2026-10-06T21:37:20Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33191484 2026-10-02T23:48:13Z category_25744 category_26524 item_type_3 month_year_10_2026

An IQ-Level Benchmark Dataset and Interference Suppression Comparison Demo for Phased-Array Dual-Polarization Weather Radar sphere nature (20731835) Meteorology Radio frequency engineering Weather Radar Dual-Polarization Radar Radio-Frequency Interference RFI Suppression IQ Data An IQ-Level Benchmark Dataset and Interference Suppression Comparison Demo for Phased-Array Dual-Polarization Weather RadarAuthors: Zhihuo Xu and Yuexia Wang Revised benchmark release: 02 October 2026 Comparison program version: 3.0.0<h2 dir="ltr">Overview</h2>This repository provides a self-contained IQ-level benchmark dataset and a reproducible comparison program for radio-frequency interference (RFI) suppression in phased-array dual-polarization weather radar.Each released NetCDF result is self-contained: it stores the Ground-truth IQ, the corresponding Observed-RFI IQ, and the Recovered IQ produced by that method. The Ground-truth and Observed-RFI IQ are duplicated across the four method files so that every result can be inspected independently while still supporting a common-input comparison.The current release contains four successive suppression models:<ol><li>PolStatJoint</li><li>PolPhaseAware</li><li>PolMomentV1</li><li>PolMomentV2</li></ol>and three interference scenarios:<ul><li>CW (<code>cw_only</code>)</li><li>LFM (<code>lfm_only</code>)</li><li>LFM+CW (<code>lfm_plus_cw</code>)</li></ul>Each method/scenario result is stored in a self-contained <code>recovered_full360.nc</code> file containing the Ground-truth IQ, Observed-RFI IQ, Recovered IQ, radar geometry, interference mask, and Level-II fields needed by the comparison workflow. For each scenario, the PolStatJoint <code>recovered_full360.nc</code> is used as the canonical source of Ground-truth IQ, Observed-RFI IQ, the interference-ray mask, azimuth/range coordinates, radar metadata, and Ground-truth/Observed Level-II fields. Each of the four method files contributes only its own Recovered IQ to the method comparison. The program verifies compatible radar geometry before combining the results.All recovered Level-II products are recalculated from each method's Recovered IQ with one common IQ-to-Level-II implementation. This avoids method-specific saved postprocessing from biasing the comparison. Run the complete inspection from <code>data</code>: <pre>python test_data.py </pre>Because this produces a long report for all 12 files, a compact release-integrity check is also available:<pre> <code>python test_data.py --summary-only</code> Running the Benchmark</pre>Open a terminal in the <code>data_code</code> directory and run: <pre>python compare_recovered_methods.py </pre>No input-path arguments are required when the directory layout above is used. Default paths are anchored to the directory containing <code>compare_recovered_methods.py</code>, not to the terminal's current working directory.<h2 dir="ltr">Suggested Citation</h2>When this benchmark dataset and comparison program are used in an experiment, cite them as:<blockquote>Zhihuo Xu and Yuexia Wang, “An IQ-level benchmark dataset and interference suppression comparison demo for phased-array dual-polarization weather radar,” figshare, 2026, dataset and comparison code, doi:10.6084/m9.figshare.33191484.</blockquote> 2026-10-02T23:48:13Z Dataset Dataset 10.6084/m9.figshare.33191484.v10 https://figshare.com/articles/dataset/An_IQ-Level_Benchmark_Dataset_and_Interference_Suppression_Comparison_Demo_for_Phased-Array_Dual-Polarization_Weather_Radar/33191484 CC BY 4.0