| Summary: | 碩士 === 國立中央大學 === 大氣科學學系 === 107 === The complex terrain in Taiwan area makes it more challenging to forecast convection initiation, intensification, and propagation. In this research, the heavy rainfall event occurring on 19 August 2014 in northern Taiwan is selected. We use a newly-developed four-dimensional variational Doppler radar assimilation system (IBM_VDRAS), which is capable of simulating the topographic effect by adopting the so-called Immersed Boundary Method, and assimilating radar observations and surface station data. The products of IBM_VDRAS are a series of frequently-updated three-dimensional analysis fields over the complex terrain. In this case study, a total of eight analysis fields times are generated with a temporal interval of 17.5 min over a period of 2.5 h.
From the surface observations and the high temporal/spatial resolution analysis fields generated by IBM_VDRAS, it is found that the rainfall process started with the initiation of individual convective cells. The outflow of one of the convective cells merged with another convective system and helped to intensify the latter. The intensified major convective cell then moved into the Taipei metropolitan area and produced 80 mm of heavy precipitation within 2.5 h. The role played by the topographic forcing on the development of the convective system is investigated. A series of experiments are also designed and conducted by moving out terrain or surface assimilated variables to examine the performance of IBM_VDRAS in short-term rainfall forecasts. The result shows that SMR prevents the outflow from propagating southward, and LKHL and MTYM increase the outflow propagation speed. The surface wind assimilation improves the QPF skill by correcting the wind speed bias and controlling the magnitude of low-level convergence.
|
|---|
