Effect of Improved Accuracy of Aircraft Reconnaissance Data in the Initialization of a Two-Dimensional Hurricane Model,

摘要

On the basis of a limited number of experiments, it appears that a significant improvement in a two-dimensional hurricane model's behavior results when the small errors associated with the proposed AWRS system are used instead of a system with temperature and wind errors about five times those of the AWRS system. The smaller errors associated with the new AWRS system are much less damaging; and the observations appear to be accurate enough to provide reasonably useful numerical forecasts, at least with this model. The greatest effects of the dynamic initialization, both positive and negative, occur when the observations are added at the middle tropospheric level rather than the boundary layer. When the observations are added in the boundary layer alone, the model storm has not adjusted very closely to the observation after 12 hours of dynamic initialization and little difference between the two sets of observations exists after 24 hours of forecast time. In contrast, when the observations are added to the middle tropospheric layer, the observations affect much more mass of the system and the hurricane adjusts much more rapidly. In this case, there is little imbalance after 12 hours of initialization and the differences in the model prediction persist for the duration of the forecast.