A real-time uncertainty-knowledge and training database

The paper describes an experimentally obtained database for training of uncertainties and data interpretation in connection with local scale accidental atmospheric dispersion scenarios. Based on remote measurement techniques using lidars, sequential 'snapshots', or movies, of the fluctuating concentration, profiles during several full scale diffusion experiments have been obtained. The aim has been to establish data sets suitable for comparison and training with the real-time atmospheric dispersion models in decision support systems, such as the RODOS system under development within the CEC. (author)