AUTOMATIC REGISTRATION FOR WIDE FIELD IMAGER (WFI) L4 IMAGERY TIME SERIES

The WFI sensor systems on board the CBERS satellite family and Amazonia-1 conduct imaging revisits of the Earth's surface at intervals of approximately five days and thus can support many remote sensing applications. A fundamental data product for such applications is the L4 image processing level corresponding to the image orthorectification. However, WFI time series may present spatial displacements and positional distortions between images from the same location. In the present work, CBERS-4, CBERS-4A and Amazonia-1 imagery time series of Aracaju, in Sergipe State, were analyzed, highlighting the displacement between images due to variations in the registration of each satellite, their respective coordinate projection and spatial resolution. The methodology used in this work consisted of developing an automated code in Python that automatically corrects the aforementioned problems in WFI L4 time series, through the co-registration technique using the library AROSICS. The results showed that images from the same sensor as the reference, presented minor displacements (average of 0,59px) while the ones that required resampling had major displacements (average of 1,17px).