Optimal alarm system applied in coffee rust

Abstract

Alarm systems have very great utility in detecting and warning of catastrophes. This methodology was applied via TARSO model with Bayesian estimation, serving as a forecasting mechanism for coffee rust disease. The coffee culture is very susceptible to this disease causing several records of incidence in most cultivated crops. Researches involving this limiting factor for production are intense and frequent, indicating environmental factors as responsible for the epidemics spread, which does not occur if these factors are not favorable. The fitting type used by the a posteriori probability, allows the system to be updated each time point. The methodology was applied to the rust index series in the presence of the average temperature series. Thus, it is possible to verify the alarm resulted or in a high catastrophe detection in points at which the catastrophe has not occurred, or in the low detections if the point was already in the catastrophe state.