The Effect of Tropical Cyclones on Southwest Monsoon Rainfall in The

Journal of the Meteorological Society of Japan, Vol. 89A, pp. 123--139, 2011. 123 DOI:10.2151/jmsj.2011-A08

The E¤ect of Tropical Cyclones on Southwest Monsoon Rainfall in the Philippines

Esperanza O. CAYANAN

Philippine Atmospheric, Geophysical and Astronomical Services Administration (PAGASA), Quezon City, Philippines

Tsing-Chang CHEN

Department of Geological and Atmospheric Sciences, Iowa State University, Ames, Iowa, U.S.A.

Joseﬁna C. ARGETE

Institute of Environmental Science & Meteorology, College of Science, University of the Philippines, Quezon City, Philippines

Ming-Cheng YEN

Department of Atmospheric Science, National Central University, Chung-Li, Taiwan

and

Prisco D. NILO

Philippine Atmospheric, Geophysical and Astronomical Services Administration (PAGASA), Quezon City, Philippines

(Manuscript received 30 May 2010, in ﬁnal form 6 October 2010)

Abstract

Intense southwest monsoon (SWM) rainfall events causing massive landslides and flash floods along the western sections of the Philippines were studied. These rainfall events, are not directly coming from the tropical cyclones (TCs) for they are situated far north to northeast of Luzon Island. The heavy rainfall is hypothesized as caused by the interaction of strong westerlies with the mountain ranges along the west coast of Luzon that pro- duces strong vertical motion and consequently generates heavy rainfall. Four of heavy SWM rainfall cases were examined to determine how the presence and position of tropical cyclones in the Philippine vicinity a¤ect these SWM rainfall events; three cases with TC of varying positions within the Philippine area of responsibility (PAR) and the fourth case without TC. Using a spatial Fourier decomposition approach, the total streamfunction is decomposed into two flow regimes: monsoon basic flow (Waves 0–1) and tropical cyclone perturbation flow (Waves 2–23) over a domain of (20E–140W, 5S–35N). The purpose of this flow decomposition is to determine the latter’s e¤ect on or contribution to the monsoon activity. The analysis utilized the NCEP Final (FNL) data with 1 long. 1 lat. resolution. Results show that the tropical cyclones over the Pacific Ocean located northeast of Luzon generate strong southwesterly winds over the west coast of Luzon. These in addition to the southwesterlies from the basic flow strengthened the southwest winds that interact with the high Cordillera Mountain ranges

Corresponding author: Esperanza O. Cayanan, Philip- pine Atmospheric, Geophysical and Astronomical Services Administration (PAGASA), Science Garden, Agham Road, Diliman, Quezon City 1100, Philippines. E-mail: [email protected] 6 2011, Meteorological Society of Japan 124 Journal of the Meteorological Society of Japan Vol. 89A

along the west coast of Luzon. When the tropical cyclone is located north or north-northwest of Luzon, it generates northwesterlies which converge with the southwesterlies from the basic ﬂow. This results to enhancement of rising motion over western Luzon. The much stronger westerlies are then forced to rise above the mountains resulting to strong vertical motion that brings about heavy rainfall.

1. Introduction to northeast of Luzon. A good example of this is the big flash floods on the last week of August Lying in the vast Pacific Ocean and regarded as 2004 which inundated four western and central typhoon belt, the Philippines experiences an aver- provinces in Luzon putting them under state of age of 19 to 20 tropical cyclones per year and the calamity. The massive floodings were reported as peak occurs during the months of July and August. caused by heavy monsoon rains induced by Ty- From later part of May to September, southwest- phoon Marce which was over the Pacific Ocean erly winds or southwest monsoon (Habagat) prevail east to northeast of Luzon Island. A similar heavy over the country (Flores and Balagot 1969; Wil- rainfall event happened on the last week of August liams and Jung 1993), hence typhoon season coin- 1990 was caused by intense SWM due to the pas- cides with the southwest monsoon months. sage of three (3) consecutive TCs north of the Phil- Monsoon is traditionally defined in terms of ippines (Cayanan 1995). Another devastating event the annual reversal of the prevailing wind regime of this kind was the historical Philippine floods in (Ramage 1971) and a contrast between a rainy July 1972 where the Lingayen Gulf and the Manila summer and a dry winter. It is driven by the di¤er- ential heating between the huge land mass of Asia (with Tibetan Plateau as a raised platform of heat source) and the huge water mass of Indian Ocean. During northern summer or southwest monsoon, winds began to blow from the cooler sea towards the land (Das 1986) arriving over the western Phil- ippines as southwesterlies. On Philippine topography, mountain ranges with heights more than 500 m cover almost half of the Luzon Island. On the east coast of northern and central Luzon, lies the Sierra Madre Range (from Cagayan to Quezon Province). Along the western coast of northern Luzon, the Cordillera Mountain Range runs down to central Luzon (the longest on Fig. 1) and linked by the Cordillera Range with Sierra Madre Range on the east. Statistics show that 43% of the average annual rainfall in the Philippines is derived from the southwest monsoon (SWM; Asuncion and Jose 1980). The average SWM rainfall distribution has a maximum rainfall of about 2,000 mm (or greater) con- centrated over mountain areas (Iba, 3,011 mm and Baguio, 2,645 mm) and heavy rains in the range 1,500–1,999 mm are found along the western portion of Luzon (Asuncion et al. 1981). During the SWM season, the heavy rainfall events in the western sections of Luzon cause flash floods over low lying areas and landslides along mountain slopes. In these disastrous events, it is noted that the heavy Fig. 1. Location of western Luzon stations rainfall are not directly coming from the tropical plotted on topographical map of the Phil- cyclones (TCs) but due to the SWM surge enhanced ippines. (Topographical map from Aka- by the presence of tropical cyclones situated north saka, 2008 PMS presentation). February 2011 E. O. CAYANAN et al. 125

Bay, due to flood waters joined to become an in- land sea (Gordon 1973). For the same flood event, Asuncion and Jose (1980) noted that this unusual condition was brought about by the presence of tropical cyclones that enhanced the monsoon activity at the time when the tropical disturbance was north to northeast of the region. This kind of weather condition was frequently observed, however, no scientific study has yet been done to ex- plain this phenomenon. It is hypothesized that the heavy rainfall in the western section of Luzon is caused by the monsoon southwesterly flow enhanced by the TC and its interaction with the mountains along the west coast. The strong southwesterlies are said to be a result of the combination of the monsoon southwesterlies and the westerlies generated by TCs present in the vicinity of Luzon Island. During the rainy season, Fig. 2. Coverage of the Philippine area of rainfall amount on the west coasts of the Philip- responsibility (PAR), bounded by 25N– pines peaks when southwesterlies are strongest 120E, 25N–135E, 5N–135E, 5N– (Akasaka et al. 2007). The strength of these wester- 115E, 15N–115E and 21N–120E). lies is modulated by the position of the northward propagating 30–60 day monsoon trough and western north Pacific anticyclone (Chen et al. 2000; by 119E–122E, 14N–19N) represented by ar- Chen and Murakami 1988). Four TCs moved rows drawn to scale. The heavy rainfall events iden- through the vicinity of the Philippines were ana- tified (shaded) in the series and the corresponding lyzed to substantiate the hypothesis introduced tracks of tropical cyclones during the events are above. also included in the figures. In the 5-year time series, 19 heavy rainfall events 2. Data and methodology were identified and investigated. Out of the 19 A five-year (2002–2006) daily rainfall data at events, 17 cases or 89.5% involve a tropical cyclone 11 western Luzon stations in the Philippines during in the PAR. This proves that tropical cyclones con- the southwest monsoon months, June–September, tribute much to the enhancement of rainfall during was analyzed. The rainfall stations used are shown SWM season. From the 17 events with tropical cy- in Fig. 1. The tracks of tropical cyclones within clones, 11 cases involved TCs over the ocean in the the Philippine area of responsibility (PAR) during vicinity of Luzon Island and only 6 events with TC the period were also considered in the study. The crossing the Philippine Islands. On the wind direc- PAR, as illustrated Fig. 2, is bounded by the lines tion, 14 events have SW winds up to 850 hPa level. joining the following points: 25N–120E, 25N– The remaining 5 events with variable winds (N, 135E, 5N–135E, 5N–115E, 15N–115E and NW, NE) involve TCs crossing the Philippine Is- 21N–120E. In the rainfall analysis, the 5-day lands. This shows that heavy rainfall events occur rainfall running mean was plotted in a time series when the winds over the western sections of Luzon shown in Fig. 3. In this study, heavy rainfall event is southwesterly. is defined or described as those with total rainfall of Four cases of heavy rainfall events are selected at least 200 mm day1 at 11 stations observed for for presentation and discussion in this study. The 5 consecutive days in the time series. The presence heavy rainfall events are selected to represent di¤er- of tropical cyclones (represented by small circles ent conditions/scenarios of heavy rainfall events. in yellow-depression, blue-storm or red-typhoon) The presence and positions of tropical cyclones are within the PAR on day to day basis is also indi- considered in the selection. Three cases with TC of cated at the top of each part of Fig. 3. Below the di¤erent positions or tracks are selected and one cyclone are the average daily wind speed and direc- case of rainfall event without TC in the PAR. This tion at 850 hPa level over western Luzon (bounded is done to show how the positions of the tropical 126 Journal of the Meteorological Society of Japan Vol. 89A

Fig. 3a–b. February 2011 E. O. CAYANAN et al. 127

Fig. 3c–d. 128 Journal of the Meteorological Society of Japan Vol. 89A

Fig. 3. Time series of 5-daily rainfall running mean (bar graph) at 11 western Luzon stations with average 1 850 hPa wind ( f ¼ 20 m s ) at 119E–122E, 14N–20N and tropical cyclones (f depression, f storm or f typhoon) within the PAR. Shaded portions in the series are heavy rain events together with corresponding track of tropical cyclones. Green shaded events are the selected cases for discussion. The lower portion of the figure is the latitude-time cross-section (y-t diagram of zonal wind (U ) at 120E, with latitudinal positions of monsoon trough (þ) at 125E. (a) June–September 2002, (b) June–September 2003, (c) June–September 2004, (d) June–September 2005 and (e) June–September 2006. cyclones a¤ect the intensity of southwest monsoon The total 5-day rainfall for all the cases are pre- rainfall. The first case (5–9 July 2002) involves a sented in Fig. 4b while Fig. 4a illustrates the aver- TC far NE of Luzon and another over Taiwan. age wind distribution and the tracks of the TCs. The second case (19–23 July 2003) has a TC cross- A spatial Fourier decomposition approach is ap- ing northern Luzon and the daily winds up to plied to decompose the wind wave pattern into the 850 hPa are not totally southwesterly. The third base flow and the tropical cyclone perturbation case (23–27 August 2004) involves a TC located flow. This is done to delineate the e¤ects or contri- east to north-northeast of Luzon over the Pacific butions of the two flow regimes on the heavy rain- Ocean and the fourth case with no TC within the fall over the western section of Luzon. Streamfunc- PAR. The rainfall time series with the selected cases tion is the parameter used in the analysis since it shaded green and the corresponding tropical cy- can represent the wind velocity field with single clones involved are shown in Fig. 3. To prove the variable and it can very well depict the circulation hypothesis that the heavy rainfall is due to strong in the tropics (Chen and Chen 1990). The NCEP westerlies, the latitude-time cross-section charts Final (FNL) gridded data at 950 hPa level with 1 (y-t diagram) of the zonal wind (U ) at 850 hPa long. 1 lat., a resolution of one (1) degree, was were prepared and presented below the time series. utilized in the diagnostic analysis. In order to February 2011 E. O. CAYANAN et al. 129

Fig. 4. (a) The average 850 hPa windﬁeld and tracks of tropical cyclones of the four cases of rainfall events. (b) distribution of observed total rainfall of the four cases of heavy rainfall event. Dark solid lines are the average trough positions.

clearly see the e¤ect of the TC and topography, the ws ¼ Vs ‘h, where ws and Vs are vertical velocity vertical velocity over the island of Luzon was com- and wind vector at surface, and h is topography. puted using the equation of continuity. As the bot- The 1 1 FNL data were interpolated into the tom boundary condition, we used the equation 18 km 18 km grid in terms of a 16-point Bessel 130 Journal of the Meteorological Society of Japan Vol. 89A interpolation scheme which was designed by Jenne also shows the daily positions the two tropical cy- (1975), following the mathematical theory devel- clones ({|) located along the monsoon trough. These oped by Whittaker and Robinson (1944). Then, TCs strengthened the monsoon trough which en- the vertical motion is computed with the 18 km res- hanced the SW flow and consequently the rainfall olution in the domain bounded by 118 E–124 E and over western Luzon as shown in Fig. 4b. 14 N–19 N. The rainfall along the windward side The Case 2 rainfall event on 19–23 July 2003 was of the western Luzon Island is proportional to ver- selected to study the impact of a TC crossing north- tical velocity, caused by the interaction of the western Luzon to the SWM rainfall. In Fig. 3b, the y-t erly flow with the north-south oriented mountains. diagram of the 850 hPa zonal wind showed strong Thus, the amount of additional rainfall derived westerlies below 15N (southern part of the ty- from the tropical cyclone westerlies is proportional phoon) and easterlies above 18N (northern part to the di¤erence of vertical motion between those of the typhoon) during the event (blue encircled generated by the base flow and the total flow area). It is also noted that strong westerlies corre- (which includes the basic and perturbation flow). spond to heavy rainfall events in the rainfall time series in the upper panel of Fig. 3b. It is again im- 3. The SWM heavy rainfall events portant to note that the rainfall only covers the sta- The four cases of heavy rainfall events selected tions from 14Nto19N latitude. The 30–60 day are described and compared to see how the pres- monsoon trough oscillation is also projected in this ence and positions of the TCs a¤ect the rainfall period by strong westerlies on the third weeks of along the western sections of Luzon. July, August and September. The average 850 hPa Case 1: July 5–9, 2002, involves two tropical cy- wind field for the 5-day event in Fig. 4a illustrates clones (the upper panel of Fig. 3a): TY Gloria with southwesterlies over the Philippines converging a north-northwest track from July 7–9 to the far with the easterlies and forming the ITCZ or mon- east of Luzon near the boundary of the PAR and soon trough. It is noted that the winds over the TS Hambalos on July 9 over Taiwan area. This western section of Luzon are more southerly than case is the heaviest and longest rainfall event in the westerly. From the rainfall pattern in Fig. 4b, it is time series of 2002 summer months. It has a dura- clearly seen that the rainfall in Case 2 is due to the tion of about 10 days (05–14 July 2002) but only TC that passed over Luzon Island. the first 5-day rainfall activity was investigated. The Case 3 rainfall episode in August 2004 is The maximum 24-hr rainfall (346 mm) was re- considered to be one of the wettest SWM events ex- corded in Baguio City, the station of highest alti- perienced in northern Luzon. The time series in the tude of about 1,500 meters above mean sea level. upper panel of Fig. 3c shows Case 3 event has The total 5-day rainfall distribution in Fig. 4b clearly the highest rainfall for the summer of 2004 as well shows that the rainfall is maximum over western as heaviest among the four cases being studied Luzon decreasing eastward. The strength and ex- (Fig. 4b). This event involves a TC which origi- tent of the 850 hPa westerlies during the heavy nated over the Pacific Ocean moved northwest and rainfall events are shown in the latitude-time cross passed north of Luzon Island. The y-t diagram in section chart (y-t diagram) of the zonal wind (U )at the bottom panel of Fig. 3c clearly shows the strong 120E longitude (the bottom panel of Fig. 3a). It is westerlies (blue encircled) at longitude 120E from noted that strong westerlies in the y-t diagram cor- 10N–22N latitude correspond to maximum rain- respond to maximum rainfall in the time series (the fall on August 23–27, 2004 in the time series. The upper panel of Fig. 3a). The latitudinal positions of second maximum (labeled B)at5N–15N latitude the monsoon trough (þ) indicated in the y-t dia- between August 1–15 is not reflected in the rainfall gram reveals that at 125E longitude, the trough is time series in the upper panel of Fig. 3c because the present to the north of the maximum westerly. It is rainfall plotted only cover stations from 14N– noted that there is about 35 days di¤erence between 19N. Above the westerlies are strong easterlies at the two maximum westerlies A and B, and around 25N–30N north of the typhoon. Troughs (de- 30 days between maxima B and C which supports noted by þ) are also noted to be present north of the 30–60 day oscillation of monsoon trough strong westerlies. The 30–60 day oscillation is re- (Chen et al. 2000). The average 850 hPa wind field flected by the maximum westerlies; B and D for 30 in Fig. 4a indicates that the southwesterly flow over days, and A and C for 60 days. For the winds, the western Philippines is stronger than 10 m s1.It strong southwesterlies prevailed over the Philip- February 2011 E. O. CAYANAN et al. 131 pines from the surface up to the 850 hPa level as in- ing mechanism resulting to heavy rainfall over the dicated in the average wind field in Fig. 4a. As indi- western Luzon. It is also important to note that cated in the figure, another cyclone, TY Chaba ap- the rainfall events follow the 30–60 day oscillations proached on Aug 25–27 but did not enter the PAR. of the monsoon trough. This information on mon- There actually are two cyclones within the domain soon trough propagation could serve as a good that are the embedded in the trough which enhanced basis for monsoon rainfall forecasting. the southwest monsoon flow during the event. 4. Diagnostic analysis For Case 4, the August 2006 case does not in- Diagnostic analysis was applied to SW monsoon volve a tropical cyclone within the PAR. The rain- heavy rainfall events with tropical cyclones of fall event which is the least among the cases studied varying positions or tracks (Cases 1–3) and with- occurred on 17–21 August 2006. The August 2006 out tropical cyclone in the PAR (Case 4). Using rainfall event as seen in the time series in the upper the spatial Fourier decomposition approach, the panel of Fig. 3e and in the total rainfall distribution 950 hPa streamflow was separated into the basic in Fig. 4b is comparably weaker than the other monsoon flow (Waves 0–1) and the tropical cy- events under study. The maximum 24-hour rainfall clone perturbation flow (Waves 2–23) to see how observed in Baguio and Dagupan (140 mm) is each of them a¤ect or contribute to the rainfall much less compared to 346 mm (Baguio) in July over western Luzon. Although the analysis in- 2002 and 400 mm (Vigan) in August 2004. The volved the expanded horizontal domain of 20E– direct relationship of the strength of westerly winds 140W, the figures (Figs. 5, 8, 10) presented only and rainfall is still evident in the y-t diagram of the cover the target domain of 100E–140E. In the zonal wind (the bottom panel of Fig. 3e) and rain- streamfunction diagram, the dark shaded portions fall time series (the upper panel of Fig. 3e). The are the positive values which correspond to anti- Case 4 westerly winds on 17–21 August 2006 in cyclonic flow and the negative areas (unshaded) as the y-t diagram (blue encircled area in the y-t dia- cyclonic flow. gram) are weaker as well as the corresponding rainfall intensity compared to the other rainfall events. 4.1 Tropical cyclone north-northeast of Luzon The 30-day oscillation is also noted to occur on the Case 1 (5–9 July 2002), involves two tropical third week of July (A) and August (B). The rainfall cyclones in the PAR: TY Gloria which originated corresponding to B is not reflected in the time series over the Pacific Ocean east of northern Luzon, because the area a¤ected is below 14N latitude moved on a NNW track from July 7–9; and TS which is not covered by the rainfall stations plotted. Hambalos which barely stay on July 9 over north Event A which involves tropical cyclone is noted to of Luzon in the vicinity of Formosa Strait. The have intense rainfall compared to Case 4. The aver- maximum rainfall occurred on July 8. In order to age 850 hPa wind field in Fig. 4a only shows a save space we shall focus our presentation on this small cyclonic circulation at the end of the mon- date. Upon application of the spatial Fourier de- soon trough located north-northeast of Luzon. composition, the streamfunction diagrams are pre- This trough is instrumental to the generation of sented in Fig. 5. The actual or total streamfunction southwest wind flow over the country. The average (cT ), the basic flow (cB), and the tropical cyclone position of monsoon trough compared to the other perturbation flow (cTC) are shown in the first row heavy rainfall events is more to the east (up to (Fig. 5a). In the basic flow, westerly to southwest- 129E) and at higher latitude (in the vicinity of erly winds flow over Luzon Island for the whole du- 25N). It is also noted that the soutwesterlies are ration of the rainfall event. A trough (heavy dashed weaker (only about 10 m s1). line) is also dominating, moving northward with From the four cases of rainfall events selected, it time extending from southern China to the Philip- is important to stress that the rainfall in Cases 1 pine Sea. On the other hand, the TC perturbation and 3 are not directly coming from the cloud bands flow (cTC) shows a TC far NNE of Taiwan with a of the tropical cyclones but due to the strong south- trough extending down to Formosa Strait. The west flow over western Luzon. As mentioned combination of the basic and the TC flow deepened earlier, the southwest flow is hypothesized to be the trough in the actual flow and the northward strengthened by the addition of the westerlies gen- movement of the TC made the east end of the erated by the tropical cyclone and the orographic trough shift to the north (Fig. 5a). The streamline e¤ect of the Cordillera mountains enhanced the lift- analysis in Fig. 5b shows southwesterly winds pre- 132 Journal of the Meteorological Society of Japan Vol. 89A

Fig. 5. (a) Stream function (c) diagram at 950 hPa level for 8 July 2002. By Fourier analysis, total stream function (cT ) is decomposed into basic flow, cB (Waves 0–1) and tropical cyclone streamflow, cTC (Waves 2–23). The heavy dashed lines represent monsoon troughs. (b) Daily mean streamline chart superimposed with p vertical velocity (o) for 8 July 2002 generated from actual windflow, basic windflow and TC wind- 1 flow. (c) Vertical motion (W,ms ) computed from Ws ¼ Vs ‘h; WT for total flow, WB for basic flow, and WTC for tropical cyclone flow for 8 July 2002 of case1. vailing over Luzon in all the regimes (total flow, The vertical motion (W in m s1) over western basic flow and TC perturbation flow) during the Luzon which is computed using zonal wind and to- whole period of rain event. Convergence of north- pography is shown in Fig. 5c; the first row, WT is westerlies was generated when the TC is located the total vertical motion, the second panel, WB north to northwest of Luzon and southwesterlies is the vertical velocity of the basic flow and third from the basic flow over western Luzon. The com- panel, WTC corresponds to vertical motion in the bination of the westerlies from the basic flow and tropical cyclone flow. The positive values (shaded) those generated by the TC flow strengthened the denote upward motion and the negative values total westerlies and the vertical wind over western mean downward motion. It is visible that in the ba- Luzon in the actual flow. sic monsoon flow, strong vertical motion occurred February 2011 E. O. CAYANAN et al. 133

Table 1. Maximum values of 950 hPa vertical motion Strong vertical motion due to westerlies and (103 ms1) at 120E derived from zonal wind and topography correspond to the increase in the topography for Case 1: 5–9 July 2002. amount of rainfall. This direct proportionality of 5 July 6 July 7 July 8 July 9 July vertical motion and rainfall is depicted in the latitude-time cross section (y-t diagram) of vertical Total 18.0 12.0 22.0 30.0 32.0 motion (Fig. 6a) and observed station rainfall (Fig. Basic 10.0 12.0 20.0 22.0 20.0 6d). Note that maximum rainfall is observed TC 6.0 0.0 2.0 8.0 12.0 around 17N latitude where strong vertical motion also exists on the 4th and 5th day (July 8–9). The increase is due to the contribution of the TCs on July 7–9. during the rainfall event. In fact, the tropical cy- Another event with a TC NE of Luzon is Case 3: clone induced vertical motion over the western sec- 23–27 Aug. 2004. It involves one TC within the tions on July 8. On the July 5, a cyclonic circulation PAR with a northwest track from the Pacific Ocean north of Taiwan enhanced the southwest flow over from Aug. 20–25. Another TC developed northeast western Luzon while the two TCs over Taiwan and outside of the PAR and moved NNW towards Philippine Sea are responsible on July 8. The nu- Japan from Aug. 25–27. The spatial Fourier de- merical contributions of the TC and the basic flow composition was also applied to Case 3 and the re- in the vertical motion over the entire period of the sulting vertical motions of the total, basic and TC event for 5–9 July 2002 are presented in Table 1. flows are summarized in Table 3. It is noted from These are the maximum point values at 120E the table that only the TC within the PAR on from the grid data of Fig. 5. The TC contributes Aug. 23–25 contributed to the vertical motion. about to 44% (6/18) for 1st day (July 5), 27% (8/ The basic southwest flow is sole responsible for the 30) for 4th day (July 8) and 32.5% (12/32) for 5th vertical motion on Aug. 26–27. Fig. 7a–c (vertical day. motion) and Fig. 7d (rainfall) show that the maxi-

Fig. 6. Time-latitude cross section of vertical motion (m s1) at 120E, 950 hPa level of (a) total flow, (b) basic flow and (c) tropical cyclone flow, and (d) rainfall (mm day1)at 120E for case 1: 5–9 July 2002. 134 Journal of the Meteorological Society of Japan Vol. 89A

Table 2. Maximum 950 hPa vertical motion Table 3. Maximum 950 hPa vertical motion (103 ms1) at 120E derived from zonal wind and (103 ms1) at 120E derived from zonal wind and topography for Case 2: 19–23 July 2003. topography for Case 3: 23–27 Aug. 2004. 19 July 20 July 21 July 22 July 23 July 23 Aug 24 Aug 25 Aug 26 Aug 27 Aug Total/Actual 10.0 0.8 6.8 14.0 1.0 Total/Actual 20.0 22.0 22.0 22.0 22.0 Basic Flow 8.0 8.3 4.6 6.0 8.0 Basic Flow 10.0 18.0 20.0 22.0 22.0 TC Flow 18.0 7.5 2.2 8.0 9.0 TC Flow 10.0 4.0 2.0 0.0 0.0

mum rainfall on Aug. 27 is mainly from the strong perturbation flow clearly illustrates the passage of basic southwest flow. Typhoon Harurot which was over northern Luzon 4.2 Tropical cyclone crossing northern Luzon on July 22. In the streamlines analysis south- For Case 2: 19–23 July 2003, which involves a westerly winds prevailed over western Luzon dur- TC crossing northern Luzon, Fig. 8a shows the to- ing the 5-day period in the basic flow. However the tal streamfunction left panel decomposed into the winds are weaker because the streamlines are far basic flow (middle panel) and tropical cyclone per- apart. The addition of the tropical cyclone flow turbation flow (right panel) on July 22. In the total strengthened the westerlies making the streamlines flow, a monsoon trough (NW-SE oriented) extends closer in the total or actual flow. Vertical winds from southern China down to the Philippines. The are only noted to be present over western Luzon trough weakens on the last two days as the two cy- during the passage of the typhoon over the island clonic circulations embedded in the trough moved on July 22 (Fig. 8b). The vertical motion generated to the northwest. The monsoon trough is also de- by the interaction of westerlies and topography are picted in the basic flow but it is relatively shallow illustrated in Fig. 8c and summarized in Table 2. It compared to the total flow. The tropical cyclone is seen in the total flow that strong positive vertical

Fig. 7. Same as Fig. 6, except for case 3: 23–27 Aug 2004. February 2011 E. O. CAYANAN et al. 135

Fig. 8. Same as Fig. 5, except for 22 July 2003. motion on the western section of Luzon are only ented. In the basic flow, Fig. 10a, the monsoon observed on July 21 (not shown) and maximum on trough on August 20 is positioned horizontally in July 22 during the passage of the TC over the W-E direction from southern China to southern island. The corresponding relationship vertical Japan. In the TC flow, the trough which is the rem- motion and rainfall is illustrated in Fig. 9. nant of the TC that moved NE expanded to SW covering Luzon. The combination of the basic and 4.3 No tropical cyclone in the PAR the TC flow resulted to a monsoon trough in the In Case 4 rain event, no tropical cyclone is within total flow which is initially extending from southern the PAR but initially, there is a cyclonic vortex far Japan oriented NE–SW becoming horizontally ori- to the northeast at 132E, 34N (over southern ented from Pacific Ocean extending to the west over Japan) on the 1st day (August 17) (total flow) and Hainan on the 5th day. (TC flow). The TC moved to the NE leaving a In the streamline analysis (Fig. 10b) and verti- trough extending from southern Japan to the SW cal wind diagram, southwesterly winds blow over passing Taiwan. The trough is initially NE-SW ori- northern Luzon in the basic flow (center panel of 136 Journal of the Meteorological Society of Japan Vol. 89A

Fig. 9. Same as Fig. 6, except for case 2: 19–23 July 2003.

Fig. 10b). Vertical winds are also noted over the minimum on August 21 (Fig. 11a). This upward western sections of Luzon. In the TC flow, wind motion is caused by the basic flow which is highest convergence are noted over northwestern Luzon on August 20 (Fig. 11b). Lifting motion is mini- and moderate wind convergence over southwestern mum on August 21 because of the sinking motion Luzon. The combination of VB and VTC resulted generated by Wave Numbers 2–23 (Fig. 11c). in slightly stronger westerlies over northwestern Based alone on the rainfall observed during the Luzon. A change in wind direction over north- four di¤erent cases of events in Fig. 4b, it is already western Luzon is also noted in the total flow; the visible that the case without TC in the PAR (Case wind is initially southwesterly to westerly then be- 4) has the least amount. Also, the rainfall distribu- came northwesterly on the last two days. tion pattern already revealed that the rainfall in Checking on the vertical motion generated by Case 2 is directly due to the crossing TC in Luzon. westerlies and topography, the total flow in Fig. After the diagnostic analysis, it is clearly shown and 10c, shows that there is a total upward motion of explained how the presence and position of tropi- about 12–14 103 ms1 over western Luzon. cal cyclone a¤ect the rainfall over western Luzon. The lifting motion is clearly seen to be due to the When the TC is located to the northeast of Luzon, basic flow. No vertical motion at all is generated it generates southwesterlies that add strength to the by Waves 2–23 (supposed to be tropical cyclone southwesterlies from the basic flow. When the TC is flow) in the target area throughout the duration of located north to northwest of Luzon, it generates the rainfall event. The values of vertical motion northwesterlies or westerlies which converge with over western Luzon are listed in Table 4. The lati- the southwesterlies from the basic flow resulting to tude-time cross section of vertical motion at 120E rising motion. In addition to the strengthened west- (Fig. 11a) clearly summarizes the daily contribution erlies accumulated from the TC and the basic flow, of the basic flow and the tropical cyclone flow on the convergence induced greater rising motion over vertical motion. There is maximum upward motion the mountain areas. It is shown in Case 4 that at latitude 17N on August 18 and August 20 and SWM rain event without TC (Fig. 11d) occurs February 2011 E. O. CAYANAN et al. 137

Fig. 10. Same as Fig. 5, except for 20 Aug 2006. due to strong southwesterlies. The westerlies are Based on previous studies and on the observational strengthened by the trough located northeast of analysis, strong westerlies up to 850 hPa over west- Luzon. ern Luzon correspond to heavy rainfall. The heavy rains are therefore attributed to the strong and hu- 5. Conclusion mid SW wind blowing over western Luzon but the In the study, it is observed that majority (about presence of the tropical cyclones also has an indi- 89.5%) of the heavy rainfall events involve tropical rect dominating e¤ect on the intensity of rainfall. cyclones. However, only 31% of these events in- It was hypothesized that the tropical cyclones over volve TCs crossing Luzon Island which means that the ocean generate westerlies in addition to the the heavy rains during rest (69%) of the events are SW monsoon airstream. The mountain range along not directly coming from the tropical cyclone rain western Luzon is also a contributing factor because bands. Excessive rainfalls were also recorded at the it serves as the lifting mechanism for the humid western stations even if the tropical cyclones are westerlies to rise and condense into clouds. over the ocean north-northeast of Luzon Island. To prove the hypothesis, the tropical cyclone 138 Journal of the Meteorological Society of Japan Vol. 89A

Fig. 11. Same as Fig. 6, except for case 4: 17–21 Aug 2006.

Table 4. Maximum 950 hPa vertical motion motion that brings about heavy rainfall. However, (103 ms1) at 120E derived from zonal wind and when the southwest flow is very strong, heavy rains topography for Case 4: 17–21 Aug. 2006. are possible even when there is only a trough in- 17 Aug 18 Aug 19 Aug 20 Aug 21 Aug stead of a cyclone northeast of Luzon. Total/Actual 12.0 14.0 12.0 14.0 6.0 Acknowledgement Basic Flow 12.0 14.0 16.0 20.0 16.0 TC Flow 0.0 0.0 4.0 6.0 10.0 This research study is undertaken through the support and assistance of the Philippine Atmo- spheric, Geophysical and Astronomical Services flow was isolated from the basic monsoon flow Administration (PAGASA), Department of Science using the spatial Fourier decomposition. This was and Technology (DOST), the Institute of Environ- done to determine the contribution of each to mental Science and Meteorology (IESM), College SWM rainfall. Results reveal that tropical cyclones of Science, University of the Philippines, the Atmo- located to the northeast of Luzon generate strong spheric Science Department, National Central southwesterlies over western Luzon which add to University, Taiwan and Department of Geological the strength of the southwesterlies from the basic and Atmospheric Sciences, Iowa State University, flow. Meanwhile, tropical cyclones located north U.S.A. Comments/suggestions made by two re- to northwest of Luzon generate northwesterlies viewers are helpful to improve the presentation of which converge with southwesterlies and enhance this paper. The technical support provided by Mr. the rising motion over the windward slopes of the Jenq-Dar Tsay to prepare this manuscript is highly mountain. These in addition to the southwesterlies appreciated. from the basic flow strengthened the southwest winds that interact with the high Cordillera Moun- References tain ranges along the west coast of Luzon. The Akasaka, I., W. Morishima, and T. Mikami, 2007: Sea- much stronger westerlies are then forced to rise sonal march and its spatial di¤erence of rainfall in above the mountains resulting in stronger vertical the Philippines, Int. J. Climatol., 27, 715–725. February 2011 E. O. CAYANAN et al. 139

Asuncion, J. F., A. M. Jose, and R. Buan, 1981: Onset, No. 613, World Meteorological Organization, peak and recession of the rains during the NE and Geneva, ISBN 92-63-10613-4, 159 pp. SW monsoons in the Philippines, PAGASA Tech- Flores, J. F., and V. F. Balagot, 1969: Climate of the nical Report, 40 pp. Philippines. Arakawa H (ed.), Climates of Northern Asuncion, J. F., and A. M. Jose, 1980: A study of the and Eastern Asia, World Survey of Climatology, 8, characteristics of the northeast and southwest mon- Elsevier, 159–213. soons in the Philippines. NRCP Assisted Project, Gordon, A. H., 1973: The great Philippine ﬂoods of 49 pp (available from PAGASA). 1972. Weather, 28, 404–415. Cayanan, E. O., 1995: A study of long duration rainfall Jenne, R., 1975: Format for Northern Hemisphere octag- episode in the Philippines during the southwest onal grid data. NCAR Tech. Doc., 8pp [available monsoon in the Philippines. M.Sc. Thesis, College online at http://dss.ucar.edu/datasets/common/ of Science, University of the Philippines, 109 pp. nmc.octagon/format]. Chen, T.-C., M.-C. Yen, and S. P. Weng, 2000: Interac- Ramage, C. S., 1971: Monsoon Meteorology. Academic tion between the summer monsoons in East Asia Press, 296 pp. and the South China Sea: Intraseasonal monsoon Whittaker, E., and G. Robison, 1944: The Calculus of modes, J. Atmos. Sci., 57, 1373–1392. Observations, 4th ed., Blackie and Son, Ltd., Chen, T.-C., and J.-M. Chen, 1990: On the maintenance 395 pp. of stationary eddies in terms of the stream function Williams, F. R., and G. H. Jung, 1993: Forecasters budget analysis, J. Atmos. Sci., 47, 2818–2824. Handbook for the Philippine Islands and Surround- Chen, T. C., and M. Murakami, 1988: The 30–50 Day ing Waters, Naval Research Laboratory, 1-1 to variation of convective activity over the Western 5-21 pp (available from http://www.nrlmry.navy Paciﬁc Ocean with emphasis on the northwestern .mil/forecaster_handbooks/Philippines2/Forecasters region, Mon. Wea. Rev., 116, 892–916. Handbook for the Philippine Islands and Surround- Das, P. K., 1986: Monsoons, Fifth WMO Lecture, WMO ing Waters.htm).