TRAMS: SVD-3DEnVar Simulations, Observation Data, and Plotting Scripts

The control forecast was initialized at 12:00 UTC on September 5, 2024, while assimilation forecasts are provided for three consecutive cycles valid at 00:00, 06:00, and 12:00 UTC on September 6, 2024. Corresponding initial condition files show both background and analysis states for each assimilation time. Observation files contain the satellite-derived sea surface wind data used in each assimilation cycle.The plot directory contains Python scripts for visualizing the experimental results.

These scripts generate the primary figures presented in the associated paper, including visualizations of the model domain, perturbation fields, parallel computing performance, assimilation increments, and typhoon forecast comparisons. The scripts are designed to work with the provided NetCDF files and require standard scientific Python libraries.File naming follows a consistent pattern throughout the dataset. Forecast files are named as forecast_YYYYMMDDHH_experiment.nc, where the timestamp indicates the initialization time and the suffix (ctl for control or da/da1/da2/da3 for assimilation cycles) indicates the experiment type.

Initial condition files follow the pattern input_YYYYMMDDHH_type, where type is either bk (background) or da (analysis). Observation files are named obs_YYYYMMDDHH.nc indicating the valid time of the observations.This dataset supports the reproducibility of the study's findings regarding the performance of the optimized SVD-3DEnVar scheme for typhoon data assimilation. It provides a valuable resource for researchers interested in evaluating and further developing data assimilation techniques for numerical weather prediction, particularly in the context of tropical cyclone forecasting.