I think the important distinction here is between searching by geometry and subsetting the actual product.
When you pass a geometry to selected_collection.search(), the geometry is mainly used as a spatial filter: it returns products whose coverage intersects the area you requested. It does not necessarily mean that the downloaded native file will be cropped to that city or bounding box.
So receiving the complete native product can actually be expected behavior.
A useful workflow would be:
-
Define a small bounding box or polygon around the city.
-
Use that geometry in selected_collection.search() to find only products covering the area.
-
Download the relevant product.
-
If you need only the pixels/data inside that city, perform a geographical subset afterwards, for example with EUMETSAT's Data Tailor or an appropriate geospatial library depending on the product format.
It is a little like selecting one level in Geometry Dash from a larger collection: the selection tells you which content you are interested in, but it does not automatically rebuild the original package containing only that selected part.
So I would first check whether your goal is finding products covering the city or actually creating a cropped dataset for the city. Those are two different operations, and search(geometry=...) mainly addresses the first one.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.