Exploiting the synergy of TanDEM-X & TerraSAR-X missions to map and monitor tropical forests
Weihing, D.; von Poncet, F.; Schlund, M.
Astrium GEO-Information Services, GERMANY

1 Introduction

Mapping and monitoring of tropical forests is important in the context of climate change. The bistatic TanDEM-X mission (Krieger et al. 2007) is a synthetic aperture radar (SAR) mission, whose aim is to create a highly accurate global digital elevation model. The two sensors, TerraSAR-X and TanDEM-X, acquire data of the entire land mass delivering a global database which can serve as a baseline for monitoring concepts of any spot of the world.
First mapping examples (Schlund et al. 2012) show that the TanDEM-X data is very useful to create accurate land cover mapping compliant with the 6 land cover categories mandatory for REDD MRV as specified by IPCC (2003). In this study, single-pass interferometric coherence proved to be the most important information source for forest-non forest and even more detailed biomass related forest mapping in X-band.
The global coverage of TanDEM-X is not only a basis to consistently map the status of forests but can also provide the reference data set for image to image change detection with up to date TerraSAR-X imagery. Even though, TanDEM-X basically provides similar technical properties as TerraSAR-X it is operated in a different mode for the global DEM mission. Therefore, the performance of amplitude change detection between TanDEM-X and TerraSAR-X has been tested to investigate if also subtle changes caused by changing canopy structure indicating forest degradation can be detected.

2 Input data

TanDEM-X data acquired on July 18th 2011 provided the reference scene. Under the same geometry TerraSAR-X StripMap data has been acquired on July 15th 2012. Both images exhibit the following acquisition parameters: HH polarisation, descending orbit, incidence angle: 32.2-35.5 °. The location of the study area in Kalimantan, Indonesia is illustrated in Figure 1.


Fig. 1: Location of the study area (left) & TanDEM-X acquisition (right)

3 Methods

For the land cover discrimination different textures and statistics are estimated amongst the active and passive amplitude and the bistatic interferometric coherence. This information together is used in an object based classification approach to restrict the search for areas and changes of interest and for the evaluation of changes.
For the image to image change detection the active amplitude from the TanDEM-X dataset served as a reference scene. The TerraSAR-X acquisition was analysed in comparison to it using the intensity ratio as a change indicator. The intensity ratio (Touzi et al. 1988) contains not only the information that a change happened but also if the backscatter increased or decreased between the acquisition dates, which is of interest especially when considering forest degradation.

4 Results & Discussion

Forest classification with X-band is a demanding task since the dynamic range of the amplitude between open land and vegetation is small compared to longer wavelength SAR data. The interferometric coherence which can be derived from a TanDEM-X acquisition improves the separability of those classes significantly. Therefore, it could be argued that the accuracy of a classification with TanDEM-X is higher than a classification based only on backscatter intensity. However, hilly terrain causes layover and shadow as well as decorrelation in those areas. In addition, the dynamic range of the interferometric coherence is dependent on the effective baseline. The terrain and the dependency on baseline certainly complicate the classification and make the algorithm more vulnerable for misclassifications. A subset of the classification result is shown in Figure 2.


Fig. 2: Mapping and monitoring results based on TanDEM-X and TerraSAR-X

The detected changes in the investigation area can be mainly lead back to large scale deforestation. However, it is possible to detect even small changes within the forest canopy using X-band. Selective logging resulting in new small shadows within the canopy is indicated by decreasing backscatter (Fig. 2). Even though TanDEM-X and TerraSAR-X are operated in different modes, this study demonstrates that the combination allows for change detection of even small scale loggings.
The classification of the reference date is very useful to restrict the search to forest areas which allows to interprete the detected changes as deforestation or forest degradation. Detections due to changing environmental conditions or changes which are not of interest for the application are minimized this way.

5 Conclusion

The study proved that even small scale selective logging activities can be detected when comparing TanDEM-X global reference datasets with up-to date TerraSAR-X StripMap imagery acquired under the same geometry.
The use of information derived from single pass interferometry significantly improves the mapping capabilities and thus enables to focus on forest changes. The combination of both, TanDEM-X based object oriented classification exploiting amplitude, phase and texture characteristics and TanDEM-TerraSAR change detection, makes use of the unique characteristics of both missions providing a powerful tool for the monitoring of tropical forests.

References

IPCC (2003): Good practice guidance for land use, land-use change and forestry. Hayama: Institute for Global Environmental Strategies (IGES).
Krieger, G., Moreira, A., Fiedler, H., Hajnsek, I., Werner, M., Younis, M., & Zink, M. (2007): TanDEM-X: A Satellite Formation for High-Resolution SAR Interferometry. IEEE Transactions on Geoscience and Remote Sensing, 45, 3317-3341.
Schlund, M., F. von Poncet, S. Kuntz & C. Schmullius (2012): Importance of bistatic SAR features from TanDEM-X for forest mapping and monitoring – ForestSAT 2012, 11-14th September 2012, Corvallis, Oregon.
Touzi, R. Lopes, A. & Bousquet, P. (1988): A statistical and geometrical edge detector for SAR images. IEEE Transactions on Geoscience and Remote Sensing, Vol 26 (6), pp. 764-773.