|TSM/TDM Science Team Meeting 2016|
|Science Team Meetings|
|How to Submit a Proposal|
|COFUR Price List (scientific use)|
|TanDEM-X Science Service System|
|Title||Evaluation of Forest Mapping Capability of TerraSAR-X Spotlight SAR Data|
|Investigator||Chen, Erxue - Chinese Academy of Forestry, Institute of Forest Resources Information Techniques|
The objective of the project is to evaluate the forest mapping capability of TerraSAR-X Spotlight SAR data in an established test site in Shandong Province of China, and to develop effective data processing and classification method to be able to improve forest mapping accuracy. The Spotlight SAR data preprocessing methodology such as radiometric calibration, GEC and GTC product generation are to be studied as the basic supporting techniques for the operationally applying TerraSAR-X Spotlight SAR data to forest mapping.
The Range-Doppler model based geocoding algorithm will be adopted to the geometric correction of the Spotlight SAR data. The developed geocoding algorithm should be able to geocode SSC and MGD to GEC and GTC product. High resolution DEM will be digitized from 1:10000 scale topographical map sheets for the test site to support the GTC image generation. Radiometric calibration program to derive backscattering coefficient for distributed target from SSC and MGD products also need to be developed firstly. For MGD and SSC SAR products, the Spotlight SAR product is firstly radiometric calibrated and output image is stored in intensity image format, which will be filtered with Lee filter. The filtered intensity image is then transformed into backscattering coefficient in db format as inputs to classifiers.
SSC product can also be processed in the way of POLSARPRO® data processing routes. The two polarization complex data can be formed into [S] matrix and then transformed into covariance matrix [C2], which can be filtered using Josen-Lee’s polarimetric SAR filter and classified using complex Wishart supervised classifier.
For multi-temporal SAR image series, multi-temporal speckle filter developed by Prof. S. Quegan should be applied before classification.
Object classifier such as that provided by eCognition® is preferred to do the classification of the high spatial resolution spotlight SAR dataset. The high resolution SAR image is firstly segmented into self-contained objects, the statistics of each object, the texture and other many parameters for an object can be thought as features supplied to a classifier. But, commonly, this kind of classification method can not handle complex SAR image directly, the phase information has to be discarded in this way. So the complex Wishart classifier will be evaluated for forest mapping.
Ground true data will be collected for the test site and used to get labeled samples for classifier training and classification accuracy evaluation. N-fold validation method will be applied to the accuracy evaluation procedure.
(1) Spotlight SAR (SL), Single Polarization (VV), 3 scenes acquired in Nov. 2006, Mar. 2007, July 2007 respectively. MGD. Descending. 45 incidence angle.
(2)Spotlight SAR (SL), Dual polarization (HH+VV), 3 scenes acquired in Oct. 2006, April. 2007, Aug. 2007 respectively. SSC. Descending. 45 incidence angle.
(3) Spotlight SAR (HS), Single Polarization (VV), 3 scenes acquired in Dec. 2006, May 2007, Sept. 2007 respectively. MGD. Descending. 45 incidence angle.
(4)Spotlight SAR (HS), Dual polarization (HH+VV), 3 scenes acquired in Nov. 2007, March. 2008, July 2008 respectively. SSC. Descending. 45 incidence angle.
Report of the progress of the project every 6 month; Publish paper in public Journals and conference proceeding and attend the symposium or workshop for TerraSAR-X research and development.
The budget is around 30000 US$ and will be supported by the Ministry of Science and Technology of China (MOST) and European Space Agency (ESA) cooperation programme named DRAGON. The team already has gotten enough funding for the Polarimetric SAR relevant MOST-ESA cooperation project.
Back to list of proposals
|© DLR 2004-2016||