Articles
Determination of Solidago spp. (goldenrod) weeds in wild blueberry fields using a Lidar sensor
Article number
1440_23
Pages
163 – 170
Language
English
Abstract
The traditional approach of a blanket application of herbicides to control weeds requires high volumes of control products at significant cost and environmental impact.
Recent developments using remote sensing approaches have allowed for possible weed detection.
This study aimed to evaluate the potential of a light detection and ranging (LiDAR) sensor to accurately measure and map goldenrod weeds (Solidago spp.). This study was conducted in the cropped season of 2024 on a commercial wild blueberry field located at Earltown which had relatively high goldenrod weed pressures.
The aerial survey was conducted at 10, 20, 50, 80, and 100 m using a DJI Matrice 300 RTK unmanned aerial system (UAS) equipped with a Zenmuse L1 laser scanner.
The resulting point cloud data set was transformed into the digital surface model (DSM) and the digital elevation model (DEM) from which normalized DSM (nDSM) was derived.
Features with height values greater than 15 cm were classified as goldenrod weeds.
Mapped weed structures at 10 and 20 m flight altitudes had a mean weed height of 47.1 and 45.2 cm in comparison with the field measured value of 45.0 cm, while the higher flight altitude data had lower weed height values.
The presence of noise observed in the lower altitude points cloud data set could introduce errors into the prescription map being derived if not pre-processed.
In summary, this approach provided a data set needed for generating prescription maps for the spot application of herbicides instead of a blanket application.
As a result, farmers may attain adequate weed control with substantially less quantities of herbicides being required, resulting in cost savings.
Recent developments using remote sensing approaches have allowed for possible weed detection.
This study aimed to evaluate the potential of a light detection and ranging (LiDAR) sensor to accurately measure and map goldenrod weeds (Solidago spp.). This study was conducted in the cropped season of 2024 on a commercial wild blueberry field located at Earltown which had relatively high goldenrod weed pressures.
The aerial survey was conducted at 10, 20, 50, 80, and 100 m using a DJI Matrice 300 RTK unmanned aerial system (UAS) equipped with a Zenmuse L1 laser scanner.
The resulting point cloud data set was transformed into the digital surface model (DSM) and the digital elevation model (DEM) from which normalized DSM (nDSM) was derived.
Features with height values greater than 15 cm were classified as goldenrod weeds.
Mapped weed structures at 10 and 20 m flight altitudes had a mean weed height of 47.1 and 45.2 cm in comparison with the field measured value of 45.0 cm, while the higher flight altitude data had lower weed height values.
The presence of noise observed in the lower altitude points cloud data set could introduce errors into the prescription map being derived if not pre-processed.
In summary, this approach provided a data set needed for generating prescription maps for the spot application of herbicides instead of a blanket application.
As a result, farmers may attain adequate weed control with substantially less quantities of herbicides being required, resulting in cost savings.
Publication
Authors
O.S. Popoola, K.E. Anku, D.C. Percival
Keywords
goldenrod weeds, remote sensing, image classification, LiDAR, digital surface models
Online Articles (74)