Reference shift between Landsat and Sentinel images

[cappendini]

Hi,
I am obtaining the shoreline position for the last decade in Northern Yucatan. There seems to be a clear shift in the shoreline positions between Landsat and Sentinel images, but do not allow to use them together to obtain long term shoreline trends. Is there a reference issue between the datasets, so that this could be fixed?
Thanks for making the scripts public!

[kvos]

wooow that is a big shift! Did you have a look at the mapped shorelines to see if there was any issue? probably the Sentinel-2 classifier is not getting the sand pixels and has a bias. You should check the images in the /jpg_files folder. You can retrain the classifier if either the Landsat or Sentinel-2 classifier is not working well at your site following the instructions here https://github.com/kvos/CoastSat/blob/master/doc/train_new_classifier.md and jupyter notebook here https://github.com/kvos/CoastSat/blob/master/classification/train_new_classifier.ipynb.

[kvos]

just referring you to issue #148 as @Noza1212 had a good go at co-registering the images from the different missions

[cappendini]

Thanks @kvos I do think the issue is more related with the co-registering as I can visually see the shift between some images. Thanks for pointing into issue #148