Page 3 of 9
CropOptimiser has been developed over a period of ten years from its humble beginnings as an Excel spreadsheet through to a stand-alone software application with inbuilt LP solver and advanced textural, charting, and spatial outputs. However, the key feature of the software throughout its development cycle is the central component of the LP model.
Many studies in agricultural management have used LP and non-LP models for problem solving, with a typical objective being to maximize seasonal benefits of cropping in an irrigation command area (Barbel and Limon, 1999). For example, Kodal (1996), Mainuddin et al. (1997) and Raju and Kumar (1999), Benli et al. (2001), Singh et al. (2001) and Reca et al. (2001) all developed linear models to optimise cropping patterns to maximize profit. Some have developed nonlinear models to optimise cropping patterns under deficit irrigation (Carvallo et al. 1998; Benly and Kodal, 2003; Kumar et al. 1998) while others have optimized over consecutive seasons to maximize net annual return (Sethi et al. 2006). Others such as Dutta and Carter (1998) and Karya (1995) have focused on optimising water use and allocation for fixed cropping options. However, in Lombok, the problem is associated with the irrigation of multiple crops in multiple cropping sequences in which water supply can be predicted using seasonal climate forecasting. None of these existing models were structured for this purpose.
Therefore a LP model was prototyped/developed in Microsoft Excel in 1999 using the inbuilt Solver engine for determining the optimum solution (Yasin et al. 200X). Because of the difficulties in visualizing and presenting the outputs in Excel, work began simultaneously on the first version of CropOptimiser to import and automatically display the Excel outputs geographically. This version of CropOptimiser was unable to manipulate the LP model directly, but was designed to coincide with it and to generate and display polygons of different outputs overlaid on to the map of Lombok.
The LP model/CropOptimiser package was never developed to exist in isolation, rather it forms part of a decision support package, or systems framework, whereby outputs from other tools are imported into the LP model for processing (Figure 1). The process requires that the hydrology of the irrigation region be modeled using the IHACRES software (GET REFERENCE) after significant patching and synthesis of meteorological and hydrological data (using tools such as Weatherman – GET REFERENCE). These outputs are then input into the water allocation model (IQQM – GET REFERENCE) to generate time-series of diverted irrigation water. Zhang et al. (200X) describes this process in detail. The seasonal climate forecasting software FlowCast (McClymont et al. 200X), which is also being developed as part of this project, is then used to estimate available irrigation volumes and effective rainfall amounts for different seasons and climate conditions (such as El Niño, La Niña and neutral conditions). Cropping models have also been used to generate crop response to water under different soil conditions. The water and cropping parameters are then input into the LP model for solution.
In 2004, CropOptimiser was rewritten to incorporate the LP model directly, including its own solver engine, and graphical user inputs to load, store and edit LP variables and constraints. Over the next four years it was enhanced and modified to simplify its coexistence with its companion decision support tools. For example, the 'stratification engine' in FlowCast was embedded into CropOptimiser to remove the need to run FlowCast separately, and to directly link with modeled water allocation and rainfall time-series data from the hydrologic models (see dotted region in Figure 1).
During this development, key design criteria have been considered and addressed, in order to arrive at the current version. This includes:
• Encapsulate the LP model for optimizing cropping choice and pattern into a standalone software product;
• Encapsulate a stratification algorithm to generate seasonal prediction information;
• Input ENSO time-series data for input into the stratification algorithm;
• Input (and link) water diversion and rainfall time-series directly from the hydrologic models to calculate available water;
• Provide a mechanism to simplify the inputting of both physical and social constraints;
• Develop a simple graphical user interface using state of the art software engineering practices; and
• Provide detailed reporting of results including textural, chart, and GIS based outputs.