SUGARCANE PRODUCTIVITY: METHODOLOGICAL MODEL BASED ON A SERIES OF BIOPHYSICAL VEGETATION INDICES

Abstract

Sugarcane represents a significant area of commercial cultivation in Brazil, and yield prediction is fundamental for guiding decision-making and anticipating scenarios in the field of sugarcane agricultural production. Thus, the study aimed to evaluate the multiple linear regression agro-model as a tool for estimating sugarcane productivity using biophysical vegetation data (Fraction of Photosynthetically Active Radiation Absorbed - fAPAR and Leaf Area Index - LAI) as independent variables and yield (tons of stalks per hectare - TSH) as the dependent variable. The multiple linear regression model was evaluated at the Serranópolis-Goiás Energy Plant during the 2020 harvest, covering 3,011.57 hectares and represented by 125 commercial sugarcane plots under rainfed cultivation (cultivars RB867515, CTC17 and CVSP7870), with cutting stages ranging from the 1st to the 13th. The resulting agro-model from the modeling process (𝒴 = 1.137 𝒳1 + 29.806 𝒳2 – 113.098) integrated the variables LAI, fAPAR, and TSH, achieving an adjusted coefficient of determination (R² = 0.912) and root mean square error (RMSE = 5.649), ensuring the effectiveness of the independent variables as modeling parameters for predicting stalk yield.