Below are the two statements from two different papers one is cited in other and they applied same process to combine multiple band images and created a 7 channel dataset I want to do the same. (" Sentinel-1 and Sentinel-2 images were finally geocoded on a common

coordinates grid and then apply amplitude normalization. In the end,

concatenate together the processed data, resulting in a five-channel

multi-sensor dataset, including B4 (red), B3 (green), B2 (blue), B8

(NIR), and VV. " )("The

SAR; slope data; and the red, green, blue, NIR, and NDWI

images from Sentinel-2 were stacked to create a seven-channel

dataset.) They stacked images then normalize the stacked-7-channel-image because because SAR band has values of -13db to 20db greater value range and others have 0 to 1 small value ranges. So for faster convergence and better performence they have normalized the stacked-7-channel-image.