Low-resolution remotely sensed images of winegrape vineyards map spatial variability in planimetric canopy area instead of leaf area index

Abstract

Background and Aims: Knowledge of the spatial variability of grapevine canopy density is useful in managing the variability of grape composition and yield. Rapid assessment of the characteristics of vineyards by remote sensing offers distinct advantages over ground-based measurements. In an effort to capture such advantages, this study aimed to assess the relative contribution to LAI of grapevine canopy density and grapevine canopy area derived from high-spatial-resolution airborne digital imagery. Methods and Results: High-spatial-resolution airborne NDVI imagery of minimally pruned, unconfined (i.e. not confined by trellising) grapevines was used to partition image pixels into grapevine-only and non-grapevine groupings. An evaluation of the relative contributions of grapevine planimetric area (number of grapevine pixels across a single row) and leaf layers (NDVI of grapevine-only pixels) found that the variability observed across the vineyard was dominated by changes in canopy area rather than grapevine-only NDVI. Conclusion: The primary predictive variable of grapevine LAI is canopy area. Low-spatial-resolution NDVI imagery of minimally pruned, unconfined vineyards is therefore effective in mapping spatial variability in planimetric canopy area, rather than LAI. Significance of the Study: The process of estimating grapevine LAI from mixed pixels has incorrectly assumed that both components of LAI within a pixel's footprint, namely the number of leaf layers and planimetric canopy area, produce a consistent response in NDVI. Correlations between NDVI and LAI reported in previous studies based on low-resolution imagery most likely relied on the proxy relationship between NDVI and canopy area.