Some thoughts on the evaluation of long-range dispersion models in the context of emergency response.

摘要

The Chernobyl reactor accident and the possibility of a similar event of equal or greater magnitude in the future suggest the need for long-range atmospheric transport and diffusion models. Assuming their ability to accurately predict the temporal and spatial distributions of a complex mixture of radionuclides, these models can provide indispensable information to emergency planners and responders as well as information to a concerned and increasingly well-informed public. Models designed for use in real-time, emergency response situations can be evaluated on the basis of accuracy, responsiveness, costs, and growth potential. Among these criteria, accuracy is paramount; without the ability to accurately determine the patterns of concentration in both space and time, all other critical measures of performance become irrelevant. The authors are in the process of evaluating the accuracy of two long-range, emergency response models -- one based on the particle-in-cell method of pollutant representation (ADPIC/US) and the other based on the superposition of Gaussian puffs released periodically in time (EXPRESS/Russia) -- using data obtained from 20 releases of perfluorocarbon tracers in January, 1987, during the Across North America Tracer Experiment (ANATEX). The purpose is to assess the current capabilities for stimulating continental-scale dispersion processes and to use these assessments as a means to improve the modeling tools. The model evaluations discussed here utilize performance measures tailored to the ANATEX field experiment. The design of a sampling network, programmatic requirements and resource limitations associated with a field experiment affect the choice of evaluation techniques. Several appropriate measures, many of which depend on the characterization of surface concentration (open quotes)footprints,(close quotes) have been identified or devised.