Analysis of ALOS PALSAR and TerraSAR-X data for protected area mapping: A case of the Bwindi Impenetrable National Park-Uganda

The main purpose of this study was to investigate the potential of Quad-pol L-band ALOS PALSAR and Dual-pol X-band TerraSAR (TSX) data, as well as derived TSX image texture for land cover mapping. A per- pixel classification was performed using non-parametric decision tree method. Classifications involving the HH, VV and HH/VV TSX polarimetric band(s) resulted in kappa indices of 0.4326, 0.3577 and 0.4657 respectively. In contrast, classifications involving the HH, HV, VV, HH/HV, VV/HV, HH/VV, HH/HV/VV and HH/HV/VH/VV bands of ALOS PALSAR data resulted in corresponding kappa indices of 0.2972, 0.3395, 0.3269, 0.7141, 0.7058, 0.4697, 0.7311 and 0.7177. A further analysis was carried out using the image textures derived from the HH polarisation of TSX data. Three different categories of textures were analysed: SAR specific (SARTEX), textures based on grey level concurrence matrices (GLCM) and textures based on SAR image histogram (HISTEX). These resulted in kappa indices of 0.6740, 0.6655 and 0.7166 respectively. Moreover, a classification using two original TSX polarisations provided a kappa index of 0.4657. This showed an improvement in the classification accuracies by 45%, 43% and 52% respectively. On the basis of the resulting accuracies, it can be concluded that analysis of data with high polarisation increases the classification accuracy of land cover information derived from SAR data. Furthermore, inclusion of derived SAR textures in the classification process, provide a potential for improved land cover identification and mapping in the tropics.

[1]  T. M. Lillesand,et al.  Remote Sensing and Image Interpretation , 1980 .

[2]  Thomas Blaschke,et al.  Land cover change assessment using decision trees, support vector machines and maximum likelihood classification algorithms , 2010, Int. J. Appl. Earth Obs. Geoinformation.

[3]  S. Saatchi,et al.  The Global Rain Forest Mapping project - A review , 2000 .

[4]  Achim Roth.  Scientific use of TerraSAR-X , 2004, IGARSS 2004. 2004 IEEE International Geoscience and Remote Sensing Symposium.

[5]  Manfred Zink,et al.  ALOS PALSAR products verification , 2007, 2007 IEEE International Geoscience and Remote Sensing Symposium.

[6]  K. Stankiewicz,et al.  The use of microwave SAR images for forest decline monitoring in mountainous area , 2002 .

[7]  Antoine Guisan,et al.  Predictive habitat distribution models in ecology , 2000 .

[8]  T. Nonaka,et al.  EVALUATION OF THE GEOMETRIC ACCURACY OF TERRASAR-X , 2008 .

[9]  Clustering versus regression trees for determining ecological land units in the Southern California mountains and foothills. , 2003 .

[10]  M. Cablk,et al.  Change in the forested and developed landscape of the Lake Tahoe basin, California and Nevada, USA, 1940-2002. , 2008 .

[11]  Qihao Weng,et al.  A survey of image classification methods and techniques for improving classification performance , 2007 .

[12]  C. Chapman,et al.  The status of anthropogenic threat at the people-park interface of Bwindi Impenetrable National Park, Uganda , 2009, Environmental Conservation.

[13]  Sang-Wan Kim,et al.  Evaluation of TerraSAR-X Observations for Wetland InSAR Application , 2010, IEEE Trans. Geosci. Remote. Sens..