Chemistry of OH and HO 2 radicals observed at Rishiri Island, Japan, in September 2003: Missing daytime sink of HO 2 and positive nighttime correlations with monoterpenes

We used laser‐induced fluorescence to measure concentrations of OH and HO 2 at Rishiri Island, Japan, during September 2003. The average maximum daytime concentrations were 2.7 × 10 6 cm −3 for OH and 5.9 pptv for HO 2 . The observed concentrations were compared to those predicted by a photochemical box model constrained by ancillary observations. During the daytime, the model overestimated HO 2 levels by an average of 89% and OH levels by an average of 35%. This overestimate of OH was rectified when the model was constrained by observed HO 2 levels, suggesting that loss processes of HO 2 were missing in the model. We calculated the loss rates of HO 2 required to bring the modeled HO 2 levels into agreement with observed levels. We then studied processes that are capable of explaining the loss rates, including halogen chemistry, heterogeneous loss of HO 2 on aerosol surfaces, and the possibility of more rapid HO 2 + RO 2 reactions than expected. In the nighttime, most of the observed hourly averaged OH and 10‐min‐averaged HO 2 concentrations were statistically significant and fell in the ranges (0.7–5.5) × 10 5 cm −3 and 0.5–4.9 pptv, respectively. Both HO 2 and OH concentrations showed strong positive correlations with total monoterpene concentrations, strongly suggesting that the radicals were produced via reactions of monoterpenes. The median nighttime modeled‐to‐observed ratios were 1.29 and 0.56 for HO 2 and OH, respectively. These ratios dropped to 0.49 and 0.29 during the evening of 25 September, possibly related to the presence of unmeasured olefinic species or chemical reactions involving RO 2 that are poorly represented in the model.