| Summary: | 碩士 === 國立中央大學 === 大氣科學學系 === 105 === In this study, the McGill Algorithm for Precipitation nowcasting using Lagrangian Extrapolation (MAPLE) system developed by McGill University is used to evaluate the very short-term forecast of precipitation in Taiwan. There are two procedures in MAPLE system. First, the motion field of precipitation is determined by Variational Echo Tracking (VET) technique. Second, a radar map is advected by means of a semi-Lagrangian backward advection scheme to be a precipitation forecast.
By using composite radar data from Central Weather Bureau (CWB), three real cases are selected for this study: a typhoon event occurred during 30-31 August 2010 (Namtheun typhoon); a Mei-Yu front case occurred during 11-12 June 2012; and a typhoon event occurred during 21-23 July 2014 (Matmo typhoon). Firstly, sensitivity tests between two parameters--vector density of motion field and time difference between each radar map in VET procedure is to determine the best setting of VET parameters for forecast. Secondly, the Quantitative Precipitation Nowcasting (QPN) would be evaluated. In general, forecasts by MAPLE system provide a certain degree capability about values and spatial patterns of the accumulated rainfall except for some rapidly growth and decay regions. Thirdly, the correlation of the VET motion field and the ECMWF reanalysis wind field is examined. It is found that most motion fields have a good relation with wind fields at each level in typhoon events, because the horizontal wind structure at one level is similar to the others. Finally, we focus on those VET motion fields that rainfall echoes are moving into Taiwan. The decrease of motion speed over Taiwan area is a good sign for extrapolation forecast, because it captures the blocking effect by orography to a certain extent. Must pay attention that main reason for rapid decrease in forecast ability is lack of growth and decay mechanism.
|
|---|
