SUB-GRID-SCALE FEATURES OF ANTHROPOGENIC EMISSIONS OF NOX AND VOC IN THE CONTEXT OF REGIONAL EULERIAN MODELS

Model uncertainty is a major issue concerning regional-scale air quali ty simulation. One major source of uncertainty in regional Eulerian mo dels is due to sub-grid-scale (SGS) effects related to anthropogenic e missions. Regional models typically have horizontal grid resolution (D elta) of 20-80 km. Since NOx chemistry in plumes is nonlinear and ofte n diffusion-limited, the sudden dilution of plumes over regional grid dimensions, as in current models, can lead to a fundamental distortion of their chemistry, resulting in over-production of ozone, peroxides, sulfates and nitrates in the source region, and a related over-deplet ion of NOx. The corresponding model uncertainty over the whole regiona l domain remains unquantified. In this paper, we use high-resolution i nformation from urban and regional emission inventories and plume fiel d studies to examine SGS features of the emissions of anthropogenic NO x and VOC (volatile organic compounds), and of their mesoscale dispers ion and chemistry. Such examination provides useful insight into some of the main sources of SGS uncertainty, as well as guidance for reduci ng it. The mesoscale chemistry of power plant plumes is very diffusion -limited, being controlled by VOC entrainment from the background. The crosswind spread of large point-source plumes typically takes 4-6 h t o reach 30 km in convective conditions, and at least a full diurnal cy cle to reach 80 km. For Delta much larger than 20-30 km, regional mode ls will not be able to capture the essence of the behavior of rural po int-source plumes even with plume-in-grid treatment, or to resolve the NOx emissions from many large power plants in urban peripheries from the urban VOC emissions. Within urban areas, there is progressive impr ovement in the resolution of the important ratio VOC/NOx as Delta is d ecreased below 20 km. The nature of these emissions-related SGS featur es suggests that significant gain in regional model accuracy should re sult by limiting Delta to 20-30 km in the regional domain, by the use of finer nested gridding in metropolitan sub-domains, and by a reactiv e plume-in-grid treatment of major point-source emissions. Copyright ( C) 1996 Elsevier Science Ltd