REAL-TIME SIMULTANEOUS PREDICTION OF AIR POLLUTION AND WEATHER DURING THE HOUSTON 2000 FIELD EXPERIMENT

摘要

The complexity of accurately forecasting changes in the large number of three-dimensional, nonlinear variables needed for weather prediction has challenged meteorologists for decades. Even more difficult is the prediction of air quality, as this requires not only an accurate weather forecast, but also knowledge and prediction of emissions, deposition, and chemical reactions of the compounds that are envolved. In addition, three-dimensional observed meteorological fields are used to initialize and verify meteorological models on a daily basis. No such option exists for air pollution applications, where three-dimensional observations are mostly restricted to field experiments. Only some surface observations of a few compunds are taken on a regular basis. Hence, the Houston 2000 field experiment was a great opportunity to not only apply complex modeling systems in real time, but also to verify results and evaluate the quality of the forecasts. Since the relevant processes contributing to photochemical air pollution are highly complex, non-linear and coupled, numerical simulation models are indispensable tool for diagnostic and prognostic analyses, and can therefore serve other scientists (such as observationalists or chemists) in diagnosing and analysing obser-vations. Most of the 'state of the art' modeling systems that are being used consist of several parts, including horizontal and vertical advection of tracers, physical processes that effect the tracers (such as turbulent mixing, convection, clouds, and radiation), and a chemical mechanism that describes the gas phase chemical interactions. The chemical processes are usually treated independently of the meteorological model, except that the transport is driven by output from a meteorological model. Due to this separation of meteorology and chemistry there is a loss of possibly important information of atmospheric processes that quite often have a time scale of much less than one hour, e.g. wind speed and direction, rainfall and cloud formation. Especially on the regional scale with gridsizes down to 1 km. the wind field and other meteorological parameters are highly variable and neglecting these variances may introduce certain errors. The modeling system that was used in this study does not separate the meteorology and chemistry, but integrates the meteorological and chemical processes simultaneously ('online'). It represents the most consistent way to couple the meteorology and chemistry.