The latter quarter of the 20th century has seen an increasing interest in the problem of air pollution and its remediation, both in developed, and, more recently, in underdeveloped countries. During the 1970's and 1980's, much attention was payed to modeling individual pollutant problems (such as "smog" in Los Angeles and "acid rain" in the eastern states). More recently, third-generation modeling systems (such as EPA's "Models-3" system and NCSC's "MAQSIP" system) have been developed with multipollutant, multidisciplinary aspects of the problem in mind. These models, based on respected predecessors, have been reformulated both from a computational and science perspective.
In the case of MAQSIP, atmospheric scientists and chemists have worked closely together with computational scientists and mathematicians to create a modular, flexible modeling system that can be used to address and simulate a variety of interrelated pollutant and atmospheric chemical problems. Further, learning from the results of such programs as NAPAP, MAQSIP developers recognized the need for much closer coupling between MAQSIP and the meteorological model(s)that were to be used as drivers.
This is particularly true of the more complex relationships between atmospheric aerosols, the radiation budget, cloudiness, and precipitation. Though not all of these processes are tightly coupled in MAQSIP yet, the quasi-operational version of MAQSIP discussed here, which is now being driven by the PSU/NCAR MM5 mesoscale model, is particularly closely coupled to MM5's convective and resolved scale moisture parameterizations. It has been known for some time that this is a key to improving the ability to simulate air quality on a regional-mesoscale basis.
In this paper, we report on advancements reaped from the development of MAQSIP and its close coupling to the PSU/NCAR MM5 Mesoscale Model. Along with vast improvements in emissions modeling, a highly optimized, parallel version of the coupled meteorology-emissions-air chemistry modeling system was used during the summer of 1998 to produce once daily experimental nested real-time numerical air quality predictions (NAQPs). These forecasts were computed on a 4-processor R10000 SGI Power Challenge compute server located at the Pennsylvania State University Department of Meteorology, and involved close cooperation between scientists at Penn State, the North Carolina Supercomputing Center, and SGI/Cray Research.
Preliminary forecast results from particularly interesting air pollution events which occurred during summer 1998 will be presented. Based on observed data, forecast successes and failures will be overviewed, and a discussion of possible causality and insights gained from the forecast experience will be reviewed.
The achievement of numerical, as opposed to empirical, statistical, or human-insight-based forecasts of lower tropospheric ozone has vast potential for the future. We mention a few of the important benefits here. First, it provides the ability to improve our understanding of the coupled atmospheric chemical-physical-dynamical system at a more rapid pace than in previous decades.
Real-time simulation enables a unique perspective and daily opportunity to gain scientific insight that cannot be achieved in any other fashion.
Second, it invites a more robust, critical dialogue and cooperation between atmospheric chemists, air pollution meteorologists, and atmospheric physicists and dynamicists. Third, it provides an avenue for the adaptation of emerging numerical weather prediction methods -- such as 3-D and 4-D VAR, adjoint modeling, ensemble forecasting, and adaptive observation, into numerical air quality prediction. Fourth, it provides a real groundwork for the close coupling between the multiple models that will be required to perform global-mesoscale climate simulations. These multiple coupled models will be required in order to simulate the projected impact of anthropogenic emissions of both organic and inorganic materials into the environment, and their potential deleterious or beneficial effects on climate.
Finally, it ushers in a new generation of forecasting tools. Modular, multiscale models such as MAQSIP can now be coupled to meteorological models ranging from the high resolution meso-gamma scale to lower resolution regional or global climate models, with a range of scales in between. As they mature,such tools can and will be used to forecast air quality in a reliable fashion. They will then gain acceptance as the front line forecasting technique used to provide early warning of poor air quality to susceptible populations.
Symposium on Interdisciplinary Issues in Atmospheric Chemistry