Implement Modified Nest Set Index for Rapid Watershed Delineation

[ptomasula]

Summary

The paper Development of a data model to facilitate rapid watershed delineation proposes a graph algorithm based approach for watershed delineation using a modified nested set index. The outlined approach would allow selection of upstream reaches from a given node (reach) based on two numerical indices, in place of more conventional techniques like adjacency lists or binary search.

Closure Criteria

• [x] Implement a modified nested set indexing algorithm to generate index for TDX hydro stream net geopackage files
• [x] Provide documentation on how to utilize output from nest set index for delineation
• [x] Run instructions for generating the modified nested set index from source files.

[ptomasula]

Thanks @aufdenkampe for connecting and helping debug this issue. As noted in the commit message for b21c352, the previous implementation's conversion from a GeoDataFrame into a dictionary and back again was resulting in the loss of critical type information (e.g. CRS). As a result, the write method was not able to function correctly. The example notebook should be working now.

[aufdenkampe]

The final revisions to examples/3_GenerateModifiedNestedSetIndex.ipynb with commit cb245a08bef4ac5f07137d5157619dfe7eb06185 closes this issue by providing the create_tdx_mnsi() function that provides a performance and storage optimized approach that:

• Opens an original streamnet GeoPackage file
• Creates a Modified Nested Set Index fields,
• Converts all LINKNO-related fields to globally unique values
• Drops useless fields that are not functionally populated with values
• Sets LINKNO as the index for to facilitate selection and interoperability with the basins datasets
• Saves a compressed streamnet GeoParquet file that is 26% of the size (~4x smaller) than the original GeoPackage file