Mapping Date Palm Structure and Age Using UAV Imagery

Abstract : Dates are oval-shaped sweet fruits growing on date palms. Date palm plantations in Saudi Arabia contribute 8.8% of the world's dates.  Saudi Arabia is the third largest date producer in the world. However, date productivity in Saudi Arabia is still relatively low e.g. due to overaged trees, use of low-quality cultivars and reduced date quality caused by suboptimal management and harvesting practices and losses from pests and diseases. Remotely sensed derived information has the potential to significantly contribute to increasing yield and quality of date production. It is important to determine the date palm density, structure and age to optimize management and production practices and to determine if mechanized harvesting methods are feasible. Narrow spacing between date palms has been linked to reduced yield and date quality and increased likelihood of spreading Red Palm Weevils, which is a major pest. Date palms usually die if they are exposed to several overlapping generations of Red Palm Weevils. Mapping date palm age is important for managing infestations, as about 75% of infested date palms are 6-15 years old. The objectives were to: (1) automatically delineate individual date palms for density mapping based on multi-spectral UAV imagery; and (2) to map structural parameters (crown projective cover, perimeter, width, area, height) and age groups. Our study area was located in the Al-Qaseem region of Saudi Arabia. Field data were collected of crown projective cover, perimeter, width, length, area and height. The age of palm trees were determined based on grower information and assessment of Google Earth time-series imagery. A multi-spectral camera mounted to a DJI Matrice-100 platform was used to collect imagery with 5 cm pixels in the visible and near infrared parts of the spectrum at three selected sites each including approximately 1000 date palms. The eCognition Developer software was used to automatically delineate date palms by first mapping their extent and then using an averaging filter on the binary extent map to identify the likely location of date palm center points. A region-growing algorithm was then used to grow date palm center points outwards to include associated palm leaves. More than 98% of all date palm trees were correctly identified. Crown projective cover showed good correlation with spectral vegetation indices (r2 > 0.75). Image derived measurements of perimeter, width, length and area of date palms corresponded well with field derived measurements, although these perimeters were in some cases over- or underestimated for date palms with leaves overlapping those of adjacent palms. Height of date palms was measured using Structure-from-Motion. The height measurements corresponded to field derived measurements, although height estimates were underestimated for sites with narrow spacing between date palms. The perimeter, area and height of individual date palms could be used to estimate age up to approximately 20 years, beyond which no significant relationship was observable. These findings may contribute to improved management practices and production of dates in Saudi Arabia. Future work will focus on scaling up the mapping result to cover larger areas based on multi-temporal satellite imagery.
Document type :
Poster communications
Complete list of metadatas

http://hal.univ-reunion.fr/hal-01959798
Contributor : Réunion Univ <>
Submitted on : Wednesday, December 19, 2018 - 7:33:08 AM
Last modification on : Tuesday, April 16, 2019 - 12:42:32 PM

Identifiers

  • HAL Id : hal-01959798, version 1

Collections

Citation

Kasper Johansen, Samir Al-Mashharawi, Yoann Malbeteau, Bruno Solorio, Matteo Ziliani, et al.. Mapping Date Palm Structure and Age Using UAV Imagery. GEOBIA 2018 - From pixels to ecosystems and global sustainability ​, Jun 2018, Montpellier, France. ⟨https://geobia2018.sciencesconf.org/⟩. ⟨hal-01959798⟩

Share

Metrics

Record views

53