Which mangroves stand between Vietnam's coast and the next storm?
758 historical cyclone tracks from NOAA IBTrACS (1950–2025), classified using
Vietnam's official tropical cyclone scale, were cross-referenced with actual mangrove
extent polygons from the Global Mangrove Watch. I first prototyped the analysis locally
using GeoPandas, then turned it into a dbt pipeline running directly on BigQuery.
Storm tracks vs. mapped mangrove extent
1950–2025. Hover a province for its mangrove hectares, or a highlighted track for the storm's intensity.
Mangrove extent
none77k ha
Storm tracks
largest mangrove exposure (top 12)
full historical record
Context
Inland Vietnam · Cambodia, Laos, China
Largest mangrove exposure
Storms ranked by mangrove hectares within 50km of their track
Most frequently exposed
Provinces by count of historical storms within 50km
Do the numbers hold up?
How does the 2020 Global Mangrove Watch extent compare with the multi-year historical mangrove
mask used in an earlier project? I compare the two datasets by province to see where the estimates
broadly agree and where they diverge.
Province
GMW 2020 (ha)
Historical max (ha)
Ratio
GMW 2020 is consistently lower than the historical maximum. This is expected, since the historical
figure is a union of a 2000 land-cover mask and GMW's 2007–2020 range (the largest
footprint ever mapped), while GMW 2020 is a single-year snapshot taken after any loss in between.
From prototype to production
The same spatial logic, tested locally in GeoPandas and rebuilt in SQL for BigQuery.
Both paths start with the same raw IBTrACS and Global Mangrove Watch data. The GeoPandas
version was used to prototype and validate the spatial-join logic locally. That same logic
was then re-implemented in SQL through a dbt pipeline running on BigQuery, where it now
powers the live production table.
Research use only
This analysis is a research prototype intended to explore spatial relationships between
historical tropical cyclone tracks and mangrove extent. The results have not undergone
sufficient validation to support operational or official decision-making. They should not
be treated as a substitute for official forecasts, warnings, hazard assessments, or
government reports.