Storm-source geometry biases mantle transition-zone water estimates from a century-scale geomagnetic archive

Seokhoon Oh 

Electromagnetic estimates of water in the mantle transition zone depend on the assumed geometry of storm-time source fields. We audit the effect of that assumption using paired processing chains that differ only by inclusion of a degree-3 zonal source term, together with pre-specified admission gates. The experiment is applied to three separately processed, partially overlapping observatory records spanning 6, 15, and 80 years and to random subnetworks drawn from a century-scale station pool. At 12 stations, 58 to 62% of random draws fail the admission gate; failed chains produce apparent transition-zone shifts of 2.5 to 3.4 decades and are not physically interpretable. Among admitted sparse draws, shifts range from −0.30 to +0.65 decades and have draw-dependent sign. By contrast, the tested curated networks at 12, 43, and 130 stations give mutually consistent legacy-minus-source-aware shifts near −0.13 decades, indicating that the degree-1-only chain underestimates conductivity. Co-estimating the degree-3 term in the 1940 to 2020 chain improves normalized misfit from 1.41 to 1.07 and yields a transition-zone conductivity of 0.386 S/m. Laboratory conductivity–water calibrations map the curated-network shift to a dry-bias factor of 1.25 to 1.55; admitted sparse networks can produce larger errors in either direction. The published conductivity range overlaps the validated sparse-network envelope, although attribution to individual studies requires controlled paired reanalysis. These results identify two source-geometry risks: a systematic, correctable dry bias in the tested curated chains and sign-unstable errors in unbalanced sparse networks.