UFRM: error-checking when LULC and biophysical table mismatch

[davemfish]

First, a user reported a cryptic IndexError when the biophysical table was missing values in the LULC raster. https://community.naturalcapitalproject.org/t/urban-flood-risk-mitigation-indexerror/2215/3

Traceback (most recent call last):
File “site-packages\natcap\invest\ui\execution.py”, line 68, in run
File “site-packages\natcap\invest\ui\model.py”, line 1649, in _logged_target
File “site-packages\natcap\invest\urban_flood_risk_mitigation.py”, line 241, in execute
File “site-packages\taskgraph\Task.py”, line 747, in add_task
File “site-packages\taskgraph\Task.py”, line 1234, in _call
File “site-packages\pygeoprocessing\geoprocessing.py”, line 438, in raster_calculator
File “site-packages\natcap\invest\urban_flood_risk_mitigation.py”, line 775, in _lu_to_cn_op
File “site-packages\scipy\sparse_index.py”, line 35, in getitem
File “site-packages\scipy\sparse_index.py”, line 139, in _validate_indices
File “site-packages\scipy\sparse_index.py”, line 171, in _asindices
IndexError: index (127) out of range

Second, when I tried to reproduce this error by modifying the sample data, I accidentally found a related bug. So long as the lucode column in the biophysical table contains a row with an lucode >= to largest int value in the LULC raster, we don't even see this IndexError. The model will complete successfully even when many/all of the LULC values are missing from the biophysical table. Presumably the results are invalid in this case.

To reproduce, edit the sample data table to be like this:

lucode,Description,NEH type (to define CN),CN_A,CN_B,CN_C,CN_D,SW_type,EMC
24,Tundra,,1,1,1,1,3000,0

This all seems a byproduct of using a sparse matrix with indices defined based on the lucode values found in the table.

[davemfish]

We should also be sure the solution includes better error messaging that can help users debug their table-LULC mismatches.