This study investigates a long-duration mesoscale system with extremely heavy rainfall over southwest Taiwan during the Terrain-influenced Monsoon Rainfall Experiment (TiMREX). This mesoscale convective system develops offshore and stays quasi-stationary over the upstream ocean and southwest coast of Taiwan. New convection keeps developing upstream offshore but decays or dies after moving into the island, dropping the heaviest rain over the upstream ocean and coastal regions. Warm, moist, unstable conditions and a lowlevel jet (LLJ) are found only over the upstream ocean, while the island of Taiwan is under the control of a weak cold pool. The LLJ is lifted upward at the boundary between the cold pool and LLJ. Most convective clusters supporting the long-lived rainy mesoscale system are initiated and develop along that boundary. The initiation and maintenance is thought to be a ''back-building-quasi-stationary'' process. The cold pool forms from previous persistent precipitation with a temperature depression of 28-48C in the lowest 500 m, while the high terrain in Taiwan is thought to trap the cold pool from spreading or moving. As a result, the orography of Taiwan is ''extended'' to the upstream ocean and plays an indirect effect on the long-duration mesoscale system.

Dual-Doppler, polarimetric radar observations and precipitation efficiency (PE) calculations are used to analyze subtropical heavy rainfall events that occurred in southern Taiwan from 14 to 17 June 2008 during the Southwest Monsoon Experiment/Terrain-Influenced Monsoon Rainfall Experiment (SoWMEX/TiMREX) field campaign. Two different periods of distinct precipitation systems with diverse kinematic and microphysical characteristics were investigated: 1) prefrontal squall line (PFSL) and 2) southwesterly monsoon mesoscale convective system (SWMCS). The PFSL was accompanied by a low-level front-to-rear inflow and pronounced vertical wind shear. In contrast, the SWMCS had a low-level southwesterly rear-to-front flow with a uniform vertical wind field. The PFSL (SWMCS) contained high (low) lightning frequency associated with strong (moderate) updrafts and intense graupel-rain/graupel-small hail mixing (more snow and less graupel water content) above the freezing level. It is postulated that the reduced vertical wind shear and enhanced accretional growth of rain by high liquid water content at low levels in the SWMCS helped produce rainfall more efficiently (53.1%). On the contrary, the deeper convection of the PFSL had lower PE (45.0%) associated with the evaporative loss of rain and the upstream transport of liquid water to form larger stratiform regions. By studying these two events, the dependence of PE on the environmental and microphysical factors of subtropical heavy precipitation systems are investigated by observational data for the first time. Overall, the PE of the convective precipitation region (47.9%) from 14 to 17 June is similar to past studies of convective precipitation in tropical regions.

A multiple-Doppler radar synthesis method is developed to recover the three-dimensional wind field. In this method the solutions are obtained by variationally adjusting the winds to satisfy a series of constraints in weak formats. Among them, the primary ones are multiple-radar radial velocity observations, the anelastic continuity equation, the vertical vorticity equation, the background wind, and spatial smoothness terms. The retrieved wind products are at two time levels, and can be readily applied to deduce the information about the pressure and temperature through the use of the thermodynamic retrieval algorithm, in which the temporal derivatives of the wind fields are required.Experiments using model-simulated data are conducted, from which two major findings are obtained. First, the wind field along and near the radar baseline can still be recovered. This is a major advantage over the traditional approach. Therefore, the proposed method is capable of providing uninterrupted observations of a weather system as it passes the baseline. This allows for more flexibility when designing the radar deployment in field experiments. Second, if the winds are applied to infer the thermodynamic fields using the traditional dynamic retrieval method, and an extra sounding (e.g., radiosonde or dropsonde) is combined in order to specify the horizontal average of the thermodynamic perturbations, the preferable place to release this sounding is within the region of weak, rather than strong, convection.In additional to the aforementioned findings, with this method there is no need to prescribe the top or bottom boundary conditions for the vertical velocities in the traditional sense. Since the computation is performed without explicit vertical integration of the continuity equation, the problem of error accumulation due to inappropriate boundary conditions for the vertical velocities is prevented. These finding are consistent with some previous publications. Furthermore, the instability that occurs during traditional iterative dual-Doppler wind synthesis based on a Cartesian coordinate can also be avoided. Finally, data from any number of radars can be easily added to the computation.This method is also tested using the radar datasets collected during the Southwest Monsoon Experiment/ Terrain-Influenced Monsoon Rainfall Experiment (SoWMEX/TiMREX), which was conducted from May to June 2008 in Taiwan, and reasonable results are obtained.

A B S T R A C TThis study develops a Doppler radar data assimilation system, which couples the local ensemble transform Kalman filter with the Weather Research and Forecasting model. The benefits of this system to quantitative precipitation nowcasting (QPN) are evaluated with observing system simulation experiments on Typhoon Morakot (2009), which brought record-breaking rainfall and extensive damage to central and southern Taiwan. The results indicate that the assimilation of radial velocity and reflectivity observations improves the three-dimensional winds and rain-mixing ratio most significantly because of the direct relations in the observation operator. The patterns of spiral rainbands become more consistent between different ensemble members after radar data assimilation. The rainfall intensity and distribution during the 6-hour deterministic nowcast are also improved, especially for the first 3 hours. The nowcasts with and without radar data assimilation have similar evolution trends driven by synoptic-scale conditions. Furthermore, we carry out a series of sensitivity experiments to develop proper assimilation strategies, in which a mixed localisation method is proposed for the first time and found to give further QPN improvement in this typhoon case.