Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Optimizing winter wheat leaf area index estimation across growth stages using UAV data

Shao, Ke LU (2024) In Student thesis series INES NGEM01 20241
Dept of Physical Geography and Ecosystem Science
Abstract
Leaf Area Index (LAI) is an important parameter for monitoring vegetation status and estimating crop yield, essential for precision agriculture, forestry and natural resource management. Therefore, estimating the LAI with high spatial resolution and temporal flexibility is crucial for providing precise and timely information to make informed decisions. This study focuses on optimizing the estimation of LAI in winter wheat across different growth stages by integrating spectral data and plant height, with a high spatial resolution of 4 cm × 4 cm and 5 cm × 5 cm respectively, which derived from Unmanned Aerial Vehicle (UAV) imagery. The research investigates which spectral bands or vegetation indices (VIs) are most effective for LAI... (More)
Leaf Area Index (LAI) is an important parameter for monitoring vegetation status and estimating crop yield, essential for precision agriculture, forestry and natural resource management. Therefore, estimating the LAI with high spatial resolution and temporal flexibility is crucial for providing precise and timely information to make informed decisions. This study focuses on optimizing the estimation of LAI in winter wheat across different growth stages by integrating spectral data and plant height, with a high spatial resolution of 4 cm × 4 cm and 5 cm × 5 cm respectively, which derived from Unmanned Aerial Vehicle (UAV) imagery. The research investigates which spectral bands or vegetation indices (VIs) are most effective for LAI estimation, the variability of LAI estimation across growth stages, and the enhancement of LAI estimation through the addition of plant height data. The result of the coefficient of determination (R2) showed that the Normalized Difference Red Edge (NDRE) was the most effective single feature for LAI estimation (R2 = 0.64), followed by the Chlorophyll index with red edge (CIrededge) with an R2 of 0.56, while red edge (REG) reflectance showed limited predictive capability (R2 = 0.01). Models incorporating multiple features generally improved estimation accuracy, demonstrating the benefit of multi-feature models. The combination of height and CIrededge achieved the highest predictive accuracy and lowest error rates for multi-feature models, with an R2 of 0.82 and a root mean square error (RMSE) of 0.37 m²/m². Growth stage-specific models further refined LAI estimation, with vegetative and productive stages showing distinct performance variations. The study concludes that NDRE can estimate LAI with a high accuracy overall while integrating the CIrededge and height data from UAVs enhances the precision of LAI estimation significantly, and the consideration of growth stages is necessary. (Less)
Please use this url to cite or link to this publication:
author
Shao, Ke LU
supervisor
organization
course
NGEM01 20241
year
type
H2 - Master's Degree (Two Years)
subject
keywords
Leaf area index, multispectral UAV, winter wheat, machine learning, random forest
publication/series
Student thesis series INES
report number
663
language
English
additional info
External supervisor: El Houssaine Bouras, Center for Remote Sensing Application, Mohammed VI Polytechnic
id
9165891
date added to LUP
2024-06-19 20:39:44
date last changed
2024-06-19 20:39:44
@misc{9165891,
  abstract     = {{Leaf Area Index (LAI) is an important parameter for monitoring vegetation status and estimating crop yield, essential for precision agriculture, forestry and natural resource management. Therefore, estimating the LAI with high spatial resolution and temporal flexibility is crucial for providing precise and timely information to make informed decisions. This study focuses on optimizing the estimation of LAI in winter wheat across different growth stages by integrating spectral data and plant height, with a high spatial resolution of 4 cm × 4 cm and 5 cm × 5 cm respectively, which derived from Unmanned Aerial Vehicle (UAV) imagery. The research investigates which spectral bands or vegetation indices (VIs) are most effective for LAI estimation, the variability of LAI estimation across growth stages, and the enhancement of LAI estimation through the addition of plant height data. The result of the coefficient of determination (R2) showed that the Normalized Difference Red Edge (NDRE) was the most effective single feature for LAI estimation (R2 = 0.64), followed by the Chlorophyll index with red edge (CIrededge) with an R2 of 0.56, while red edge (REG) reflectance showed limited predictive capability (R2 = 0.01). Models incorporating multiple features generally improved estimation accuracy, demonstrating the benefit of multi-feature models. The combination of height and CIrededge achieved the highest predictive accuracy and lowest error rates for multi-feature models, with an R2 of 0.82 and a root mean square error (RMSE) of 0.37 m²/m². Growth stage-specific models further refined LAI estimation, with vegetative and productive stages showing distinct performance variations. The study concludes that NDRE can estimate LAI with a high accuracy overall while integrating the CIrededge and height data from UAVs enhances the precision of LAI estimation significantly, and the consideration of growth stages is necessary.}},
  author       = {{Shao, Ke}},
  language     = {{eng}},
  note         = {{Student Paper}},
  series       = {{Student thesis series INES}},
  title        = {{Optimizing winter wheat leaf area index estimation across growth stages using UAV data}},
  year         = {{2024}},
}