Science Service System

Summary of Proposal LAN0126

TitleChange detection in boreal forest
Investigator Eriksson, Leif - Chalmers University of Technology, Dept. of Earth and Space Sciences
Team Members
Professor Ulander, Lars - Swedish Defence Research Agency (FOI), Radar Systems
Dr. Fransson, Johan - Swedish University of Agriculutral Sciences, Forest Resource Management and Geomatics
Mr. Persson, Anders - Swedish National Board of Forestry, GIS and Remote Sensing
Eriksson, Leif - Chalmers University of Technology, Dept. Of Radio and Space Science
Smith-Jonforsen, Gary - Chalmers University of Technology, Dept. Of Radio and Space Science

The objectives for this project are to perform a scientific evaluation of TerraSAR-X for detecting and mapping changes in boreal forests. The primary aim is to evaluate whether TerraSAR-X can be used for mapping of wind throw as this is an application which requires the all weather day & night capability of SAR to provide timely maps, but to-date no satellite systems have been able to provide data with sufficient accuracy. As a secondary, related topic we will also evaluate the ability to map clear-cutting of forest in Sweden using TerraSAR-X, and compare price and accuracy with alternative techniques. At present optical satellites are used, however uncertainties in the future availability of this data have increased the interest to find alternatives.

The basic approach will be controlled cutting of the forest in small test-areas to provide data for a full evaluation. Firstly trees will be felled, but left in place to simulate wind thrown forest. Then the same areas will be cleared to provide data for mapping clear-cuts. The use of controlled experiments will allow a time series from before and after the changes to be collected, and analysed for changes in interferometric coherence as well as image texture both techniques which have shown more promise than measuring backscatter intensity alone.

The project will operate over two small test-sites; in the south of Sweden on relatively flat terrain, and in the north in a more topographic area. The differences in weather conditions, topography and forest properties are important to show that the methods developed are robust for future operational use, particular with respect to the large seasonal changes occurring at high latitudes. Since the image texture depends greatly on the effect of shadows, we would like to analyse the importance of incidence angle, which will be performed at one of the test-sites in an initial phase. This will also include analysis of the trade-off between higher resolution and the use of polarimetric data. At the end of this stage the most promising imaging mode and incidence angle will be chosen and remain fixed throughout the project.

For change detection we plan to collect images from each test-site in 4 time intervals, i.e. twice before cutting (in winter and summer), after felling, and after clearing the trees. In each interval we request 4 high-resolution spotlight images to test the use of interferometry, and possibilities of temporal filtering, as well to evaluate the robustness to changes in the weather. In total 38 images are requested. The importance of resolution will be investigated by degrading the resolution to test whether other products may offer satisfactory performance (e.g. stripmap instead of spotlight to allow better coverage). Using this technique reduces the number of images required for this project, but also separates the effects of resolution from other effects (such as weather conditions on acquisition date), that could occur if using data acquired with the different imaging modes.

The results from this project will be published in scientific journals, and presented as recommendations for future use of TerraSAR-X data for meeting the operational mapping requirements of the government authorities in Sweden. If the results are good these experiences can be transferred to authorities in other countries (e.g. Finland, Canada and Russia), and to commercial forestry enterprises as a means of updating their inventories.

Funding for this project is hoped to be obtained from the Swedish National Space Board, and will be applied for in May 2006, with a decision being reached in December 2006 allowing for a project start in January 2007. The ground activities are planned to be synchronised with plans for similar projects using data from ALOS and RADARSAT-2, hence improving the cost-effectiveness of the study, allowing a comparison of the sensors and the possibility to investigate synergy.

Back to list of proposals

DLR 2004-2016