Detecting Dubas bug infestations using high resolution multispectral satellite data in Oman

Al Shidi, Rashid H; Kumar, Lalit; Al-Khatri, Salim A H

doi:10.1016/j.compag.2018.12.037

Author(s)

Al Shidi, Rashid H

Kumar, Lalit

Al-Khatri, Salim A H

Publication Date

2018-12-20

Abstract

The Dubas bug, Ommatissus lybicus de Bergevin, is one of the major pests of the date palm, Phoenix dactylifera, in Oman, reducing its production by 28%. In addition to the important annual costs to control this pest nationwide, effort, cost and time are spent surveying and spotting O. lybicus infestations. Several studies have indicated the possibility of using remote sensing technology to identify plants stressed by pest infestation. The aim of the present study is to detect O. lybicus infestations by quantifying reflectance changes of different infestation levels and calculating different vegetation indices (VIs) using high-resolution multispectral (MS) images. An image of an area with different sub-locations that had varying levels of infestation was acquired in March 2017 using the WorldView-3 satellite. The reflectance of 8 bands, 32 spectral VIs and maximum likelihood classification (MLC) were derived from the image, and then the correlation was tested using ground-infestation data. The results revealed that the reflectance decreased in the red edge and near-infrared (NIR) bands as the infestation level increased. High levels of infestation showed a significant difference in three bands, red edge, NIR1 and NIR2, compared to no, low and medium levels of infestation. Nineteen out of 32 VIs showed a significant relation with the infestation levels. The relation ranged between r = −0.12, p < 0.05 using the Normalized Difference Mud Index (NDMI) and r = −0.39, p < 0.000 using the Transformed Difference Vegetation Index (TDVI)) and Tasselled Cap – Non - Such Index (TC-NSI). The location affected the relation between the infestation and VIs, where the correlation coefficients increased. The maximum correlation found was r = 0.64 using the Visible Atmospherically Resistant Index (VARI) in Al’Ayn Village. The overall accuracy of the supervised classification for detecting the infestation level was 68.3%, and the Kappa coefficient was 0.50.

Citation

Computers and Electronics in Agriculture, v.157, p. 1-11

ISSN

1872-7107

0168-1699

Link

https://hdl.handle.net/1959.11/29988

Language

en

Publisher

Elsevier BV

Title

Detecting Dubas bug infestations using high resolution multispectral satellite data in Oman

Type of document

Journal Article

Entity Type

Publication

Author(s)	Al Shidi, Rashid H Kumar, Lalit Al-Khatri, Salim A H
Publication Date	2018-12-20
Abstract	The Dubas bug, <i>Ommatissus lybicus</i> de Bergevin, is one of the major pests of the date palm, <i>Phoenix dactylifera</i>, in Oman, reducing its production by 28%. In addition to the important annual costs to control this pest nationwide, effort, cost and time are spent surveying and spotting <i>O. lybicus</i> infestations. Several studies have indicated the possibility of using remote sensing technology to identify plants stressed by pest infestation. The aim of the present study is to detect <i>O. lybicus</i> infestations by quantifying reflectance changes of different infestation levels and calculating different vegetation indices (VIs) using high-resolution multispectral (MS) images. An image of an area with different sub-locations that had varying levels of infestation was acquired in March 2017 using the WorldView-3 satellite. The reflectance of 8 bands, 32 spectral VIs and maximum likelihood classification (MLC) were derived from the image, and then the correlation was tested using ground-infestation data. The results revealed that the reflectance decreased in the red edge and near-infrared (NIR) bands as the infestation level increased. High levels of infestation showed a significant difference in three bands, red edge, NIR1 and NIR2, compared to no, low and medium levels of infestation. Nineteen out of 32 VIs showed a significant relation with the infestation levels. The relation ranged between r = −0.12, p < 0.05 using the Normalized Difference Mud Index (NDMI) and r = −0.39, p < 0.000 using the Transformed Difference Vegetation Index (TDVI)) and Tasselled Cap – Non - Such Index (TC-NSI). The location affected the relation between the infestation and VIs, where the correlation coefficients increased. The maximum correlation found was r = 0.64 using the Visible Atmospherically Resistant Index (VARI) in Al’Ayn Village. The overall accuracy of the supervised classification for detecting the infestation level was 68.3%, and the Kappa coefficient was 0.50.
Citation	Computers and Electronics in Agriculture, v.157, p. 1-11
ISSN	1872-7107 0168-1699
Link	https://hdl.handle.net/1959.11/29988
Language	en
Publisher	Elsevier BV
Title	Detecting Dubas bug infestations using high resolution multispectral satellite data in Oman
Type of document	Journal Article
Entity Type	Publication

Detecting Dubas bug infestations using high resolution multispectral satellite data in Oman

Files: