Support GeoTIFF?

[mdbartos]

Support GeoTIFF?

[mdbartos]

Rasterio supports geotiff, which means that they should be supported: https://rasterio.readthedocs.io/en/latest/

Did you try using grid.read_raster or grid.from_raster on your geotiff? It seems to work for me.

[mdbartos]

@lz80521: Were you able to resolve this problem?

[ShashankBice]

Hi @mdbartos ! What if I do not have grid defined from beforehand and want to read the tiff file and define the grid file based on the properties of the tiff file?
Is there a way to define the grid geospatial properties manually?
Please let me know..
I tried grid = Grid(['affine=dataset.transform','shape=(219,564)','nodata=-1.79769313486231571e+308','crs=dataset.crs'])
but it fails!
Please let me know.....

[mdbartos]

Greetings,

If you call:

grid = Grid.from_raster('/path/to/file.tiff', 'dataset_name')

It should populate the affine transform, shape, and other metadata of grid automatically.

Alternatively, read_raster will populate the metadata if it is the first dataset in the grid:

grid = Grid()
grid.read_raster('/path/to/file.tiff', 'dataset_name')

These are probably the most preferred ways of doing it.

If you need to set attributes manually, you can also do so after instantiating a grid:

grid.affine = dataset.transform
grid.shape = dataset.shape
grid.crs = dataset.crs
grid.nodata = -1.79769313486231571e+308

There are some other ways that may be easier or harder depending on your situation, but see if any of those help.

[ShashankBice]

Hi @mdbartos !
Thanks for the quick reply. After going through other issues with pysheds, I found the problem to be in version of rasterio. I use anaconda's rasterio and the latest release on anaconda does not play well with gdal. I did a pip install of rasterio and it works. And thankfully, it does not messes my other gdal related packages :)
Thanks for the help, I think I can now get my assignment done tomorrow morning!

[mdbartos]

lol, no problem

[ShashankBice]

1 issue though, grid.view('dem') instead of plotting a raster gives this:-

Raster([[-1.79769313e+308, -1.79769313e+308, -1.79769313e+308, ...,
-1.79769313e+308, -1.79769313e+308, -1.79769313e+308],
[-1.79769313e+308, -1.79769313e+308, -1.79769313e+308, ...,
-1.79769313e+308, -1.79769313e+308, -1.79769313e+308],
[-1.79769313e+308, -1.79769313e+308, -1.79769313e+308, ...,
-1.79769313e+308, -1.79769313e+308, -1.79769313e+308],
...,
[-1.79769313e+308, -1.79769313e+308, -1.79769313e+308, ...,
-1.79769313e+308, -1.79769313e+308, -1.79769313e+308],
[-1.79769313e+308, -1.79769313e+308, -1.79769313e+308, ...,
-1.79769313e+308, -1.79769313e+308, -1.79769313e+308],
[-1.79769313e+308, -1.79769313e+308, -1.79769313e+308, ...,
-1.79769313e+308, -1.79769313e+308, -1.79769313e+308]])
Any workaround for this?

[mdbartos]

To plot, it just call:

from matplotlib import pyplot as plt

plt.imshow(grid.view('dem'))

This was omitted in the readme for brevity, but can be seen in the quickstart:
https://github.com/mdbartos/pysheds/blob/master/examples/quickstart.ipynb

grid.view returns a Raster object (essentially an ndarray with metadata) that represents a "view" of the data at the grid's current affine transform, shape, crs, and nodata value.

[bmalbusca]

issue closed? @mdbartos