Agrometeorological-spectral models based on artificial intelligence for coffee cercosporiosis estimation

• Grant number: 21/03746-0
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: July 01, 2021
• End date: June 30, 2022
• Field of knowledge: Physical Sciences and Mathematics - Probability and Statistics - Applied Probability and Statistics
• Principal Investigator: Glauco de Souza Rolim
• Grantee: Gislaine Ferreira Barbosa
• Host Institution: Faculdade de Ciências Agrárias e Veterinárias (FCAV). Universidade Estadual Paulista (UNESP). Campus de Jaboticabal. Jaboticabal , SP, Brazil

Abstract

Coffee is one of the most consumed and important drinks in the world. Climate variables influence their quality, production, and incidence of diseases, such as cercosporiosis. The objective is to estimate the incidence of coffee cercosporiosis for the main producing regions of the country, based on agrometeorological-spectral models that are tools that help in monitoring and in more sustainable management strategies. Weather data from Franca, Araxá, Patrocínio, and Araguari from 2015 to 2020, such as air temperature, precipitation, relative humidity, wind speed, and radiation balance will be obtained from the NASA-POWER platform. Based on these data, Haythornthwaite-mather's sequential water balance will be estimated. Ndvi and EVI data will be extracted from the SATveg platform (Embrapa). On the other, the data of cercosporiosis will be obtained from the monthly monitoring of the Procafé Foundation system. The disease estimation will be performed using the Multiple Linear Regression (RLM) models and the Random Forest (RF) learning machine. The actual incidence of the disease, agrometeorological data, and NDVI will be the independent variables and the estimated incidence will be the dependent variable. For the performance of the models, the coefficients of adjusted determination (adjusted R2), systematic error (SE), mean square error (RMSE), and mean percentage absolute error (MAPE) will be analyzed. It is expected to obtain a sufficiently accurate agrometeorological-spectral model that allows the producer to make the most assertive decision in relation to the agricultural system. (AU)