<![CDATA[Fertilizer recommendation is yearly activities for oil palm nutrient requirement correction by fertilization activities. Oil palm leaf nutrients content in fertilization recommendation activities is a small part of several parameters used. In a row of the remote sensing gains, the researchers have been studying the oil palm leaf nutrients content prediction from multispectral satellite data and field spectral from spectroradiometer and they have a variance of results. According to the existing research of the oil palm leaf nutrients content prediction, the unmanned aerial vehicle (UAV) assembled by the multispectral camera does not yet use for oil palm leaf nutrients. Therefore, this research is necessary to do. The objectives of this research were to compare the model predictor of oil palm leaf nutrient contents with several variations of variables from three bands of the multispectral camera and several vegetation indices and also to determine the best model predictors from them. The image from the multispectral camera of Mapir Survey 3 that consists of green, red, and near-infrared, and also simple ratio, normalized difference vegetation index, as well as green NDVI were used as the independent variable of regression analysis included simple regression, polynomial regression, multivariant regression with selected variables from linear regression and random forest methods of recursive feature elimination technic, and polynomial multivariant regression. The responses were the leaf nutrient analysis of N, P, K, Ca, Mg, and B from twenty samples. The results showed the best model predictor was the regression model from polynomial multivariant regression with variables from RF method of RFE technique. The model just predicted the N, P, K, and Mg oil palm leaf nutrients with 0.9415 to 0.9991 R2 value, 0.7223 to 0.9837 adjusted R2 value, and 0.0045 to 0.0340 residual standard error value (RSE).]]>
Copyrights © 2021
