Flood-Map Updates, Insurance Premiums, and the Formalization of Disaster Risk

# Flood map paper replication package

This package reproduces the tables, figures, and checked in-text numbers for the IJDRR manuscript on FEMA flood-map updates, NFIP insurance outcomes, and mortgage-market responses.

## Repository structure

- `code/run_cleaned.py`: reproduces all manuscript outputs from frozen cleaned data.

- `code/run_raw.py`: rebuilds the cleaned data from uploaded public raw files and then reproduces outputs.

- `data/cleaned/`: frozen cleaned tract-year panel and simplified map geometry (77.4 MiB).

- `docs/raw_files_to_upload.csv`: required raw-upload checklist with file sizes, checksums where available, and source/download method.

- `docs/table_figure_crosswalk.csv`: maps manuscript items to generated output files.

- `docs/manuscript_number_crosswalk.csv`: checked in-text/table numbers.

- `reference_outputs/`: expected tables and figures used for the submitted manuscript.

## Computational environment

Use either conda:

```bash

conda env create -f environment.yml

conda activate fema-ijdrr-replication

```

or venv/pip:

```bash

python -m venv .venv

source .venv/bin/activate

pip install -r requirements.txt

```

## One-line reproduction from frozen cleaned data

```bash

python code/run_cleaned.py --analysis-data data/cleaned/fema_auxiliary_tract_year_panel.parquet --map-data data/cleaned/treated_tract_map_geometry.parquet --output outputs_from_cleaned && python code/check_outputs.py --outputs outputs_from_cleaned

```

## One-line reproduction from uploaded raw data

Upload every file listed in `docs/raw_files_to_upload.csv` to the exact `required_upload_path` location inside this package. The required raw-upload files total approximately 32.5 GiB. Then run:

```bash

python code/run_raw.py --output outputs_from_raw && python code/check_outputs.py --outputs outputs_from_raw

```

The raw route first checks that all required files are present. If files are missing, it writes `docs/missing_raw_files.csv` and stops.

The cleaned-data route has been run end to end. The raw-data route is provided for full reconstruction and requires the 96 public raw files listed in `docs/raw_files_to_upload.csv`.

## Expected outputs

Generated outputs are written to the chosen output directory with `tables/`, `figures/`, and `sha256_outputs.csv` subfiles. The `check_outputs.py` script verifies that all manuscript-table and figure files exist and that the checked manuscript numbers match `docs/manuscript_number_crosswalk.csv`.