[EPIC] Flood Watch Workflow

[CarsonPruitt-NOAA]

This Epic's purpose is to track the development of an experimental FIM-based flood watch functionality.

[ZahraGhahremani]

The presence of inf values is due to the way we aggregate branches. We compute the median value across all branches. When Branch 0 has a valid value, but the two GMS branches have values of 0, the median becomes 0, resulting in a surface area of 0. We can use the mean instead of the median to avoid this issue. The HUC12s shown in yellow in the left figure represent areas with inf values.

[ZahraGhahremani]

The flood watch land-use layer for the April 2nd event is available here: fim-home/zahra.ghahremani/Flood_Watch/20250402/flood_watch_pct_land_use_0402.gpkg.
Each HUC includes the percentage of each land use class. I grouped the original NLCD classes into broader categories. The final classes are:

NLCD Class Groupings

Group	NLCD Classes
Urban	Open Space, Low Intensity, Medium Intensity, High Intensity
Agriculture	Pasture, Cultivated Crops
Forest	Deciduous Forest, Evergreen Forest, Mixed Forest
Wetlands	Woody Wetlands, Emergent Wetlands
Water	Open Water
Barren	Barren Land
Shrubland	Shrub/Scrub
Herbaceous	Grassland

There's a new attribute called score for each HUC, which is a weighted average of the NLCD classes (the weights reflect levels of human impact). Please let me know if you have any suggestions for adjusting those weights.

Land Use Class Weights

Land Use Class	Weight
Urban	1.00
Agriculture	0.60
Forest	0.30
Wetland	0.05
Water	0.05
Barren	0.20
Shrublands	0.30
Herbaceous	0.30

The density_nlcd value is currently based on the percentage area of each land use class across the entire HUC. If we're all on board with this workflow, I’ll move forward and calculate it based only on the flooded areas (by computing the percentage area for flooded catchments and then averaging across them).