Measuring East Asian Summer Monsoon Rainfall Contributions by Different Weather Systems Over Taiwan

2068 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47

NOTES AND CORRESPONDENCE

Measuring East Asian Summer Monsoon Rainfall Contributions by Different Weather Systems over Taiwan

SHIH-YU WANG AND TSING-CHANG CHEN* Atmospheric Science Program, Department of Geological and Atmospheric Sciences, Iowa State University, Ames, Iowa

(Manuscript received 2 July 2007, in final form 24 October 2007)

ABSTRACT

The east Asian summer monsoon (EASM) is characterized by a distinct life cycle consisting of the active, break, and revival monsoon phases. Different weather systems prevail in each phase following the change of large-scale flow regime. During the active phase, midlatitude cold, dry air moving equatorward into the tropics develops eastward-propagating fronts and rainstorms. The western ridge of the North Pacific subtropical anticyclone, which leads to the break phase, suppresses the development of synoptic-scale weather systems but enhances the diurnal heating. In the revival phase, the monsoon trough displaces the anticy- clonic ridge northward and increases typhoon activity. This study examines quantitative measurements of climatological rainfall contributed by these weather systems that will help to validate simulations of the EASM climate system and facilitate water management by government agencies. To accomplish this goal, rain gauge measurements in Taiwan were analyzed. It was found that in the active phase (late spring–early summer), mei-yu rainstorms forming over northern Indochina and the South China Sea contribute one-half of the total rainfall, and cold-frontal passages account for about 15%. During the synoptically inactive break spell (midsummer), rainfall is produced mainly by diurnal convection (51%) along the western mountain slopes of the island. In the revival phase (late summer–early autumn), the impacts of diurnal convection and typhoons become comparable, with each accounting for about 40% of the total rainfall. The diurnal and typhoon contributions are separated by the north–south-oriented mountain range, with more than 50% of diurnal (typhoon) rainfall over the western (eastern) half of the island. Rainfall contributions from diurnal convection over the entire summer and mei-yu rainstorms in the active monsoon phase are also significant.

1. Introduction ery and the western tropical Pacific (WTP) monsoon trough along its southwest rim, composes the lower- Monsoon rainfall in eastern Asia—in particular, dur- tropospheric circulation of the EASM. Seasonal north- ing summer—is beneficial and affects nearly one-quar- ward progression of the EASM circulation results in ter of the world’s population. The east Asian summer sequential passages of the mei-yu rainband, the ridge of monsoon (EASM) is unique for its distinct active– the anticyclone, and the WTP monsoon trough. These break–revived life cycle of precipitation (Ramage 1952; periods cause a rapid increase in rainfall (active phase), Matsumoto 1989; Chen et al. 2004, hereinafter CWHY). a synoptically inactive dry spell (break phase), and an- The western North Pacific subtropical anticyclone, be- other period of large rainfall caused by the northwest- tween the mei-yu rainband along its northwest periph- ward propagation of tropical cyclones (revival phase), respectively (CWHY). The evolution of this monsoon life cycle is depicted in Fig. 1a by a latitude–time cross * Current affiliation: Visiting Scientist at National Central Uni- versity in Taiwan sponsored by a National Science Council Chair. section of the Climate Prediction Center Merged Analysis of Precipitation (CMAP; Xie and Arkin 1997) data between 120° and 125°E. Corresponding author address: Tsing-Chang (Mike) Chen, At- With the passage of each monsoon phase, a change in mospheric Science Program, Department of Geological and At- mospheric Sciences, Iowa State University, 3010 Agronomy Hall, the flow regime results in different weather systems Ames, IA 50011. affecting eastern Asia. During the active monsoon E-mail: [email protected] phase (also known as mei-yu or baiu), deep cumulus

DOI: 10.1175/2007JAMC1821.1

JAMC1821 JULY 2008 NOTES AND CORRESPONDENCE 2069

rainfall is closely related to frontal activity in the active phase and tropical disturbances in the revival phase but have not categorized rainfall contributions from all weather systems during the monsoon season. Quantita- tive measurement of the monsoon rainfall produced by different weather systems will facilitate water management and land-use planning by the local government. For example, new reservoirs should be constructed over areas where rainfall amounts are largest and yet least hazardous (Lin 2005). In addition, rainfall measurements will help in the understanding of the EASM climate system and will validate regional climate models. Isolating the contributions of rainfall from different weather systems is generally complicated. Nevertheless, with proper criteria, numerous studies over different regions have shown that reasonable estimates of precipitation amounts produced by different weather systems can be obtained. For instance, it was found that mesoscale convective systems (MCSs) contribute one- quarter of the rainy season precipitation in Sahelian Africa (Adelekan 1998; Laing et al. 1999) and about 75% of the summer rainfall in the central United States (e.g., Changnon 2001; Fritsch et al. 1986) and Spain (Llasat and Puigcerver 1997). In Israel, previous studies have shown that cold fronts produce up to 45% of the cold-season rainfall while other cloud systems contribute another 40% (Goldreich et al. 2004). In addition, Indian monsoon depressions have been found to bring

FIG. 1. (a) Latitude–time cross section of pentad-mean CMAP in about one-half of the monsoon rain over India (Yoon (Xie and Arkin 1997) rainfall averaged over the longitudinal zone and Chen 2005), and satellite observations indicate that of 120°–125°E (adopted from Fig. 3a of CWHY). (b) Half-month tropical cyclones during the hurricane season contrib- rainfall averaged over the 1993–2005 period estimated with 427 ute 10%–20% of the total rainfall over the North Pa- rain gauges over Taiwan as shown in Fig. 2, below. The three EASM phases (active, break, and revival) are defined in (a) and (b). cific (Rodgers et al. 2000) and the North Atlantic (Rodgers et al. 2001) Oceans. Surprisingly little effort has been made to measure convection frequently occurs over the region from rainfall contributions by eastern Asian weather sys- northern Indochina and southwest China to the north- tems, although previous studies have analyzed the ern South China Sea and quickly develops into orga- physical nature of individual rainfall events in this re- nized convective storms. Mei-yu rainstorms (the defini- gion (e.g., Li et al. 1997; Johnson and Ciesielski 2002; tion is given in section 2, called rainstorms hereinafter) Wang 2004, and many others). Using a field experimen- frequently propagate eastward and often lead to disas- tal dataset during May–June 1987, Yeh and Chen trous floods in Vietnam, southern coastal China, and (1998) discussed rainfall characteristics associated with Taiwan. In the revival monsoon phase, frontal activity various weather events over Taiwan, but the analysis is generally restricted to higher latitudes and allows for was confined to a short season in an unusually weak west-northwestward-moving tropical cyclones to bring monsoon year (Krishnamurti et al. 1989). Therefore, heavy rainfall to eastern Asia (CWHY). In addition to their results were likely not climatologically represen- these synoptic disturbances, topographic heating in tative of the entire EASM period. Assessing rainfall Asia provides forcing for diurnally varying rainfall (e.g., contributions from different weather systems in eastern Chen et al. 1999; Krishnamurti and Kishtawal 2000; Asia is complicated because of inhomogeneous observ- Chen 2005). The variety of weather systems in eastern ing locations and highly variable terrain. However, Tai- Asia complicates the understanding of the monsoon wan, a subtropical island located between the eastern rainfall characteristics. CWHY noted that the EASM Asian continent and the western North Pacific, has a 2070 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47

FIG. 2. Orography of Taiwan and distribution of 427 rain gauges administered by the Environ- mental Protection Administration (EPA; open circles), the Weather Bureau (asterisks), the Automatic Rainfall and Meteorological Telemetry System (ARMTS; filled circles), and the military (plus signs) in Taiwan. The scale of the orography is given in the lower-left corner. fine-resolution observational network with 427 rain half-month mean rainfall measured by this network gauges in an area of 36 000 km2. Administrated by sev- (Fig. 1b). Because the EASM produces, on average, eral governmental agencies, this rain gauge network over 1600 mm of rainfall during the warm season (May– consists of the Automatic Rainfall and Meteorological September) in Taiwan, it is a very important part of the Telemetry System (ARMTS) of more than 300 rain climate and inspires us to understand further the gauges (Chen et al. 1999), 25 conventional stations weather systems involved in its production. managed by the weather bureau of Taiwan, 72 stations The objectives of this study are to measure quantita- monitored by the Taiwan Environmental Protection tively the EASM rainfall produced by various weather Administration, and several military-based surface ob- systems, to discover the relationship between these serving sites (denoted by dots, asterisks, open circles, weather systems and total precipitation, and to examine and plus signs, respectively, in Fig. 2). All stations in the spatial rainfall distribution based on the observa- operation since 1993 provide hourly observations. The tions in Taiwan. In particular, our goal is to estimate three-phase EASM life cycle is clearly depicted by the these rainfall contributions from a climatological per- JULY 2008 NOTES AND CORRESPONDENCE 2071 spective. Data observed by the rain gauge network in this category. An example of a diurnal rainfall event Taiwan for a 13-yr period from 1993 to 2005 are ana- that occurred on 8 July 2003 is shown in Fig. 3a. With- lyzed to reach this goal. Surface weather maps of every out any active weather system nearby, cumulus convec- 6 h issued by the Japan Meteorological Agency (JMA), tion started at 1400 LST (0600 UTC) and produced a hourly infrared (IR) images from the Geostationary rainfall rate of more than 80 mm hϪ1 at 1500 LST over Meteorological Satellite (GMS) provided by the JMA, the Taipei basin and the southwest plains (0700 UTC; and 6-h typhoon records archived by the JMA Regional indicated by the inset in the lower lower-right corner of Specialized Meteorological Center were used to trace the top panel). all weather systems that produced rainfall across Tai- wan. Section 2 lists the guidelines of how each type of b. Typhoon weather system is identified and classified. In section 3, A typhoon event is identified when 1) Taiwan is at the rainfall characteristics and contributions from each least partially embedded within the radius of 30-kt weather system in the three distinct phases of the Ϫ (ϳ15ms 1) sustained wind speeds of a typhoon (pro- EASM life cycle are discussed. A summary of the vided in the JMA maps) or 2) typhoon cloud bands analysis is offered in section 4. covered any part of Taiwan. Tropical depressions with weaker sustained winds (Ͻ30 kt) are also included in 2. Rain-producing weather systems and this category because they may produce heavy rain identification procedures when passing through the island. Such depressions can be easily identified with surface weather maps and by Diurnal convection (Chen et al. 1999), typhoons, the organized convection in IR images. Typhoon Herb (31 mei-yu seasonal rainstorms (e.g., Johnson and Ciesiel- July 1996) is an example (Fig. 3b) of a typhoon event. ski 2002), and cold fronts are found to be the most Even though its center was located slightly north of common rain producers. The remaining weather sys- Taiwan, Herb’s cloud bands engulfed most of the is- tems that do not belong to these four types are grouped land. into one category because of their small contributions to seasonal rainfall (as shown in Fig. 4, below). A rain- c. Rainstorm fall event is recorded for a day when the amount of A rainstorm event is identified when 1) a cloud shield precipitation received by more than three adjacent sta- with temperature lower than Ϫ50°C and a domain tions during any time of the day is larger than 1 mm Ϫ larger than 150 km ϫ 150 km covers Taiwan and 2) h 1. Each rainfall event in Taiwan is classified into one local rainfall rate reaches 50 mm within6hatanysta- of these five categories of weather systems based on a tion. The first criterion is modified based on general subjective analysis of JMA weather maps and GMS IR characteristics of rainstorms from Maddox’s (1980) images. Examples of these weather systems depicted by identification scheme for mesoscale convective com- IR images and surface weather maps are illustrated in plexes and specifically separates rainfall produced by Fig. 3. The classification procedures for the five catego- rainstorms from nearby frontal cloud bands (if any). ries are given below. Because there is not a commonly accepted definition for rainstorms in eastern Asia, the second criterion fol- a. Diurnal convection lows that of Chen et al. (1998), which specifies that a A diurnal rainfall event is identified when 1) rainfall convective storm resulting in precipitation rates of at rate of more than 1 mm hϪ1 was detected in an after- least 50 mm within6hiscalled a rainstorm. A typical noon/evening (1200–2100 LST) by at least three sta- rainstorm (7 June 1998) over the northern South China tions, 2) no rainfall was detected in the morning (0600– Sea is shown in Fig. 3c (indicated by a thick arrow). 1100 LST) of that day, and 3) synoptically calm condi- When this storm reached Taiwan, it generated over 400 tions were present within 100 km of Taiwan. Calm mm dayϪ1 of precipitation in the southwest plains. conditions are declared when the GMS IR images show The interaction between southwesterly monsoon either cloudlessness or a lack of convective clouds flows and midlatitude frontal systems may develop around the island before the afternoon/evening convec- these convective storms (e.g., Lau et al. 2000; Wang tion occurs. This identification rule is derived from 2004). As a result, rainstorms sometimes coexist with a Chen et al.’s (1999) approach to differentiate local frontal cloud band and are often regarded as MCSs forcing from synoptic forcing. As a consequence of embedded in the “mei-yu front” (e.g., Lin and Kueh this rule, potential diurnal variations in any synoptic 2003). However, they do not exhibit the typical features weather system (fronts, typhoons, etc.) are excluded in of midlatitude MCSs (e.g., tilted, front-to-rear updraft 2072 ORA FAPIDMTOOOYADCIAOOYV CLIMATOLOGY AND METEOROLOGY APPLIED OF JOURNAL

FIG. 3. The (top) GMS IR images and (bottom) surface weather charts of (a) a diurnal convection event at 0600 UTC (1400 LST) 8 Jul 2003, (b) Typhoon Herb at 1200 UTC 31 Jul 1996, (c) a mei-yu rainstorm over the northern South China Sea at 0600 UTC 7 Jun 1998, and (d) a cold front at 0000 UTC 9 May 2004 provided by the JMA. The inset in (a) is a visible-channel cloud image magnifying the afternoon convection over Taiwan at 0700 UTC. The thick arrows in (c) indicate the location of the rainstorm. OLUME 47 JULY 2008 NOTES AND CORRESPONDENCE 2073 and rear-to-front downdraft) as described by Moncrieff each weather system in terms of days is shown in Fig. (1992). Mei-yu rainstorms exhibit deep, vertical up- 4b. Diurnal rainfall events occur most frequently, aver- drafts (e.g., Johnson and Ciesielski 2002; Wang 2004) aging 60 days per warm season, and produce 31% of the that resemble tropical hot-tower convection. Thus, seasonal rainfall (3.2 mm dayϪ1). Although typhoon treating rainstorms as part of a front is not an appro- passages are often disastrous in Taiwan, they only con- priate practice. tribute 11 rainfall days and about 28% (ϳ3mmdayϪ1) of the total rainfall. Rainstorms contribute 25% (2.8 Ϫ1 d. Cold front mm day ) of rainfall during the warm season despite being the least frequent (ϳ8 days) in the four catego- A cold-frontal case is identified when an equator- ries. Cold fronts (13 days) are slightly more frequent ward-intruding polar front migrates to within 50 km of than typhoons but contribute just 9% (Ͻ1mmdayϪ1) Taiwan (detected by examining the JMA weather of the rainfall. It is revealed from these statistics that maps) and its frontal cloud band reaches the island. The the local diurnal mode contributes to the climatological time period and rainfall rate are recorded until the rainfall as much as do typhoons. In addition, precipita- cloud band moves out of the entire island or the front tion contributed by rainstorms during the relatively propagates outside 50 km of Taiwan. A similar proce- short period of late spring–early summer is comparable dure has been used by Goldreich et al. (2004) for cat- to the production by typhoons during the entire rainy egorizing cold-frontal rainfall in Israel. A typical ex- season. These results disagree with the general percep- ample of a cold front (9 May 2004) moving across Tai- tion that fronts contribute the most to the mei-yu sea- wan is shown in Fig. 3d. This front is depicted by the sonal rainfall (Lin and Kueh 2003; Chen 1998) and that surface weather map (bottom panel), and its cloud typhoons generate the majority of late-summer precipi- band is about 400 km in width (top panel). The frontal tation in Taiwan (e.g., Chen 1998). Next, the fluctua- cloudband was not associated with rainstorms because tions of these rainfall contributions during each EASM no organized convection was embedded in the front. phase as a result of different flow regimes are examined. The three phases of the EASM life cycle based on e. Others the Taiwan rainfall are defined in Fig. 1b. When the criteria for the previous four weather sys- In the active phase (1 May–15 June), rainstorms and ϳ tems do not apply, the rainfall event is placed in the fronts occur at approximately the same frequency ( 7 “others” category. These usually happen when no dis- days) across Taiwan (Fig. 4d). However, rainstorms tinct weather system is present in weather maps and IR produce nearly 50% of rainfall while midlatitude cold images during a rainfall event. The weather conditions fronts contribute only 14%. At times, Taiwan is cov- in this category are usually strong environmental flow ered partially by a rainstorm and partially by a cold interacting with the terrain (revealed by station winds), front. When both systems are present (about 18% of land breeze–induced rainbands moving inland, or other cold-frontal cases), categorizing rainfall into either difficult situations to identify. rainstorm or cold front generates different rainfall con- Although some subjective bias could be involved in tributions from each weather system. Such differences the identification processes, operational weather result in a 6% change of the rainfall contribution, which maps and satellite images help to minimize this bias. means that the rainfall contribution by cold fronts Once all rainfall events are categorized, the precipita- (rainstorms) may fluctuate between 11% and 17% tion amounts are accumulated for each category (45% and 51%). Nevertheless, these two weather sys- ϭ ϩ throughout the EASM life cycles. Rainfall distributions tems contribute the majority [62% ( 48% 14%)] of are then obtained using Cressman’s (1959) objective rainfall during the active phase (Fig. 4c). Diurnal rain- ϳ analysis with a 5-km radius of influence over the entire fall events are 2 times as frequent ( 15 days) as rain- rain gauge network shown in Fig. 2. storms but generate only one-half of the rainstorm rainfall (24%). The low activity of typhoons in this period contributes less than 10% of the rainfall, and other 3. Rainfall characteristics associated with the weather systems not belonging to these four types (oth- eastern Asian weather systems ers) contribute just 6%. The monsoon break (16 June–15 July) is a synopti- a. Rainfall statistics cally inactive period relative to other EASM phases. For the summer monsoon season (May–September) Diurnal convection, which occurs on approximately 12 during the 1993–2005 period, the average rainfall in days, is the major rain producer and contributes 51% of Taiwan was 10.5 mm dayϪ1 (Fig. 4a). The frequency of the total rainfall (Fig. 4e). In the break phase, typhoons 2074 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47

Ϫ FIG. 4. (a) Total rainfall P (mm day 1) averaged over Taiwan and rainfall contributions from diurnal rainfall events, typhoons, mei-yu rainstorms (“M. Rainstorm”), midlatitude cold fronts (“C. Front”), and others for May–September (1993–2005), and (b) rainfall days D of each weather system affecting Taiwan. As in (a), (b), but for the (c), (d) active, (e), (f) break, and (g), (h) revival phases. The time period of each monsoon phase is given under the labels in (c), (e), and (g). begin developing more frequently as the South Asia– In the revival phase (16 July–30 September), the WTP monsoon trough deepens in midsummer (CWHY). South Asia–WTP monsoon trough has advanced to its The average number of days that typhoons affect Tai- maximum eastward extension in this season, and the wan increases to 2.1 (Fig. 4f) from less than 1.0 in the enhanced frequency of typhoons is channeled to east- active phase. On occasion during the monsoon break, ern Asia along this trough (CWHY). The frequency strong typhoons may impact Taiwan and produce very of typhoon passages across Taiwan increases to 8 heavy rainfall. A recent example is Supertyphoon Min- days (Fig. 4h) and contributes 41% of the rainfall. Di- dulle in 2004, which produced record-high precipitation urnal convection remains the most frequent rain pro- amounts of 1500 mm dayϪ1 in some areas over a 3-day ducer (33 days) with a 43% contribution of rainfall. period and also triggered severe landslides and debris Despite the general perception that typhoons produce flows. This kind of typhoon does not occur regularly most of the rainfall during the typhoon season of late but can notably change the climatological rainfall. As a summer, it is found that diurnal convection produces result, typhoon passages contribute 32% of the rainfall rainfall amounts comparable to typhoons. Fronts and during the monsoon break (Fig. 4e). category others occur least frequently and contribute JULY 2008 NOTES AND CORRESPONDENCE 2075

FIG. 5. Rainfall distributions of (a) total amount PT, (b) diurnal rainfall events PD, (c) rainstorms PRS, and (d) midlatitude fronts PFt over Taiwan during the active monsoon phase. Rainfall contributions by (e) diurnal rainfall events (PD /PT), (f) rainstorms (PRS/PT), and (g) fronts (PFt/PT), expressed in terms of percentage. Orography is depicted by dots, with the scale shown in the middle right. Contour intervals of rainfall and contribution ratio are given in (a)–(g). only 9% and 6% of the late-summer rainfall, respec- along the west and southwest slopes of the mountain tively. range (Fig. 5a). This rainfall pattern is similar to that of rainstorms (Fig. 5c), implying that rainstorms are the b. Spatial distribution major rain producer. After the monsoon onset over the Spatial distributions of rainfall generated by different South China Sea, strong low-level southwesterly flow weather systems during the three monsoon phases are tends to form upslope winds against mountains in Tai- shown in Figs. 5–7. To avoid redundancy, only weather wan (Yen and Chen 2000). Although more than 60% of systems that contribute more than 10% of the total rainfall over the southwest plains is produced by rain- precipitation are focused on. storms (Fig. 5f), rainfall over the windward slopes in The spatial distribution of the total rainfall in the southwest Taiwan is enhanced by this flow–terrain inactive-monsoon-phase period exhibits two maxima teraction (Figs. 5a,c). Frontal passages produce about 2076 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47

FIG. 6. Rainfall distributions of (a) total amount PT, (b) diurnal rainfall events PD, and (c) typhoons PTY over Taiwan during the break monsoon phase. Rainfall contributions by (d) diurnal rainfall events (PD /PT) and (e)

typhoons (PTY/PT), in terms of percentages. Orography and contour intervals of rainfall and contribution ratio are the same as in Fig. 5.

15% of total precipitation over northern Taiwan (Figs. in agreement with that of Yeh and Chen (1998) for the 5d,g) but produce less significant amounts relative to early summer of 1987. Note that rainfall produced from rainstorms. In addition to these synoptic weather sys- weather systems in the others category is also influ- tems, diurnal convection produces rainfall primarily enced by the flow–terrain interaction as precipitation along the western mountain slopes (Fig. 5b), which ac- concentrates on the windward side of mountains facing counts for as much as 30% of the rainfall contribution southwest (not shown). (Fig. 5e). The distribution of diurnal rainfall is generally The total amount of rainfall decreases during the JULY 2008 NOTES AND CORRESPONDENCE 2077

FIG. 7. As in Fig. 6, but for the revival monsoon phase. break-monsoon-phase period, though the dual-center afternoon/evening thunderstorms along the western rainfall pattern persists (Fig. 6a). Both centers are mountain slopes. This process contributes over 50% of formed by the combination of diurnal (Fig. 6b) and rainfall along the western slopes. Typhoons produce a typhoon (Fig. 6c) rainfall, but not by rainstorms as in pair of rainfall maxima over tall mountains (Fig. 6c) the active phase (not shown). The pattern of diurnal with more than 40% of the total precipitation (Fig. 6e). rainfall is characterized by a rain belt that extends from Because the circulation of an approaching typhoon is the northwestern slopes to the southwest plains. This forced to move across the terrain, precipitation is en- diurnal rain belt supports the observation by Chen et al. hanced over the windward slopes of the mountains (1999) that the interaction between sea breezes from (e.g., Bender et al. 1985, 1987) despite the typhoon tra- the west coast and terrain fosters the development of jectory or the location of landfall. 2078 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47

FIG. 8. East–west sections across southern Taiwan depicting the analyzed mean station precipitation (meridional average between 22° and 24°N) and error bar (range of 1 std dev), together with the meridionally averaged topography (shadings), for (a) diurnal rainfall

PD and (b) typhoon rainfall PTY during the revival monsoon phase.

After the break spell, rainfall in Taiwan increases slopes and that precipitation by typhoons is concen- and persists for another 2.5 months (Fig. 1b). The re- trated over the mountain top. Rainfall contributions established monsoon southwesterlies in the revival- from these two weather systems (Figs. 7d,e) are virtu- monsoon phase supply moisture and intensify the flow– ally divided into two halves by the north–south- terrain interaction over the windward (southwest fac- oriented mountain range, with more than 50% of rain- ing) slopes (Yen and Chen 2000). The distribution of fall over the western (eastern) portion of the island total rainfall during this period (Fig. 7a) reveals two contributed by diurnal convection (typhoons). This asymmetric maxima, with the strong (weak) one over character of rainfall distribution corresponds to the re- the southwest (central west) mountain slopes. The rain- sults shown in Fig. 4g that the diurnal mode and ty- belt generated by the diurnal convection (Fig. 7b) re- phoons generate the majority of rainfall in this mon- sembles that in the previous phases except with much soon phase. These results reveal the significance of di- larger rainfall maxima over the southwest foothills in- urnal rainfall. The general perception that typhoons are tensified by the monsoon southwesterly flow (Chen et the dominant rain producer in late summer (Chen al. 1999). In the meantime, typhoons contribute signif- 1998) is only true in the eastern half of the island. icant rainfall on elevated terrain (Fig. 7c). To ensure that the relatively coarse station density over moun- 4. Summary tains (Fig. 2) does not cause bias in depicting the rainfall pattern, a test without any interpolation was per- Using the fine-resolution rainfall measuring network formed. The east–west sections across southern Taiwan with 427 rain gauges in Taiwan for a 13-yr period (Fig. 8) depict the analyzed mean station precipitation (1993–2005), a study was conducted that investigated (meridional average between 22° and 24°N), together the rainfall contributions from weather systems com- with the meridionally averaged topography, for the di- mon in eastern Asia: typhoons, rainstorms, cold fronts, urnal (Fig. 8a) and typhoon (Fig. 8b) rainfall. Figures and the local diurnal convection. It was revealed from 8a,b and Figs. 7b,c are consistent in revealing that di- these rainfall analyses that local diurnal convection ap- urnal rainfall is maximized over the western mountain pears to be equally important (31%) as typhoons (28%) JULY 2008 NOTES AND CORRESPONDENCE 2079 and rainstorms (25%) in producing the summer mon- large-scale monsoon circulation. This information is soon rainfall over Taiwan. Based on the clear life cycle useful to validate regional climate simulations of the of the EASM consisting of three distinct phases (active, eastern Asian monsoon and its hydrological cycle. In break, and revival), quantitative measurements of the addition to their scientific relevance, these rainfall mea- rainfall contributions from four common weather sys- surements should also facilitate the local government in tems during different monsoon phases were examined. improving its water management and urban planning. Results are summarized as follows. For example, the water supply of major reservoirs in Taiwan currently depends on rainfall produced by a. Active phase mei-yu rainstorms or typhoons, but pronounced interannual variability of circulation over eastern Asia and Contributing one-half of the rainfall, mei-yu rain- the western North Pacific often suppresses the activity storms appear to be the most important rain-producer of these weather systems (e.g., Wang et al. 2001) and, in in this period. Another one-quarter of the precipitation turn, leads to droughts. The population of Taiwan has is generated by the afternoon/evening convection pri- reached over 23 million. To cope with water-supply marily forced by topographic heating and sea-breeze deficits, future reservoirs should be built in locations circulations along the western mountain slopes. Cold that promote effective collection of diurnal rainfall. fronts contribute less than 15% of the rainfall, and, because typhoons seldom strike Taiwan in this mon- Acknowledgments. This study was supported by the soon phase, they contribute only 6% of the rainfall. Cheney Research Fund and was finished during the second author’s visit to National Central University b. Break phase (NCU). The visit was partially supported by the NCU Development Program of Internationally Top-Ranked As the frequencies of fronts and rainstorms decrease, University sponsored by the Ministry of Education in rainfall in the break phase is mainly produced by diur- Taiwan, and Grants NSC95-2811-M-008-009, NSC95- nal convection (51%) along the western mountain 2111-M-008-007, and NSC96-3114-P-492-002-Y of Tai- slopes. The occasional passages of typhoons generate, wan. Numerous comments offered by anonymous re- in a climatological sense, 32% of the rainfall in Taiwan. viewers were helpful in improving this paper. Editorial Because individual diurnal convection events do not assistance provided by Adam Clark and Daniel Rajew- produce as much rainfall as do rainstorms or typhoons, ski is appreciated. the inactive synoptic condition during the monsoon break results in a pronounced reduction of total rainfall. REFERENCES

Adelekan, I. O., 1998: Spatio-temporal variations in thunderstorm c. Revival phase rainfall over Nigeria. Int. J. Climatol., 18, 1273–1284. After the break phase, typhoon passages across Tai- Bender, M. A., R. E. Tuleya, and Y. Kurihara, 1985: A numerical study of the effect of a mountain range on a landfalling tropi- wan become more frequent and produce 41% of the cal cyclone. Mon. Wea. Rev., 113, 567–583. rainfall. It was estimated by Rodgers et al. (2000, 2001) ——, ——, and ——, 1987: A numerical study of the effect of that tropical cyclones contribute up to 20% of the total island terrain on tropical cyclones. Mon. Wea. Rev., 115, 130– rainfall over the western North Pacific and the North 155. Atlantic. In comparison with Rogers et al.’s estimates, Changnon, S. A., 2001: Thunderstorm rainfall in the conterminous the typhoon–terrain interaction over Taiwan consider- United States. Bull. Amer. Meteor. Soc., 82, 1925–1940. ably enhances rainfall by typhoons. Another 43% of Chen, L.-F., 1998: Speaking about the spring rainfall in Taiwan (in Chinese). J. Earth Sci., 5, 2–11. [Available online at http:// the rainfall is produced from local diurnal precipitation. earth.fg.tp.edu.tw/learn/esf/magazine/980301.htm.] The diurnal rainfall occurs steadily over the western Chen, S.-J., Y.-H. Kuo, W. Wang, Z.-Y. Tao, and B. Cui, 1998: A slopes and the southwestern plains (as shown in Figs. modeling case study of heavy rainstorms along the mei-yu 5b, 6b, and 7b). The contributions of rainfall by ty- front. Mon. Wea. Rev., 126, 2330–2351. phoons and diurnal convection appear to be separated Chen, T.-C., 2005: Variation of the Asian monsoon water vapor well by the north–south-oriented mountain range, with budget: Interaction with the global-scale modes. The Asian more than 50% of rainfall generated by the former over Monsoon, B. Wang, Ed., Springer Verlag, 417–458. ——, M.-C. Yen, J.-C. Hsieh, and R. W. Arritt, 1999: Diurnal and the east and by the latter over the west. seasonal variations of the rainfall measured by the Automatic These results provide additional information to the Rainfall and Meteorological Telemetry System in Taiwan. climatological rainfall description of the EASM life Bull. Amer. Meteor. Soc., 80, 2299–2312. cycle observed by CWHY from a perspective of the ——, S.-Y. Wang, W.-R. Huang, and M.-C. Yen, 2004: Variation 2080 JOURNAL OF APPLIED METEOROLOGY AND CLIMATOLOGY VOLUME 47

of the east Asian summer monsoon rainfall. J. Climate, 17, tive rainfall in Catalonia, Spain. Int. J. Climatol., 17, 1683– 744–762. 1695. Cressman, G. P., 1959: An operational analysis system. Mon. Wea. Maddox, R. A., 1980: Mesoscale convective complexes. Bull. Rev., 87, 367–374. Amer. Meteor. Soc., 61, 1374–1387. Fritsch, J. M., R. J. Kane, and C. R. Chelius, 1986: The contribu- Matsumoto, J., 1989: Heavy rainfalls over east Asia. Int. J. Clima- tion of mesoscale convective weather systems to the warm- tol., 9, 407–423. season precipitation in the United States. J. Climate Appl. Moncrieff, M. W., 1992: Organized convective systems: Archetyp- Meteor., 25, 1333–1345. al dynamical models, mass and momentum flux theory, and Goldreich, Y., H. Moses, and D. Rosenfeld, 2004: Radar analysis parameterization. Quart. J. Roy. Meteor. Soc., 118, 819–850. of cloud systems and their rainfall yield in Israel. Israeli J. Ramage, C. S., 1952: Variation of rainfall over south China Earth Sci., 53, 63–76. through the wet season. Bull. Amer. Meteor. Soc., 33, 308– Johnson, R. H., and P. E. Ciesielski, 2002: Characteristics of the 311. 1998 summer monsoon onset over the northern South China Rodgers, E. B., R. F. Adler, and H. F. Pierce, 2000: Contribution Sea. J. Meteor. Soc. Japan, 80, 561–578. of tropical cyclones to the North Pacific climatological rain- Krishnamurti, T. N., and C. M. Kishtawal, 2000: A pronounced fall as observed from satellites. J. Appl. Meteor., 39, 1658– continental-scale diurnal mode of the Asian summer mon- 1678. soon. Mon. Wea. Rev., 128, 462–473. ——, ——, and ——, 2001: Contribution of tropical cyclones to the North Atlantic climatological rainfall as observed from ——, H. S. Bedi, and M. Subramaniam, 1989: The summer mon- satellites. J. Appl. Meteor., 40, 1785–1800. soon of 1987. J. Climate, 2, 321–340. Wang, B., R. Wu, and K.-M. Lau, 2001: Interannual variability of Laing, A. G., J. M. Fritsch, and A. J. Negri, 1999: Contribution of the Asian summer monsoon: Contrasts between the Indian mesoscale convective complexes to rainfall in Sahelian Af- and the western North Pacific–east Asian monsoons. J. Cli- rica: Estimates from geostationary infrared and passive mi- mate, 14, 4073–4090. crowave data. J. Appl. Meteor., 38, 957–964. Wang, J.-J., 2004: Evolution and structure of the mesoscale con- Lau, K.-M., and Coauthors, 2000: A report of the field operations vection and its environment: A case study during the early and early results of the South China Sea Monsoon Experi- onset of the Southeast Asian summer monsoon. Mon. Wea. ment (SCSMEX). Bull. Amer. Meteor. Soc., 1261–1270. 81, Rev., 132, 1104–1120. Li, J., Y.-L. Chen, and W.-C. Lee, 1997: Analysis of a heavy rain- Xie, P., and P. A. Arkin, 1997: Global precipitation: A 17-year fall event during TAMEX. Mon. Wea. Rev., 125, 1060–1081. monthly analysis based on gauge observations, satellite esti- Lin, S.-C., 2005: The materials flow analysis of the water supply mates, and numerical model outputs. Bull. Amer. Meteor. status from Nan-shih and Bei-shih River (in Chinese). M.S. Soc., 78, 2539–2558. thesis, National Taiwan University, 108 pp. [Available online Yeh, H.-C., and Y.-L. Chen, 1998: Characteristics of rainfall dis- at http://etds.lib.ntu.edu.tw/main/index.] tributions over Taiwan during the Taiwan Area Mesoscale ——, and M.-T. Kueh, 2003: A modeling diagnosis of the devel- Experiment (TAMEX). J. Appl. Meteor., 37, 1457–1469. opment of mesoscale convective systems over the South Yen, M.-C., and T.-C. Chen, 2000: Seasonal variation of the rain- China Sea during the summer monsoon onset in 1998. Terr. fall over Taiwan. Int. J. Climatol., 20, 803–809. Atmos. Oceanic Sci., 14, 369–399. Yoon, J.-H., and T.-C. Chen, 2005: Water vapor budget of the Llasat, M. C., and M. Puigcerver, 1997: Total rainfall and convec- Indian monsoon depression. Tellus, 57A, 770–782.