Formulario de búsqueda

Characterization of variables for modeling agroclimatic and phytosanitary events in agricultural crops using deep learning models uri icon

Abstracto

  • Abstract Digital transformation is increasingly present in the agricultural sector thanks to the development and availability of information technologies. However, the agricultural sector is one of the least digitized worldwide, including in countries with highly competitive and technologically advanced agricultural chains and processes. To optimize production using technological tools, it is necessary to combine variables associated with soil, climate, and plant type to estimate crop yields. Therefore, it is important to identify the physical variables that are of interest for the modeling of agroclimatic and phytosanitary events in agricultural crops. This study presents an exploratory type of documentary research to determine what variables are of interest for the modeling of agroclimatic and phytosanitary events in agricultural crops, using a critical analysis based on the results of related studies. Results show the characterization of (i) agroclimatic and biotic variables required to model the mentioned events, (ii) detection of physical evidence associated to biotic factors by means of the analysis of anomalies in the wavelengths of the spectral reflectance of the productive units, (iii) the computational model based on deep learning for the processing of these variables. Furthermore, a data pipeline is proposed that indicates the flow that the characterized variables must go through the analytical tasks.