Assessment of Soil and Vegetation Index Variability in a Traditional Olive Grove: A Case Study

Precision Agriculture (PA) is nowadays one of the most feasible solutions to reduce input and pollution in agriculture [1]. One of PA's most important techniques is identifying in-field spatial variability. Multiple techniques can be used to identify the variability: soil analysis and vegetation index identification are two of the most applied. In this study, the variability of an olive grove was assessed to evaluate the best option for creating variability maps and assessing the relationship between soil and vegetative variability in the case study. The study was carried out in two olive orchards, with a 6 × 4 m planting layout; the mean plants’ height was 3 m. To assess soil variability an EMI (Electromagnetic Induction) analysis was performed. The field was completely mapped at 0–50, 100 cm deep. After the evaluation of the electric resistivity, maps were created. The proximal OptRx Crop Sensor (Ag Leader, Iowa, USA) was used to assess the plant vegetation index. This sensor was mounted on a tractor and positioned at a height of 2 m from the ground to assure the acquisition of the vegetation index for all the assessed plants. NDVI and NDRE indexes were measured. To georeferenced all the acquisitions, a GNSS system was installed on the tractor (Ag Leader GPS6500 GNSS receiver, Ag Leader, Iowa, USA). Through this method, site-specific olive canopy NDVI NDRE data gathering was performed. Soil characterization maps revealed significant in-field differences in electric resistivity for all the evaluated deeps, through this analysis a homogeneous resistivity-value map was created. The data points of every sampling were interpolated within the whole plots by ordinary kriging through the GIS software QGIS (GNU General Public License). NDVI and NDRE predictive maps were developed using ordinary kriging fitting the best variogram. An exploratory correlation analysis was performed between NDVI, soil proximal sensing (EMI 0.5, EMI 1, RP), and soil strength, to highlight the statistical relationships between the main parameters used for this study. Collected data were analysed and interpolated by k-means clustering to make thematic maps.