Air Quality Modeling
7.6 Relevant Applications
7.6.6 Conclusion
Successful model applications require careful planning and execution. First and foremost is the necessity for understanding the problem, in terms of both the health impacts and the atmospheric science. This includes knowing which pollutant or pollutants are important, and what measure of the pollutants is appropriate.
Modeling long-term effects, as opposed to studying short term episodes, presents difficulties. Apart from the additional input data requirements, running a model for a year takes 30 times as long (in computer time) as it does to run a 12-d episode. When this is compounded over the study of a number of scenarios, the cost quickly becomes prohibitive. Advances have been made in aggregation schemes, which assemble a long-term average from a suitable combination of representative short-term episodes, but further development is required.
Clear problem definition amounts to generation of a conceptual model of the situation. Based on this conceptual model, the most appropriate mathematical modeling tool can be chosen. Considerations that enter into the choice of model or modeling system include matching the model chemistry to the pollutants of concern, and ensuring that the spatial scale of the model is appropriate.
Along with the model, an appropriate infrastructure is also essential. This will include all of the appropriate input data, a suitable computing platform, and skilled modelers to carry out the runs, and provide quality assurance and interpretation. These runs will involve scenarios, usually based on altered emissions, but possibly including other changes, such as different treatments of chemistry in the model, or altered meteorological conditions.
Model results will be scrutinized, analyzed, and interpreted, before being presented to the policymakers. If at all possible, the analysis should include incorporation of measurements. Above all, we make a plea for the involvement of all parties— people who make measurements, people who run models, and people who make policy—at all stages of the process. If all of these steps are carried out, the guidance provided by atmospheric models to the policy development process can be used with confidence.
Modeling long-term effects, as opposed to studying short term episodes, presents difficulties. Apart from the additional input data requirements, running a model for a year takes 30 times as long (in computer time) as it does to run a 12-d episode. When this is compounded over the study of a number of scenarios, the cost quickly becomes prohibitive. Advances have been made in aggregation schemes, which assemble a long-term average from a suitable combination of representative short-term episodes, but further development is required.
Clear problem definition amounts to generation of a conceptual model of the situation. Based on this conceptual model, the most appropriate mathematical modeling tool can be chosen. Considerations that enter into the choice of model or modeling system include matching the model chemistry to the pollutants of concern, and ensuring that the spatial scale of the model is appropriate.
Along with the model, an appropriate infrastructure is also essential. This will include all of the appropriate input data, a suitable computing platform, and skilled modelers to carry out the runs, and provide quality assurance and interpretation. These runs will involve scenarios, usually based on altered emissions, but possibly including other changes, such as different treatments of chemistry in the model, or altered meteorological conditions.
Model results will be scrutinized, analyzed, and interpreted, before being presented to the policymakers. If at all possible, the analysis should include incorporation of measurements. Above all, we make a plea for the involvement of all parties— people who make measurements, people who run models, and people who make policy—at all stages of the process. If all of these steps are carried out, the guidance provided by atmospheric models to the policy development process can be used with confidence.
