Exploitation of CBERS-02B as auxiliary data in recovering the Landsat7 ETM+ SLC-off image

The scan-line corrector (SLC) for the Enhanced Thematic Mapper Plus (ETM+) sensor, on board the Landsat7 satellite, have failed probably permanently since May 31, 2003. The consequence of the SLC failure (denoted SLC-off) is that approximately 20% of the pixels in an ETM+ image are not scanned, which hampers the use of the data accordingly. To improve the usability of the ETM+ SLC-off data, several researches relating to the gap-fill algorithms have been conducted with acceptable accuracy. Due to the limitation of data acquisition, e.g. temporal resolution and atmospheric condition, there are always a large number of overlapping areas filled with un-scanned pixels in two cloud-free ETM+ SLC-off images which are close in time. Consequently, the recovering procedure would be unavailable through general gap-fill algorithms. The bands similarity between China Brazil Earth Resources Satellite-02B (CBERS-02B) and Landsat7 ETM+, particularly for the visible/near-infrared bands, makes it possible to estimate the un-scanned pixels in the ETM+ SLC-off image considering the close time CBERS-02B as auxiliary data. In this paper, CBERS-02B(acquired on January 17, 2009) was taken as auxiliary data to fill the un-scanned pixels in two ETM+ SLC-off images(acquired on January 21, 2009 and November 5, 2009 respectively) with the suitable gap-fill algorithms. Four gap-fill algorithms were practiced and compared, which called simple filling (Simple), global linear histogram match (GLHM), localized linear histogram match (LLHM), and adaptive window linear histogram match (AWLHM) respectively. In contrast to Simple, GLHM and LLHM, AWLHM took into account the number of effective pixels as well as the impact of local conditions. Therefore, while ignoring computing speed or time consumption, AWLHM was a generally superior method with higher accuracy to others. Lastly, taking Xiamen Island as a study region, we used the recovered ETM+ data filled by AWLHM to extract urban impervious surface - - (UIS) at sub-pixel scale, adopting the selective endmember linear spectral mixture model (LSMM). The accuracy of UIS estimation was validated using a sharpened IKONOS image with spatial resolution of 1m(acquired on January 18, 2009). Results indicated that there was no significant difference between the scanned and filled (un-scanned in original ETM+ SLC-off image) pixels, in view of the estimation accuracy of UIS. In conclusion, CBERS-02B should be regarded as usefully auxiliary data so as to recover the ETM+ SLC-off image and enable more scientific use of the data.