The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Publications Copernicus
Download
Citation
Articles | Volume XL-1/W1
https://doi.org/10.5194/isprsarchives-XL-1-W1-195-2013
https://doi.org/10.5194/isprsarchives-XL-1-W1-195-2013
02 May 2013
 | 02 May 2013

AUTOMATIC DETECTION OF STORM DAMAGES USING HIGH-ALTITUDE PHOTOGRAMMETRIC IMAGING

P. Litkey, K. Nurminen, and E. Honkavaara

Keywords: Change Detection, Hazards, DSM, Image Matching, Forestry, Photogrammetry

Abstract. The risks of storms that cause damage in forests are increasing due to climate change. Quickly detecting fallen trees, assessing the amount of fallen trees and efficiently collecting them are of great importance for economic and environmental reasons. Visually detecting and delineating storm damage is a laborious and error-prone process; thus, it is important to develop cost-efficient and highly automated methods. Objective of our research project is to investigate and develop a reliable and efficient method for automatic storm damage detection, which is based on airborne imagery that is collected after a storm. The requirements for the method are the before-storm and after-storm surface models. A difference surface is calculated using two DSMs and the locations where significant changes have appeared are automatically detected.

In our previous research we used four-year old airborne laser scanning surface model as the before-storm surface. The after-storm DSM was provided from the photogrammetric images using the Next Generation Automatic Terrain Extraction (NGATE) algorithm of Socet Set software. We obtained 100% accuracy in detection of major storm damages. In this investigation we will further evaluate the sensitivity of the storm-damage detection process. We will investigate the potential of national airborne photography, that is collected at no-leaf season, to automatically produce a before-storm DSM using image matching. We will also compare impact of the terrain extraction algorithm to the results. Our results will also promote the potential of national open source data sets in the management of natural disasters.