TROPICAL STORMS IN VIETNAM: DATASET FOR FORECASTING

This dataset provides a multi-source collection of meteorological and environmental data designed for tropical cyclone (TC) forecasting research over the South China Sea (6°S–24°N, 102°E–117°E).It integrates ERA5 reanalysis data from the European Centre for Medium-Range Weather Forecasts (ECMWF) and CMA/IBTrACS best track data, offering synchronized, standardized, and preprocessed inputs suitable for machine learning and deep learning experiments.

The dataset was developed as part of the research project “DEEP LEARNING-BASED FORECASTING OF TROPICAL STORM TRACKS AND INTENSITY IN VIETNAM”, aiming to enhance regional TC forecasting through spatio-temporal data fusion.

Dataset Components:

File
Description
Size

CMA_BESTTRACK_VN_FINAL.zip
Processed best track data (1949–2024) from CMA, including latitude, longitude, wind speed, pressure, and storm ID for all tropical cyclones affecting Vietnam and nearby regions.
237 KB

ERA5_Pressure_Level.zip
ERA5 reanalysis at pressure levels (200–850 hPa), including u/v wind, temperature, geopotential, and relative humidity fields extracted around storm centers (±15°).
16.6 GB

ERA5_Single_Level.zip
ERA5 single-level variables, such as mean sea level pressure (msl), 10m winds, and sea surface temperature (SST).
4.1 GB

ENV_DATA.zip
Derived environmental features used for model input (e.g., shear, humidity index, SST anomaly).
10 MB

Data Coverage and Format:

Temporal coverage: 1949–2024 (6-hour intervals, 00/06/12/18 UTC)

Spatial coverage: South China Sea region

Format: NetCDF (.nc) and CSV

Coordinate system: Latitude/Longitude (EPSG:4326)

Temporal alignment: Synced between IBTrACS and ERA5 at 6-hour resolution

Usage:

The dataset is ideal for:

Tropical cyclone track and intensity forecasting

Spatio-temporal modeling (Conv3D, Transformer, U-Net)

Data fusion research combining track and reanalysis data

Educational or benchmarking purposes in meteorological deep learning

Acknowledgements:

ERA5 data were retrieved from the Copernicus Climate Data Store (CDS), and IBTrACS/CMA data from the NOAA National Centers for Environmental Information (NCEI).Processing and alignment were performed using Python (xarray, pandas, numpy).