Hi @JuliaNiemeyer,
I have asked the science team about this and haven’t gotten a lot of feedback yet (please see the latest post here). It sounds like some people have come up with values that work, but they may not be citable, peer-reviewed sources. For the mining areas, I would guess that’s mostly bare ground. Googling for bare soil Kc values did turn up some sources, but I can’t speak to their quality.
Since I have a large area that comprises very different vegetation types, this is not feasible.
Do you mean that each pixel covers a large area? Or that within each urban land class, there are too many different vegetation types? If there is that much variation within the urban land class, I would see if you can split it into multiple classes according to the amount/type of vegetation present, if any.
I’m sorry we don’t have a straightforward answer for you, and thanks for working to improve the quality of data available for Brazil!