Técnicas de engenharia de software aplicadas à modelagem e simulação de doenças de plantas

Abstract

Faced with increased demands for food by a growing population, threats to the environment due to increasing pressures on land and water resources, and globalization of markets and trade, many agricultural production systems around the world are rapidly changing. Methods are needed to evaluate options for development, impacts of changes in production systems, and stability and sustainability of proposed changes in practices so that decisions can be made to achieve societal goals A systems approach, integrating knowledge from different disciplines, is necessary. In this direction, simulation models are presented as a research tool for addressing problems involving the agricultural sector. The development of these models requires a deep analysis of the system in study, being crucial the choice of methods and tools of software engineering. In the last few decades, the mathematical modeling of plant diseases, for example, has expanded very quickly, contributing for disease integrated management. The recent advances in the computer science have assisted in this development, bringing operational advantages and usefulness for the construction of models representing complex processes. The modularity and the generic form are terms that describe the new and widely accepted methodology to surpass the complexities in the development and re-use of models. This work had as objective to apply project patterns towards the development of crop simulation models, using MVC (Model-View-Controller), to develop a generic plant disease model, using UML and object oriented programming languages, and to apply the proposed architecture for the development of a novel approach for delivering decision supporting to plant disease management (SISALERT). The application of software engineering techniques revealed to be appropriate and robust to guide in the development of plant disease simulation models. Combining technologies proved to be possible to use existing knowledge legacy. The programming language R, allowed to carry through the communication between Fortran and Java. The simulation model of the pathosystem Puccinia-Wheat reasonable well, representing leaf rust progress accordingly to the wheat leaf area growth. It can be used as a tool for studying pathogen/host relationship and for damage quantification. The platform SISALERT, showed to be feasible the application of decision support systems on a large scale. Moreover, it was demonstrated that it can be expanded for other uses/crops