Evaluation of Ground Surface Models Derived from Unmanned Aerial Systems with Digital Aerial Photogrammetry in a Disturbed Conifer Forest

Abstract

Detailed vertical forest structure information can be remotely sensed by combining technologies of unmanned aerial systems (UAS) and digital aerial photogrammetry (DAP). A key limitation in the application of DAP methods, however, is the inability to produce accurate digital elevation models (DEM) in areas of dense vegetation. This study investigates the terrain modeling potential of UAS-DAP methods within a temperate conifer forest in British Columbia, Canada. UAS-acquired images were photogrammetrically processed to produce high-resolution DAP point clouds. To evaluate the terrain modeling ability of DAP, first, a sensitivity analysis was conducted to estimate optimal parameters of three ground-point classification algorithms designed for airborne laser scanning (ALS). Algorithms tested include progressive triangulated irregular network (TIN) densification (PTD), hierarchical robust interpolation (HRI) and simple progressive morphological filtering (SMRF). Points were classified as ground from the ALS and served as ground-truth data to which UAS-DAP derived DEMs were compared. The proportion of area with root mean square error (RMSE) <1.5 m were 56.5%, 51.6% and 52.3% for the PTD, HRI and SMRF methods respectively. To assess the influence of terrain slope and canopy cover, error values of DAP-DEMs produced using optimal parameters were compared to stratified classes of canopy cover and slope generated from ALS point clouds. Results indicate that canopy cover was approximately three times more influential on RMSE than terrain slope.

Assessing the Ability of Image Based Point Clouds Captured from a UAV to Measure the Terrain in the Presence of Canopy Cover

“…The only exceptions to this were at the PNT01 and ENS01 sites where the minimal leaf area allowed the flat terrain to be easily identified by the filters. It is acknowledged within the literature that the filters often applied to identify ground with image based point clouds were developed for ALS datasets [13]. The PNT01 and ENS01 datasets contained even distributions of ground points across the plots which is similar to what ALS data can capture.…”

“…The ability to represent the terrain is crucial in extracting structural information from UAS captured image based point clouds [13,14]. Whilst previous studies have developed strategies to overcome the lack of ground points present in this type of data, the availability of ground information in the presence of different canopy conditions has only been quantified on a few occasions (see Wallace et al [1], Graham et al [13], Goodbody et al [14] for examples). In these studies reference ground information is based on an alternate remote sensing dataset.…”

“…For example, in ALS data these errors can be up to 1 m [17]. In these studies, errors in the co-registration of the two point clouds are also likely propagate into terrain height errors [13,14]. In order to counter these effects, this study utilised a local coordinate system.…”

“…Nevertheless, several studies have extracted terrain description from UAS image-based point clouds, primarily in open canopy forests, to provide accurate terrain measurements. Wallace et al [1], Giannetti et al [2], Graham et al [13], Goodbody et al [14] for example showed that terrain information could be retrieved with similar accuracy (root mean square error (RMSE) of between 0.5-1.5 m) to that obtained using laser scanning datasets.…”

“…While using novel approaches to coregister the two datasets, studies such as Jayathunga et al [18], and Puliti et al [19] have successfully provided measures of forests metrics, timber volume, Lorey's mean height, stem number and basal area. Although the use of alternative sources of terrain information can be successful, often this data is not available and when it is, the co-registration of two datasets is complicated by the use of low cost positioning sensors onboard UAS [13].…”