In field quantification and discrimination of different vineyard water regimes by on-the-go NIR spectroscopy

Abstract

Precise and rapid methods to assess plant water status are needed in agriculture. The goal of this work was to evaluate the capability of a new plant-based method based on proximal near-infrared (NIR) spectroscopy acquired on-the-go from a moving vehicle to quantify and discriminate different water regimes in a commercial vineyard. Proximal on-the-go NIR spectroscopy (1100–2100 nm) was acquired at solar noon on five days from veraison (onset of ripening) to harvest 2015 in a commercial Tempranillo vineyard. Spectral measurements were taken at ∼0.30 m from the canopy, on both canopy sides, from a vehicle moving at 5 km h −1 . Measurements of midday stem water potential (Ψ s ) and leaf stomatal conductance (g s ) were simultaneously acquired to be used as reference indicators of plant water status. Partial least squares (PLS) was used to build calibration, cross validation and predictive models for Ψ s and g s. The determination coefficients of prediction (R 2 p ) were above 0.86 for Ψ s and above 0.66 for g s, while the root mean square errors of prediction (RMSEP) were less than 0.18 MPa and 93.7 mmol [H 2 O] m −2 s −1 , respectively. PLS-Discriminant Analysis (PLS-DA) was applied to classify the data into three different water regimes, according to Ψ s or g s . The average correctly classified percentage was greater than 72% for Ψ s and g s . This discriminant capability, together with the large number of measurements that the on-the-go NIR spectroscopy can provide, enables the quantification and mapping of the variability of a vineyard water status and may help to define precise irrigation strategies in viticulture.