Taiwan is extremely vulnerable to flash floods resulting from heavy rain produced by local thunderstorms. These warm season thunderstorms typically occur in a weak synoptic environment by interactions among convergence boundaries (e.g. sea breeze fronts, gust fronts, mountain/valley circulations, etc) and Taiwan's complex terrain. NCAR is collaborating with the Taiwan Central Weather Bureau (CWB) to develop a short–term heavy rainfall nowcasting system to be used by Taiwan forecasters for providing township rainfall forecasts. It is critical for the Central Weather Bureau (CWB) forecasters to be able to issue short–term forecasts (or nowcasts) on where thunderstorms will initiate, their likely motion and dissipation, and rainfall rate.

RAL staff will begin transferring the NCAR thunderstorm nowcasting (Auto_Nowcaster; Mueller et al., 2003) system to Taiwan's CWB in February 2010. This will be a two–year effort that will include research into the evolution of heavy rainfall events in Taiwan and the tuning of this system for Taiwan's locale using high resolution datasets collected during the Terrain–influenced Monsoon Rainfall Experiment (TiMREX ) in 2008. Collaborative research between forecasters and scientists at CWB, Taiwan Universities and NCAR is underway to develop means for forecasting thunderstorm initiation over Taiwan. This collaborative effort is supported by the CWB and through the NCAR STEP program. The lower right–hand panel in Fig. 1 below shows Taiwan's complex terrain. This figure also demonstrates the importance of wind direction on the location of thunderstorm initiation over complex terrain. On each of these three days during TiMREX the wind direction is almost exactly the same from a direction of 200 degree. Wind velocity plays a role in the intensity of convection that results due to winds impinging on the terrain.

Figure 1.

The 4–D Variational Doppler Radar Analysis System (VDRAS; Sun and Crook; 1997; Sun 2005) is being used to assimilate radar reflectivity and velocity data from Taiwan into a mesoscale numerical model STEP project. Output fields from VDRAS include high resolution, 3–D wind fields in convective boundary layer and temperature perturbation fields as shown in Fig. 2 below. These fields will be critical to understanding the flow regimes and thermodynamics associated with the start of intense rainfall events. The VDRAS system will be installed at Taiwan's CWB during 2010.

Figure 2. VDRAS analysis on a domain over southern Taiwan. (a) Horizontal temperature gradient (color shading), 25 and 35 dBZ reflectivity (black contour), and wind vector at the height of 0.187 km above ground at 0258 UTC on May 31, 2008. (b) Same as (a) but at 0715 UTC. (c) Same as (a) but for vertical velocity (color shading) at the height of 1.3 km. (d) Same as (c) but at 0715 UTC. The 200 m terrain height is outlined by cyan contour in (a) and (b) and black contour in (c) and (d)

During TiMREX, radiosondes were launched every 3 hours, providing a dense set of measurements prior to and during heavy rainfall events. Hourly gridded CAPE fields, such as those shown below, have been produced using this unique sounding dataset and a network of over 100 surface stations. Additional factors important to the timing of convection initiation include the warming of the atmosphere to convective temperature and monitoring the levels of instability of the atmosphere. Figure 3 below illustrates the change in Convective Available Potential Energy (CAPE) observed during the afternoon over Taiwan on 8 June 2008 from 8:00 (upper panel) to 16:00 (lower panel) local time; a day when the convection was locally driven (i.e., non–synoptically triggered). Taiwan has an abundance of boundary layer moisture available for convective activity. Thus, prediction of intense convection is more sensitive to reaching convective temperature and to the presence of dry layers aloft. Radar storm echoes present at 8:00 and 16:00 overlaid onto these images. The most intense convection occurs over the higher terrain in regions of highest CAPE and where convective temperature was already reached.

Figure 3.