1. Introduction
Warm-season precipitation in Taiwan is mainly contributed by mei-yu fronts and tropical cyclones (TC). Besides the short-period heavy precipitation of TC, the relatively continuous rain brought by the mei-yu fronts, which usually occurs from mid-May to mid-June (the so-called mei-yu season), is often associated with the strongest average rain intensity in a year, especially in southern Taiwan (e.g., Ding 1992; Chen et al. 1999; Chen and Chen 2003; Ding and Chan 2005; Henny et al. 2021). During mei-yu seasons, various weather systems at different scales come into play and can exert complex interactions with each other (e.g., Hirasawa and Yasunari 1990; Choi and Moon 2012; Huang et al. 2019; Tai et al. 2020; Yao et al. 2020), potentially leading to nonlinear variations in precipitation. The mei-yu front is a nearly stationary convergence zone between different air masses, with significant moisture and potential temperature gradients on both sides of the front (Lin et al. 1992; Li et al. 1997). Low-level jets (LLJ) with wind speeds stronger than 15 m s−1 at low levels (e.g., 700–850 hPa) are frequently observed south of the front and play an important role in heavy rain by transporting warm and moist air from the tropical ocean to the frontal zone (Chen and Yu 1988; Li et al. 1997; Chen et al. 2005). When this warm, humid air is lifted along the front, abundant rainfall occurs near the front. Strong mesoscale convective systems (MCS) can also develop over the ocean on the south side of the front, and as they move inland, heavy rain may occur in Taiwan (Lau et al. 1988; Kuo and Chen 1990; Zhang et al. 2003; Chien 2015).
The southwesterly (SW) flow, forming under the influence of the East Asian summer monsoon, is another important weather system during mei-yu seasons (Chien and Chiu 2019; Chien et al. 2021). Similar to the LLJ (e.g., Lin et al. 1992; Li et al. 1997; Tu et al. 2019; Chen et al. 2022) but operating on a broader scale, the SW flow significantly influences the intensity and distribution of precipitation in Taiwan during mei-yu seasons (e.g., Li et al. 1997; Chen and Yu 1988; C.-S. Chen et al. 2005; G. T.-J. Chen et al. 2008; Chien and Chiu 2019; Chien et al. 2021; Wang et al. 2022; Chen et al. 2022). The SW flow can act like an atmospheric river (Ralph et al. 2011; Paltan et al. 2017; Ralph et al. 2019), exhibiting pronounced moisture flux that transports warm and humid air northward. It can also play a role in sustaining and enhancing the frontal systems (Tu et al. 2020; Chien et al. 2021; Chen et al. 2022; Volonté et al. 2022). Chen et al. (2022) pointed out that the moisture brought by the SW flow from the southwest is transported to the frontal zone where the moist air is lifted, releasing latent heat and further intensifying the front. The intensification of the front increases the pressure gradient, forming the marine boundary layer jet (MBLJ) that provides more moisture to the frontal zone, creating a positive feedback and further enhancing precipitation. Additionally, strong SW winds of the SW flow can slow down the southward movement of the mei-yu front, resulting in more rainfall in the area of frontal influence (Wang et al. 2022; Chien and Chiu 2023). A positive feedback of moisture transport by the SW flow can lead to more fronts, stronger SW flow, and increased rainfall during the mei-yu season (Chien et al. 2021).
The interaction between low pressure systems on the northwest side of Taiwan and the SW flow can also play a crucial role in Taiwan’s precipitation. For example, when a TC moves to the northwest of Taiwan, it increases the pressure gradient near Taiwan, further strengthening the SW flow (Tao et al. 2011; Chien and Chiu 2021). This enhanced SW flow, coupled with the vast amount of moisture brought by the TC, creates favorable conditions for heavy precipitation in Taiwan. The southwest vortex (SWV), which is a low-level mesoscale cyclonic system forming on the leeward side of the Tibetan Plateau (e.g., Kuo et al. 1988; Feng et al. 2016), is another low pressure system that can enhance the strength of the SW flow when moving to the northwest of Taiwan, thereby greatly affecting the precipitation intensity in Taiwan (Chien and Chiu 2023).
In addition, terrain also plays a vital role in influencing Taiwan’s precipitation. The topography of Taiwan can greatly add to the complexity by modifying the airflow that has already been complexified by the interaction among the aforementioned weather systems. For the most part, terrain significantly affects the location of convective systems, consequently influencing the intensity and spatial distribution of precipitation in Taiwan (Chen and Yu 1988; C.-S. Chen et al. 2005; G. T.-J. Chen et al. 2008; Xu et al. 2012; Chien 2015; Sever and Lin 2017; Tai et al. 2020; Wang et al. 2021). Idealized experiments for the SW flow during mei-yu seasons by Wang et al. (2022) illustrate that when the wind speed of the SW flow exceeds 12.5 m s−1, precipitation is primarily dominated by terrain lifting of low-stability air. Wind directions favorable for heavy rainfall range from southerly (180°) to westerly (270°), with the most intense precipitation occurring when the wind direction is between 240° and 255°.
Chien and Chiu (2023) revealed a significant heavy rainfall event (hereafter referred to as Y20R1) that occurred in Taiwan between 21 and 23 May 2020. Observations of 48-h accumulated rainfall (Fig. 1) showed two rain hotspots: one over the plain and mountainous areas in southern Taiwan, and the other mainly over the central mountainous region, while the eastern coastal areas received less rain. A large area in southern Taiwan observed rainfall exceeding 330 mm in the 2-day period, with the maximum rainfall of ∼900 mm occurring in the southern mountainous region. During this time period, an SWV moving out from South China passed through northern Taiwan, enhancing the SW flow and causing the front to stall in the southern regions of Taiwan (Fig. 2a), consequently influencing the precipitation distribution during the Y20R event. Compared to the mean atmospheric conditions (hereafter Y20; Figs. 2b,c) in the first half of the mei-yu season (15–31 May, hereafter FHMY) in 2020 in which there was no SWV, this SWV not only strengthened the SW winds, but also increased the low-level moisture near Taiwan, especially on the southwestern side. The combination of these factors created a favorable environment for strong precipitation. Additionally, the environmental conditions during FHMY of 2020 (Y20) were exceptional. Chien and Chiu (2023) illustrated that the western North Pacific subtropical high (PSH) was particularly strong and extended southwestward toward the Indochina Peninsula during this time period, leading to the increased geopotential height gradient from the South China Sea (SCS) to the northeast of Taiwan and the significant enhancement of the southwest monsoon near Taiwan (Fig. 2b). The extremely strong PSH was primarily affected by the interdecadal increase in the sea surface temperature of the tropical Indian Ocean (Takaya et al. 2020; Guo et al. 2021). Compared to the climatological mean (CM) of FHMY in 42 years (Fig. 2d), the strong southwest monsoon in Y20 not only intensified the SW flow near Taiwan, but also transported more moisture from the northern SCS and southern China, resulting in increased moisture and a conducive environment for heavy rain in the vicinity of Taiwan. Under the favorable conditions provided by the SWV and Y20, Taiwan experienced the extreme Y20R precipitation event.
The 48-h accumulated rainfall (mm) from rain gauge stations (black dots) in Taiwan ending at 0000 UTC 23 May 2020. The number in the lower-right corner denotes the maximum rainfall.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
(a) The mean 850-hPa geopotential height (contours; interval: 10 gpm), wind (vectors; m s−1), and specific humidity (color; g kg−1) of Y20R, using ERA5. (b) As in (a), but for Y20. (c) The differences of 850-hPa geopotential height (contours; interval: 5 gpm), wind (vectors; m s−1), and specific humidity (color; g kg−1) between Y20R and Y20, i.e., Y20R-Y20. (d) As in (c), but for the differences between Y20 and CM, i.e., Y20-CM. The color and vector scales of each panel are denoted in the right and the top right, respectively. Gray color indicates the area below topography.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
However, the individual degrees of the impact of the Y20 and SWV on the precipitation and whether there were interactions between the two systems, and perhaps with others, remain unknown. Therefore, this paper aims to investigate their roles and the roles of the possible interactions in the Y20R precipitation event in a more quantitative way by conducting several ensemble experiments, and to answer the following key questions:
-
What favorable conditions do the Y20 provide and yet are absent in climatology?
-
What roles does the SWV play in heavy rain of Y20R?
-
What is the impact of the interaction between the Y20 and the SWV on precipitation?
The data sources, model settings, and experimental design are detailed in section 2. Section 3 presents the impacts of the SWV, the Y20, and their interactions on precipitation. The SW flow and moisture flux are discussed in section 4, and the SWV and the front are explored in section 5. The last section offers a summary and conclusions.
2. Data and experiment design
This paper utilizes the fifth generation of atmospheric reanalysis (ERA5; Hersbach 2016) data provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) during FHMY from 1979 to 2020 for climate statistics. The data have a horizontal spatial resolution of 0.5° × 0.5° and a temporal resolution of 6 h, resulting in a total of 2856 samples, i.e., 42 years × 17 days × 4 times per day. Observed precipitation data from Taiwan’s Central Weather Administration (CWA) rain gauge stations (marked as black dots in Fig. 1) were also used.
The WRF (Weather Research and Forecasting) Model version 4.1.2 (Skamarock et al. 2019) was employed in this paper for ensemble simulations based on Chien and Chiu (2023). The model setting includes three nested domains (Fig. 3a), with horizontal resolutions of 27, 9, and 3 km and at 45 vertical levels. A two-way interaction was applied between two adjacent nested domains. The Goddard scheme (Tao et al. 1989) was used for microphysics, and the Yonsei University scheme (Hong and Lim 2006) for boundary layer parameterization. During the data assimilation (DA) process, two cumulus parameterization schemes (CPS) were used: the Tiedtke scheme (Tiedtke 1989) and the Betts–Miller–Janjić scheme (Janjić 1994). However, only the Betts–Miller–Janjić CPS was used during the simulation period. The ERA5 data with a horizontal resolution of 0.25° × 0.25° and a temporal resolution of 6 h were used for the model’s initial (IC) and boundary conditions (BC), respectively.
(a) Domain setting of the model, which includes three nested domains at 27-, 9-, and 3-km horizontal resolutions. (b) The relationships among four experiments (Y1V1, Y1V0, Y0V1, and Y0V0). The definition and interpretation of the differences between two selected experiments (AYC, EY20, ESWV, EY20*, ESWV*, and RES) are explained in the main text. (c) Time frames of the ensemble simulations of Y1V1, Y1V0, Y0V1, and Y0V0. The ETKF DA is performed nine times (eight cycles) in domain 1 from 1200 UTC 18 May to 1200 UTC 20 May 2020. The 64 ensemble member runs in the three nested domains are initialized at 1200 UTC 20 May 2020 and end at 0000 UTC 23 May 2020.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
A 32-member ensemble was created using the ensemble transform Kalman filter (ETKF; Barker et al. 2012) DA that performed nine times (or eight cycles) at a 6-h interval during 1200 UTC 18 May–20 May 2020 (Fig. 3b), during which only domain 1 was run. The DA process, which followed the method developed by Bishop et al. (2001) and used the inflation factor strategy described by Wang et al. (2007), was repeated twice using the aforementioned two different CPS schemes to create a 64-member ensemble. The assimilated observational data included traditional observations from surface stations, soundings, aircraft, and ships. Subsequently, the 64 ensemble member runs were carried out in the 3 nested domains during the 60-h period from 1200 UTC 20 May to 0000 UTC 23 May 2020. This ensemble, which represents the real simulations of the Y20R event and is the same as the control (CTL) experiment in Chien and Chiu (2023), is referred to as Y1V1 in this paper.
The results of Chien and Chiu (2023) demonstrated that the ensemble simulations of Y1V1 captured the timing and spatial distributions of precipitation in southern Taiwan well. Moreover, they successfully reproduced the enhanced SW flow due to the eastward movement of the SWV and the interaction between the SW flow and the frontal system, helping the rain simulations in southern Taiwan. In this regard, Y1V1 can be utilized as a control experiment to be compared with other experiments in this paper.
To evaluate the impacts of the Y20 and the SWV on the Y20R event, this paper followed the method of Tai et al. (2020) and further conducted three sensitivity experiments (Table 1) by adopting different environmental states, including the state of the Y20 without the SWV (Y1V0), the climate state with the SWV (Y0V1), and the climate state without the SWV (Y0V0). In the control experiment (Y1V1), the initial conditions (ICs) of each member at 1200 UTC 20 May 2020 were obtained from the ensemble mean of the WRF inputs of the 64 members (EMWI52012) plus the perturbations of that particular member (ETKFptb) through the ETKF DA process. The boundary conditions (BCs) were acquired from the ERA5 from 1200 UTC 20 May to 0000 UTC 23 May 2020. In the Y1V0 experiment, the ICs were derived by adding the ICs of Y1V1 to the difference between the average WRF inputs of the Y20 (EMWIY20) and EMWI52012. In other words, all the background field variables of EMWI52012 were replaced by those of the Y20 (EMWIY20) such that this experiment contained the environmental state of the 2020 FHMY, but no SWV. The BCs of Y1V0 were constructed in a way to ensure that the diurnal cycle of the 2020 FHMY could be maintained. Specifically, the BCs at 0000 UTC used the average fields at 0000 UTC of all 17 dates during the 2020 FHMY. The same procedure was repeated for the times of 0600, 1200, and 1800 UTC; the BCs were then generated. In the Y0V0 experiment, the design was similar to Y1V0, but the mean state of the Y20 was replaced by the average state of the FHMY from the CM of 42 years. This replacement was designed to use the climate state for the background fields such that the Y20 were removed. Finally, in the Y0V1 experiment, the ICs and BCs were derived by adding Y1V1 to the difference between Y0V0 and Y1V0. In other words, this experiment was constructed to retain the SWV while removing the Y20.
Initial conditions (ICs) and boundary conditions (BCs) used in the 64-member ensemble runs of the control experiment (Y1V1) and the three sensitivity experiments of Y1V0, Y0V1, and Y0V0. See text for more details.
Since the four experiments contain 256 simulations, it was necessary to develop automatic methods for detecting the mei-yu front and the SWV. The former will be introduced in a later section. As for the automatic detection of the SWV center and intensity, the 850-hPa geopotential height fields from domain 1 of each simulation were used. The detection time span was the 48 h of Y20R, with a frequency of every six h, resulting in nine samples for each simulation. At each detection time, the ensemble-mean geopotential height of each experiment was used to manually locate the SWV center, and then a 30° longitude × 16° latitude rectangular detection area centered on the mean SWV center was defined. Within this area, each ensemble member was independently analyzed. We first computed the mean and standard deviation of the geopotential height for each simulation using all the grid points in the detection area. The region with the geopotential height below the areal mean minus one standard deviation was then defined as BSWV, representing the boundary of SWV. Next, the areal mean and standard deviation of the geopotential height in BSWV were calculated. Subsequently, within BSWV, the region where the geopotential height was below the new areal mean minus 1.5 standard deviations (of BSWV) was defined as CSWV, representing the central region of the SWV. Finally, the average longitude (latitude) of all grid points within CSWV was considered as the SWV center longitude (latitude). The SWV intensity was then defined as the difference between the areal-mean geopotential height within a radius of 50 km from the SWV center and the areal-mean geopotential height within the detection area. In other words, it was the 850-hPa geopotential height anomaly within a radius of 50 km from the SWV center. Since the SWV is a low pressure system, smaller values indicate stronger intensity.
3. Precipitation simulations
The ensemble-mean 48-h accumulated precipitation of Y1V1 (Fig. 4a) successfully captures the heavy rain in the southern and central mountainous regions of Taiwan as observed (Fig. 1), as well as the lower precipitation along the eastern coast. Although the maximum rain of the ensemble mean (740 mm) is slightly lower than that of the observations (906 mm), which is the basic characteristic of an ensemble-mean forecast (Ebert 2001), the ensemble of Y1V1 does perform well from the perspective of overall precipitation distribution. Additionally, Y1V1 exhibits relatively large standard deviation (ensemble spread) in the southern mountainous region (Fig. 4e), reaching a maximum of 157 mm. When adding this value to the maximum of mean precipitation (740 mm), the total becomes 897 mm, indicating that some ensemble members with larger precipitation may produce rainfall close to the observation. The other three sensitivity experiments (Figs. 4b–d) show that precipitation is, overall, underestimated over most areas, with only Y0V1 (Fig. 4c) showing a notable overestimation in the northeastern and eastern regions. The distribution of standard deviation (Figs. 4f–h) is similar to that of the mean precipitation (Figs. 4b–d), indicating a positive correlation between the two variables, i.e., higher mean precipitation regions also exhibit larger spread.
The ensemble-mean 48-h accumulated rainfall (mm) ending at 0000 UTC 23 May 2020 in (a) Y1V1, (b) Y1V0, (c) Y0V1, and (d) Y0V0. (e)–(h) As in (a)–(d), but for standard deviation (mm) in the ensemble. (i)–(l) As in (a)–(d), but for the exceedance probability (%) of hourly rain intensity greater than 10 mm h−1 in the ensemble at every full hour from 0000 UTC 21 May to 0000 UTC 23 May 2020. The number in the lower-right corner denotes the maximum in each panel. Boxes Ts and Te in (a) show the regions for areal-mean calculation.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
To gain a more intuitive understanding of the different precipitation characteristics across Taiwan, this study conducted quantitative analyses for three regions: the extended Taiwan region (all grid points in Fig. 4a, TW), southern Taiwan (box Ts in Fig. 4a), and eastern Taiwan (box Te in Fig. 4a). In the extended Taiwan region (Fig. 5a) and southern Taiwan (Fig. 5c), the areal-ensemble-mean 48-h accumulated precipitation of Y1V1 is significantly higher than those of the other three experiments. In eastern Taiwan (Fig. 5e), Y0V1 has notably larger rainfall than the other three experiments. Regardless of the region, the experiments with SWV (Y1V1 and Y0V1) consistently exhibit larger precipitation than the experiments without SWV (Y1V0 and Y0V0). In addition to accumulated precipitation, frequency of precipitation is also an important feature of a precipitation event. Therefore, the exceedance probability (%) of rain intensity greater than 10 mm h−1 in the ensemble of the four experiments at every full hour during 0000 UTC 21 May–0000 UTC 23 May 2020 was analyzed. The total number of samples evaluated in each experiment is 3136, i.e., 64 members × 49 times. Overall, the pattern (Figs. 4i–l) shows similarities to (or positive spatial correlations with) the mean accumulated precipitation distribution (Figs. 4a–d), suggesting that regions with higher accumulated precipitation are a result of accumulation from continuous rainfall rather than from intermittent extreme rainfall.
(a) The areal-ensemble-mean 48-h rainfall (colored bars; mm) ending at 0000 UTC 23 May 2020 for Y1V1, Y1V0, Y0V1, and Y0V0. Regions for the areal-mean include all grid points in Fig. 4a (TW). The range of one standard deviation from the ensemble is denoted by the error bars. The values of mean and standard deviation are also shown above the plot for the respective bar. (b) As in (a), but for AYC, EY20, ESWV, EY20*, ESWV*, and RES. (c), (d) As in (a) and (b), but for Ts region in Fig. 4a. (e),(f) As in (a) and (b), but for Te region in Fig. 4a.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
Figure 6 presents the exceedance probability of hourly rain intensity greater than various thresholds using samples at all the grid points of the three regions. Counting the 64 members and the 49 hourly times, the total amounts of samples evaluated in each experiment is 3136 times the number of grid points in each region (TW: 40663, Ts: 8214, Te: 4107). For the extended Taiwan region (Fig. 6a), across all thresholds, the ranking order of precipitation occurrence for the four experiments is nearly identical, from the highest to the lowest: Y1V1, Y0V1, Y0V0, and Y1V0. In southern Taiwan (Fig. 6b), Y1V1 stands out as having significantly higher probabilities compared to the other three experiments. In eastern Taiwan (Fig. 6c), Y0V1 consistently exhibits the highest precipitation probabilities across all thresholds. Below approximately 10 mm h−1, Y1V1 shows higher probabilities than Y0V0, while above this threshold, the situation is reversed. For Y1V0, the precipitation probabilities remain the lowest across all thresholds. Across all thresholds and regions, Y1V1 has a higher precipitation occurrence probability than Y1V0, and Y0V1 higher than Y0V0, indicating the significant contribution of the SWV to the occurrence of precipitation. Additionally, in all regions, Y1V0 exhibits the lowest probability among the four experiments for rain intensity exceeding 30 mm h−1; this result implies that an environment with strong SW flow conditions of the 2020 FHMY, but without SWV, reduces the probability of heavy rain and thus the accumulated precipitation. In contrast, the experiments with SWV consistently show larger total precipitation and higher probability of heavy rain occurrence compared to their counterparts without SWV. This finding indicates that in an environment with SWV present, the likelihood of heavy precipitation increases, resulting in larger total precipitation. Furthermore, the locations of precipitation are influenced by the large-scale background fields, with the Y20 favoring precipitation in southern Taiwan, whereas the climate conditions favor precipitation in eastern Taiwan.
(a) The exceedance probability (%; in logarithmic scale) vs hourly rain intensity (mm h−1) at all grid points in the region of TW (Fig. 4a) in all 64 ensemble members of Y1V1 (green), Y1V0 (red), Y0V1 (blue), and Y0V0 (black) at every full hour from 0000 UTC 21 May to 0000 UTC 23 May 2020. (b) As in (a), but for Ts region in Fig. 4a. (c) As in (a), but for Te region in Fig. 4a.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
To reveal the roles of the Y20 and the SWV in precipitation, Fig. 7 presents the spatial distributions of ensemble-mean 48-h accumulated precipitation of AYC, EY20, ESWV, EY20*, ESWV*, and RES. Almost throughout Taiwan, AYC has a positive precipitation anomaly (Fig. 7a). Particularly in precipitation-rich southern Taiwan, large areas exhibit a positive anomaly of over 200 mm and many mountainous regions even exceed 300 mm, with a maximum close to 700 mm. Both EY20 and ESWV (Figs. 7b,c) positively contribute to rainfall in the western and southern regions of Taiwan, with a pattern similar to that of AYC. In eastern Taiwan, EY20 makes a negative contribution to rainfall. These results underscore the notable influence of the Y20 and the SWV on precipitation in western and southern Taiwan. However, their counterparts (EY20* and ESWV*, Figs. 7d,e), in which only either Y20 or SWV exists in the simulations, show insignificant positive contribution to rainfall in western and southern Taiwan, resulting in large RES (Fig. 7f). It is thus evident that neither the Y20 alone nor the isolated influence of the SWV would suffice to generate substantial rainfall in southern Taiwan. In the eastern part of Taiwan, EY20* has small negative contributions (Fig. 7d), whereas ESWV* shows larger positive contributions (Fig. 7e).
The ensemble-mean 48-h rainfall (mm) differences of (a) AYC, (b) EY20, (c) ESWV, (d) EY20*, (e) ESWV*, and (f) RES from 0000 UTC 21 May to 0000 UTC 23 May 2020. Only the areas where the difference exceeds the 95% significance level are presented in color shading. The number in the lower-right corner denotes the maximum in each panel.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
Quantitative analyses were also conducted to assess the impact of the Y20 and the SWV on precipitation within the three regions of TW, Ts, and Te. In the extended Taiwan region (TW), AYC on average has positive rain anomaly of 118.1 mm (Fig. 5b). This amount of mean rain can be interpreted in three ways: EY20 plus ESWV*, ESWV plus EY20*, or the sum of EY20*, ESWV*, and RES. EY20 is about 63% of the AYC and ESWV is slightly larger than the AYC. This difference is also related to the positive ESWV* (approximately 37% of AYC) and the small negative value of EY20*. These results indicate that SWV is more important than Y20 in determining rainfall in the extended Taiwan region. RES is about 65% of AYC. In southern Taiwan (Fig. 5d), AYC has a positive anomaly of 230 mm. Both EY20 and ESWV contribute largely (about 87%) to this rain anomaly, while both ESWV* and EY20* have small positive values. As a result of large differences between EY20 and EY20* (or between ESWV and ESWV*), the RES term accounts for a high proportion (about 76%) of AYC. The above results demonstrate that both the SWV and the Y20 have significant impact on the rainfall and the interaction between the SWV and the Y20 (RES) plays a crucial role in Taiwan’s precipitation, especially in southern Taiwan. An atmosphere environment having either the SWV only or the Y20 only would not produce heavy precipitation as observed in Y20R. In eastern Taiwan (Fig. 5f), ESWV* is positive, while both the EY20* and RES terms are negative, leading to a positive ESWV and a negative EY20; this result suggests that rainfall in eastern Taiwan is primarily influenced by the SWV.
4. The SW flow and moisture flux
The next goal of this paper is to examine the causes of the precipitation differences among the experiments. According to Chien and Chiu (2023), precipitation of Y20R is mainly influenced by the SW flow intensity, the frontal location, the SWV location and intensity, and the distribution of moisture. Therefore, this paper first examines the roles of the SW flow by comparing the 2-day-mean low-level-averaged environmental fields of the ensemble mean among the four experiments during the Y20R event (Fig. 8). In Y1V1 (Fig. 8a), an SWV is passing the north side of Taiwan, while the PSH extends toward the Philippines, creating a pronounced geopotential height gradient near Taiwan. Consequently, Taiwan and its surrounding regions experience predominantly strong SW flow. As the SWV moves eastward during the 2 days, the strong SW flow with abundant moisture on the south of the SWV produce large moisture fluxes over the coastal areas of South China, the northern SCS, and the southwest of Taiwan. These favorable low-level environmental conditions led to the significant precipitation events of Y20R.
The 2-day-mean low-level-averaged (surface to 700 hPa) mixing ratio (color shading; g kg−1) and wind (vectors; m s−1), and 850-hPa geopotential height (contours; interval: 10 gpm) from the ensemble mean of (a) Y1V1, (b) Y1V0, (c) Y0V1, and (d) Y0V0 from 0000 UTC 21 May to 0000 UTC 23 May 2020. Boxes A and B in (a) denote the regions of areal mean.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
In Y1V0 (Fig. 8b), the PSH extends farther southwestward compared to that of Y1V1 (Fig. 8a), with the 1510-gpm isoheight approaching the region near Indochina. As a result, the monsoonal circulation in Y1V0 shifts to a more northward latitude, creating a SW wind belt extending from the Indochina Peninsula to the northeast of Taiwan. This environment, unlike that of Y1V1, is not favorable for frontogenesis near Taiwan as will be discussed in the next section. Although there is no SWV in Y1V0, SW flows are still present near Taiwan, but weaker than those of Y1V1. However, due to the absence of the moisture supply by the SWV, Y1V0 has much less moisture to the southwest of Taiwan, leading to less rainfall in southern Taiwan (Fig. 4b). Compared with Y1V1 (Fig. 8a), Y0V1 (Fig. 8c) has a more eastward position of the PSH and a more southwestward position of the SWV. Therefore, the monsoonal circulation on land is stronger in Y0V1, with its center positioned on South China. Influenced by this cyclonic circulation, the wind direction in southern Taiwan is SW, while in northern Taiwan, it is southeasterly, with overall smaller wind speed. Furthermore, as a result of a more southwestward location of the SWV, Y0V1 has less moisture near Taiwan than Y1V1. It can thus be inferred that Y0V1 exhibits fewer favorable conditions for precipitation in southern Taiwan but increased conditions in northeastern Taiwan, resulting in a rainfall pattern shown in Fig. 4c. In Y0V0 (Fig. 8d), both the monsoonal circulation to the northwest of Taiwan and the PSH to the southeast are weak, leading to a small geopotential height gradient and weak south-southwesterly winds over the northern SCS. Without the SW flow, moisture from South China cannot be transported to the area near Taiwan, resulting in little moisture in the vicinity of Taiwan. Overall, compared to other experiments, Y0V0 has significantly fewer favorable conditions for precipitation in Taiwan, making it less prone to rainfall (Fig. 4d).
The above results show that precipitation in this case is closely determined by moisture transport. Since the sources of moisture for southern Taiwan (box Ts) and northeastern Taiwan (box Te) differ, we need to analyze the mean upstream moisture flux in boxes A and B of Fig. 8a, respectively. Since the moisture flux is evaluated for the 64 ensemble members at every full hour from 0000 UTC 21 May to 2300 UTC 22 May 2020, each experiment contains 3136 samples. For Y1V1, moisture fluxes are mostly large and with a SW direction upstream of Ts (Fig. 9a) during the 48 h. Many samples have a moisture flux larger than 200 m s−1 g kg−1. Heavy rain samples (green dots) are mostly associated with west-southwesterly winds and large moisture flux, with rain intensity increasing generally with intensifying moisture flux. The larger probability of heavy rain for winds having a more westerly component is related to the orientation of the Central Mountain Range of Taiwan where westerly winds are more conducive to precipitation (Wang et al. 2022). Wind directions in Y1V1 exhibit large variations upstream of Te (Fig. 9b), resulting in significant differences in moisture transport directions among the samples. Nevertheless, there are still samples with large moisture flux toward the northeast, while samples with moisture flux having a westward component mostly show small magnitudes below 100 m s−1 g kg−1, resulting in only a few samples of heavy precipitation (green dots).
(a) The areal-mean (in box A of Fig. 8a) low-level-averaged (surface to 700 hPa) moisture flux (gray dots; m s−1 g kg−1) of the 64 ensemble members of Y1V1 at every full hour from 0000 UTC 21 May to 2300 UTC 22 May 2020. The total number of samples is 3136. The abscissa and ordinate represent the x and y components of the moisture flux, respectively. Green dots denote the sample in which the areal-mean rain intensity in Ts (Fig. 4a) during the 1 h after the sample time exceeds 10 mm h−1, with the dot size denoting rain intensity shown on the right and the number of green dots shown in the top right. (b) As in (a), but for the areal-mean region in box B (Fig. 8a) and the rain intensity in Te (Fig. 4a). (c),(d) As in (a) and (b), but for Y1V0. (e),(f) As in (a) and (b), but for Y0V1. (g),(h) As in (a) and (b), but for Y0V0.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
In Y1V0, moisture fluxes upstream of Ts (Fig. 9c) exhibit small variations in both the direction and magnitude among samples, with the direction being primarily toward the northeast. However, all the magnitudes are smaller than 200 m s−1 g kg−1, resulting in no occurrences of heavy precipitation. Upstream of Te, moisture fluxes are still mainly toward the northeast (Fig. 9d). None of the samples has a moisture flux exceeding 200 m s−1 g kg−1, leading to only one sample experiencing heavy precipitation (red dot). For Y0V1 (Fig. 9e), most samples upstream of Ts have south-southwesterly winds, and most of the moisture fluxes do not exceed 200 m s−1 g kg−1, resulting in only a few samples experiencing stronger precipitation under the influence of west-southwesterly winds (blue dots). Upstream of Te (Fig. 9f), most samples have moisture fluxes toward the west. Some of them have moisture fluxes exceeding 100 m s−1 g kg−1, leading to numerous occurrences of heavy precipitation (blue dots). In these samples, the rain intensity also tends to increase with increasing moisture flux. The heavy precipitation of Y0V1 occurs primarily in northeastern Taiwan (Fig. 4c) where airflow encountering the terrain is often a major factor of precipitation (Yeh and Chen 2004; Su et al. 2022). In Y0V0 (Fig. 9g), most samples upstream of Ts have larger southerly wind components, and none of their moisture fluxes exceeds 150 m s−1 g kg−1, resulting in no occurrences of heavy rain. Upstream of Te (Fig. 9h), moisture is mainly transported toward the northeast and northwest directions, and some heavy rain samples (black dots) are associated with moisture transported toward the northwest direction. Compared to those in other experiments, these samples have much smaller moisture fluxes probably because the primary region of heavy precipitation in Y0V0 occurs over the ocean (Figs. 4d,l), and box B is not its upstream.
Inter-comparisons of moisture fluxes among the four experiments show that in the southern region, experiments based on the Y20 (Y1V1 and Y1V0) produce winds with a larger westerly component than their counterparts, and experiments with the SWV present (Y1V1 and Y0V1) have in general larger moisture fluxes than their counterparts. In northeastern Taiwan, experiments based on the Y20 show more samples with SW winds and fewer samples with northeasterly winds, and experiments with the SWV present have more samples with northeasterly winds. These results overall suggest that the Y20 enhances the SW wind component near Taiwan, increasing (reducing) favorable conditions for precipitation in the southern (eastern) part of Taiwan. The presence of the SWV increases the geopotential height gradient and moisture near Taiwan, thereby enhancing moisture transport in southwestern Taiwan. Additionally, when the SWV gradually approaches Taiwan, the eastern portion of its circulation also influences the wind direction in the eastern part of Taiwan, increasing the eastward component of the wind and thus enhancing precipitation in the northeastern region.
As the moisture flux can be separated into moisture and winds, we further examined the effect of the SWV and the Y20 in the low-level moisture and wind fields (Fig. 10) and performed a quantitative analysis upstream of the Ts and Te regions (Fig. 11). The former is done first by computing 2-day mean of low-level-averaged (surface to 700 hPa) fields, whereas the latter is done by evaluating the same fields at every full hour of the 2 days from 0000 UTC 21 May to 2300 UTC 22 May 2020. In AYC (Fig. 10a), large positive moisture anomalies are present in the vicinity of Taiwan, with areal-mean moisture anomalies of 1.47 g kg−1 in box A (Fig. 11a) and 0.72 g kg−1 in box B (Fig. 11d). This surrounding area is also associated with negative geopotential height anomaly, with a minimum center located to the northeast of Taiwan. As a result, AYC has westerly to west-southwesterly wind anomalies in box A, with westerly more significant than southerly wind anomalies (Figs. 11b,c), whereas in box B, AYC has northeasterly wind anomalies, with northerly more significant than easterly wind anomalies (Figs. 11e,f). With both moisture and wind anomalies favorable for precipitation, AYC overall shows positive precipitation anomalies in Taiwan (Fig. 7a), particularly in the southern region (Fig. 5d).
The 2-day-mean low-level-averaged (surface to 700 hPa) mixing ratio (color shading; g kg−1) and wind (vectors; m s−1), and 850-hPa geopotential height (contours; interval: 10 gpm) differences of (a) AYC, (b) Y20, (c) SWV, (d) Y20*, (e) SWV*, and (f) RES from 0000 UTC 21 May to 0000 UTC 23 May 2020. Only the anomalies (or differences) exceeding the 95% significance level are presented. White color denotes the areas where the 95% significance level is not met for moisture. Boxes A and B in (a) are as in Fig. 8a.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
(a) The mean (colored bars) and standard deviation (error bars) of areal-mean (in box A of Fig. 10a) low-level-averaged (surface to 700 hPa) mixing ratio (color shading; g kg−1) differences of AYC, EY20, ESWV, EY20*, ESWV*, and RES. The number of samples is 3136, including 64 ensemble members and 49 hourly times from 0000 UTC 21 May to 0000 UTC 23 May 2020. The values of mean and standard deviation are also shown above the plot for the respective bar. (b),(c) As in (a), but for u and υ (m s−1), respectively. (d)–(f) As in (a)–(c), but for the areal-mean region in box B of Fig. 10a.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
In EY20 (Fig. 10b), the area of maximum positive moisture contribution is located far northeastward from Taiwan’s vicinity, resulting in a limited moisture contribution to the area surrounding Taiwan. On average, the moisture contribution of EY20 is 0.56 g kg−1 to the southwest of Taiwan (Fig. 11a), and 1.45 g kg−1 to the northeast (Fig. 11d). The geopotential height exhibits negative (positive) contributions on the northeast (south) side of Taiwan, leading to positive contributions to the SW winds extending from the northern SCS to the east of Taiwan, with the contribution to westerly winds (Figs. 11b,e) more significant than that of southerly winds (Figs. 11c,f). The SW winds cause an increase of moisture flux in southwestern Taiwan but a decrease in eastern Taiwan, leading to the precipitation pattern shown in Fig. 7b. Since AYC is the sum of EY20 and ESWV*, it is not surprising to see that ESWV* (Fig. 10e) has a large positive moisture contribution extending from South China, through the northern SCS, to the ocean southeast of Taiwan where the contribution from EY20 is small. Comparisons of these two terms suggest that the SWV contributes to the moisture increase, whereas the Y20 primarily contributes to the SW flow strengthening.
The positive moisture contribution of ESWV (Fig. 10c) is relatively smaller and shifts more southward compared to the moisture anomaly of AYC (Fig. 10a). On average, ESWV has a mean moisture contribution of 0.99 g kg−1 in box A (Fig. 11a) and 0.23 g kg−1 in box B (Fig. 11d). The geopotential height exhibits a large area of negative contributions extending from the northern SCS, through Taiwan, to the east of Taiwan, resulting in cyclonic circulations near Taiwan. In box A, ESWV positively contributes to westerly winds but insignificantly to northerly winds (Figs. 11b,c). In box B, ESWV provides strong northeasterly winds (Figs. 11e,f), leading to an increase in northeasterly moisture flux over ocean northeast of Taiwan. Consequently, ESWV makes a positive precipitation contribution not only in southern Taiwan (Figs. 7c and 5d), but also in northeastern Taiwan (Figs. 7c and 5f). The above moisture and wind patterns of ESWV can become the same as those of AYC after adding EY20* which contributes a large part to the SW flow strengthening and a small fraction to the moisture increase in the vicinity of Taiwan (Fig. 10d).
Except in the area to the northeast of Taiwan where positive moisture contributions associated with negative geopotential height and cyclonic circulation contributions are found, RES (Fig. 10f) is in general small. This small residual is due to the similar moisture and wind patterns between EY20 and EY20* and also between ESWV and ESWV*. As a result, the average RES is small in box A (Figs. 11a–c). The main contribution of moisture transport to AYC comes from ESWV* and EY20* in which ESWV* mainly contributes to the moisture increase, whereas EY20* primarily to the SW flow strengthening (similar to those shown in Figs. 10d,e). However, as shown in Fig. 7f, RES accounts for a major part of AYC in precipitation, particularly in southern Taiwan. The reason is because both the SWV and the Y20 alone (i.e., ESWV* and EY20*) cannot produce heavy rain as observed. For winds and moisture, however, the contributions from both the ESWV* and EY20* are significant, leading to the small RES in southern Taiwan. It is therefore suggested that the increased winds and moisture caused individually by the SWV and the Y20 provide favorable conditions for rainfall in Y1V1. However, if only one of these two exists, there would be little rainfall in southern Taiwan. Only when they come together do a nonlinear interaction between the SWV and the Y20 and large precipitation in southern Taiwan occur.
5. The interaction between the SWV and the Y20
The previous section showed that the nonlinear interaction between the SWV and the Y20 (RES) does not play a role in providing favorable conditions for rainfall upstream of southern Taiwan. There must be some other factors, caused by the interaction, that offer the assistance. For example, RES exhibits northerly flow in the Taiwan Strait north of box A (Fig. 10f), which may promote frontogenesis when the SWV and the Y20 coexist. This paper thus digs further into this issue by quantitatively analyzing the differences in the front and SWV activities among the experiments. The definition and detection methods of the SWV were already introduced in section 2. As for the front detection, the same approach of Chien and Chiu (2023) was adopted and is briefly described in the following; this method defined the latitude of the front based on the thermal and dynamic criteria, using the 1000-hPa horizontal wind convergence (∇ ⋅ V < −7.1 × 10−4 s−1) and virtual potential temperature gradient (|∇θυ| > 0.14 K km−1). As the first step, if both of these criteria are met at a grid point, the point is identified as a front point. The second step is to ensure that front points occur continuously with a relatively long distance to resemble a real front (see that paper for details). The intensity of the front was determined by the average |∇θυ| near the frontal region. The detection area was on the Taiwan Strait (21°–24.5°N, 118°–120°E; box C in Fig. 12b) in that paper as it primarily investigated precipitation in southern Taiwan. However, since the precipitation areas differed among the experiments in this paper, it was necessary to examine frontal activities on the eastern side of Taiwan (21°–25.5°N, 122°–124°E; box D in Fig. 12b) as well. Figures 12 (the rightmost column) and 13 present, for the four experiments, the percentage of front points and the mean latitude of the front detected in the 64 ensemble members at every full hour during 0000 UTC 21 May–0000 UTC 23 May 2020 (3136 samples), respectively.
The 1000-hPa (a) streamlines and horizontal divergence (color shading; 1 × 10−4 s−1), and (b) winds (m s−1), virtual potential temperature (θυ; contours; interval: 1 K), and horizontal θυ gradient (|∇θυ|; color shading; K 100 km−1) in the ensemble mean of Y1V1 at 0000 UTC 22 May 2020. The divergence and θυ gradient are computed in each individual member before an ensemble mean is taken. (c) The percentage (%) of front points detected in the 64 ensemble members of Y1V1 at every full hour from 0000 UTC 21 May to 0000 UTC 23 May 2020. The number of samples is 3136. (d)–(f),(g)–(i),(j)–(l) As in (a)–(c), but for Y1V0, Y0V1, and Y0V0, respectively. Boxes C and D in (b) denote the regions of frontal detection.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
(a) The appearance frequency (times, the abscissa) of mean latitude of the front (the ordinate) detected in the 64 ensemble members of Y1V1 (green), Y1V0 (red), Y0V1 (blue), and Y0V0 (black) at every full hour from 0000 UTC 21 May to 0000 UTC 23 May 2020. The number of samples is 3136 for each experiment. The interval of frontal latitude is 0.5°. Circles and crosses represent frontal detection in boxes C (west) and D (east) of Fig. 12b, respectively. (b) As in (a), but showing a close-up view for frequencies smaller than 20 times.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
Owing to the presence of the SWV in Y1V1 (Figs. 12a,b), Taiwan experiences significant low-level convergence on both sides due to the influence of strong SW monsoonal flow and the northeasterly flow. The convergence zone on the western side is at a more southern position than on the eastern side (Fig. 12a). The warm, moist air brought by the SW winds and the dry and cold air by the northeasterly winds create a significant virtual potential temperature gradient near the low-level convergence zone (Fig. 12b), highly conducive to mei-yu front formation (Fig. 12c). In the experiment without the SWV (Y1V0; Figs. 12d,e), strong southerly winds are present in the vicinity of Taiwan, with large convergence and virtual potential temperature gradient occurring near the coastline, indicating that the low-level environment near Taiwan is mainly influenced by the land-sea contrast, rather than mei-yu frontogenesis (Fig. 12f).
In Y0V1 (Figs. 12g,h), due to the weaker SW winds and the further southern extent of the northeasterly flow, compared to those of Y1V1, the favorable conditions for frontogenesis shift southward. The regions of a slightly larger virtual potential temperature gradient are located to the southwest and southeast of Taiwan. Therefore, the low-level environment in Y0V1 is less ideal for mei-yu front generation, except over ocean south to Taiwan (Fig. 12i). Y0V0 exhibits unfavorable frontogenesis conditions (Figs. 12j–l) that are similar to those of Y1V0, except that the flow to the south of Taiwan is southeasterly in Y0V0 and SW in Y1V0, leading to different locations of higher percentage of front points.
Fronts in Y0V1 (blue circles, Fig. 13a) occur much less frequently and are overall situated at more southern latitudes than those in Y1V1 (green circles, Fig. 13a) in box C, leading to much less precipitation in Y0V1 (Fig. 4c) than in Y1V1 (Fig. 4a) in southern Taiwan. Y1V1 has a considerably larger number of front events, about 20 times that of the Y0V1 (Table 2), which is also evident by comparing Figs. 12c and 12i. The peak latitude of front appearance in Y1V1 is around 23°N, with an average latitude of 22.73°N, coinciding with the offshore area southwest of Taiwan with the strongest precipitation (Fig. 4a). In box D, there are fewer front events in Y0V1 (blue crosses), with an even more southern frontal position. North of 23°N, there is almost no front, favoring the sustaining northeasterly winds bringing precipitation to northeastern Taiwan. In Y1V1 (green crosses), the peak of front events is located north of 24°N. Due to the more northern front position, the northeasterly winds poleward of the front also shift northward in Y1V1, leading to weaker winds and consequently less precipitation in northeastern Taiwan than in Y0V1. Table 2 further shows that no matter whether on Taiwan’s eastern or western side, the average frontal intensity in Y1V1 is much greater than in Y0V1.
The frontal occurrence (N; times), the mean frontal latitude (
On the other hand, in the two experiments without SWV, Y1V0 (red in Fig. 13b), and Y0V0 (black in Fig. 13b), there are very few front events in both the western and eastern regions, which is also evident in Figs. 12f and 12l. This result of Y1V0 indicates that in the absence of an SWV, strong southwest monsoon alone is not enough to provide favorable conditions for frontogenesis. It also illustrates that the influence of EY20* on frontogenesis is minimal (Fig. 14d), and most of the frontal activity differences between Y1V1 and Y0V1 (i.e., EY20, Fig. 14b) discussed above come from the nonlinear interaction between the SWV and the Y20 (RES, Fig. 14f). This is one of the reasons why the interaction between the Y20 and the SWV can play an important role in the aforementioned large precipitation in southern Taiwan. In addition, comparisons between ESWV (Fig. 14c) and EY20 (Fig. 14b) and between ESWV* (Fig. 14e) and EY20* (Fig. 14d) clearly show that the SWV plays a more important role in frontogenesis contribution to AYC (Fig. 14a) than the Y20 over the Taiwan Strait.
The percentage difference (%) of front points detected in the 64 ensemble members at every full hour from 0000 UTC 21 May to 0000 UTC 23 May 2020 (3136 samples) for (a) AYC, (b) Y20, (c) SWV, (d) Y20*, (e) SWV*, and (f) RES. White color denotes the areas where the 95% significance level is not met.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
To gain a deeper understanding of the roles played by the front in environmental fields and rainfall patterns in southern Taiwan, we utilized the simulations of the 64 ensemble members in Y1V1 at every full hour during 0000 UTC 21–23 May 2020 (3136 samples) to determine whether there is a front or not in box C for a particular member at a particular time. In total, we identified 1128 samples with a front and 2008 samples without a front. The composite means of these two groups show that when fronts are present in box C, on average the SWV center is located to the north of Taiwan with a length of ∼1000 km in an east-northeast–west-southwest orientation, judged from the 1460-gpm isoheight of the 850-hPa level (Fig. 15a). The SWV provides a favorable environment for frontogenesis and strong moisture-laden southwesterly flow over the southern Taiwan Strait, leading to heavy rainfall in southern Taiwan. In contrast, for samples without fronts, the SWV on average is situated over South China (Fig. 15b). In this no-front group, convergence is weaker, and the southwesterly flow is both weaker and drier over the southern Taiwan Strait compared to the front group, leading to minimal rainfall in southern Taiwan. The areal-mean rain intensity in box Ts is 5.92 mm h−1 for the front group, and 4.70 mm h−1 for the no-front group. These results once again indicate the pivotal role of the SWV and its associated front in influencing rainfall, particularly when the SWV moves to the north of Taiwan.
(a) The low-level-averaged (surface to 700 hPa) mixing ratio (color shading; g kg−1) and wind (vectors; m s−1), and 850-hPa geopotential height (contours; interval: 5 gpm) from the composite mean of the 1128 samples with fronts, using data at every full hour from 0000 UTC 21 May to 0000 UTC 23 May 2020 of the 64 ensemble members of Y1V1. (b) As in (a), but for the 2008 samples without fronts.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
The second reason why the interaction between the Y20 and the SWV can play a role in rainfall of southern Taiwan is related to the locations of the SWV. Figure 16a shows that most Y1V1 members have their SWV centers distributed northwest of Taiwan and with stronger intensities, averaging at an anomaly of −40.4 gpm. In contrast, most Y0V1 members have their SWV centers distributed in South China with weaker intensities, averaging at an anomaly of −31.0 gpm. The distributions of SWV centers and intensities over time (Fig. 16b for Y1V1 and Fig. 16c for Y0V1) show that the initial SWV center positions (lightest-colored dots) of both experiments are close, mostly located within 24°–27°N, 105°–115°E. However, as time progresses to the later stages of the simulations (darker-colored dots), the SWV in Y1V1 rapidly moves eastward, passing north of Taiwan and turning northeastward, while the SWV in Y0V1 slowly moves southeastward and eventually reaches the coastlines of South China. Such quasi-stationary nature of the SWV in normal years is consistent with many previous SWV studies (e.g., Rui et al. 2014; Feng et al. 2016). These final locations of the SWV are also where the negative geopotential height contributions of ESWV* occur (Fig. 10e). As EY20* and RES come into play, the SWV shifts to a more northern latitude and has an east-northeastward moving direction as shown in AYC (Fig. 10a). This movement is because EY20* provides a northeastward pathway for the SWV with the uniquely strong southwest monsoon (Fig. 10d), and the interaction between the SWV and the Y20 results in an eastward movement of the SWV suggested by the negative geopotential height difference northeast to Taiwan and positive geopotential height difference near South China in RES (Fig. 10f). Therefore, although the interaction between the SWV and the Y20 does not directly contribute to winds and moisture in southern Taiwan, it contributes to the eastward movement of the SWV, resulting in the SWV passing through northern Taiwan. The SWV then provides increased moisture and greater SW winds, which help the heavy rain in southern Taiwan.
(a) The 2-day-mean SWV centers of the 64 ensemble members in Y1V1 (green dots) and Y0V1 (blue dots) from 0000 UTC 21 May to 0000 UTC 23 May 2020. Three dot sizes represent different average intensities of the SWV; the larger the size, the stronger the SWV (scale shown in the right; gpm). (b) The SWV centers (color dots) of the 64 ensemble members in Y1V1 at every 6 h from 0000 UTC 21 May to 0000 UTC 23 May 2020, with lighter colors denoting the time closer to the beginning and darker colors closer to the end. Sizes also represent different intensities of the SWV, with the scale shown on the right. (c) As in (b), but for Y0V1.
Citation: Monthly Weather Review 152, 3; 10.1175/MWR-D-23-0198.1
6. Summary and conclusions
This paper extends the investigation of Chien and Chiu (2023) and aims to be the first study discussing the influence of interactions between weather systems at different scales on heavy precipitation in southern Taiwan in late May 2020. We quantitatively analyzed the impact of the environmental fields during the first half of the 2020 mei-yu season (Y20), the southwest vortex (SWV), and their interactions on this intense precipitation event, using four WRF ensemble experiments based on the ensemble transform Kalman filter (ETKF) data assimilation technique. Each experiment comprised of 64 members, resulting in a total of 256 simulations. In contrast to a single-member deterministic simulation, ensemble simulations incorporate diverse initial conditions, thereby mitigating potential systematic biases, affording more comprehensive information, and ultimately elevating the overall accuracy of assessments.
The control experiment demonstrated its ability to capture heavy precipitation in southern and central Taiwan, along with the lower precipitation along the eastern coast, in agreement with observations of the rain event. Exceedance probability analyses highlighted the positive spatial correlation between precipitation occurrence probabilities and mean accumulated precipitation, implying a dominance of continuous rainfall accumulation over intermittent extreme events. Comparisons of the control experiment with other sensitivity experiments determined the roles played by the Y20, the SWV, and their interaction in the precipitation. First, the Y20 were unique in that the western North Pacific subtropical high was strong and extended farther southwestward, resulting in an increased pressure gradient near Taiwan. This pattern caused the prevailing winds to shift from southerly to SW, with strengthening wind speeds and augmented moisture in the vicinity of Taiwan. Such environmental conditions favored precipitation in the windward regions of southern Taiwan but were unfavorable for the eastern, leeward areas, leading to increased precipitation in southern Taiwan and reduced precipitation in the eastern part. The SWV created a moist, vortical environment near Taiwan, supplying significant moisture and enhancing SW winds to the southwest region of Taiwan. This alteration, in turn, increased the southwest-to-northeast moisture flux, promoting precipitation formation in southern Taiwan, and also induced a north-south moisture contrast, augmenting the north-south virtual potential temperature gradient and thus increasing the frontal activity. The front tended to form at a more southward latitude such that the northeasterly winds behind the front enhanced the northeast-to-southwest moisture flux to the northeast of Taiwan, favoring precipitation in eastern Taiwan. The SWV therefore contributed to increased precipitation across the entire Taiwan region.
Although the interaction between the SWV and the Y20 had little contributing effect on the wind and moisture fields to the southwest of Taiwan, it contributed significantly to the precipitation (75.8%) in southern Taiwan. The reason is that the interaction was defined as a residual and the residual accounted for a major part of precipitation anomaly because both the SWV and the Y20 alone cannot produce heavy rain as observed in this case. For winds and moisture, however, the contributions from both the SWV and the Y20 alone were significant, leading to a small residual. The SWV contributed primarily to the moisture increase and the Y20 mainly to the SW flow enhancement, providing favorable conditions for rainfall. However, only when they came together did a nonlinear interaction between the SWV and the Y20 and large precipitation in southern Taiwan occur. Although this interaction did not play a role in providing favorable wind and moisture fields for rainfall, it increased frontal activity over the Taiwan Strait and influenced the movement of the SWV, intensifying the SW winds and the southwest-to-northeast moisture flux in southwestern Taiwan.
The results of this study not only underscore the importance of considering multiple interacting factors, but also provide valuable insights into the unique meteorological conditions that led to the extraordinary Y20R heavy precipitation event. However, this paper still has some limitations. While ensemble simulations are used to simulate various scenarios, the study scope is limited to a single precipitation event and cannot cover all possible cases. Additionally, there may be some other factors contributing to the interaction between the SWV and the Y20, or even other interactions between other types of weather systems and the strong SW winds that are not considered in this paper. These issues need more attention in future studies.
Note that the period of Y20R in this paper includes 48 h from 0000 UTC 21 May–23 May 2020, which is 12 more hours than that defined in Chien and Chiu (2023).
Acknowledgments.
The data used in this study are obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) and the Central Weather Administration of Taiwan (CWA). This research was supported by the Ministry of Science and Technology of Taiwan (Grants: MOST 111-2111-M-003-004, MOST 111-2625-M-003-002, and NSTC 112-2111-M-003-003). We thank the National Center for High-performance Computing (NCHC) for providing the computational and storage resources. We also acknowledge three anonymous reviewers for providing constructive and insightful comments.
Data availability statement.
The ERA5 dataset is available at https://www.ecmwf.int/en/forecasts/dataset/ecmwf-reanalysis-v5. The datasets of surface weather stations and rain gauge stations are available at https://asrad.pccu.edu.tw/dbar/.
REFERENCES
Barker, D., and Coauthors, 2012: The Weather Research and Forecasting Model’s Community Variational/Ensemble Data Assimilation System: WRFDA. Bull. Amer. Meteor. Soc., 93, 831–843, https://doi.org/10.1175/BAMS-D-11-00167.1.
Bishop, C. H., B. J. Etherton, and S. J. Majumdar, 2001: Adaptive sampling with the ensemble transform Kalman filter. Part I: Theoretical aspects. Mon. Wea. Rev., 129, 420–436, https://doi.org/10.1175/1520-0493(2001)129<0420:ASWTET>2.0.CO;2.
Chen, C.-S., and Y.-L. Chen, 2003: The rainfall characteristics of Taiwan. Mon. Wea. Rev., 131, 1323–1341, https://doi.org/10.1175/1520-0493(2003)131<1323:TRCOT>2.0.CO;2.
Chen, C.-S., W.-C. Chen, Y.-L. Chen, P.-L. Lin, and H.-C. Lai, 2005: Investigation of orographic effects on two heavy rainfall events over southwestern Taiwan during the mei-yu season. Atmos. Res., 73, 101–130, https://doi.org/10.1016/j.atmosres.2004.07.005.
Chen, G. T.-J., and C.-C. Yu, 1988: Study of low-level jet and extremely heavy rainfall over northern Taiwan in the mei-yu season. Mon. Wea. Rev., 116, 884–891, https://doi.org/10.1175/1520-0493(1988)116<0884:SOLLJA>2.0.CO;2.
Chen, G. T.-J., C.-C. Wang, and S.-W. Chang, 2008: A diagnostic case study of mei-yu frontogenesis and development of wavelike frontal disturbances in the subtropical environment. Mon. Wea. Rev., 136, 41–61, https://doi.org/10.1175/2007MWR1966.1.
Chen, T.-C., M.-C. Yen, J.-C. Hsieh, and R. W. Arritt, 1999: Diurnal and seasonal variations of the rainfall measured by the automatic rainfall and meteorological telemetry system in Taiwan. Bull. Amer. Meteor. Soc., 80, 2299–2312, https://doi.org/10.1175/1520-0477(1999)080<2299:DASVOT>2.0.CO;2.
Chen, Y.-L., C.-K. Wang, C.-C. Tu, F. Hsiao, and P.-L. Lin, 2022: Revisiting a mei-yu front associated with heavy rainfall over Taiwan during 6–7 June 2003. Atmosphere, 13, 644, https://doi.org/10.3390/atmos13050644.
Chien, F.-C., 2015: The role of southwesterly flow in MCS formation during a heavy rain event in Taiwan on 12–13 June 2005. Terr. Atmos. Oceanic Sci., 26, 411–429, https://doi.org/10.3319/TAO.2015.03.11.01(A).
Chien, F.-C., and Y.-C. Chiu, 2019: A composite study of southwesterly flows and rainfall in Taiwan. J. Meteor. Soc. Japan, 97, 1023–1040, https://doi.org/10.2151/jmsj.2019-057.
Chien, F.-C., and Y.-C. Chiu, 2021: Assessing the impact of dropsonde data on rain forecasts in Taiwan with observing system simulation experiments. Atmosphere, 12, 1672, https://doi.org/10.3390/atmos12121672.
Chien, F.-C., and Y.-C. Chiu, 2023: Factors leading to heavy rainfall in Southern Taiwan in the early mei-yu season of 2020. Mon. Wea. Rev., 151, 1885–1908, https://doi.org/10.1175/MWR-D-22-0226.1.
Chien, F.-C., Y.-C. Chiu, and C.-H. Tsou, 2021: A climatological study of southwesterly flows and heavy precipitation in Taiwan during mei-yu seasons from 1979 to 2018. J. Meteor. Soc. Japan, 99, 913–931, https://doi.org/10.2151/jmsj.2021-044.
Choi, K.-S., and I.-J. Moon, 2012: Influence of the western Pacific teleconnection pattern on western North Pacific tropical cyclone activity. Dyn. Atmos. Oceans, 57, 1–16, https://doi.org/10.1016/j.dynatmoce.2012.04.002.
Ding, Y., 1992: Summer monsoon rainfalls in China. J. Meteor. Soc. Japan, 70, 373–396, https://doi.org/10.2151/jmsj1965.70.1B_373.
Ding, Y., and J. C. L. Chan, 2005: The East Asian summer monsoon: An overview. Meteor. Atmos. Phys., 89, 117–142, https://doi.org/10.1007/s00703-005-0125-z.
Ebert, E. E., 2001: Ability of a poor man’s ensemble to predict the probability and distribution of precipitation. Mon. Wea. Rev., 129, 2461–2480, https://doi.org/10.1175/1520-0493(2001)129<2461:AOAPMS>2.0.CO;2.
Feng, X., C. Liu, G. Fan, X. Liu, and C. Feng, 2016: Climatology and structures of southwest vortices in the NCEP Climate Forecast System Reanalysis. J. Climate, 29, 7675–7701, https://doi.org/10.1175/JCLI-D-15-0813.1.
Guo, Y., R. Zhang, Z. Wen, J. Li, C. Zhang, and Z. Zhou, 2021: Understanding the role of SST anomaly in extreme rainfall of 2020 meiyu season from an interdecadal perspective. Sci. China Earth Sci., 64, 1619–1632, https://doi.org/10.1007/s11430-020-9762-0.
Henny, L., C. D. Thorncroft, H.-H. Hsu, and L. F. Bosart, 2021: Extreme rainfall in Taiwan: Seasonal statistics and trends. J. Climate, 34, 4711–4731, https://doi.org/10.1175/JCLI-D-20-0999.1.
Hersbach, H., 2016: The ERA5 atmospheric reanalysis. 2016 Fall Meeting, San Francisco, CA, Amer. Geophys. Union, Abstract NG33D-01.
Hirasawa, N., and T. Yasunari, 1990: Variation in the atmospheric circulation over Asia and the western Pacific associated with the 40-day oscillation of the Indian summer monsoon. J. Meteor. Soc. Japan, 68, 129–143, https://doi.org/10.2151/jmsj1965.68.2_129.
Hong, S.-Y., and J.-O. J. Lim, 2006: The WRF Single-Moment 6-Class Microphysics Scheme (WSM6). Asia-Pac. J. Atmos. Sci., 42, 129–151.
Huang, W.-R., P.-Y. Liu, J.-H. Chen, and L. Deng, 2019: Impact of boreal summer intra-seasonal oscillations on the heavy rainfall events in Taiwan during the 2017 meiyu season. Atmosphere, 10, 205, https://doi.org/10.3390/atmos10040205.
Janjić, Z. I., 1994: The step-mountain eta coordinate model: Further developments of the convection, viscous sublayer, and turbulence closure schemes. Mon. Wea. Rev., 122, 927–945, https://doi.org/10.1175/1520-0493(1994)122<0927:TSMECM>2.0.CO;2.
Kuo, Y.-H., and G. T.-J. Chen, 1990: The Taiwan Area Mesoscale Experiment (TAMEX): An overview. Bull. Amer. Meteor. Soc., 71, 488–503, https://doi.org/10.1175/1520-0477(1990)071<0488:TTAMEA>2.0.CO;2.
Kuo, Y.-H., L. Cheng, and J.-W. Bao, 1988: Numerical simulation of the 1981 Sichuan flood. Part I: Evolution of a mesoscale southwest vortex. Mon. Wea. Rev., 116, 2481–2504, https://doi.org/10.1175/1520-0493(1988)116<2481:NSOTSF>2.0.CO;2.
Lau, K., G. Yang, and S. Shen, 1988: Seasonal and intraseasonal climatology of summer monsoon rainfall over East Asia. Mon. Wea. Rev., 116, 18–37, https://doi.org/10.1175/1520-0493(1988)116<0018:SAICOS>2.0.CO;2.
Li, J., Y.-L. Chen, and W.-C. Lee, 1997: Analysis of a heavy rainfall event during TAMEX. Mon. Wea. Rev., 125, 1060–1082, https://doi.org/10.1175/1520-0493(1997)125<1060:AOAHRE>2.0.CO;2.
Lin, Y.-J., R. W. Pasken, and H.-W. Chang, 1992: The structure of a subtropical prefrontal convective rainband. Part I: Mesoscale kinematic structure determined from dual-Doppler measurements. Mon. Wea. Rev., 120, 1816–1836, https://doi.org/10.1175/1520-0493(1992)120<1816:TSOASP>2.0.CO;2.
Paltan, H., D. Waliser, W. H. Lim, B. Guan, D. Yamazaki, R. Pant, and S. Dadson, 2017: Global floods and water availability driven by atmospheric rivers. Geophys. Res. Lett., 44, 10 387–10 395, https://doi.org/10.1002/2017GL074882.
Ralph, F. M., P. J. Neiman, G. N. Kiladis, K. Weickmann, and D. W. Reynolds, 2011: A multiscale observational case study of a pacific atmospheric river exhibiting tropical–extratropical connections and a mesoscale frontal wave. Mon. Wea. Rev., 139, 1169–1189, https://doi.org/10.1175/2010MWR3596.1.
Ralph, F. M., and Coauthors, 2019: A scale to characterize the strength and impacts of atmospheric rivers. Bull. Amer. Meteor. Soc., 100, 269–289, https://doi.org/10.1175/BAMS-D-18-0023.1.
Rui, Z., L. Zhong, L. Hua, and S. Feng, 2014: A climatology of the southwest vortex during 1979–2008. Atmos. Oceanic Sci. Lett., 7, 577–583, https://doi.org/10.1080/16742834.2014.11447227.
Sever, G., and Y.-L. Lin, 2017: Dynamical and physical processes associated with orographic precipitation in a conditionally unstable uniform flow: Variation in basic wind speed. J. Atmos. Sci., 74, 449–466, https://doi.org/10.1175/JAS-D-16-0077.1.
Skamarock, W. C., and Coauthors, 2019: A description of the Advanced Research WRF Model version 4. NCAR Tech. Note NCAR/TN-556+STR, 145 pp., https://doi.org/10.5065/1dfh-6p97.
Su, S.-H., and Coauthors, 2022: Observing severe precipitation near complex topography during the Yilan Experiment of Severe Rainfall in 2020 (YESR2020). Quart. J. Roy. Meteor. Soc., 148, 1663–1682, https://doi.org/10.1002/qj.4271.
Tai, S.-L., Y.-C. Liou, S.-F. Chang, and J. Sun, 2020: The heavy rainfall mechanism revealed by a Terrain-resolving 4DVar data assimilation system—A case study. Mon. Wea. Rev., 148, 2307–2330, https://doi.org/10.1175/MWR-D-19-0244.1.
Takaya, Y., I. Ishikawa, C. Kobayashi, H. Endo, and T. Ose, 2020: Enhanced meiyu‐baiu rainfall in early summer 2020: Aftermath of the 2019 super IOD event. Geophys. Res. Lett., 47, e2020GL090671, https://doi.org/10.1029/2020GL090671.
Tao, W.-K., J. Simpson, and M. McCumber, 1989: An ice-water saturation adjustment. Mon. Wea. Rev., 117, 231–235, https://doi.org/10.1175/1520-0493(1989)117<0231:AIWSA>2.0.CO;2.
Tao, W.-K., and Coauthors, 2011: High-resolution numerical simulation of the extreme rainfall associated with Typhoon Morakot. Part I: Comparing the impact of microphysics and PBL parameterizations with observations. Terr. Atmos. Oceanic Sci., 22, 673–696, https://doi.org/10.3319/TAO.2011.08.26.01(TM).
Tiedtke, M., 1989: A comprehensive mass flux scheme for cumulus parameterization in large-scale models. Mon. Wea. Rev., 117, 1779–1800, https://doi.org/10.1175/1520-0493(1989)117<1779:ACMFSF>2.0.CO;2.
Tu, C.-C., Y.-L. Chen, P.-L. Lin, and Y. Du, 2019: Characteristics of the marine boundary layer jet over the South China Sea during the early summer rainy season of Taiwan. Mon. Wea. Rev., 147, 457–475, https://doi.org/10.1175/MWR-D-18-0230.1.
Tu, C.-C., Y.-L. Chen, P.-L. Lin, and P.-H. Lin, 2020: The relationship between the boundary layer moisture transport from the South China Sea and heavy rainfall over Taiwan. Terr. Atmos. Oceanic Sci., 31, 159–176, https://doi.org/10.3319/TAO.2019.07.01.01.
Volonté, A., A. G. Turner, R. Schiemann, P. L. Vidale, and N. P. Klingaman, 2022: Characterising the interaction of tropical and extratropical air masses controlling East Asian summer monsoon progression using a novel frontal detection approach. Wea. Climate Dyn., 3, 575–599, https://doi.org/10.5194/wcd-3-575-2022.
Wang, C.-C., M.-S. Li, C.-S. Chang, P.-Y. Chuang, S.-H. Chen, and K. Tsuboki, 2021: Ensemble-based sensitivity analysis and predictability of an extreme rainfall event over northern Taiwan in the mei-yu season: The 2 June 2017 case. Atmos. Res., 259, 105684, https://doi.org/10.1016/j.atmosres.2021.105684.
Wang, C. C., P. Y. Chuang, S. T. Chen, D. I. Lee, and K. Tsuboki, 2022: Idealized simulations of mei-yu rainfall in Taiwan under uniform southwesterly flow using a cloud-resolving model. Nat. Hazards Earth Syst. Sci., 22, 1795–1817, https://doi.org/10.5194/nhess-22-1795-2022.
Wang, X., T. M. Hamill, J. S. Whitaker, C. H. Bishop, and X. Wang, 2007: A comparison of hybrid ensemble transform Kalman filter–Optimum interpolation and ensemble square-root filter analysis schemes. Mon. Wea. Rev., 135, 1055–1076, https://doi.org/10.1175/MWR3307.1.
Xu, W., E. J. Zipser, Y.-L. Chen, C. Liu, Y.-C. Liou, W.-C. Lee, and B. Jong-Dao Jou, 2012: An orography-associated extreme rainfall event during TiMREX: Initiation, storm evolution, and maintenance. Mon. Wea. Rev., 140, 2555–2574, https://doi.org/10.1175/MWR-D-11-00208.1.
Yao, X., Q. Zhang, and X. Zhang, 2020: Potential vorticity diagnostic analysis on the impact of the easterlies vortex on the short-term movement of the subtropical anticyclone over the western Pacific in the mei-yu period. Adv. Atmos. Sci., 37, 1019–1031, https://doi.org/10.1007/s00376-020-9271-y.
Yeh, H.-C., and G. T.-J. Chen, 2004: Case study of an unusually heavy rain event over eastern Taiwan during the mei-yu season. Mon. Wea. Rev., 132, 320–337, https://doi.org/10.1175/1520-0493(2004)132<0320:CSOAUH>2.0.CO;2.
Zhang, Q.-H., K.-H. Lau, Y.-H. Kuo, and S.-J. Chen, 2003: A numerical study of a mesoscale convective system over the Taiwan Strait. Mon. Wea. Rev., 131, 1150–1170, https://doi.org/10.1175/1520-0493(2003)131<1150:ANSOAM>2.0.CO;2.
