Peering through the thicket : effects of UAV LiDAR scanner settings and flight planning on canopy volume discovery

2022

Article référencé dans le Web of Science WOS:000907048500006

Brede B., Bartholomeus H. M., Barbier Nicolas, Pimont F., Vincent Grégoire, Herold M.

International Journal of Applied Earth Observation and Geoinformation, 2022, 114, p. 103056 [14 p.] ISSN 1569-8432

Unoccupied aerial vehicle laser scanning (UAV-LS) has been increasingly used for forest structure assessment in recent years due to the potential to directly estimate individual tree attributes and availability of commercial solutions. However, standardised procedures for campaign planning are still largely missing. This study investigated scanner properties and flight planning to provide recommendations on minimising forest canopy occlusion and thereby maximise exploration of canopy volume. A flight campaign involving two UAV-LS systems was conducted over a dense, wet tropical forest at the Paracou research station (French Guiana). Four experiments on scanner properties and flight planning were conducted, analysed and recommendations derived. First, the scanner pulse repetition rate (PRR) should be at least 100 kHz per 1 m s-1flight speed based on 360 degrees FOV for exploration of middle canopy strata (5 m to 20 m). Higher PRR are beneficial for exploration of lower canopy (<5 m) but would need to be increased exponentially to achieve linear improvement. Alternatively, flight speed could be reduced within the constraints given by the inertial measurement unit (IMU), but would increase flight time. Second, the scanner maximum range was identified as a proxy for the laser pulse power, which positively impacts canopy exploration. This was particularly the case when using multi-return capabilities. No saturation could be observed when increasing the laser power, suggesting that this is currently a limiting factor. Additionally, a smaller laser beam divergence and pulse width were plausible reasons for better exploration of the upper canopy just below the top of canopy. Third, off-nadir scanning angles up to 20 degrees were found to result in similar occlusions, suggesting a practical FOV of 40 degrees in the investigated dense forest. This number might be larger for open canopies. UAV-LS systems with viewing geometries that focus laser pulses downwards and within optimal ranges should be preferred. Fourth, using different horizontal flight directions in the mission planning favours minimisation of occlusion. A minimum of two different flight directions is suggested. However, specific optimal yaw angles were not possible to predict before flight. Therefore, including multiple directions ensures coverage of all possible configurations. Many of these investigated features can be optimised independently from each other, and should be considered before acquisition of new UAV-LS systems and flight mission planning. These results support the establishment of general guidelines for the investment in UAV-LS systems and optimal mission planning for forest structure assessment.

Etudes, transformation, conservation du milieu naturel [082] ; Télédétection [126]

Fonds IRD [F B010086810]

fdi:010086810