Motivation

The current FIM database indexes flood maps using a single value: discharge at the downstream point of the modeled domain. This works for models like HAND, which are anchored on discharge alone.

However, HEC-RAS models (such as those in the National Water Center's library) are generated using two boundary conditions per reach: - Upstream BC: discharge entering the reach - Downstream BC: a water surface elevation (WSE), normal depth, or other condition at the downstream end

A single-discharge index cannot distinguish between two HEC-RAS runs with the same flow but different downstream BCs — resulting in different flood extents. This gap prevents systematic indexing, querying, and comparison of HEC-RAS flood map libraries.

## What was done

### 1. Database schema updated for dual BC indexing

The relational schema was extended with three new columns (shown in red in the diagram below):

Each unique (model x downstream BC) combination now gets its own rating curve, so a HEC-RAS reach with 8 downstream WSE values produces 8 separate rating curves — all under one FIM source.

### 2. Converted Colab notebooks to standalone Python scripts

The original workflow ran in Google Colab notebooks. To support local execution and automation, the pipeline was converted to three standalone Python scripts driven by YAML config files:

Each script is configured via a corresponding .yaml file. A conda environment.yml is included for reproducibility.

### 3. HEC-RAS folder structure supports dual indexing

HEC-RAS flood maps are organized as:

HECRAS-1D//z_/f_.tif

For example:

HECRAS-1D/447692/z_nd/f_982.tif ← normal depth BC HECRAS-1D/447692/z_90_0/f_982.tif ← downstream WSE = 90.0 ft HECRAS-1D/447692/z_85_0/f_982.tif ← downstream WSE = 85.0 ft

The preprocessing script automatically parses this structure and assigns each downstream BC its own rating curve in the database.

### 4. Visualization with dual-indexed scenarios

#### Scenario visualization

The pipeline generates a visualization JSON compatible with the [FIM visualization app] (https://hydroinformatics.tulane.edu/lab/fims/) developed by the Hydroinformatics Lab at Tulane University. Users can select a model and downstream BC combination from the dropdown menu and use the slider to step through different discharge scenarios, instantly viewing the corresponding flood extent and depth map.

Currently, the downstream boundary condition is embedded in the dropdown label alongside the model name (e.g., "HEC-RAS1D | DS: WSE 90 ft"). This is because the current work focuses on the database side, and the visualization app is maintained independently. In the future, the app could be extended to include separate controls — one slider for discharge and another for downstream BC — for a more intuitive user experience.

#### Comparison with HAND FIM

The workflow also generates HAND-based FIMs for the same reach using FIMserv, enabling direct comparison between HAND and HEC-RAS1D flood maps within the same viewer. HAND models are anchored on discharge alone (no downstream BC), so they serve as a useful baseline.

This side-by-side comparison is valuable for evaluating how HAND and HEC-RAS1D complement each other, and could inform how both models are used together in NOAA's operational flood inundation mapping services

## Why this matters

The National Water Center maintains a large library of HEC-RAS flood maps generated with varying boundary conditions. This dual-indexing extension enables:

• Systematic cataloging of HEC-RAS flood maps by both upstream flow and downstream BC
• Querying and comparison across boundary condition scenarios for the same reach
• Integration of HEC-RAS maps alongside HAND, FEMA, and other models in one unified database
• Scalability — the same framework could index hundreds of reaches across the NWC library

## Sample data and reproducibility

The branch includes a small sample dataset (reach 447692 with multiple downstream BCs) so the full pipeline can be tested end-to-end:

```bash conda env create -f environment.yml conda activate fim cd scripts python run_fimserve.py --config run_fimserve.yaml python run_preprocessing.py --config run_preprocessing.yaml python run_populate_db.py --config run_populate_db.yaml

Next steps

• Gather feedback from stakeholders on the dual-indexing approach
• Test with a larger subset of the NWC HEC-RAS library
• Evaluate integration with the existing visualization platform at scale