Advances in monitoring vineyard with multiscale and multiplatform data for precision agriculture systems

Within a vineyard, the variation in plant water status is intricately tied to the spatial variability of the soil, where the physical attributes of the soil govern the processes shaping the soil water balance. As the soil and its characteristics exhibit inhomogeneity, horizontally and vertically, the productivity and qualitative response within the vineyard become less uniform. In this context, employing proximal sensing to gauge the apparent soil Electrical Conductivity (ECa) and monitoring it throughout the growing season becomes instrumental in understanding the nature of spatial variability within the vineyard. This not only aids in viticultural microzoning, identifying Homogeneous and functional Homogeneous Zones (HZs and fHZs), but also supports field experiments. We propose a machine learning approach that works as a predictive model for soil ECa, involving spatially predicting ECa based on discrete measurements obtained from a network of Time Domain Reflectometry (TDR) probes capable of measuring ECa. This methodology enables the spatial prediction of ECa values across the surveyed area. The main purpose is to create a process that using multiscale and multiplatform measurements helps the farmer monitoring and interacting with the crop in a better way, reducing resources and improving the crop productivity. Records on soil and atmosphere systems, in-vivo plant monitoring of eco-physiological parameters in 2020 and 2021, and spatial variability of plant status monitored through UAV multispectral images were used to test this approach, on a Greco di Tufo grapevines (white) in southern Italy. The apparent EC measurements were obtained using a PROFILER EMP 400 in both dipole modes and with 3 different frequencies (5, 10 and 15 kHz), exploring different depths of the soil. The predictive model shown a good performance, with results that are in good agreement with previous knowledge of the area.