Data and code for "Geographically Organized Near-Surface Winds in AIFS Forecasts: Retained Patterns and Weak Amplitudes"

Processed data and Python code supporting "Geographically Organized Near-Surface Winds in AIFS Forecasts: Retained Patterns and Weak Amplitudes", by Yulong Ma, Shu Wang and Qilin Wan. The archive contains processed data for Figures 1–9, aggregate station comparisons, and the analysis and plotting functions used in the study. Deterministic comparisons cover June–July 2026; the ensemble comparison also includes August 2026. The forecast and analysis products are ECMWF AIFS Single v2, AIFS ENS v2, IFS Cycle 50r1 control forecasts, and IFS operational analyses. The included processed data reproduce the figures directly. With Python 3.10 or later, install the packages in requirements.txt and run python code/plot_figures.py. Plotting uses only the included data and requires no network access. Array units and dimensions are documented in data/README.md. Recalculating diagnostics from raw forecasts and observations requires the provider data and array inputs specified in the analysis functions. Surface and radiosonde observations were supplied by the China Meteorological Administration. The archive includes station-aggregated diagnostics and station locations rounded to the 0.25° plotting grid; raw observations and station-level wind records are not distributed. Original data remain subject to their providers' terms. Processed data and documentation are licensed under CC BY 4.0; Python code is covered by the MIT License in LICENSE.txt. Contact: Yulong Ma, [e-mail].