Customizing unmanned aircraft systems to reduce forest inventory costs: can oblique images substantially improve the 3D reconstruction of the canopy?

GASTÓN MAURO DÍAZ; DIEGO MOHR-BELL; MARTÍN GARRETT; LUCAS MUÑOZ; JOSÉ DANIEL LENCINAS

INTERNATIONAL JOURNAL OF REMOTE SENSING

TAYLOR & FRANCIS LTD

Lugar: Londres; Año: 2019

0143-1161

Photogrammetry makes it possible to estimate forest canopy surface at lower costs than light detection and ranging (LiDAR), which is considered the best data source to evaluate forest structure. Recent studies even suggest that points of forest understories can be obtained by means of unmanned aerial photogrammetry. However, little is known about how the characteristics of image sets, processing workflows, and forest openness affect understory point surveying. For forest inventories, unmanned aircraft systems (UASs) of fixed-wing type are preferred because they can survey large areas. It has been shown that the accuracy of UAS photogrammetry tends to increase by adding oblique images but acquiring them with fixed-wing UASs is challenging. To address this challenge, we proposed a multi-camera array for acquiring oblique images with fixed-wing UASs. To test our idea, we built two customized UAS and surveyed an open pine plantation and a tall deciduous forest with variable overstory density. The open plantation was selected for optimizing the measurement of reference canopy surface points using a terrestrial laser scanner. In the deciduous forest, we obtained reference understory points from a leaf-off photogrammetry survey. We confirm that including oblique images in the image set is a good practice for forestry applications. Using the UASs to test whether a multi-camera system is better than a single-camera system for acquiring nadir-oblique image sets, we conclude that the advantages are (1) more efficient acquisition of oblique images and (2) better understory modelling in open canopies. The multi-camera acquisition of oblique images increases the understory point density, making the estimation of crown cover percentage and maximum canopy height more accurate, by 33 % and 50 %, respectively.