Disturbance feedbacks on the height of woody vegetation in a savannah: a multi-plot assessment using an unmanned aerial vehicle (UAV). Issue 14 (1st August 2018)

• Record Type: Journal Article
• Authors: Mayr, Manuel J.; Malß, Sophia; Ofner, Elisabeth; Samimi, Cyrus
• Is Part Of: International journal of remote sensing. Volume 39:Issue 14(2018)
• Page Start: 4761
• Page End: 4785
• Publication Date: 2018-08-01
• Subjects: Remote sensing -- Periodicals; Télédétection -- Périodiques; 621.3678
• DOI: 10.1080/01431161.2017.1362132
• Languages: English
• ISSNs: 0143-1161

Abstract

ABSTRACT: Disturbances affect the woody, i.e. trees and shrubs, and herbaceous vegetation in savannah ecosystems worldwide. In Northern Namibia, livestock grazing and fires depict two prominent agents of disturbance. These affect the structural parameters of vegetation such as the height of woody species. Remote sensing is a tool to quantify such structural parameters. In particular, Image-Based Point Clouds (IBPCs) obtained from unmanned aerial vehicles (UAVs) are nowadays increasingly used for three-dimensional (3D) remote-sensing applications. Here we aim at deriving the height of woody stands through a multi-plot UAV campaign ( n = 19) carried out at the end of the dry season. We use direct georeferencing from the navigation-grade instruments on board the UAV in a Structure-from-Motion (SfM) approach. Watershed segmentation is applied to derive plot-scale height metrics (maximum, mean, and median) based on delineated individuals. Fire and grazing – both individually and synergistically – are then investigated for their impacts on UAV-derived height metrics. The results indicate good agreement between the UAV-derived and in situ -measured height metrics on the plot scale (coefficient of determination ( R 2 ) approximately 0.7, root mean square error (RMSE) <1.9 m). Underestimations of height are apparent with large, leafless trees. Clumping of equally sized individuals complicated their correct delineation. Grazing was found to be significant for all height metrics as well as in combination with fire for the plots' maxima. We conclude that the approach applied here is able to reproduce the plot-scale heights of woody vegetation with acceptable accuracy. We attribute the observed height reductions with the simultaneous presence of disturbances to legacy effects.