Abstract: Precision agriculture benefits from rapidly evolving geospatial information technologies, including global positioning system, geographic information system, yield monitors, remote sensing, and electronic sensors and controllers for variable rate application technology. The potential benefits of precision agriculture include: (1) determination of spatially referenced data for improved understanding of agricultural systems; (2) precise placement of agricultural inputs to improve net economic return, environmental quality, and global competitiveness; and (3) accurate measurement of outputs such as yield. By allowing optimal use of farm inputs on a location-specific basis, precision agriculture shows promise to be economically consistent with, and in many cases superior to, conventional farming techniques. This paper describes the components and application of a spatially explicit decision support system developed to enhance the implementation and adoption of precision agriculture management at the whole-farm level. The system integrates biophysical simulation models, multivariate statistical analysis techniques, ArcView GIS, and other decision support tools for improved synthesis, analysis, and modeling of data related to crop production and environmental quality. The decision support system enables users to examine several scenarios of site-specific resource management and to elucidate the potential benefits of precision agriculture.
Keywords: precision agriculture, modeling, decision support system, spatial statistics
Research and development in agronomy has over the past decade aimed at developing less intensive and integrated farming systems with less use of machinery and lower inputs of fertilizers and pesticides (Olesen et al., 1997). Such ideas and practices can be categorized as precision agriculture. The main objective of precision agriculture is to increase agricultural profit by increasing crop yield and minimizing costs, with as low negative environmental impact as possible. Precision agriculture technology can be used quite effectively to adjust vary agricultural inputs (e.g., lime, fertilizer, seeding, and tillage) to match agronomic requirements at different locations within the field and to account for field spatial variability.
Precision agriculture has benefited quite extensively from rapidly evolving geospatial information technologies, such as the global positing systems (GPS), geographic information systems (GIS), remote sensing (RS), and electronic sensors and intelligent controllers for variable rate application technology. Using the GPS technology, soil samples can be taken across a field and their locations recorded at desired accuracy and precision. Intelligent controllers and sensors allow the producer to vary the rate of inputs according to agronomic needs to enhance productivity. GIS allows the integration, analysis, and display of all types of precision agriculture information (e.g., data on crop rotation, tillage, nutrient and pesticide application, yield, soil type, and drainage files). What is needed in precision agriculture is a tool that integrates these component technologies with bioeconomic models for tactical, strategic, and policy-level decision making.
In this study, an integrated spatial decision support system has been developed to facilitate data management and enhance implementation of precision agriculture at the whole-farm level. The decision support system incorporates interactions between the managerial, agronomic, climatic, environmental, and landscape factors that influence crop production and integrates bioeconomic models, statistical analysis software, GIS and knowledge-based systems for improved decision-making in precision agriculture. In the sections below, the components of the decision support system are described and an example application of the system is presented.
The complexity of making routine, coherent, and cost-effective farm management decisions presents a formidable challenge to the adoption and implementation of precision agriculture. In crop production, for example, these decisions must be technically defensible, environmentally sound, and acceptable, while production risks and enhancing net farm profits. Effective crop management decision making also requires an underlying knowledge of agroecosystem structure and functioning, the accumulation of quantitative information for modeling, and the selection of appropriate management options that meet environmental and economic constraints and the accurate interpretation of the model results for effective decision making and risk reduction.
Over the past several decades, computer-based information systems have been utilized to alleviate bottlenecks associated with crop production decision-making. Interactive computer-based decision support systems that help decision makers utilize data and models to resolve unstructured problems have been developed. In resource management, these systems have evolved to encompass interactive and integrated multi-component systems that include various combinations of biophysical and economic simulation modeling, optimization and statistical techniques, heuristics and artificial intelligence techniques, geographic information systems and associated spatial and non-spatial databases, and graphical user interface components. The seamless integration of these components within a decision support system enhances the solution of semi-structured and unstructured problems by bringing together human judgement and computerized information, supports a variety of decision-making processes and styles, and improves the effectiveness of decision making. The scope of decision-making in natural resources can range from decisions related to crop production to those involving ecosystem structure, function and sustainability.
Figure 1 shows the overall structure of the integrated decision support system for precision agriculture or IDSSPA. The system is developed within an integrated systems research on precision agriculture and incorporates the various interactions and interrelationships among the various parts of crop production agriculture. It enhances the manipulation, analysis, modeling, and display of the large volumes of information collected under site-specific crop production and can be used as a decision support system both to optimize agricultural inputs (fertilizer, pesticides, irrigation, etc.) and to evaluate the environmental impacts of production practices.
As with many spatial decision support systems, IDSSPA is composed of the following modules: (a) spatial and non-spatial data management module, (b) biophysical and economic modeling module, (c) a decision-aid and knowledge-based module, and (d) an intuitive user-interface module. The spatial and non-spatial data management module handles an extensive set of data relevant to the successful implementation of precision agriculture. It is designed to manipulate multiple level (farm, field and regional) data and can import and export data from and to standard precision agriculture and GIS software such as SSToolbox, AgLink for Windows, and ArcView GIS. The spatial and non-spatial data management module also stores, manipulates, and manages weather data, field-farm data collected from GPS and terrestrial sensors as well as remotely sensed data. A typical weather data comprises of historical data describing the local climate, meteorological data describing past climatic conditions, and weather forecasts generated by a climate generator such as CLIGEN. The farm data describes the general crop production system of the farm and contains farm-specific prices for relevant input factors, field data ranging from data on terrain, soils, and management to relevant data on pesticide and nutrient application rates, weed/pest scouting reports and soil nutrient analysis.
The model-based module of IDSSPA consists of biophysical, economic, statistical, or other quantitative models that provide the system’s analytical and pragmatic modeling capabilities. These analytical and pragmatic models are fully coupled with other modules or subsystems of IDSSPA. The biophysical models incorporated into IDSSPA include: Groundwater Loading Effects of Agricultural Management Systems or GLEAMS (Leonard et al., 1987), a field-scale model for evaluating the effects of crop production and management practices on runoff, erosion, and chemical transport in agricultural soils; Root Zone Water Quality Model or RZWQM (USDA-ARS, 1992), one-dimensional process-based biophysical model that simulates major physical, chemical, and biological processes in an agricultural crop production system. Simulation of plant growth processes and crop yields is provided by these models and by CROPGRO (Boote et al., 1996), a group of dynamic, physiologically based crop simulation model that can be used to optimize planting density, maturity type, fertilizer input and irrigation as well as to determine which genetic traits would maximize yields and economize returns across latitudes, soils and climate zones (Kiniry et al., 1997; Boote et al., 1996). The economic model consists of an input-output model that estimates changes in management decisions and profits for a farm under site-specific management as opposed to a similar farm under conventional management. The model identifies the critical amount of variability that justifies a prescribed level of investment in precision agriculture technology and practices. Other analytical and optimization models incorporated into the model-based module are designed to enhance either the analysis or the summarization of data. For instance, the module incorporates prognostic models for creating farm input recommendations (e.g., lime and nutrient prescriptions) and S-PLUS statistical software that enhances interpolation and exploratory analysis of data. Data analysis capabilities of the IDSSPA ranges from simple univariate and bivariate analysis to more complex multivariate analysis of different data series.
The decision-aid and knowledge-based module of IDSSPA provides intelligence and support to augment the user’s knowledge. This component is intended to meet the strong needs for a tool that supports the producer in critical decisions that have to be made before and during implementation of precision agriculture. Occupying the primary core of this module are heuristics and production rules provided by an expert system and data mining tools that enhance the interpretation of data and modeling results in various phases of ecological, economic, and agronomic decisions. The decision-aid and knowledge-based component also provides expert advice on routine, site-specific resource management strategies, such as fertilizer or herbicide application, and in the selection of risk-efficient crop production decision alternatives.
Communication between the user and the interrelated modules described above is facilitated by the user interface, which not only emphasizes case-of-use and accessibility, but also addresses desirable factors in human-computer interactions. Constituted on the basis of standard graphical user interface models, the IDSSPA user interface module comprises of objects (pull-down menus and buttons) and dialogue boxes designed using object-oriented programming languages and Avenue scripts. It is uniquely designed to accommodate various information representations and the action languages that enable the user to manage the data inputs and outputs in the form of dialogues or processes. Figure 2 is the main panels and pull-down menus of the IDSSPA, and Figure 3 -- 5 show the interfaces to configure running parameters for the RZWQM and to perform spatial data analysis.
Components of the decision support system are designed to provide a synergistic and seamless environment for evaluating sustainable production issues related to precision agriculture. As indicated previously, the database and data management module is uniquely constructed to be interoperable with standard precision farming and GIS software. Many of the data storage, manipulation, analysis, modeling, visualization, and decision-making functions required by the precision agriculture community industry are incorporated into the system. Figure 3 shows a typical result that can be derived by using the integrated decision support system. This result relates to the estimation of crop yield obtained from the biophysical model. As seen in Figure 3, the grid-based display of model results enhances integration of results with other data sets or coverages, and facilitates the incorporation of other derived GIS coverages such as remotely sensed imagery of nutrient or water deficit areas of the field.
Precision agriculture is the term used to describe crop production practices and management strategies that maximize net farm income (through enhanced yield and reduced farm inputs) and minimize environmental pollution through site-specific variable-rate management of chemicals. Many in the industry contend that for precision agriculture to meet these sustainable production goals, new tools for effective and efficient decision making are needed. This paper describes the major components of a spatial decision support system designed to enhance data management, modeling, visualization, and decision making in precision agriculture. The Integrated Decision Support System for Precision Agriculture couples ArcView GIS, biophysical and economic models, spatial statistical models, expert or knowledge-based system and user interface. It can enhance understanding of agricultural systems by determining spatially referenced data, and improves analysis of the trade-off between economic returns and environmental quality. IDSSPA can be a useful research and decision-support tool for precision agriculture and natural resource management.
Boote, K.J., J.W. Jones, and N.B. Pickering. 1996. Potential uses and limitations of crop models. Agron. J. 88:704-716.
Kiniry, J.R., J.R. Williams, R.L. Vanderlip, J.D. Atwood, D.C. Reicosky, J. Mulliken, W.J. Cox, H.J. Mascagni, J.S.E. Hollinger, and W.J. Wiebold. 1997. Evaluation of two maize models for nine U.S locations. Agron. J. 89:421-426.
Leonard, R.A., W.G. Knisel, and D.A. Still. 1987. GLEAMS: Groundwater Loading Effects of Agricultural Management Systems. Trans. ASAE 30:1403-1418.
Olesen, J.E., L. Pedersen, S. Christensen, B.J.M. Secher, and J. Petersen. 1997. An integrated decision support system for management of winter wheat. First European Conference for Information Technology in Agriculture. Copenhagen.
USDA-ARS. 1992. Root zone water quality model (RZWQM). User’s manual. Great Plains System Research Unit, Fort Collins, Co.
U.Sunday Tim:
Associate professor in Agricultural and Biosystems Engineering Department of Iowa State University and my specialty is water quality modeling and decision support systems.
Xixi Wang: Ph.D student in Agricultural and Biosystems Engineering Department of Iowa State University. I am a graduate research assistant, and mainly study GIS and its applications on precision resource management.