Regresar

Yield pbkp_rediction for precision territorial management in maize using spectral data

Abstract:

A multinominal logistic regression-based machine learning algorithm was applied to pbkp_redict yield. Leaf area index extracted from on-field spectrometer readings and normalized difference vegetation index extracted from satellite images at two crop growth stages were used: full leaf development and beginning of tassel emergence. At crop maturity, yield information was collected from each farm. A model using polynomial regression and four explanatory variables estimated best the yield. Pbkp_redictions could serve to make recommendations to increase the yield, such as replanting where the density is low, increasing fertilization, and use of pesticides. Pbkp_redicted yield can also provide an early warning to the government for decision making on imports of maize, to avoid overlapping with the national production.