Articles
Study of intra-block water stress variability in a vineyard based on multispectral satellite imagery
Article number
1409_2
Pages
7 – 14
Language
English
Abstract
Monitoring vine water status during the growing season is essential for achieving the desired balance between wine grape quality and yield.
Traditional ground-based measurements such as stem water potential (ΨS) incorporates the soil, plant, and meteorological conditions, but this method is labour intensive, time-consuming and cannot be implemented as a monitoring tool for grapevine water status management on a block level.
In this sense, remote sensing methods offer non-destructive alternatives by indirect estimations.
The objective of this study was to explore relationships between vegetation indices (VIs) obtained by satellite imagery and ground-based midday stem water potential (ΨS) measurements as indicators for intra-block variability of water stress.
The field experiment consisted of 42 georeferenced target vines, located 10 m apart in a 2.42 ha commercial ‘Cabernet Sauvignon’ vineyard block trained on a vertical shoot-positioning (VSP) system in Stellenbosch, South Africa.
Multispectral images at 3 levels of resolution: i) low resolution – Landsat (30 m), ii) middle resolution – Sentinel 2 (10 m), iii) high-resolution – PlanetScope (3 m) were obtained at veraison where ΨS variability was maximum. ΨS values were grouped into 3 water stress classes and VIs extracted from images were assessed for sensitivity to detect similar patterns of water stress.
First step was to compare VIs trends calculated across the block for each image.
Secondly, point extraction was performed per target vine location.
Results indicated similar VIs trends observed between the 3 levels of spatial resolutions, but with clear differences in the range of the VIs calculated.
Minimum and maximum values increased 1.5-fold from low resolution (30 m) to high resolution (3 m) satellite images.
The comparison between ΨS and VIs classes was done by means of a confusion matrix, which indicated an accuracy level below 50% for all 3 spatial resolutions.
Timely based mapping of natural spatial and temporal variability of plant water status is critical for aiding site-specific management decisions during important phenological stages according to predefined stress levels.
Traditional ground-based measurements such as stem water potential (ΨS) incorporates the soil, plant, and meteorological conditions, but this method is labour intensive, time-consuming and cannot be implemented as a monitoring tool for grapevine water status management on a block level.
In this sense, remote sensing methods offer non-destructive alternatives by indirect estimations.
The objective of this study was to explore relationships between vegetation indices (VIs) obtained by satellite imagery and ground-based midday stem water potential (ΨS) measurements as indicators for intra-block variability of water stress.
The field experiment consisted of 42 georeferenced target vines, located 10 m apart in a 2.42 ha commercial ‘Cabernet Sauvignon’ vineyard block trained on a vertical shoot-positioning (VSP) system in Stellenbosch, South Africa.
Multispectral images at 3 levels of resolution: i) low resolution – Landsat (30 m), ii) middle resolution – Sentinel 2 (10 m), iii) high-resolution – PlanetScope (3 m) were obtained at veraison where ΨS variability was maximum. ΨS values were grouped into 3 water stress classes and VIs extracted from images were assessed for sensitivity to detect similar patterns of water stress.
First step was to compare VIs trends calculated across the block for each image.
Secondly, point extraction was performed per target vine location.
Results indicated similar VIs trends observed between the 3 levels of spatial resolutions, but with clear differences in the range of the VIs calculated.
Minimum and maximum values increased 1.5-fold from low resolution (30 m) to high resolution (3 m) satellite images.
The comparison between ΨS and VIs classes was done by means of a confusion matrix, which indicated an accuracy level below 50% for all 3 spatial resolutions.
Timely based mapping of natural spatial and temporal variability of plant water status is critical for aiding site-specific management decisions during important phenological stages according to predefined stress levels.
Authors
A. Berry, M.A. Vivier, C. Poblete-Echeverría
Keywords
grapevines, stem water potential, vegetation index, water stress
Online Articles (64)