Measured Validation Dataset for Credibility Calibration of Multi-Fidelity Radar Plot Data in Equipment Digital Twins (171 Target-CPI Raw Echoes, Flight-Log Truth, and Derived Calibration Products)

Data were collected in an anti-UAV field trial with a Ku-band pulse-Doppler radar (16.3 GHz carrier, 12.5 MHz sampling, 12 MHz waveform bandwidth, 64-pulse coherent integration), observing a cooperative target (DJI Air 3S) on a radial flight segment (azimuth ~142.5° north-referenced, elevation 2°–6°, range 0.5–2.2 km).

The dataset contains: (1) raw echo files of 171 target-beam-position CPIs (204,800 bytes each: a 768-byte header plus 64-pulse IQ samples), which constitute the complete raw dependency of the paper's measured results; (2) 10 Hz flight-log truth (latitude/longitude/altitude/velocity), with 21 CPIs truth-aligned to reference plots from the full 64-pulse trusted processing chain, forming real high-fidelity anchors; (3) derived calibration products, including unified plot-quality rows for five sources (full signal-level chain, effect-equivalent probabilistic chain, analytic radar-equation model, and two reduced-integration chains), the five-source calibration summary, theoretical-source parameter calibration, a cross-mechanism consistency report, and a K=100 seed-stability sweep; and (4) preregistration documents (main experiment plan, measured-experiment requirements and instantiation mapping, and nine preregistered expectation checks).