Overview |
Crop Production |
Crop Pattern |
Crop Yield |
Irrigation |
Soil Management | Precision Farming |
Relevant Products |
Relevant Links
Web-Based Expert System for supporting Precision Agriculture Thirumurugan Ponnusamy Department of Geographical and Environmental Studies Faculty of Humanities and Social Sciences, University of Adelaide, South Australia, 5005 Australia Email: thirumurugan.ponnusamy@adelaide.edu.au Abstract Precision agriculture demands wide array of knowledge from diverse disciplines such as soil, landuse, groundwater etc. In these disciplines, the available expert knowledge is scarce and is distributed amongst the scientists, researchers and progressive farmers. There exists a knowledge gap between scientific and farming communities. The knowledge bases could be connected to create a centralised virtual knowledge-based expert system, which will answer queries from users (farmers). Expert systems are machine-equivalent of human experts and rely on borrowed human intelligence to give recommendations for the management of resources. In addition to the knowledge derived from geo-spatial technologies (remote sensing, Global Positioning System (GPS), Geographic Information System (GIS)), expert systems also have an advantage of explicitly representing the cognitive knowledge (rules of thumb, knowledge from experience, empirical findings) used by human experts for solving problems. This proposed web-based expert system will act like a multi-domain agriculture expert, accessing various knowledge bases to support precision agriculture. Respective domain experts (Knowledge administrators) could maintain these knowledge bases, so that they will remain accurate and up-to-date. An inference engine, which accepts inputs from users, requests the corresponding knowledge bases (could be one or more depending on the problem) on an on-demand basis to provide solutions (farm recommendations). The paper also elucidates the possible roadblocks in implementing such an expert system in the developing countries. 1. Introduction Historically, if there is one profession that had symbolized the continent of Asia, it is agriculture. Almost 42 percent of world's cropped area (including fodder crops) falls in this continent (FAO 1995) and it houses 60 percent of world's population (United Nations 2005). The growing population pose a great threat to food security in the near future. The agricultural production is decreasing due to land degradation and improper farming practices. The profession of agriculture has become a gamble with the nature. With irregular monsoon rains, frequent drought conditions and declining soil productivity, farmers are clueless about increasing agricultural productivity. Increased use of chemical fertilizers and pesticides have also caused tremendous damage to the environment. Conventional farming methods are no longer sustainable and will not satisfy the ever-growing need for food. The only available solution is to devise a sustainable agricultural strategy by which both the economic and environmental aspirations can be met. Precision agriculture, "which is based on the philosophy that production inputs have to be applied only as needed and where needed for the most economic production" as observed by Searcy (1997), seems to be a perfect solution. 2. Precision Agriculture Precision Agriculture (PA) refers to site-specific farming practice in which land units are managed optimally considering precisely their inherent capabilities and constraints, crop information, advanced technologies to maximize agricultural production. It aims at increased efficiency, improved quality of agricultural products, low pollution and conservation of resources without sacrificing economic benefits. Precision agriculture calls for timely information and knowledge from diverse disciplines at various stages and utilizes various technical tools. Precision agriculture uses geospatial technologies, like Remote Sensing (RS), Global Positioning Systems (GPS), Geographic Information Systems (GIS) at both strategic and field-level planning to arrive at an action plan for crop production. These technologies take into account of the unique natural resource characteristics which could be mapped spatially. With the help of advanced machineries, the application of fertilizers and pesticides can be automatically varied spatially across a land unit. However, precision agriculture does not necessarily makes use of advanced tools and technologies. This is particularly valid in most of the developing countries where poor farmers can not afford advanced equipments such as variable-rate controllers and auto steering systems. The motivation of these farmers lies in increasing the crop yield and maximizing the economic returns with minimum production inputs such as fertilizer and pesticides. Without fulfilling this motivation, it is difficult to convince the farmers to undertake sustainable management practices in their land. The practice of precision agriculture is very minimal in the third-world countries. Even though, the weighed benefits are comparatively higher, farmers are reluctant because of the following factors. First, they cannot afford to purchase technical equipments as their counterparts in the developed countries. Secondly, the lack of expert support (guidance) for providing timely and relevant information (farm advice) throughout the crop growing period. The expert knowledge which is accessible to commercial farmers do not reach poor, individual farmers who also share similar social and economic aspirations. Traditional information dissemination systems such as Newspapers, Radio or TV broadcast have their inherent shortfalls as pointed out by Reddy & Ankaiah (2005). The recommendations are too generic and does not cater to the needs and aspirations of individual farmers. Finding quality agricultural experts is often a daunting task and they are not readily available for consultations with the farmers. Bringing experts to a commonplace for interacting with the farmers involve considerable time and cost. In countries like India and China which has a high population of small and marginal farmers, it would not be feasible for an expert to have one-to-one interaction with farmers. 3. Need for Information Support Information is a primary requirement and considered as the heart of precision agriculture. In this information age, the available information from various sources is growing at a phenomenal rate. Numerous researches are being carried out, new results are discovered at the research institutions and they continue to accumulate in the form of reports and dissertations. Most of these findings, recommendations do not reach end users (farmers) at the implementation level. There exists a gap between researchers and farmers, particularly in the developing countries. The lack of 'proper decision-support systems' (DSS) to disseminate timely, relevant farming advice has been observed as a major roadblock for adopting precision agriculture (McBratney et al. 2005). However, there are some good evolving frameworks, such as the one proposed by Reddy & Ankaiah (2005) for disseminating agricultural information using advances in Information and Communication Technology (ICT), are being tested and developed in many South Asian countries. The need of the hour is an virtual expert who can give personalised expert advice to a large community of farmers specific to their needs and aspirations considering various knowledge bases. The information and knowledge available for decision making are enormous and they are growing continuously at a staggering pace. It is almost impossible for any human expert to consider every piece of available information before arriving at an optimal decision. To ease this information overload and to help in the process of decision-making, there exists a few decision support systems (DSS) which deliver an action plan in the form of maps and reports. However, these decision support systems have some inherent drawbacks and they do not provide recommendations like a human expert. 4. Expert Systems for Decision-Making Conventional decision support systems depend on algorithms for solving a problem. and require a pre-defined set of input data to begin the analysis. With a step-by-step approach, as dictated by the algorithm they proceed to analyse the input data to reach conclusions. The knowledge about solving the problem is represented in the DSS's algorithm. When the knowledge of problem gets changed, the algorithm needs to be improved or rebuilt, which means developing a better decision-support system. Solving a new problem in the same discipline or domain may need a new system to be developed. However, human experts tend to follow a cognitive approach rather than an algorithmic approach. They rely on an extensive knowledge base (in their mind) which may contain facts, assertions, past mistakes, trial-by-error methods and conclusions to solve new problems. The machine equivalent of human experts are expert systems. Expert systems forms a branch of artificial intelligence (AI) science which solves problems by cognitive approach. Expert systems derive the power from the extensive knowledge bases. A typical knowledge base stores information and knowledge in the form of rules. These rules may contain the cognitive knowledge derived from human experience or scientific knowledge derived from research or both. The knowledge bases are kept separately from the program, which allows easy modification of knowledge base. The knowledge base can be updated without disturbing the whole structure of the expert system. An expert system could solve or provide recommendations to more than a single problem related to the knowledge domain (only limited by the knowledge base). Another advantage of expert systems lies in their reasoning capability. They can provide reasons for arriving at a particular decision. Expert systems can work with partial information and can handle incomplete inputs (unlike the decision support systems). It is proposed to develop an expert system which will provide personalised farming advice specific to a farmer's need at any time and place. The knowledge bases from various sources can be integrated to create a web-based online expert system, which would answer agricultural queries from farmers. This system can accommodate empirical (cognitive) knowledge from human experts apart from other scientific information bases. A web portal will be developed to deliver customised farm recommendations which is powered by the expert system in the background. 5. Web-based Expert System for Precision Agriculture (WESPA) Conventionally, the user (owner of a land unit) approaches a group of experts to get farm advice or recommendation. The experts analyse resource potentials and constraints of the land unit to arrive at a feasible plan of action. This model will be emulated by having a real-time Web-based Expert System for Precision Agriculture (WESPA). The proposed expert system (Fig.1) would consist of an inference engine which accepts queries from farmers. Farmers can either type in a specific agriculture query or ask for a general recommendation by providing the survey number of their land unit. The inference engine will analyse the query and send information requests to the central knowledge base and match them with knowledge rules stored in the knowledge base.  Fig. 1. Web-based Expert System for Precision Agriculture (WESPA) The central knowledge base will contain a compiled link to the networked servers from various disciplines such as soil science, geology, hydrology which will cater to the needs of the expert system on an on-demand basis. Depending on the problem, the corresponding servers (knowledge bases) will be accessed and possible solutions will be drawn. These servers (knowledge bases) are maintained at the respective organisations and departments by the discipline experts (knowledge administrators). These knowledge bases contain a collection of 'If-Then' rules specific to that particular domain or discipline. For example, if a specific set of symptoms are observed in a crop, then deliver remedial measures. A knowledge administrator who is also a domain expert can update or modify the knowledge base of a particular domain. The knowledge administrator keeps an eye on the various information sources such as world wide web, information bulletins, research reports, case study outcomes related to a particular domain in order to update the knowledge base. Since, the reliability of information from some information sources may be questionable or may be applicable to particular geographic regions, a knowledge administrator's intervention is required to update only the relevant information to the knowledge base. In this way, it is ensured that the knowledge bases remain up-to-date, accurate and need not be duplicated at multiple locations. Some of these updates can be automated for example, the web server of a meteorology department may automatically update the local weather status periodically. The expert system is served by a geo-spatial server that has detailed spatial information at cadastral level such as landuse, soil, water, geology etc. When the recommendations are made, the expert system will check the capabilities and constraints of a land unit before delivering the farm advices and action plans. Location-specific farming advice can be delivered based on the geographic location of the land unit (which could be identified by the survey number) with the help of a large scale digital cadastral database. Such recommendations considering weather pattern, rainfall, market information and socio-economic status of land owner would be more relevant than the conventional, generalized farm recommendations. The expert system will also give reasons and explanations for arriving at a particular recommendation. 6. Development Phases The proposed project consists of four phases, which are listed below: 1. Developing a Model for supporting Precision Agriculture. This phase includes determining information required for supporting precision agriculture and listing the information sources with their reliability (accuracy and relevancy). A comprehensive model will be developed considering the information required by farmers and consulting some of the agricultural experts. This phase forms the foundation on which the expert system will be developed. 2. Designing of Expert System accommodating the Model. The model has to be transformed into a schematic diagram which includes various parameters required for supporting precision agriculture. This phase explicitly mentions the schema of the knowledge base, methods of knowledge acquisition, roles of an expert and the knowledge administrator and methods for analysis, reasoning and explanation. 3. Development of the Expert System The third phase involves collection of knowledge rules from various discipline experts and coding them into a knowledge base. A prototype will be developed to test the model with experimental data. A series of iterative testing, feedback and improvement will build a satisfactory expert system. When the expert system is ready, real world information can be fed into it and tested for validity. The expert system will improve its performance by continuos feedback from users. 4. Development of a Web Portal This is the final and the crucial phase as it involves the delivering the personalized expert advice to farmers. A web portal will be developed with a login screen which will ask the user some form of identification. For example, this may be an unique survey number of the land unit. Depending on the location of the land unit (which can be read from a geo-spatial server), farmers can view farm recommendations specific to their land unit. The user profile can be built by the user by filling an online form with crop details, soil test results etc. By this way, recommendations are tailor-made by the expert system to deliver only relevant expert knowledge as and when required by the farmer throughout the crop growing period. This web portal has to be delivered in local language to enable user-friendliness. Every page will have a feedback form which can be used to send feedbacks and suggestions for improving the expert system. 7. Types of Recommendations (Farm Advices) a. General Recommendations: Some recommendations are general in nature, suggesting a comprehensive land use plan for a land unit considering its inherent constraints and capabilities. For example, what crop a farmer can grow considering various resources and constraints, water availability, management plans and economic status of farmer. b. Specific to a region or a crop: Expert systems can answer queries specific to a geographical region (advice, warnings, precautions and notices for farmers in a particular geographic region) or to a particular crop (possible pest attacks, application of fertilisers and pesticides, efficient methods of irrigation, cultivation throughout the life cycle). c. Multi-media content: The use of multi-media content like colour images, videos, animations showing symptoms of crop diseases, application of pesticides, effective methods of farming practices will improve the impact of advice and recommendations to the farmers. d. Special notices: Periodical notices, government price policies, market demand forecasts, availability of high-yielding seeds, timely pest warnings and remedies can be posted periodically to help the farmers. 8. Possible Roadblocks a. Access to Technology The access to computers and the internet is still a distant dream for poor and marginal farmers in the developing nations. However, state governments are already setting up village information kiosks for delivering citizen-centric services across rural areas (John 2003). This agriculture expert system can be part of the village information kiosk. When the recommendations delivered by the expert system improves the yield, farmers will pay a small amount of money for consultation. Alternatively, state local governments can set up try-for-free systems for initial stages to get the acceptability and feedback. b. Revenue Model The experts, who develop, maintain and update the knowledge bases, have to suitably rewarded for their information support. A revenue model has to be worked out to compensate the time and efforts spent by the domain experts (knowledge administrators) who agree to participate by sharing their knowledge. c. User Acceptance A user requirement survey has to be carried from the farmers to know exactly what kind of information support is needed. Once developed, the system needs to be user-friendly and tailor-made to the user's requirements. The system should also consider traditional practices (that are sustainable) of farming community (use of green manure or any other specific farming practices). d. Feedback and Support The system relies on incremental development and hence feedback from the users plays a crucial role. A mechanism needs to be worked out to gather the feedback from farmers who can not use computer to send their feedback. For example, a village administrative officer can take this job of disseminating such vital information and getting feedback from farmers (sometimes in the form of photographs, if needed). 9. Conclusion The advent of various information and communication technologies have shrunk the world into a small market place. The market for a agricultural product is no longer restricted by geographical boundaries. Now, the farmers of Asian countries have to compete with the likes of Australian, European countries. They should concentrate on improving the quality of agricultural products as well as maximize quantity of production. Increasing consumer awareness about the environmental balance also thrust the farmers towards sustainable farming. The proposed expert system will make use of knowledge from various sources and provide advice suited to a geographic location taking the local weather, rainfall pattern, market information and traditional knowledge into consideration. The continuous support for practicing precision agriculture will be delivered in the form of recommendations for a particular crop throughout its life cycle. The best practices from the past successful case studies, traditional knowledge from farming community or from other countries will be part of the proposed system. The technical advancements such as high-resolution imaging (remote sensing), large scale cadastral databases, expert systems, networking, interoperability and hand-held computer devices will help the farming communities to harness the power of Information and Communication Technology (ICT) and to share knowledge among themselves towards realising a sustainable future. References
Page 1 of 1
|