A replacement for simple back trajectory calculations in the interpretation of atmospheric trace substance measurements

Trajectory calculations are often used for the interpretation of atmospheric trace substances measurements. However, two important effects are normally neglected or not considered systematically: first, measurements of trace substances sample finite volumes of air, whereas a trajectory tracks the path of an infinitesimally small particle; second, turbulence and convection. Advection by the deformative synoptic-scale atmospheric flow is responsible for the fact that a compact measurement volume is in fact turned into filamentary structures at earlier times, which a single trajectory cannot represent. Turbulence and convection add to this by causing a growth of the volume (backward in time) where processes such as emissions can affect measured concentrations. In this paper, we show that both effects are substantial and may be the largest sources of error when using trajectory calculations to establish source–receptor relationships. We use backward simulations with a Lagrangian particle dispersion model (LPDM) and cluster analysis of the particle positions to derive more representative single “trajectories” (transport paths) and trajectory ensembles. This reduces errors caused by filamentation and backward growth of the measurement volume to a few percent as compared to using a single, mean-wind trajectory and also yields estimates of the spread of the region of influence. Thus, we recommend to replace simple back trajectory calculations for interpretation of atmospheric trace substances measurements in the future by backward simulations with LPDMs, possibly followed by the clustering of particle positions as introduced in this paper.