Nutritional status assessment of olive crops by means of the analysis and modelling of multispectral images taken with UAVs

Abstract

This research was aimed at developing an efficient method for Nitrogen, Phosphorus, and Potassium (NPK) foliar content retrieval in olive trees by means of the analysis and modelling multispectral images taken by an unmanned aerial vehicle (UAV) under field conditions. To this end, an experiment was carried out in a super hight density olive orchard. The fertirrigation system of the experimental area was sectorized to obtain plots with different status of NPK. The orchard was overflown with a UAV equipped with a multispectral camera that photographed the entire experimental surface. A new image analysis approach was developed for integrating all the spectral images gathered during the flight in orthomosaics from which to automatically extract information from discrete points. Finally, several retrieval techniques (partial least squares regression, artificial neural network (ANN), support vector regression and Gaussian process regression) were evaluated for NPK leaf content retrieval by using the spectral data as input variables, and the results of chemical analyses as reference. Among all, the best results were obtained by ANN approach (N (R 2 = 0.63), P (R 2 = 0.89), K (R 2 = 0.93)). These results showed the suitability of the proposed image processing approach and indicate ANN as the best recovery technique for the experimental conditions evaluated. However, the approach must be validated under other environmental conditions, olive varieties and plant vegetative stages before making fertilization recommendations. • Method for NPK leaf content retrieval in olive trees by means of multispectral data. • Multispectral images taken by a UAV under field conditions. • Image analysis algorithm to automatically extract information from discrete points. • ANN showed better results than PLSR, SVR and GPR.