Flood Inventory Cleaning Before Analysis

Flood datasets often combine field reports, remote sensing, news records, and administrative summaries. Cleaning is required before modelling or dashboard visualization.

1. Remove duplicates

Check incident date, location, source, and description to avoid counting the same event multiple times.

2. Standardize locations

Convert local names into district, municipality, ward, basin, and coordinate fields when possible.

3. Separate hazard and impact

Flood occurrence, inundation area, damaged houses, casualties, and road closure are different fields and should not be mixed.

4. Check time windows

For satellite flood extents, separate pre-event, event, and post-event windows clearly.

5. Keep raw and cleaned copies

Store the original dataset and the cleaned analysis version separately for transparency.