Dear Said, I understand that you want to study the risk of sanding using the remote sensing as a tool. Generally, this could be simplified in conducting the dune mobility and geodynamics using bi-temporal or multi-temporal geo-spatial data. So it is important to extract the sand-cover classe which refers to each date and, therefore doing your comparison. In your case, the 1977’s image should correspond to an MSS Landsat image, while the 2016's one is an OLI Landsat 8 or a TM which is still provided by NASA. Several solutions are available : Classification, target detection, the infra red band (band 4) ... As well, you can determine the dune form, trend, migration rate and direction from the satellite image of medium resolution at least (30 m) or, fine resolution for a better accuracy. If you work with TM or ETM+ Landsat data, you can also try to use some useful index such as: the normalized difference sand dune in order to extract the sand-cover classe.