Optimizing airborne emission rate retrievals with sub-hectometre resolution numerical modelling

A comprehensive model-based study is designed to provide optimal flight paths for airborne top-down emission rate retrieval methodologies. The meteorology and plume dispersion were modelled using the Weather Research and Forecasting (WRF) modelling platform with the Advanced Research WRF (ARW) dynamical core at 50-m resolution. Multiple flight path designs and parameters were investigated to determine emission rate retrieval accuracy for emissions of a trace gas as a function of downwind distance and transect spacing, which are ultimately related to flight time and cost. Three unique source types (multiple smokestack plumes, small area sources, and a large area source) were investigated for 4 summer afternoon flight cases over 2 d. The results demonstrate that emissions estimate uncertainty is primarily due to storage and release, with uncertainties as high as 30 % at optimal downwind distances (which vary by source type). Interpolation of the sparse measurements can be a significant source of error close to the source, but the uncertainty is ≤17 % for D ≥6 km. The average advective flux estimates are within 12 % of the known emissions for downwind distance of D ≥4 km. Variability between flights decreases with D . For stack sources the variability near D =10 km is approximately half that at D =4 km. For small area sources, there is less reduction with D , and for the large area source, variability reaches a minimum at D =8 km. For stack sources, vertical spacing of transects is optimized at 100 m, while for area sources, a spacing of 50 m reduces uncertainty. Error due to extrapolation below the lowest flight path is less than 20 % for stack sources and less than 30 % for area sources for non-dimensionalized downwind distance of D′≥3. Results demonstrate the need for surface sampling coincident with the flights to reduce extrapolation error, and the use of modeling with reanalysis data to account for storage and release effects.