Methodological Bias between Trajectory Geometric Co-location and Air-Parcel Lagrangian Co-location: Evidence from UAV Re-penetration Observations over the Western Sichuan Plateau

Li Huijing

Whether "trajectory geometric co-location" is equivalent to "air-parcel Lagrangian co-location" remains a methodological blind spot in evaluating UAV cloud-seeding re-penetration (the UAV performed a repeated cloud penetration—hereafter, re-penetration—to re-sample the same cloud volume). Using silver iodide (AgI) seeding observations from a Wing Loong-2 UAV flight over the Western Sichuan Plateau on 15 February 2026, together with three-station soundings, ERA5 reanalysis, and airborne measurements, we quantify the bias between the two frameworks. At the ERA5 horizontal wind speed of 13.1 m s⁻¹ at the seeding altitude (≈6.68 km), the plume advective displacement over 53 min is about 41.6 km, comparable to the 49.1 km longitudinal span of Window A; the geometrically co-located window therefore very likely no longer enclosed the same air mass (i.e., geometric co-location ≠ Lagrangian co-location). Three-dimensional geometric co-location yields a size spectrum ratio R(D) ≈ 1.0, which, given the available data, cannot be distinguished as either (i) persistence of the seeded plume or (ii) sampling of fresh upwind background air whose concentration happens to match that of the initial seeding leg. Lon-only co-location, by contrast, produces a spurious decay (R(D) = 0.67/0.60/0.58). We propose methodological protocols for re-penetration evaluation and operational recommendations for Lagrangian tracking (wind-offset flight tracks, dual-aircraft coordination, inert tracers, and FLEXPART-type dispersion modeling), and stress that, because Leg1 is itself a seeding leg with no unseeded baseline, re-penetration evaluations must state this limitation explicitly. The results provide a quantitative basis for the methodological design of UAV seeding-effect verification over the plateau.