One approach to adjust earth observation data

There are more than hundreds of different earth observation sensors out there. Added to the different resolutions many sensors are sensitiv to different wave lenghts and therefore producing different observations from the same place. Added to this problem is, that the enviroment changes a lot. The sun shines from different directions, the amount of clouds differ and the vegetation changes a lot.

A supvervised deep learning approach can only produce good results, when the amount of clutter (the bias inside the dataset) is not too high. Therefore to produce good results the data should be as much adjusted as possible.

##