Vineyard water status assessment by non-destructive, proximal, NIR spectroscopy

Abstract

Assessing and mapping the spatial variability of vine water status within vineyards constitute a major challenge. An effective sampling system, capable of providing a large amount of accurate observations, is needed to define accurate irrigation decision-making. In this work, a new, proximal, non-destructive method based on NIR spectroscopy is presented and validated against a traditional reference indicator of the plant water status, such as the stem water potential (y_stem) over two seasons (2015 and 2016). With the aim of creating a greater water status variability, three different irrigation treatments were imposed in a four-block completely randomized experimental layout in a commercial vertical shoot positioned (VSP) Tempranillo vineyard (N-S orientation). On-the-go NIR spectroscopy (1100-2100 nm) was acquired at ~30 cm from the canopy (east side), from a moving vehicle (5 km/h) at solar noon (between 2 and 3 p.m.) on 15 days from veraison to harvest in years 2015 and 2016. Simultaneously, a total of 540 measurements of stem water potential (Ψ_stem) were taken and used as reference values. A wide range of Ψ_stem was found in both seasons, ranging from -0.55 to -2.25 MPa in 2015 and from -0.75 MPa to -1.95 MPa in 2016. Calibration, cross-validation, and predictive regression models were built using modified partial least squares (MPLS). Values of R² above 0.83, 0.75 and 0.66 were obtained for calibration, cross-validation and prediction, respectively, while the sensitivity of the predictive models in terms of standard error (SEP) oscillated between 0.139 and 0.152 MPa in the two seasons. These results are promising and evidence the potential of non-invasive, on-the-go NIR spectroscopy to become a helpful tool to appraise the vineyard water status variability and support irrigation scheduling decisions in viticulture in the short term.