Evaluation of agronomic variables and Sentinel-2 satellite images for sugarcane yield estimation using the Random Forest algorithm

Main Article Content

Rafaella Pironato Amaro
Ana Cláudia dos Santos Luciano

Abstract

The objective of this project was to evaluate the importance of agronomic variables and Sentinel-2 satellite images to estimate sugarcane yield using the Random Forest algorithm. We used agronomic data referring to the variety, cutting stage, soil type and relief, in addition to data from satellite images referring to the average, maximum NDVI and the standard deviation of the NDVI of each field. Two empirical models were created considering: i) Agronomic variables, ii) Agronomic variables and Sentinel-2 images. The model to estimate sugarcane yield showed R² equal to 0.64 and 0.83, RMSE of 10.17 and 7.0 ton/ha for models i and ii, respectively. The evaluation of the importance of the variables indicated that the variable cutting stage was the most important, followed by the variables variety and average NDVI of the field. The combination of agronomic variables and satellite images brought improvements to estimate sugarcane productivity.

Downloads

Download data is not yet available.

Article Details

How to Cite
Pironato Amaro, R., & dos Santos Luciano, A. C. (2023). Evaluation of agronomic variables and Sentinel-2 satellite images for sugarcane yield estimation using the Random Forest algorithm. Revista Militar De Ciência E Tecnologia, 39(4), 65-78. Retrieved from http://ebrevistas.eb.mil.br/CT/article/view/9497
Section
Ciência dos Materiais