A review on methodology in O3-NOx-VOC sensitivity study

Gaining insight into the response of surface ozone (O3) formation to its precursors plays an important role in the policy-making of O3 pollution control. However, the real atmosphere is an open and dissipative system, and its complexity poses a great challenge to the study of nonlinear relations between O3 and its precursors. At present, model-based methods based on reductionism try to restore the real atmospheric photochemical system, by coupling meteorological model and chemical transport model in temporal and spatial resolution completely. Nevertheless, large inconsistencies between predictions and true values still exist, due to the great uncertainty originated from emission inventory, photochemical reaction mechanism and meteorological factors. Recently, based on field observations, some nonlinear methods have successfully revealed the complex emergent properties (long-term persistence, multi-fractal, etc) in coupling correlation between O3 and its precursors at different time scales. The emergent properties are closely associated with the intrinsic dynamics of atmospheric photochemical system. Taking them into account when building O3 prediction model, is helpful to reduce the uncertainty in the results. Nonlinear methods (fractal, chaos, etc) based on holism can give new insights into the nonlinear relations between O3 and its precursors. Changes of thinking models in methodology are expected to improve the precision of forecasting O3 concentration. This paper has reviewed the advances of different methods for studying the sensitivity of O3 formation to its precursors during the past few decades. This review highlights that it is necessary to incorporate the emergent properties obtained by nonlinear methods into the modern models, for assessing O3 formation under combined air pollution environment more accurately. Moreover, the scaling property of coupling correlation detected in the real observations of O3 and its precursors could be used to test and improve the simulation performance of modern models.