Machine learning and image analysis on photos of a solitary tree in a complex background

extraction and analysis of key properties for wind-tree interaction

Master thesis (2020)

Authors

C.C.A. Bekkers Aerospace Engineering

Contributors

S.J. Watson Wind Energy - Aerospace Engineering (supervisor 1)
A.C. Viré Wind Energy - Aerospace Engineering (supervisor 1)
E. Dellwik (supervisor 1)
A.B. Dahl (supervisor 1)
Faculty
Aerospace Engineering
Copyright: © 2020 Casper Bekkers
More Info
expand_more

Published Date

18-08-2020

Language

English

Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Faculty

Aerospace Engineering

Abstract

A method is developed to automatically analyze over 200000 images of a single mature tree. This method is assessed and applied to a tree located at Risø in Roskilde, Denmark, next to two meteorological masts. Image aspects such as the surface area and the center of area are extracted from the images and they are related to the wind statistics. Concurrence is found between expected behaviors and observations to validate the method and unexpected behaviors are documented. Expected behaviors include the decrease in surface area with an increase in wind speed, swaying directional dependency with wind direction and swaying magnitude dependency with wind speed. New behaviors include a non-linear relationship between porosity and wind deficit and an increase in surface area with wind speed from infrequent wind directions. Furthermore, a specific period is analyzed which determined that for this tree, roughly half of the Vogel exponent is attributed to the decrease in surface area.