A re-implementation of the Kienholz et al., 2014 algorithm in python.
This algorithm has been implemented succesfully in OGGM since several years already. This code, however, has a few drawbacks:
- it requires OGGM to be installed to run.
- the OGGM developpers have made certain choices with respect to algorithm design and grid handling which made sense for the model, a bit less for the centerline tool
- the agorithm code is a bit hidden in the large OGGM codebase, maybe impeding innovation and adaptations from the community.
The main goal of this project is to adress these shortcomings and develop a single tool with a simple purpose: given glacier outlines and a digital elevation domain, compute the centerlines of glaciers.
Here are two files to test things on:
- https://cluster.klima.uni-bremen.de/~oggm/tutorials/Norway_Inventory_sel.zip
- https://cluster.klima.uni-bremen.de/~oggm/tutorials/Norway_DEM_sel.tif
Workflow:
- assume that the DEM and the outlines are in a cartesian projection (units: m)
- open the geotiff with rioxarray
- start with one outline, and crop the DEM to the outline + a few grid point
- compute a mask of the glacier.
- use the OGGM code to compute the heads, terminus, without simplifying geometries as done in OGGM.
The tools you will need:
- rioxarray to read geotiff
- geopandas to read and write geometries
- shapely to do the geometrical stuff (as OGGM does)
- scipy for the routing algorithm (as OGGM does)
See documentation in \docs
Now the tool is pip installable. The tool has become an entity task that can be called using oggm already existing functions:
(e.g. from snippet_run_rgi_centerlines.py)
#import general execution for oggm taks
from oggm.workflow import execute_entity_task
#new package where to take the task from:
import glacier_centerlines as gc
# run
execute_entity_task(gc.centerlines_rgi.compute_centerlines_rgi, gdirs)
The new centerlines will be stored at the original oggm glacier directory as centerlines_rgi.tar.gz