Use Tandem-X based Forest Non-Forest map as global forest mask

[felixcremer]

We could use the tandem-X based forest non forest map to mask the global forest.
The data is from 2011 to 2015 and has a spatial resolution of 50 meter
https://www.dlr.de/en/hr/research-transfer/projects/forest-non-forest-map

[danlooo]

I like the radar part of the dataset. Is this dataset a time series or is it just one map?

[felixcremer]

I think just a map and abit earlier than our data, but nonetheless a good starting point.

[felixcremer]

To use this as a mask for our data we would have to bring it to the projection of the Sentinel-1 data and resample it to the same grid.
I am not entirely sure, whether we should try to do this resampling for the whole dataset before doing our analysis or whether we should try to resample it on the fly. But this would mean, that we would have to do the resampling multiple times.

I am going to explore resampling the data to the Sentinel-1 grid in equi 7 and saving the data in the same tile structure.
To do this, I would have to fetch the right tiles of the tandem-x mask for every tile in equi7 T3.

[danlooo]

Thanks for your effort! I think its a matter of how much storage the forest map will consume in equi7. it's only binary and will be used in every call. Chances are high to just cache the results.