Embry-Riddle Aeronautical University
Unmanned Aircraft Systems, UAS, Drone, Agriculture, Normalized Difference Vegetation Index, NDVI, 3D, Point Cloud, Huanglongbing, HLB, Citrus, Orange
Three-dimensional NDVI point clouds can be an innovative method for detecting Huanglongbing (HLB) disease in citrus trees. In February 2018, an Unmanned Aircraft System (UAS) captured narrow-band multispectral images to detect healthiness variations of infected citrus trees. A 30-acre section of a citrus grove in Florida with a known HLB infection was examined to determine if three-dimensional Normalized Difference Vegetation Index (NDVI) point clouds can indicate healthiness variations in HLB-infected citrus trees and how three-dimensional NDVI point clouds compared to two-dimensional NDVI reflectance maps for detecting healthiness variations in HLB-infected citrus trees. Wilcoxon Sign Rank testing compared Whole-Tree Vegetation Indices (WTVI) comprising of point or pixel proportions within five NDVI classifications between three-dimensional NVDI point clouds and two-dimensional NDVI reflectance maps. The results indicated significant differences between three-dimensional and two-dimensional points, grouped at the tree level, for suspected HLB-infected trees (p = 0.000). The data suggests three-dimensional NDVI point cloud points were more sensitive to less healthy levels of NDVI values by 2.7% compared to two dimensional NDVI data for suspected HLB-infected trees and by 10.6% (p = 0.000) for non-suspected HLB-infected trees. Researchers concluded three-dimensional NDVI point clouds could be used to determine healthiness variations in suspected HLB-infected citrus trees. Three-dimensional NVDI point clouds had a wider distribution of five index classifications than two-dimensional NDVI reflectance maps for suspected HLB-infected trees. The vertical structure of the citrus tree may contribute to the difference in distribution. There was a 10.01% (p = 0.021) increase in 3D NDVI point cloud points for non-suspected HLB-infected trees compared to the suspected HLB-infected trees. Additionally, there was a 9.04% (p = 0.032) increase in tree crown dimension for non-suspected HLB-infected trees compared to suspected HLB-infected trees. These data suggest non-suspected HLB-infected trees were larger than suspected HLB-infected trees.
Scholarly Commons Citation
Cerreta, J., Hanson, A., Martorella, J. E., & Martorella, S. (2018). Using 3 Dimension Health Vegetation Index Point Clouds to Determine HLB Infected Citrus Trees. Journal of Aviation/Aerospace Education & Research, 28(1). https://doi.org/10.15394/jaaer.2018.1776