Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Etats Unis → 2016 → Quantification of Ecological Change Using Repeat Photography and Ground Based Lidar

Oregon State University (2016)

Quantification of Ecological Change Using Repeat Photography and Ground Based Lidar

Batchelor, Jonathan L.

Titre : Quantification of Ecological Change Using Repeat Photography and Ground Based Lidar

Auteur : Batchelor, Jonathan L.

Université de soutenance : Oregon State University

Grade : Master of Science (MS) 2016

Remote sensing techniques have long been useful in quantifying changes in ecosystems and the field of remote sensing is constantly evolving to better assess and describe changes, both spatially and temporally. In this thesis I explored the novel use of two remote sensing methods to quantify ecosystems ; repeat photograph to describe change over time, and ground based lidar to describe change spatially. Using repeat photography, I assessed the effects of the removal of livestock in riparian systems at Hart Mountain National Antelope Refuge in southeastern Oregon, 23 years after the cessation of cattle grazing. I compared photos taken before grazing was ended with later retake photos. Two methods were used for this assessment : 1) a qualitative visual method comparing seven cover types and processes and 2) a new quantitative method of inserting digital line transects into photos. Results indicated that channel widths and eroding banks decreased in 64% and 73% of sites, respectively. I found a 90% decrease in the amount of bare soil (p<0.001) and a 63 % decrease in exposed channel (p<0.001) as, well as a significant increase in the cover of grasses/sedges/forbs (15% increase, p=0.037), rushes (389% increase, p=0.014) and willow (388% increase, P<0.001). I also assessed the accuracy of the new method of inserting digital line transects into photo pairs. An overall accuracy of 91% (kappa 83%) suggests digital line transects can be a useful tool for quantifying vegetation cover from photos. My results indicate that the removal of cattle can result in dramatic changes in riparian vegetation, even in semi-arid landscapes and without active restoration treatments. I used ground based lidar to quantify forest structure in a spatially explicit manner. The structural complexity of a forest has profound influences on its ecological functioning and overall health. The arrangement and amount of aboveground biomass are two important components of this structural complexity that influence the biodiversity in a forest ecosystem. In this paper, I explored and develop novel depth and openness metrics derived from single point ground-based lidar scans, which can quantify this complexity to a higher level of detail. Depth is a 3D, radial measure of the visible distance in all directions from a location (e.g., at the scanner origin). Openness is the percent of scan pulses in the near-omnidirectional view without a return,. To derive these metrics, I collected 243 scans at 27 forested stands in the Pacific Northwest region of the United States, representing a broad range of forest structure types. I created structural signatures using depth and openness for each stand and determined that our metrics could reliably differentiate forests. These metrics were compared to several traditional metrics : diameter at breast height (DBH), basal area, and diameter diversity index (DDI). The mean and variance in DBH of trees, basal area, and DDI of each stand were determined by generating isovists (polygons measuring visible space from a location) derived from point cloud cross sections. Interestingly, there was only weak to moderate correlation observed between depth or openness metrics when compared with the mean and variance in DBH, basal area, and DDI, suggesting that our new metrics (depth and openness) quantify a wider range of aspects of structure at the stand and plot level that are not captured by those traditional metrics. The proposed metrics can quantify forest structure at a high level of precision, reduce observer bias, and preserve a level of complexity lost in simple indices. The proposed metrics have great potential for quantifying change in forested systems, and describing habitat for organisms.

Présentation ->

Version intégrale (115 Mb)

Page publiée le 22 décembre 2018, mise à jour le 7 mars 2019