Radical chemistry at a UK coastal receptor site - Part 2: experimental radical budgets and ozone production.

In our companion paper (Woodward-Massey et al., 2022), we presented measurements of radical species and OH reactivity (k'OH) made in summer 2015 during the ICOZA (Integrated Chemistry of OZone in the Atmosphere) field campaign at the Weybourne Atmospheric Observatory, a site on the east coast of the UK. In the present work, we used the simultaneous measurement of OH, HO₂, total RO₂, and k'OH to derive experimental (i.e., observationally determined) budgets for all radical species as well as total RO_x (= OH + HO₂ + RO₂). Data were separated according to wind direction: prevailing SW winds (with influence from London and other major conurbations), and all other winds (NW-SE; predominantly marine in origin). In NW-SE air, the RO_x budget could be closed during the daytime within experimental uncertainty but OH destruction exceeded OH production, and HO₂ production greatly exceeded HO₂ destruction while the opposite was true for RO₂. In SW air, the RO_x budget analysis indicated missing daytime RO_x sources but the OH budget was balanced, and the same imbalances were found with the HO₂ and RO₂ budgets as in NW-SE air. For HO₂ and RO₂, the budget imbalances were most severe at high NO mixing ratios. We explored several mechanistic modifications to the experimental budgets to try to reconcile the HO₂ and RO₂ budget imbalances: (1) the addition of generic radical recycling processes, (2) reduction of the rate of RO₂ x HO₂ conversion, (3) inclusion of heterogeneous HO₂ uptake, and (4) addition of chlorine chemistry. The best agreement between HO₂ and RO₂ production and destruction rates was found for option (2), in which we reduced the RO₂ + NO rate constant by a factor of 5. The rate of in situ ozone production (P(O_x)) was calculated from observations of RO_x, NO, and NO₂ and compared to that calculated from MCM-modelled radical concentrations. The MCM-calculated P(O_x) significantly 35 underpredicted the measurement-calculated P(O_x) in the morning, and the degree of underprediction was found to scale with NO. [ABSTRACT FROM AUTHOR]