An early burst-shape fit reduces later within-day DM error by 47 percent for FRB 20220912A in CHIME Catalog 2

We test whether burst properties predict dispersion measure (DM) differences within an observing day for FRB 20220912A. We use CHIME/FRB Catalog 2 release 20251231 and count each event once. A linear fit to peak frequency, bandwidth and log width uses 42 early days. It reduces mean squared error by 47.5 percent on 42 later days. A day bootstrap gives a nominal 95 percent skill interval of 0.093 to 0.627. This test removes a separate mean from each day. Thus, a slow change in electron column cannot by itself explain the measured relation. Common fit errors, burst physics and instrument effects remain possible. Applying the same coefficients across days is a separate, unproved assumption. We test that correction and two event cuts, but do not claim a physical DM trend or a new source. Published independent microshot data support concern about burst-shape bias but do not confirm this numerical relation. Code, frozen catalog rows, provenance, controls and figures are included.

astronomy · CHIME · FRB 20220912A · dispersion measure · burst morphology · time validation