GDAL Error in REMMaker.get_river_elev()

[phaehnel]

Version: 1.1.2

When using the REMMaker.make_rem() function, I get the following error message:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[3], line 2
      1 rem_maker = REMMaker(dem = path_dgm1, out_dir = path_res, centerline_shp = path_streamdiss, workers=12)
----> 2 rem_maker.make_rem()
      3 rem_maker.make_rem_viz(cmap = 'winter')

File ~/miniforge3/envs/mfenv/lib/python3.13/site-packages/riverrem/REMMaker.py:432, in REMMaker.make_rem(self)
    430 else:
    431     self.read_centerline_input()
--> 432 self.get_river_elev()
    433 self.interp_river_elev()
    434 self.detrend_dem()

File ~/miniforge3/envs/mfenv/lib/python3.13/site-packages/riverrem/REMMaker.py:334, in REMMaker.get_river_elev(self)
    332 res = f"-tr {self.cell_w} {self.cell_h}"
    333 # res = f"-tr {self.cell_w} {self.cell_h *-1}"
--> 334 gdal.Rasterize(self.centerline_ras, self.river_shp, options=f"-a id {extent} {res}")
    335 # raster to numpy array same shape as DEM
    336 r = gdal.Open(self.centerline_ras, gdal.GA_ReadOnly)

File ~/miniforge3/envs/mfenv/lib/python3.13/site-packages/osgeo/gdal.py:2197, in Rasterize(destNameOrDestDS, srcDS, **kwargs)
   2194     srcDS = OpenEx(srcDS, gdalconst.OF_VECTOR)
   2196 if isinstance(destNameOrDestDS, (str, os.PathLike)):
-> 2197     return wrapper_GDALRasterizeDestName(destNameOrDestDS, srcDS, opts, callback, callback_data)
   2198 else:
   2199     return wrapper_GDALRasterizeDestDS(destNameOrDestDS, srcDS, opts, callback, callback_data)

File ~/miniforge3/envs/mfenv/lib/python3.13/site-packages/osgeo/gdal.py:9471, in wrapper_GDALRasterizeDestName(*args)
   9469 def wrapper_GDALRasterizeDestName(*args):
   9470     r"""wrapper_GDALRasterizeDestName(char const * dest, Dataset srcDS, GDALRasterizeOptions options, GDALProgressFunc callback=0, void * callback_data=None) -> Dataset"""
-> 9471     return _gdal.wrapper_GDALRasterizeDestName(*args)

TypeError: in method 'wrapper_GDALRasterizeDestName', argument 3 of type 'GDALRasterizeOptions *'

This seems to be related to the DEM metadata that is used to rasterize the centerline shape with gdal in get_river_elev() (line 334). According to the GDAL documentation the resolution provided with option -tr has to be all positive values. In my case my DEM provides a negative value for the pixel size for the cell height (y coordinate) because the y coordinate values decrease from top to bottom in the coordinate system the DEM uses.

When I changed self.cell_h to be positive everything worked fine.

[klarrieu]

Thank you for sharing this! Did taking the absolute value of cell height make your outputs still properly georeferenced? I will test this myself too to make sure. Feel free to open a PR, otherwise I'll make the adjustment myself and put out a patch release.

[phaehnel]

Thanks for the reply! Using the absolute value resulted in correctly georeferenced output because the output REM again had a negative pixel size for the y coordinate as it should have. Thus, it could be that using the absolute value for self.cell_h could fix the issue in general.

I tried but could not open a pull request myself.

[klarrieu]

Can you please share the output of conda list in the environment you are using for this? I just tested a freshly created environment with riverrem 1.1.2 and gdal 3.8.5, using an input raster with (cell_w, cell_h) = (3, -3) and could not reproduce this error.

If you could also share your input files, that could be helpful for me to try reproducing this issue as well. If you can't attach them here you can email them to me: kglarrieu@ucdavis.edu

[collinjroland]

I experienced this same error in an environment running riverrem 1.0.4 and gdal 3.10.3. I ended up hardcoding line 332 of REMMaker.py as res = f"-tr {self.cell_w} {np.abs( self.cell_h )}", but not sure if this is the best long-term bug fix. I imagine if I switched up the package versions as suggested I could run without errors.

[phaehnel]

Sorry for the late reply. This got lost between other tasks. @klarrieu I'll send you the DEM file that caused the issue and the conda list info via mail this week. For me, the error occured with gdal 3.10.2 with Python 3.13.3.

[klarrieu]

Hello, sorry for the late reply. Thanks for sharing your files.

In the pinned dependency versions for this package (e.g. in environment.yml and meta.yaml), I have the gdal version pinned to gdal>=3.7, <3.9. I used this range because there are other breaking changes in the dependencies that would require more involved refactoring to get working, namely with gdal >=3.10, numpy >= 2.0, and the latest osmnx, which have changed function names and syntax that this package uses.

Creating a fresh environment following the installation instructions should work fine as-is, automatically installing dependencies in the pinned ranges.

To fix this issue with the latest gdal, I'd like to refactor and test the code base with the latest versions of these other dependencies as well, which I haven't had the time to work on lately. In the meantime, I'd be happy to review any PRs to make it compatible with the latest versions of gdal, numpy, and osmnx.

[phaehnel]

Hello @klarrieu, thanks for the reply. Okay, then a separate environment seems like the best solution for now.