RasterDataset with RBG Tif

[graceebc9]

Hello, I'm creating a new RasterDataset class for the NOAA high res RGB satellite imagery (https://storms.ngs.noaa.gov/)

but struggling to get the sampler to work, think it may be an issue with the TIF being multi-channel/ not sure how to extract the RGB bands from a single TIF thats not labelled.

The data comes in TIF tiles of 2.5km, 3 x 9415 x 9415, and looks like this in XARRAY

This is my current class , for tiles with names like '20210830aC0932315w294800n.tif'

class NOAA(RasterDataset):
    filename_glob = "*n.tif"
    filename_regex = "^(?P<date>\d{8})a.*$" 
    date_format = "%Y%m%d"
    
    seperate_files= False
    
    def __init__(
            self,
            root: str = "data",
            crs: Optional[CRS] = None,
            res: Optional[float] = None,
            # run: Sequence[str] = [],
            # bands: Optional[str] = [],
            transforms: Optional[Callable[[Dict[str, Any]], Dict[str, Any]]] = None,
            cache: bool = True,
        ) -> None:
            """Initialize a new Dataset instance.

            """
            # self.bands = bands if bands else self.all_bands

            super().__init__(root, crs, res, transforms, cache,)

Thanks

[adamjstewart]

Could this be the same issue as #466 (comment)?

[weiji14]

Could you point to a download link for the GeoTiff file (somehow can't get the download working on my browser)? Or maybe post the output of gdalinfo 20210830aC0932315w294800n.tif

My guess is that you'll need to read the RGB tif using rasterio's color interpretation, see https://rasterio.readthedocs.io/en/latest/topics/color.html. Note that xarray can't read the colorinterp properly, though I've been trying to get it to work via rioxarray (see corteva/rioxarray#414).

[adamjstewart]

@graceebc9 does #466 (comment) solve your issue?

[adamjstewart]

Assuming that this issue has been resolved, but if not feel free to reopen this issue.