Science Service System
You are here : Home : Proposals_Summary

Summary of Proposal LAN0591

TitleTesting, improving and validating algorithms for the retrieval of target forestry parameters using Terra-SAR X and Radarsat-2 data
Investigator Poncos, Valentin - Kepler Space Inc. / University of Calgary, SAR systems and exploitation
Team Member
Mr. Poncos, Valentin - Kepler, SAR Research & Development
SummaryThe objective of the proposal is to extract as much information as possible from dual and eventually quad-polarization Terra-SAR X data with respect to backscaterring mechanisms of boreal forests. Terra-SAR X is one of the few operational missions that provide high quality dual-polarization data and it is perceived as a well supported, robust service. Having this in mind, there is a lot of interest to pursue research and development of Terra-SAR X related techniques that could evolve to an operational level. The starting point of the R&D activity is to develop a methodology for generation of high quality interferograms that will serve as an input step to PolInSAR analysis. Although interferograms generation is seen as a standard process, there are still many ways to improve the quality of the phase and coherence. And since these parameters are very important to PolInSAR, it is expected to also improve the output from the current algorithms. The next step will consist in trying to process the X band data with PolSARPro, if this will be available within a reasonable time. If not, we will try to develop code that would customize PolSARPro processing for Terra-SAR X data. We hope that here we will get valuable information from Irena Hajnsek and maybe from other PolSARPro authors. In parallel, we will process Radarsat-2 quad-polarization data with PolSARPro and decide on eventual improvement. We will try to get a good understanding of the processing approach for C-band data and find ways to improve it. We would like to apply/update our findings to Terra-SAR X data. We will need data to be acquired in the month of July when the forest is the most developed and stable and the chances for temporal decorrelation are lower. We would like to acquire 2 sets of at least 3-4 images each, with interferometric geometry. One set could be acquired with a steep look angle and another with a wider look angle, to analyze differences in coherence levels, eventually related to the penetration depth. We would like at least 3 images with the same geometry to increase the chances for a low atmospheric interferograms and form optimum baselines. Also multiple interferometric scenes may be used to apply multi baseline PolInSAR or Polarization Coherence Tomography. The deliverables will consist in integrated results on forest parameters obtained with diffferent methods and data, out of which we hope that Terra-SAR X will play an important role.
Final ReportThe objective of this project was to estimate the possibility of using TerraSAR-X data to derive forest height maps using PolInSAR, PolSAR or InSAR. The PolInSAR task was assigned to DLR (Dr. Irena Hajnsek), the PolSAR task was assigned to Gamma Remote Sensing (Dr. Maurizio Santoro) and the InSAR task was assigned to Kepler Space (Valentin Poncos). The test site was a boreal forest in Timmins, Ontario, Canada. Ancillary data (Lidar DEM, high resolution optical images, tree species information) was provided by Dendron Resources Inc. TerraSAR-X was acquired in two periods (summer and winter). The summer data was processed by DLR and the coherence level was found to be too low for PolInSAR. Kepler estimated the coherence levels for the winter TerraSAR-X data and provided the results to DLR; the conclusion was that the coherence level of the winter data wasn't significantly higher than the one from the summer data and PolInSAR analysis was still impossible. In the same time, quad-polarization Radarsat-2 data was acquired for the same site; Gamma Remote Sensing underwent the task of analysing this data using PolSAR, coherence based methods and PolInSAR. PolInSAR was dismissed for the same reason invoked by DLR. PolSAR and coherence based analysis was also inconclusive and a forest height map wasn't generated. Kepler processed the data for interferometry; although the coherence levels are low, it was shown that there is some valid information in the data. A proper unwrapping process was carried on and height maps were generated. The height maps resemble the Lidar and C-band SRM in general. It is very difficult to exactly compare the three DEMs because of the different penetration depths and also temporal decorrelation from TerraSAR-X. The task is not impossible though and it will be carried on by Kepler as part of future projects (the current ESA project was ended with the DLR and Gamma remarks as conclusions). Kepler also tried to process the TerraSAR-X data with ESA PolSARPro software. It turns out that PolSARPro has problems in the first part, when interferograms are created. It is missing a good coregistration step, all the required filtering and the orbits correction is not available. As a result, the interferograms created with PolSARPro are not correct and any further PolInSAR analysis is due to fail. Kepler contacted ESA to report on the problems and it turns out that PolSAR Pro is designed for ALOS Palsar, which is less sensitive to InSAR processing problems. The results are not publishable. Kepler Space and Dendron Resource Survey appreciate the support from DLR for both data acquisition and processing.

Back to list of proposals

DLR 2004-2016