Inconsistent Channel Heads

[mdp0023]

Something is happening during "calculate_geodesic_distance" with certain basins to cause the geodesic distance to be 0, resulting in erroneous channel heads. Only occurring for specific basins in specific catchments, not sure what the pattern is, but narrowed down to something within fast_marching algorithm in tools.py. See image for example of 0 geodesic distance with too many channel heads. (catchment 5781905 in Austin AOI for reference)