Post on 21-Jul-2016




1 download

Embed Size (px)


<ul><li><p>ABSTRACT: This study presents three optimization techniques foron-farm irrigation scheduling in irrigation project planning: namelythe genetic algorithm, simulated annealing and iterative improve-ment methods. The three techniques are applied to planning a394.6 ha irrigation project in the town of Delta, Utah, for optimiz-ing economic profits, simulating water demand, and estimating thecrop area percentages with specific water supply and planted areaconstraints. The comparative optimization results for the 394.6 hairrigated project from the genetic algorithm, simulated annealing,and iterative improvement methods are as follows: (1) the seasonalmaximum net benefits are $113,826, $111,494, and $105,444 perseason, respectively; and (2) the seasonal water demands are3.03*103 m3, 3.0*103 m3, and 2.92*103 m3 per season, respectively.This study also determined the most suitable four parameters ofthe genetic algorithm method for the Delta irrigated project to be:(1) the number of generations equals 800, (2) population size equals50, (3) probability of crossover equals 0.6, and (4) probability ofmutation equals 0.02. Meanwhile, the most suitable three parame-ters of simulated annealing method for the Delta irrigated projectare: (1) initial temperature equals 1,000, (2) number of moves equal90, and (3) cooling rate equals 0.95.(KEY TERMS: genetic algorithm; simulated annealing; iterativeimprovement; optimization; irrigation planning.)</p><p>Kuo, Sheng-Feng, Chen-Wuing Liu, and Shih-Kai Chen, 2003. ComparativeStudy of Optimization Techniques for Irrigation Project Planning. J. of theAmerican Water Resources Association (JAWRA) 39(1):59-73.</p><p>INTRODUCTION</p><p>Irrigation planners must analyze complex climate-soil-plant relationships and apply mathematical opti-mization techniques to determine optimally beneficialcrop patterns and water allocations. A computerbased model which simulates the climate-soil-plant</p><p>systems with a novel mathematical optimization tech-nique could help irrigation planners make sound deci-sions before each crop season.</p><p>Recently, external influences such as environmen-tal concerns and global trade have been creating newchallenges for the agricultural engineers. Precisionagriculture (PA) is a management strategy that usesinformation technologies to bring data from multiplesources to bear on decisions associated with crop pro-duction (NRC, 1997). Agricultural engineers usinginformation technologies such as genetic algorithm(GA) and simulated annealing (SA) methods will playan increasingly important role in natural resourcemanagement and crop production to meet the newchallenges in the 21st Century (NRC, 1997). There-fore, this preliminary study used the GA and SAmethods to solve the linear programming problem foroptimizing the net benefit of on-farm irrigated projectand the uncertainty expected when extending the on-farm irrigation scheduling to more complicated waterresources management problems in the future, suchas: (1) optimizing combinations of surface and groundwater for irrigation applications, and (2) extendingon-farm irrigation scheduling to optimize reservoiroperation to conserve water resources. Furthermore,the traditional optimization method, iterativeimprovement, was used to compare the GA and SAmethods to distinguish the results from different ran-dom search procedures.</p><p>Many models (Hill et al., 1982; Keller, 1987; Smith,1991; Prajamwong, 1994) simulate on-farm irrigationwater demands based on climate-soil-plant systems.The traditional optimizing irrigation planning model</p><p>1Paper No. 00134 of the Journal of the American Water Resources Association. Discussions are open until August 1, 2003.2Respectively, Associate Professor, Department of Leisure Management and Graduate Institute of Resource and Environment Manage-</p><p>ment, Leader College, Tainan, Taiwan 709, ROC; Professor, Department of Bioenvironmental System Engineering, National Taiwan Universi-ty, No. 1, Section 4, Roosevelt Road, Taipei, Taiwan 106, ROC 10617; and Researcher, ChiSeng Water Management R&amp;D Foundation, 2F, No.39, Lane 9, Section 4, Mu-Cha Road, Taipei, Taiwan, ROC (E-Mail/Liu:</p><p>JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 59 JAWRA</p><p>JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATIONFEBRUARY AMERICAN WATER RESOURCES ASSOCIATION 2003</p><p>COMPARATIVE STUDY OF OPTIMIZATION TECHNIQUESFOR IRRIGATION PROJECT PLANNING1</p><p>Sheng-Feng Kuo, Chen-Wuing Liu, and Shih-Kai Chen2</p><p>Used Mac Distiller 5.0.x Job OptionsThis report was created automatically with help of the Adobe Acrobat Distiller addition "Distiller Secrets v1.0.5" from IMPRESSED GmbH.You can download this startup file for Distiller versions 4.0.5 and 5.0.x for free from</p><p>GENERAL ----------------------------------------File Options: Compatibility: PDF 1.2 Optimize For Fast Web View: Yes Embed Thumbnails: No Auto-Rotate Pages: No Distill From Page: 1 Distill To Page: All Pages Binding: Left Resolution: [ 600 600 ] dpi Paper Size: [ 612 792 ] Point</p><p>COMPRESSION ----------------------------------------Color Images: Downsampling: Yes Downsample Type: Bicubic Downsampling Downsample Resolution: 151 dpi Downsampling For Images Above: 227 dpi Compression: Yes Automatic Selection of Compression Type: Yes JPEG Quality: Medium Bits Per Pixel: As Original BitGrayscale Images: Downsampling: Yes Downsample Type: Bicubic Downsampling Downsample Resolution: 300 dpi Downsampling For Images Above: 450 dpi Compression: Yes Automatic Selection of Compression Type: Yes JPEG Quality: Medium Bits Per Pixel: As Original BitMonochrome Images: Downsampling: Yes Downsample Type: Bicubic Downsampling Downsample Resolution: 600 dpi Downsampling For Images Above: 900 dpi Compression: Yes Compression Type: CCITT CCITT Group: 4 Anti-Alias To Gray: No</p><p> Compress Text and Line Art: Yes</p><p>FONTS ---------------------------------------- Embed All Fonts: Yes Subset Embedded Fonts: Yes Subset When Percent Of Characters Used is Less: 100 % When Embedding Fails: Cancel JobEmbedding: Always Embed: [ ] Never Embed: [ ]</p><p>COLOR ----------------------------------------Color Management Policies: Color Conversion Strategy: Leave Color Unchanged Intent: DefaultDevice-Dependent Data: Preserve Overprint Settings: No Preserve Under Color Removal and Black Generation: No Transfer Functions: Preserve Preserve Halftone Information: No</p><p>ADVANCED ----------------------------------------Options: Use and No Allow PostScript File To Override Job Options: No Preserve Level 2 copypage Semantics: No Save Portable Job Ticket Inside PDF File: No Illustrator Overprint Mode: No Convert Gradients To Smooth Shades: Yes ASCII Format: NoDocument Structuring Conventions (DSC): Process DSC Comments: No</p><p>OTHERS ---------------------------------------- Distiller Core Version: 5000 Use ZIP Compression: Yes Deactivate Optimization: No Image Memory: 524288 Byte Anti-Alias Color Images: No Anti-Alias Grayscale Images: No Convert Images (&lt; 257 Colors) To Indexed Color Space: Yes sRGB ICC Profile: sRGB IEC61966-2.1</p><p>END OF REPORT ----------------------------------------</p><p>IMPRESSED GmbHBahrenfelder Chaussee 4922761 Hamburg, GermanyTel. +49 40 897189-0Fax +49 40 897189-71Email: info@impressed.deWeb:</p><p>Adobe Acrobat Distiller 5.0.x Job Option File</p><p> /ColorImageDownsampleType /Bicubic /GrayImageDict &gt; /CalCMYKProfile (U.S. Web Coated (SWOP) v2) /ParseDSCComments false /PreserveEPSInfo false /MonoImageDepth -1 /AutoFilterGrayImages true /SubsetFonts true /GrayACSImageDict &gt; /ColorImageFilter /DCTEncode /AutoRotatePages /None /PreserveCopyPage false /EncodeMonoImages true /ASCII85EncodePages false /PreserveOPIComments false /NeverEmbed [ ] /ColorImageDict &gt; /AntiAliasGrayImages false /GrayImageDepth -1 /CannotEmbedFontPolicy /Error /EndPage -1 /TransferFunctionInfo /Preserve /CalRGBProfile (sRGB IEC61966-2.1) /EncodeColorImages true /EncodeGrayImages true /ColorACSImageDict &gt; /Optimize true /ParseDSCCommentsForDocInfo true /GrayImageDownsampleThreshold 1.5 /MonoImageDownsampleThreshold 1.5 /AutoPositionEPSFiles false /GrayImageResolution 300 /AutoFilterColorImages true /AlwaysEmbed [ ] /ImageMemory 524288 /OPM 0 /DefaultRenderingIntent /Default /EmbedAllFonts true /StartPage 1 /DownsampleGrayImages true /AntiAliasColorImages false /ConvertImagesToIndexed true /PreserveHalftoneInfo false /CompressPages true /Binding /Left&gt;&gt; setdistillerparams&gt; setpagedevice</p></li><li><p>seeks to optimize the values to satisfy the objectivefunction and constraints. Traditional optimizationmodels have received extensive attention in irrigationplanning (Lakshminarayana and Rajagopalan, 1977;Maji and Heady, 1978; Matanga and Marino, 1979;Jesus et al., 1987; Paudyal and Gupta, 1990; Ramanet al., 1992). Jesus et al. (1987) developed a linearoptimization model for managing Irrigation DistrictNo. 38 in Sonora, Mexico. Meanwhile, Paudyal andGupta (1990) employed a multilevel optimizationtechnique to resolve the complex problem of irrigationmanagement in a large heterogeneous basin. Further-more, Raman et al. (1992) presented a linear pro-gramming model to generate optimal croppingpatterns based on data tracking previous droughts. Areview of the literature reveals that traditional opti-mization methods have limitations in finding nearglobal optimization results and are difficult to applyto complex irrigation planning problems since theyseek the optimization by searching point to point.Meanwhile, more recently proposed optimizationmethods, such as the genetic algorithm and the simu-lated annealing method, search the entire populationinstead of moving from one point to the next and thusmay overcome the limitations of traditional methods.</p><p>A genetic algorithm (GA) is a search procedure thatuses random choice as an effective means of directinga highly exploitative search through a numerical cod-ing of a given parameter space (Goldberg, 1989). Agenetic algorithm has been applied to several opti-mization problems (Wang, 1991; Wentzel et al., 1994;Fahmy et al., 1994; McKinney and Lin, 1994; Reddy,1996; Montesinos et al., 2001). Wang (1991) proposeda genetic algorithm for function optimization andapplied it to the calibration of a conceptual rainfall-runoff model. Wentzel et al. (1994) used GAs to opti-mize a pipe network pumping strategy at New MexicoState University. Fahmy et al. (1994) used GAs foreconomic optimization of river management. Accord-ing to the results of that study, GAs generated nearoptimum solutions for large and complex waterresource problems more efficiently than dynamic pro-gramming techniques. Meanwhile, McKinney and Lin(1994) incorporated GAs into a ground water simula-tion model to solve three ground water managementproblems: (1) maximum pumping from an aquifer, (2)minimum cost in water supply development, and (3)minimum cost in aquifer remediation. Furthermore,Reddy (1996) developed a nonlinear optimizationmodel based on genetic algorithms for land gradingdesign of irrigation fields. Additionally, Montesinos etal. (2001) designed a seasonal furrow irrigation modelwith genetic algorithm to determine a quasi optimumirrigation season calendar based on economic profitmaximization.</p><p>On the other hand, simulated annealing (SA) is astochastic computational technique derived from sta-tistical mechanics for finding near global solutions to large optimization problems (Davis,1991). Severalworks have applied simulated annealing to waterresource management (Dougherty and Marryott,1991; Marryott et al., 1993; Mauldon et al., 1993) andirrigation scheduling (Walker, 1992). Dougherty andMarryott (1991) applied simulated annealing to fourproblems of optimal ground water management: (1) adewatering problem, (2) a dewatering problem withzooming, (3) a contamination problem, and (4) con-taminant removal with a slurry wall. Walker (1992)applied the simulated annealing method to a peanutgrowth model to optimize irrigation scheduling. Thepeanut growth model was first applied to determinethe number of irrigation days and the amount of irri-gation during the season. Later, simulated annealingwas implemented in the peanut model.</p><p>This study first adopts an irrigation simulation andplanning model to simulate the on-farm surface irri-gation system and obtain the water demand and cropyield information. Three optimization methods, name-ly genetic algorithm, simulated annealing, and inter-active improvements are then applied to maximizethe net benefit of a sample irrigation project in Utah.Various constraints, including minimum and maxi-mum planted crop area and an upper limit on thewater supply, are imposed to satisfy the field condi-tions. Results obtained from the three optimizationmethods are then evaluated to determine the besttechnique for optimizing economic benefits, determin-ing the ideal crop area for a given water supply, andgenerally assisting irrigation planner in makingsound decisions before each crop season.</p><p>METHODS</p><p>Irrigation Simulation and Planning Model</p><p>The irrigation simulation and planning model isdominated by six basic modules: (1) the main module,which directs the running of the model with pull-down menus ability; (2) the data module, for dataentry via a user friendly interface; (3) the weathergeneration module, which generates the daily weath-er data; (4) the on-farm irrigation scheduling module,which simulates the daily water requirements andrelative crop yields; (5) the three optimization tech-niques module, which optimize the project maximumbenefit; and (6) the results module, which presentsresults using tables, graphs, and printouts and subse-quently sends these results to the three optimization</p><p>JAWRA 60 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION</p><p>KUO, LIU, AND CHEN</p></li><li><p>models for economic optimization and determiningthe optimal patterns and irrigation water application.</p><p>Figure 1 presents the framework and logicemployed in the farm irrigation simulation and plan-ning model. As presented, the model requires sixbasic types of data: (1) project site and operation data,(2) command area data, (3) seasonal water supplydata, (4) monthly weather data, (5) soil propertiesdata, and (6) crop phenology and economic data. Theweather generation module from CADSM (Prajam-wong, 1994) is adopted herein to generate daily refer-ence crop evapotranspiration and rainfall data byusing the monthly mean and standard deviationsdata.</p><p>On-Farm Irrigation Scheduling</p><p>The daily on-farm irrigation scheduling modulesimulates the on-farm water balance and estimatesthe relative crop yield and irrigation water require-ments using the...</p></li></ul>