This study intorduces the Southwest Monsoon Experiment (SOWMEX) in 2007 around Taiwan first. We then focus on the weather diagnosis of a 6-day stationary front case which caused local flood events. The weather Radar network, geostationary satellites images, dropsonde from Gulfstream-100 jet aircraft, balloon radiosonde and Viasala total lightning detection system (TLDS) were coordinated to discuss the movement of thunderstorm cells during these flood events.

The six morning flights with droposnde launched provided an observed wind and moisture cross section of 500-km width on the frontal system. They showed that the deep humid air with low-level westerly jet played the important role for inducing the thurderstorms developed. The numerical model simulations revealed a meso-scale cyclone genesis for several days by favorite synoptic factors, such as low-level wind-share line and upper-air trough line. But the onshore multi-path of thunderstorms every morning could not be simulated well from these mesoscale models.

The geolation mapping between TLDS lightning data and weather Radar echo had good relationship, but TLDS could be used to trace higher time resolution of strom movements than Radar. TLDS also gave precise information for flight safety warning in SOWMEX 2007. But we found TLDS has system bias at south corner of Taiwan nonregularly. The remove of TLDS bias is uner studied.