Nicolas Ackermann Supervisor : Prof. Christiane Schmullius

1. TSX Science Team Meeting Investigations of TerraSAR-X and Cosmo-SkyMed backscatter intensity and interferometric coherence over temperate forested areas Nicolas Ackermann Supervisor: Prof. Christiane Schmullius Co-supervisors: Dr. Christian Thiel, Dr. Maurice Borgeaud DLR Oberpfaffenhofen, the 16th February 2011 ENVILAND2 is sponsored by the Space Agency of the German Aerospace Center with federal funds of the German Federal Ministry of Economics and Technology on the basis of legislation by the German Parliament grant no. 50 EE 0844 - 50 EE 0847

2. Presentationoutline • Introduction • Test site & available data • Pre-processing & preliminary investigations • Analysis of the data • Conclusions

3. Introduction • Context: • The monitoring of forested areas represents a great challenge in the context of the actual Global Warming. • The launch of TerraSAR-X spaceborne system in June 2007 has been pioneer of new capabilities for the retrieval of forest biophysical parameters. • Objectives: • Investigate the TSX and CSK backscattter intensity and interferometric coherence. • Underline the scattering and decorrelation mechanisms occuring in the temperate forest. • Provide some new scientific knowledge for future researches.

4. Test site & available data

5. Test site • Thuringia Forest (Germany) • Surface: 110 km x 50 km • Terrain variations • Tree species composition • Scots pines • Norway Spruce • European Beech • Climate • cool and rainy • frequently clouded • Peculiarities • logging for forest exploitation • Kyrill storm (February 2007)

6. Test site Topography

7. Test site Forest understory

8. Available data • SAR data • ALOS PALSAR (L-Band, 46 days) • TerraSAR-X (X-Band, 11 days) • Cosmo-SkyMed (X-Band, 1 day) • Optical data • RapidEye • Kompsat-2 • Ancillary data • Laser DEM 5 m • Laser points (2004), • Orthophotos (2008) • HyMap (2008,2009) • Forest inventory (1989-2009) • Photos with GPS coord. (2009) • Weather data • Field work 25 km

9. Satellite data 80 scenes Thuringia Forest test site (state February 2010)

10. Pre-processing & preliminary investigations

11. High geometrical distorsions by steep incident angle. Topographic normalisation • Relief 5 x 10 [km] LiDAR DEM shaded relief TSX HS 23° HH Asc. 16jul09 Non normalised

12. Topographic normalisation • Correction main components (Castel et al., 2001) Local incident angle Ground scattering area

13. High intensity for steep slopes facing radar. Topographic normalisation • Experiment - Polarplot Slope oriented in radar flight direction Aspect [°] Aspect [°] Gamma nought [dB] Sensor heading : 350° Sensor azimuth angle : +90° 0° TSX SM 38° HH Asc. 21aug10 Non normalised TSX SM 38° HH Asc. 21aug10 Normalised

14. Overcorrection? • Crown optical depth? • Other effects? Topographic normalisation • Relief normalisation Gamma nought [dB] Aspect [°] TSX SM 38° HH Asc. 21aug10 Normalised

15. Weather data • Weather data • DWD: Deutsche Wetter Dienst • Acquisition period: 2006-2010 • Parameters: Precipitation, Snowdepth, Water-equivalent, Wind, Temperature, Sunshine duration, relative Humidity • Pre-processing • Collaboration with FSU geoinformatic institute • JAMS (Jena Adaptable Modelling System) Software • 2 temporal scales : daily / hourly Station 2 Station 1 Station 3 Station 5 Station 4 90m JAMS Software Daily input conversion Daily Selected [ .dat ] Daily Regionalisation Daily Regionalised [ .txt ] Daily output generation Daily [ raster ] Weather data [ .xml ] Hourly Input conversion Hourly Selected [ .dat ] Hourly Regionalisation Hourly Regionalised [ .txt ] Hourly output generation Hourly [ raster ]

16. Weather data • Weatherparameters outputs • Raster data • 90m spatial resolution • Excel table • Describemeanweather values of the overlappingselectedforest stands and satellite data Precipitations [mm] Temperature [°C] High: 10 0 High: 36 Example Daily: 31mar08 - 03apr08 Daily: 4 days Hourly: 4 hours Table weather data (simplified version) 0

17. Analysis of the data

18. Fields of investigations • Backscatter intensity • Multi-temporal • Polarisation • Frequency • Pass • Interferometriccoherence • Visual interpretations • Stem volume

19. Good species separability over the acquisition time. • Precipitations increase the backscatter. Backscatterintensity • TerraSAR-X multi-temporal datasets Precip [mm] Temp [°] Wind [m/s] Gamma nought [dB] Daily Acquisition dates Series of TSX HS, 34.4, HH, Asc.

20. HV do not allow tree species discrimination. Backscatterintensity • TerraSAR-X polarisation Gamma nought [dB] – TSX SM HV. Gamma nought [dB] – TSX SM VV Precip 2.3 [mm] Precip 1.3 [mm] Gamma nought [dB] – TSX SM HH Gamma nought [dB] – TSX SM HH - 05may10 - - 22mar09 -

21. A ratio X-band / L-band improves the separation between Norway Spruce and European Beech. Backscatterintensity • TerraSAR-X Frequency TSX (X-band) Precipdaily: 0.8 [mm] Preciphourly: 0.0 [mm] PALSAR (L-band) Precipdaily: 1.1 [mm] Preciphourly: 0.0 [mm] - 27jun10 - Gamma nought [dB] – FBD 34° HH Asc. Gamma nought [dB] – TSX SM 33° HH Asc. - 16jun10 -

22. Dew effect is not clearly observable. Intensities differ due to the varying penetration in Ascending and Descending. Backscatterintensity • TerraSAR-X pass direction Descending: 6 a.m (morning) Precipdaily: 1.1 [mm] Preciphourly: 0.0 [mm] Ascending: 6 p.m (afternoon) Precipdaily: 1.7 [mm] Preciphourly: 0.0 [mm] Gamma nought [dB] – TSX HS HH 46° Desc. - 26aug09 - Gamma nought [dB] – TSX HS HH 48° Asc. (mon - 28aug09 -

23. Interferometriccoherence • Coherence • The degree of coherence can be related to several factors, each expressing a specific source of decorrelation. • The volume decorrelation is related to objects presenting a vertical extension. This factor is spatialbaseline dependent. The temporal decorrelation is related to thestability of the objects between the two acquisitions.

24. Interferometriccoherence • Visual interpretations RapidEye RGB – 13jun09 CSK Himage 28sep10 - 29sep10 - 1 day temporal baseline - TSX HS 28may10 - 08jun10 - 11 days temporal baseline -

25. As expected, no correlation between stem volume and TSX coherence. Interferometriccoherence • TerraSAR-X coherence vs Stem volume r2S=0.01 r2B=0.01 Coherence – TSX HS HH 37° Asc. - 06oct09 – 17oct09 - Stem volume [m3/ha]

26. Increase of correlation with higher perpendicular baseline. Interferometriccoherence • Cosmo-SkyMed coherence versus Stem volume Perpendicular baseline r2S=0.01 r2S=0.00 r2S=0.11 r2B=0.01 r2B=0.01 r2B=0.01 r2P=0.01 r2P=0.01 r2P=0.01 Interferometriccoherence 1 day temporal baseline r2S=0.05 r2S=0.03 r2S= 0.15 r2B=0.00 r2B=0.01 r2B=0.02 r2P=0.01 r2P=0.00 r2P=0.06 Stem volume [m3/ha]

27. Interferometriccoherence • Cosmo-SkyMed coherence - Weather information Not yet available Weather data for Cosmo-SkyMed acquisitions

28. Conclusions and outlook

29. Conclusions and outlook • Different investigations of the TSX and CSK backscattter intensity and interferometric coherence have been conducted. • Conclusions and open issues • The topography influence is not fully understood. • The signal for Norway Spruce and European beech can be separated with TSX HH or VV. • Weather conditions and particularly precipitations affect the backscatter intensity and interferometric coherence. • Interferometric coherence in X-band with 1 day temporal baseline may show a potential for estimating biomass. • Future work • Further investigations for the topographic normalisation • Analysis of ALOS PALSAR intensity and interferometric coherence • Derivation of indices and ratios between X-band and L-band

30. Acknowledgements I wouldliketothank DLR, ESA, ASI and RESA forthedistributionofthedata ENVILAND2 is sponsored by the Space Agency of the German Aerospace Center with federal funds of the German Federal Ministry of Economics and Technology on the basis of legislation by the German Parliament grant no. 50 EE 0844 - 50 EE 0847

31. Questions ? Vielen Dank für Ihre Aufmerksamkeit! Thuringia Forest – July 2010