Framework para simulação do crescimento da cultura da soja utilizando o modelo CROPGRO-Soybean

Abstract

The agricultural sector has sought to employ new technologies in its daily life with the aim to add practicality and ease in carrying out tasks, aiming to increase production, control diseases and pests, as well as improve the final quality of the product, saving time and resources. Agricultural simulation models, such as DSSAT-CROPGRO, are important tools for predicting crop yields and evaluating the impacts of changes in agricultural practices and environmental conditions. Mobile applications in the middle agriculture also deserve to be highlighted, as they are technologies that have been growing due to adoption of new computational techniques to process data, making it easier to carry out of agronomic analyses, which can reach the producer in real time. However, for use these models, the demand for collecting and organizing this data for input model and presenting the output of the results are complex, which can become a problem for those looking to use this technology. In this way, this work aimed to present "Soybean Alert", a framework for simulating crop growth soy using the CROPGRO-Soybean model, composed of two modules interconnected. The first module is an API, responsible for collecting, organizing and process data used to simulate soybean cultivation, bringing data about the entire period of culture development. The second module is an application, responsible by requesting new simulations and presenting the results directly, abstracting all the complexity for the use of these simulation models. The framework developed predicts the future behavior of the crop for the user, simulating the entire cycle, using weather forecast data, emergency dates, start of reproductive stage, physiological maturation and harvest forecast. A comparative study pilot, with field data from the Brasmax Ativa RR and NA5909RG cultivars, showed that the proposed solution estimated values ​​very close to reality for the final productivity (Kg/ha), with a variation of only 4.69% and 2.96% for each cultivar, respectively. This reinforces the potential of the "Soybean Alert" framework as predictive reference and planting scenario simulation tool for farmers.