Assessing date palm health using Remote Sensing: The case of the red palm weevil

Organisator Laboratory of Geo-information Science and Remote Sensing

di 6 juni 2017 10:30 tot 11:00

Locatie Gaia, building number 101
Droevendaalsesteeg 3
6708 PB Wageningen
+31 317 48 16 00
Zaal/kamer 1

By Maggie Mully

Date palms are an extremely valuable crop, primarily cultivated in the Middle East and North Africa where they can be grown in highly saline, hot conditions. Due to cultivation practices and global transport of palms, red palm weevil (Rhynchophorus ferrugineus) has emerged as a destructive force in many date-growing countries. Whilst progress has been made in managing this pest, it still remains difficult to detect in the early stages of infestation. Remote sensing has been proposed as a potential technological solution for early detection and assessment of the entire field. A few studies used thermal imagery to successfully detect unhealthy date palms. However, these findings need to be replicated, and other remote sensing imaging techniques have not been investigated in the date palm case. In this study, LiDAR, thermal, hyperspectral and RGB images have been recorded for a date palm plantation in Saudi Arabia. Height from LiDAR and temperature from thermal images have been extracted for several sections of the date palm plantation in order to determine whether the general health of palms can be assessed on a block level. Height area ratio (HAR) was proposed as a proxy for health, and was used to determine healthy and unhealthy trees. Using ANOVA, a statistically significant P value of 0.00061 was recorded when comparing mean temperature of healthy and unhealthy trees - further analysis using Tukey’s HSD revealed that inter-block variation was the cause of this trend. Due to geometric distortions in the hyperspectral data, only a small subset of healthy and unhealthy trees was used for analysis. Higher reflectance in the visible spectrum for unhealthy trees suggests that hyperspectral imagery has high potential for use in this case. In conclusion, it was possible to see health trends on a block level using temperature and height data. Results on an individual tree level are inconclusive due to the lack of field data. In the future, field data should be collected for validation purposes, and intensity of backscatter reflectance for LiDAR, hyperspectral imagery, and temporal and spatial analysis should be explored.