Surface ozone and its precursors at Summit, Greenland: comparison between observations and model simulations

摘要

Recent studies have shown significant challenges for atmosphericmodels to simulate tropospheric ozone (O) and its precursors in theArctic. In this study, ground-based data were combined with a global 3-Dchemical transport model (GEOS-Chem) to examine the abundance and seasonalvariations of O and its precursors at Summit, Greenland(72.34° N, 38.29° W; 3212 m a.s.l.). Model simulations foratmospheric nitrogen oxides (NO), peroxyacetyl nitrate (PAN), ethane(CH), propane (CH), carbon monoxide (CO), and Ofor the period July 2008–June 2010 were compared with observations. The modelperformed well in simulating certain species (such as CO and CH),but some significant discrepancies were identified for other species andfurther investigated. The model generally underestimated NO and PAN (by  ∼  50 and 30 %, respectively) for March–June. Likelycontributing factors to the low bias include missing NO and PANemissions from snowpack chemistry in the model. At the same time, the modeloverestimated NO mixing ratios by more than a factor of 2 inwintertime, with episodic NO mixing ratios up to 15 times higher thanthe typical NO levels at Summit. Further investigation showed thatthese simulated episodic NO spikes were always associated withtransport events from Europe, but the exact cause remained unclear. The modelsystematically overestimated CH mixing ratios by approximately20 % relative to observations. This discrepancy can be resolved bydecreasing anthropogenic CH emissions over Asia and the US by  ∼ 20 %, from 5.4 to 4.4 Tg year. GEOS-Chem was able to reproduce theseasonal variability of O and its spring maximum. However, comparedwith observations, it underestimated surface O by approximately 13 %(6.5 ppbv) from April to July. This low bias appeared to be driven by severalfactors including missing snowpack emissions of NO and nitrous acid inthe model, the weak simulated stratosphere-to-troposphere exchange flux ofO over the summit, and the coarse model resolution.