Automatic UAV-based detection of Cynodon dactylon for site-specific vineyard management

Area of Science:

• Precision agriculture
• Remote sensing
• Weed management

Background:

• Bermudagrass (Cynodon dactylon) is a pervasive weed in vineyards, difficult to distinguish from grapevines using traditional spectral analysis due to their similar spectral signatures.
• Effective weed mapping is crucial for targeted management strategies to mitigate crop yield loss and reduce reliance on broad-spectrum herbicides.

Purpose of the Study:

• To develop an automated, accurate, and rapid method for mapping bermudagrass infestations in vineyards.
• To design site-specific management maps for bermudagrass control.
• To assess the potential of Unmanned Aerial Vehicle (UAV) technology combined with object-based image analysis (OBIA) for precision weed management.

Main Methods:

• Acquisition of multispectral aerial imagery (RGB and RGNIR) using UAVs in two vineyard sites.
• Spectral analysis to identify the optimal vegetation index (VI) for distinguishing bermudagrass from bare soil.
• Development and application of a VI-based OBIA algorithm to automatically classify grapevines, bermudagrass, and bare soil.

Main Results:

• The VI-based OBIA algorithm achieved high classification accuracies (greater than 97.7%) for mapping grapevines, bermudagrass, and bare soil.
• The study successfully generated site-specific management maps, enabling quantification of grapevine growth and identification of expanding bermudagrass-infested areas.
• Combining UAV imagery with OBIA provided valuable geospatial information for targeted weed control.

Conclusions:

• UAV-based OBIA offers an effective solution for the automatic and accurate mapping of bermudagrass in vineyards, overcoming spectral similarity challenges.
• The developed methodology supports the design of well-programmed, site-specific weed management strategies, potentially leading to reduced herbicide use, optimized fuel consumption, and lower operational costs.
• This approach provides farmers with crucial data for improving vineyard management efficiency and sustainability.