from a drought. (S)he may have several options to accomplish this mission of irrigation demand reduction. We tentatively proposed four possible scenarios from the perspective of distance from water supply source, the conveyance loss, the deep percolation loss and the cropping patterns: The first scenario is to fallow the area most distant to the water supply diversion; the second one is to fallow the irrigated areas of two poorly maintained canals. The third scenario is to fallow the area with mostly sandy soil and the last one is to fallow the area with double paddy crops. As shown in Fig. 7, the three scenarios are having roughly the same irrigation demand reductions but with quiet different fallow areas.

By incorporating water manager decision logic into scenario generation, the scenario-based planning framework provides comprehensive information and performance indices for comparing various scenarios. This scenario-based planning framework can generate spatial scenarios and allows water managers to compare the associated decision rules with uncertainties of stochastic phenomena and various evaluation indices, including ef?ciencies of saving water, and the impact of leaving land fallow. It can be used to determine suitable cropping patterns rather than merely generating an optimal solution. Furthermore, the decision process based on scenario-based planning can satisfy water managers who realize that interactions of these factors impact speci?c spatial water demand scenarios. A GIS is incorporated into this scenario-based planning framework for capturing spatial variations in the environment and to enable spatial orientation and visualization for effective scenario set up. The spatial impact of each scenario can be visually analyzed. Integrating the scenario-based planning model with spatial analysis in the GIS generates additional understanding of spatial and temporal characteristics during irrigation planning.

STRATEGIC WATER RESOURCES MANAGEMENT STUDIES IN TAIWAN

As mentioned above, the agricultural water use sector in Taiwan is asked to reallocate part of its water right to municipal and industrial sectors due to water supply shortage pressure. The request becomes keener in the recent years from its ever decreasing GDP share. The agriculture sector is also thought to have higher water shortage tolerance than municipal and industrial sectors. While the municipal and industrial sectors enjoy their benefits, the reallocation of water supply from agriculture sector will definitely increase its water supply shortage risk. It is necessary to assess the relationships between water right reallocation and the water deficit risk for reasonable policy implementation. The Chia-Nan Region is used as a pilot study area for this relationship of water right and risk shift among sectors. Other than the irrigation demand model as mentioned above, models for both municipal and industrial water demand estimations are enclosed into the system framework. A linear programming model is also included for regional water supply water allocation. The detailed water system scheme is also needed to be incorporated into the system for this study.

Water supply deficit rates of agriculture and municipal and industrial sectors are firstly simulated under climate scenarios with different occurrence probabilities as shown in Fig. 8 as the base line for comparison. The flat portion of the agriculture cure shows that the water supply is normally lower than what is actually needed for irrigation.

The whole routine is run again with certain percentage of agricultural water right reallocated to the municipal and industrial sector. The water supply deficit risk in agriculture sector is expected to be increased while that in the municipal and industrial sector decreased from this water right reallocation as shown in Fig. 9. The relationship between the amount of reallocation water and the risk changed can then be identified for the implementation of this water right reallocation policy. This information is also helpful for assessing the loss and gain in different sectors from this water right reallocation, and to establish a scheme for proper economic compensation in the agriculture sector.

CONCLUSION

Water is a scare and precious natural resource not only for agricultural production but also for municipal, industrial, and environmental use. The competition of limited water supplies among all water use sectors are very acute and this situation is expected to get worse in the future. Agriculture sector (mostly irrigation) traditionally uses the largest portion of the total water supply but its GDP contribution is usually low, especially in the developing and developed countries. It is necessary for the irrigation planners and managers to estimate the irrigation demands more accurately and also prepare for water supply reduction when limited water supplies are to be used more economically and efficiently. A scenario-based multi-sectorial regional water demand estimation and water supply allocation model is introduced for this purpose.

REFERENCES

  • Buras, N. (2000), "Building new water resources projects or managing existing systems?" Water Int.,25(1), 110_114.
  • Deaton, M. L., and Winebrake, J. J. (1999). Dynamic modeling of environmental systems, Springer, New York.
  • Klein, M. 1994. "Reengineering methodologies and tools." Inf. Sys., 11(2), 30_35.
  • Knox JW, Weatherhead EK, Bradley RI (1996) Mapping the spatial distribution of volumetric irrigation water requirements for main-crop potatoes in England and Wales. Agricultural Water Management 31, pp.1_15.
  • Knox JW, Weatherhead EK, Bradley RI (1996) Mapping the spatial distribution of volumetric irrigation water requirements for main-crop potatoes in England and Wales. Agricultural Water Management 31, pp.1_15.
  • Knox JW, Weatherhead EK (1999) The application of GIS to irrigation water resources management in England and Wales. Geographical Journal 165(1), pp.90_98.
  • Levite, H., Sally, H., and Cour, J. (2003). "Testing water-demand management scenarios in a water-stressed basin in South Africa: Application of the WEAP model." Phys. Chem. Earth, Part B,28(20_27), 779_786.
  • Martin, M., Warren, R., and Stanley, G. (2003). "Rural water use, a system approach to three years of challenges with triumphs, a Queensland dairy perspective." Proc., 2003 APEN National Forum, Hobart, Australia, 26_28.
  • Morton FI (1983) Operational estimates of areal evaportranspiraiton and their significance to science and practice of hydrology. Journal of Hydrology 66, pp.1_76
  • Raman, H., Mohan, S., and Rangacharya, N. C. V. (1992). "Decision support for crop planning during drought." J. Irrig. Drain. Eng., 118, 229_241.
  • Singh, D. K., Jaiswal, C. S., Reddy, K. S., Singh, R. M., and Bhandarkar, D. M. (2001). "Optimal cropping pattern in canal command area." Agricultural Water Management, 50, 1_8.
  • Weatherhead EK, Knox JW (1999) Predicting and mapping the future demand for irrigation water in England and Wales. Agricultural Water Management 43, pp.203_218.
  • Wen JH, Su MD, Yeh YL, (2004), A GIS-based Framework of Regional Irrigation Water Demand Assessment, Paddy and Water Environment 2:33-39.
  • Wen JH, Lin CH, Chen CT, Su MD, (2007) Analysis of Spatial Scenarios Aiding Decision-Making for Regional Irrigation Water-Demand Planning, Journal of Irrigation and Drainage Engineering ASCE 133(5):455-467 (SCI).


Index of Images

  • Fig. 1 Sectorial water demands in Taiwan

    Fig. 1 Sectorial water demands in Taiwan

  • Fig. 2 The regional irrigation water demand model framework (Wen, et al. 2007)

    Fig. 2 The regional irrigation water demand model framework (Wen, et al. 2007)

  • Fig. 3 A scenario-based simulation model framework (Wen, et al. 2007)

    Fig. 3 A scenario-based simulation model framework (Wen, et al. 2007)

  • Fig. 4 Model interface for scenario generation

    Fig. 4 Model interface for scenario generation

  • Fig. 5 Different decision scenarios for fallow planning

    Fig. 5 Different decision scenarios for fallow planning

  • Fig. 6 Fallow impact on regional irrigation demand reduction

    Fig. 6 Fallow impact on regional irrigation demand reduction

  • Fig. 7 Scenario comparison for regional fallow decision

    Fig. 7 Scenario comparison for regional fallow decision

  • Fig. 8 Current exceeding probability of water supply deficit

    Fig. 8 Current exceeding probability of water supply deficit

  • Fig. 9 Exceeding probability of water supply deficit after water reallocation<BR>

    Fig. 9 Exceeding probability of water supply deficit after water reallocation

  • Table 1 Water-demand reduction under different decision scenarios

    Table 1 Water-demand reduction under different decision scenarios

Download the PDF. of this document (536), 1,917,919 bytes (2 MB).