Sector-dependent shallow atmospheric ducting during tropical cyclones in coastal boundary layers of the western North Pacific

Tropical cyclones can reorganize coastal thermodynamic structure at scales that control microwave propagation, yet storm effects on shallow atmospheric ducts remain poorly quantified because decisive modified-refractivity gradients often occur within the lowest tens of metres. We diagnose storm-conditioned shallow ducts across western North Pacific coastal sectors using IBTrACS tracks, ERA5 context, 46,485 radiosonde profiles from 12 stations, two Hong Kong typhoon soundings, 3 GHz parabolic-equation propagation experiments, a Krosa radar-constrained admissible-state library, and published microwave-link observations. The primary comparison pairs a duct-prone South China Sea margin with a two-station Pacific-facing Philippine aperture. At the 100 m start-height threshold, the Pacific aperture showed an 11.6 percentage-point increase relative to same-station, same-month non-storm controls (storm-clustered 95% confidence interval: 1.3–22.8), whereas the South China Sea margin showed no positive matched response. Direct Saola and Mangkhut soundings resolved 45 and 27 m launch-level ducts, respectively, and both produced positive propagation-factor anomalies throughout 40–100 km. A signed decomposition of d𝑀∕d𝑧 shows why storm intensity alone is an insufficient predictor: inversions and drying aloft favour trapping, while turbulent homogenization can erase the same interface. The results are consistent with background-state gating by frictional inflow, warm/dry capping or descending flow, rainband cold-pool/downdraft processes, and storm-relative asymmetry. In Krosa, admissible trapping-layer tops descended from 130–150 m to 50–60 m within 90 min. Retrospective 12–48 h discrimination was strong, but prospective forecasting and dynamical attribution require targeted storm-relative observations.