Matched-Filter Stability Under Real Observational Noise

This technical note presents a null stress test of a matched-filter structural detection framework applied to real observational data from HST ACS F814W skycell mosaics. The objective is strictly methodological: to verify that the framework does not generate artificial statistical significance when applied to complex, non-ideal imaging data containing realistic background structure, PSF variation, mosaic discontinuities, and instrumental artifacts. The analysis includes: Sigma-clipping source masking Gaussian smoothing (σ = 12 pixels) Zero-mean halo template construction (σ = 10″) Local nuisance-plane regression Empirical local-control amplitude estimation within a 60–120″ annulus The matched-filter amplitude at the nominal center is evaluated against an empirical null distribution constructed from locally sampled positions under identical processing conditions. The observed amplitude lies well within the bulk of the local-control distribution (p_low ≈ 0.40), indicating no spurious detection under real observational noise. This record serves as a robustness validation of the matched-filter structural detection framework under real observational conditions. The fully reproducible Colab implementation used for this analysis is archived alongside this record as a Jupyter notebook file.