A nowcasting tool for the evolution of convective cells using the water vapor absorption and infrared window channels of the Meteosat Second Generation

摘要

At the Italian Air Force Meteorological Service a neural network model (NN) was defined in order to forecast the convective systems evolution in the Mediterranean area. This model, composed by a system of NNs, uses combination of water vapour absorption (WV) and infrared window (IR) data of Meteosat Second Generation (MSG). We realized that cloud top temperature, from IR window channel, does not give enough information to forecast the evolution of convective systems. As a consequence we introduced information about middle troposphere humidity content, from water vapor absorption band. We had preliminary results using the Meteosat rapid scan (RS) data. The use of WV and IR data from Meteosat-6 RS service, with a time sampling of 10 minutes, allowed us to track satisfactorily the evolution of convective cells and improved the detection of the beginning of the cell life. We can say that information of IR channel temperature only is not enough, for example, to evaluate the dissolving phase of the convective cell. A small decrease of the cloud top temperature (detected in the IR channel) it is not a unique indication for the beginning of that phase. It is known that, during mature phase, a convective cell may have a pulsating behaviour, so its top increases and decreases for an unknown time interval. After having defined two main evolution phases on the base of the features deduced from IR and WV channels, a specific NN algorithm was set up for nowcasting convective cells, using first RS data and then MSG data. A statistical analysis of cross-correlation between time series of different channels was performed for different areas of the Mediterranean region. From these statistics we may conclude that the performance of the NN system is more than satisfactory. This allows us to improve the operational automatic nowcasting application with the insertion of a NN module which gives information on the evolution of convective systems. In this way the forecasters are able to evaluate the probability of an increase or decrease of the severe convective activity.