Aim : The objective of this work was to test an approach to extrapolate the vine water status (Ψ) over a whole denomination. The method has already been tested at the field level, but it has never been validated on a spatial scale that large.
Methods and results : The extrapolation model is based on an empirical approach that relies on a reference measurement and a spatial model of Ψ. Experiments were carried out over the 2946 ha of the denomination of Tavel (Gard, France). This site exhibits significant spatial variability in soil composition (three different soil units) and elevation, it is not irrigated and has a Mediterranean climate. This context leads to strong spatial variability in Ψ. Focusing on the most common grape variety used in the denomination, Grenache, a data base of predawn leaf water potential (PLWP) was built over three consecutive years (2008, 2009 and 2010). Respectively 10, 24 and 24 sites located throughout the denomination have been monitored for PLWP. The monitoring consisted respectively in 7, 10 and 9 measurement dates throughout the growing season. The spatial model was calibrated from 2009 and 2010 data and the prediction was tested on 2008 data. The results showed that the approach was completely transferable to a large territory like Tavel. The model significantly improved the quality of the prediction (R² = 0.9) compared to a conventional non-spatial method based on the mean values (R² = 0.61).
Conclusion: On the basis of a single measurement taken at time “t” on a reference site, the model makes it possible to estimate Ψ on any site of the study area. The results, therefore, demonstrated the ability to extrapolate a PLWP measurement made on a reference site over the whole denomination constituted of three different soil units.
Significance and impact of the study: This study represents a significant step since it demonstrates the existence of a spatial model of Ψ at a scale larger than the single field. It provides critical knowledge to consider the use of an empirical spatial model across a relatively large territory. This step opens up the possibility to provide spatial extrapolation model of Ψ based on data assimilation or existing data base.