Diagnosing sugarcane crop problems through satellite images

Abstract

The development of new products and techniques that reduce production costs is a permanent concern of sugarcane producers. Production losses related to delayed control of diseases or pests can cause decreases in the order of 85% of the productive potential of the plots. Therefore, the early diagnosis of the problem is fundamental to improve pest control procedures in the field, especially when the areas to be monitored are extensive. Thus, this project proposes the monitoring and diagnosing in areas of low agricultural productivity caused by pest infestations and phytopathogenic diseases, using free orbital images and spatial dynamic modeling. The proposed method uses weights of evidence to parameterize the model and verify the influence of each of the selected predictive variables and cellular automata to study the spatial and geometric distribution of new areas with low production. To verify the spatial independence between the input maps, the Joint Information Uncertainty statistic will be used. The validation of the model will occur through the exponential decay and constant decay methods, which use the concept of fuzziness of location, more efficient than the pixel to pixel comparison method. (AU)