1. Introduction
The summertime intraseasonal oscillation (ISO) is a phenomenon that influences weather in East Asia (Wang and Rui 1990; Hsu 2005; Hsu and Weng 2001; Ko and Hsu 2009; Ko and Liu 2016a,b). Of the 122 tropical intraseasonal convective events investigated by Wang and Rui (1990) for the 1975–85 period, 27 moved northward and 18 moved westward. The northward-propagating events prevailed from May to October, and the Indian Ocean and western Pacific were favorable locations for northward-propagating events to occur. Hsu (2005) and Hsu and Weng (2001) have documented the evolution of summertime 30–60-day disturbances and found that these disturbances propagated northwestward from the equatorial western North Pacific (WNP) toward Taiwan and southern China.
Another phenomenon affecting the weather in East Asia is the 7–30-day wave pattern (Ko and Hsu 2006, 2009; Ko and Liu 2016a,b). Ko and Hsu (2006) identified a wave pattern propagating from the tropical WNP to the East China Sea between July and August. The wave pattern reappeared every 7–30 days and comprised a major cyclonic anomaly followed by an anticyclonic anomaly. Tropical cyclones (TCs) generally accompanied the cyclonic anomaly of the wave pattern. TC tracks associated with the 7–30-day wave patterns tended to follow a recurving path and turned to midlatitude areas. In another composite study, Ko and Liu (2016a) applied a circulation index in East Asia and linked the 7–30-day wave pattern to a 3–8-day (synoptic) wave discovered by Lau and Lau (1990). The synoptic wave tended to move northwestward from the equatorial WNP to the northern portion of the South China Sea. Chang et al. (1996) found that TCs tended to move along with cyclonic anomalies of the northwestward-propagating synoptic wave. Ko and Liu (2016a) also examined Taiwan rainfall patterns associated with these two types of waves and found that a heavy rainfall area typically appeared in northern Taiwan when the 7–30-day wave patterns passed by, whereas the synoptic wave usually caused rainfall in eastern Taiwan. The 7–30-day wave patterns and synoptic waves associated with TCs together became part of the East Asian summer monsoon system.
Multiscale interactions of the East Asian summer monsoon system have been a crucial topic since the 2000s. The intraseasonal modulation effect on the higher-frequency wave patterns has been among the most important topics in research on multiscale interactions (Maloney and Dickinson 2003; Ko and Hsu 2009; Ko et al. 2012; Hsu et al. 2011; Ko and Chiu 2014; Ko and Liu 2016b). Maloney and Dickinson (2003) separated the top 20 synoptic wave cases into westerly and easterly events on the basis of the equatorial 850-hPa zonal wind representing the Madden–Julian oscillation (MJO). They found the kinetic energy to be stronger for tropical disturbances in the westerly phase than in the easterly phase. Ko and Hsu (2009) linked the 7–30-day wave pattern and northwestward-propagating 30–80-day ISO. Composite results showed that the ISO westerly phase circulation provided conditions that are more conducive for the wave pattern than did the easterly phase. Ko and Chiu (2014) applied a similar ISO modulation technique on southern Taiwan’s summer monsoon rainfall cases. The resultant ISO westerly monsoon cases exhibited stronger circulations and convection, but there was less rainfall in southern Taiwan than in the easterly phase. Subsequent analyses indicated the northern limit of the northward-propagating ISO to be near Taiwan, and this could be the cause of the reversed rainfall amounts. Following a previous study, Ko and Liu (2016b) applied a 5–16-day (i.e., submonthly) filtering technique and divided submonthly cases into ISO westerly and easterly events. The results indicated that the ISO westerly flow could move the northward-propagating cyclonic anomaly farther west, causing an increase in rainfall in Taiwan. Because the background flow for these two ISO phases yielded separated tracks for the cyclonic anomalies, which could impact the weather in nearby areas, analyzing these ISO-modulated wave patterns could improve medium-range forecasting.
Another issue in multiscale interactions is the interannual variability of ISOs. Hendon et al. (2007) examined the relationship between MJO and El Niño–Southern Oscillation (ENSO) by season. They discovered that MJO activity in late boreal spring led to an ENSO event in the subsequent fall–winter for the 1979–2005 period. Moreover, the enhanced MJO would trigger low-frequency westerly anomalies that could favor a developing ENSO event in spring. In a model experiment study, Tam and Lau (2005) examined the effects of ENSO on MJO circulation and convection. They found that during warm events, the MJO-related convection penetrated farther east over the central Pacific. They further conducted a moisture budget study and found that the eastward shift of the MJO convection could be attributed to increased low-level humidity caused by warming over the central Pacific, and they suggested that the increased humidity would lead to enhanced moisture accumulation.
Most previous multiscale interaction studies have focused on two phenomena of different time scales. Less attention has been given to interactions that link phenomena of more than two time scales, namely, interannual variability, ISO, and associated smaller-scale phenomena such as wave patterns and TCs. Additionally, interactions among these phenomena would influence the local weather in East Asian countries such as Taiwan. Therefore, the present study investigated the interactions among these phenomena and how they influence Taiwan’s weather.
The remainder of this paper is organized as follows. Section 2 describes the data and analysis procedures, and section 3 discusses the mean-state and composite results. In section 4, TCs associated with the submonthly wave patterns are examined. Tracks of the submonthly cyclonic anomaly centers associated with large-scale circulation changes are discussed in section 5 along with their impact on rainfall patterns in Taiwan. Finally, a summary and discussion of the study findings are offered in section 6.
2. Data and analysis procedures
The description of datasets and filtering techniques in the current study parallel that of Ko and Liu (2016b) as follows in the next paragraph and the following text is derived from there with minor modifications. Circulation data were extracted from the National Centers for Environmental Prediction (NCEP) Reanalysis 1 dataset (Kalnay et al. 1996). The dataset contains 6-hourly (0000, 0600, 1200, and 1800 UTC) temperature, humidity, horizontal wind, vertical velocity, and geopotential height data on a 2.5° × 2.5° latitude–longitude grid. Another dataset used in this study included interpolated outgoing longwave radiation (OLR) data from the National Oceanic and Atmospheric Administration (NOAA)/Climate Diagnostics Center. OLR data were daily observations with the same grid spacing as the NCEP dataset. Monthly sea surface temperature (SST) data from NOAA with 2.5° × 2° latitude–longitude grid spacing were also used to compare with the interannual variability of the circulation. The best track TC data compiled by the Joint Typhoon Warning Center (JTWC) at Pearl Harbor, Hawaii, were also included to examine links between large-scale circulation features and TCs. Taiwan precipitation data were obtained from the Taiwan Climate Change Projection and Information Platform (TCCIP). They are daily databased on a 1 × 1 km2 grid, with the spatial domain being 21.67°–25.4°N, 120°–122.14°E, which covers all of Taiwan. This dataset contains precipitation data from all observation stations in Taiwan; the data were fitted to grid points over land areas following the latent Gaussian variable method described by Glasbey and Nevison (1997). The temporal range of the data used in the current study spans a 35-yr period, from July to October (JASO) for the 1979–2013 period. The Butterworth bandpass filter (Kaylor 1977) employed by Hsu and Weng (2001) was used to isolate the periodic signals.
Before case selection procedures were performed, Fig. 1 was plotted to determine the 35-yr JASO mean state OLR, streamfunction, and winds over the East Asia and Pacific basin. A considerably strong subtropical high was centered near 32°N, 155°W, and a ridge extended westward from the center of the subtropical high toward central China. The tropical central Pacific was occupied by easterly winds connected by strong southeasterlies to the east of Taiwan. Most of the OLR minimal area was located in the western Pacific and Southeast Asia, where strong southwesterlies and the southern flank of a summer monsoon trough were collocated. Additionally, most of the East Asian monsoon region was occupied by the monsoon trough separating the southeasterlies and southwesterlies and extending from central China southeastward to the eastern portion of the OLR minimal area.
Mean state streamfunction (contours) and winds at 850 hPa during July–October for a 35-yr period (1979–2013). The contour interval for the streamfunction is 20 × 105 m2 s−1. Also shown is the OLR (shaded). Wind vectors less than 1.5 m s−1 are omitted.
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
Selected cases were organized into three tiers: submonthly, ISO, and interannual cases. The first two tiers were identical to those employed by Ko and Liu (2016b) and the following text is derived from there with minor modifications. A Japan–South China Sea (JSCS) index derived from a circulation pattern (JSCS pattern; Ko and Liu 2016a, b) similar to the Pacific–Japan pattern (Nitta 1987) was defined as the 850-hPa geopotential height difference between two grid points (i.e., 27.5°N, 130°E and 20°N, 115°E). A 5–16-day filter was adopted to focus on the banded signals, because the JSCS index time series exhibited spectral maxima over an approximately 5–16-day period (Ko and Liu 2016a). The submonthly cases were selected on the basis of these 5–16-day filtered JSCS index time series. When a filtered JSCS index maximum exceeded the average of the 35-yr filtered JSCS indices +1.0 standard deviation (SD), the time for the filtered JSCS index maximum was denoted as day 0 for a JSCS high case. Similarly, day 0 of a JSCS low case was selected when a filtered JSCS index minimum was less than the average of the 35-yr filtered JSCS indices −1.0 SD. These day-0 times were subsequently used as the central times of selected cases for the composite technique described by Ko and Hsu (2006, 2009). A spatial bias of 14.5 m, defined as the 35-yr JASO mean unfiltered JSCS index, was applied to select the JSCS high cases as a threshold value. The resultant number of submonthly cases was 241 for the JSCS high and 225 for the JSCS low cases (Table 1).
Case numbers for the JSCS high and low index cases, ISO westerly and easterly events, MaxV and MinV years, and ISO events divided by MaxV and MinV years.
The second tier of cases comprised selected ISO events. A base area (5°–15°N, 105°–125°E) at the location of the maximal 30–80-day 850-hPa zonal wind variance in JASO for the 1979–2013 period was chosen for selecting the ISO events, which were selected on the basis of the zonal wind over the base area. Maxima and minima of the zonal wind averaged over the base area were considered to be ISO westerly and easterly events if the 30–80-day filtered areal-averaged zonal wind deviated from the mean of the 35-yr filtered 30–80-day 850-hPa zonal wind by ±0.5 SD. In total, 85 ISO westerly events and 99 easterly events were selected. After selection of the second-tier cases, the submonthly JSCS high and low cases were divided into two opposite ISO phases (i.e., westerly and easterly). A JSCS high or low case was categorized as the westerly phase if the 30–80-day zonal wind averaged over the ISO base region at the central time (day 0) of the submonthly case was greater than 0.5 SD of the 30–80-day zonal wind time series. Conversely, an easterly JSCS high or low case was selected when the 30–80-day filtered zonal wind averaged over the ISO base region was less than −0.5 SD of the 30–80-day filtered zonal wind time series at day 0 of a submonthly case.
Before the cases in the final tier were selected, the 35-yr time series of the JASO variance for the 30–80-day filtered areal-averaged zonal winds at 850 hPa over the ISO base area (5°–15°N, 105°–125°E) was plotted, as shown in Fig. 2. The purpose of the current study was to investigate the interannual variability of the 850-hPa ISO to determine whether the lower-tropospheric submonthly anomalies and associated TCs are modulated under the ISO interannual variability. Figure 2 was plotted to facilitate comparing the interannual variability and selecting the final tier. In total, 8 of the 35 observation years were selected as the maximal variance (MaxV) years, because their 4-month variances were greater than the 35-yr mean variance +0.5 SD, and 15 observation years were defined as minimal variance (MinV) years, as their 4-month variances were lower than the 35-yr mean variance −0.5 SD. Table 1 lists the MaxV and MinV years. Although no MaxV year was selected for the 1984–95 period, the possible interdecadal change in the ISO variability is beyond the scope of the current study.
Variance of area-averaged 850-hPa zonal wind over 5°–15°N, 105°–125°E during JASO for a 35-yr period (1979–2013).
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
Figure 3 illustrates the OLR and 850-hPa circulation anomalies from the 35-yr JASO mean OLR and circulation in the lower troposphere. Only the negative OLR anomalies are shown because they usually represent convection anomalies (Ko and Hsu 2006). A lower-tropospheric anticyclonic anomaly emerged in the WNP for the MaxV years (Fig. 3a), driving significantly stronger anomalous southwesterlies on its western flank from northern Luzon toward the oceanic area south of Japan. Another secondary anticyclonic anomaly at 850 hPa was located near the center (35°N, 155°W) of the subtropical anticyclone (Fig. 1), indicating that the lower-tropospheric subtropical anticyclone was stronger and extended farther southwestward toward the Philippines in the MaxV years. Therefore, the 850-hPa easterlies were enhanced near the equatorial central Pacific. The OLR negative anomalies exhibited an enhanced convective anomaly near the anomalous southwesterlies south of Japan and another convective anomaly near the Maritime Continent. By contrast, in the MinV years, the circulation anomalies shown in Fig. 3b present a nearly mirrored image of those in the MaxV years. The WNP was occupied by a cyclonic anomaly from the center of the subtropical high to the east of the Philippines, where anomalous northeasterlies were located. An area of anomalous westerlies was also observed in the equatorial central Pacific. The OLR negative anomalies shifted eastward to the central and eastern tropical Pacific, forming an ENSO-like pattern. Therefore, the JASO SST anomalies (Fig. 4) were constructed using the NOAA monthly SST data. The SST anomalies implied a La Niña–like pattern in the MaxV years (Fig. 4a) and an ENSO-like pattern in the MinV years (Fig. 4b). In other words, the mean state of the MaxV (MinV) years exhibited a subtropical anticyclonic (cyclonic) anomaly and generated anomalous southwesterlies (northeasterlies) near East Asia.
Anomalies of JASO-averaged OLR (shaded) and 850-hPa streamfunction (contours) and winds for (a) MaxV and (b) MinV years from the 35-yr (1979–2013) climatological JASO mean. The contour interval for the streamfunction anomalies is 3 × 105 m2 s−1.
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
Anomalies of JASO-averaged SST for (a) MaxV and (b) MinV years from the 35-yr (1979–2013) climatological JASO mean. The contour interval is 0.1°C.
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
The ISO events were further divided into four categories on the basis of the interannual variability: ISOW-MaxV, ISOW-MinV, ISOE-MaxV, and ISOE-MinV. Table 1 lists the corresponding number of events.
After the first two tiers of cases were selected, the procedure for selecting the cases in the final tier was to separate the selected cases into the MaxV and MinV years. In other words, the JSCS high and low cases were classified into the aforementioned four ISO categories on the basis of interannual variability. Table 2 lists the resultant number of cases and the corresponding number of TCs for all eight categories. Although there were more ISO easterly events than westerly events, there were more JSCS ISO westerly cases than easterly cases. This agrees with the results reported in Ko and Liu (2016b).
Case numbers for all eight categories of JSCS submonthly cases and the associated TCs.
3. Mean state and composite results
The propagation and development of circulation anomalies accompanied by TCs are usually connected with the background flow (Ko and Liu 2016b). Figure 5 shows the mean-state maps, which were averaged between day 0 and −5 of the submonthly cases in the four ISO phases in the MaxV years. Also shown are the TC genesis positions that emerged during the 5-day period of the submonthly cases. The TC genesis positions are the first-time positions obtained from the JTWC data archives. The circulation patterns exhibited a monsoon trough extending from the northern portion of the South China Sea to the oceanic area southeast of Taiwan in the westerly phases, as shown in Figs. 5a and 5c. In addition, the monsoon trough in the westerly low phase extended farther eastward, generating a stronger southerly flow on the eastern flank of the trough. Most of the TCs formed in the confluence zone between the monsoon trough and subtropical anticyclone where the convergence areas were located. However, the subtropical ridge dominated most of the East Asian monsoon region in the easterly high phase, and the TC genesis positions were scattered around the southwestern quadrant of the subtropical high. The subtropical anticyclone and a weak monsoon trough formed a significantly strong southerly flow area in the easterly low phase, and TCs mostly formed near the confluence zone. The aforementioned results are in agreement with those of Ko and Liu (2016b).
Mean state of the MaxV-yr 850-hPa streamfunction, winds averaged between day −5 and 0 for (a) westerly high index cases, (b) easterly high index cases, (c) westerly low index cases, and (d) easterly low index cases. The contour interval for the streamfunction is 10 × 105 m2 s−1. Only the winds exceeding the 95% confidence levels are shown. Also shown are the TC genesis positions (blue triangles) and convergence areas (orange contours with yellow shaded areas ≤−1 × 10−6 s−1).
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
The mean-state circulation in the MinV years (Fig. 6) revealed different patterns. The monsoon trough in the westerly high phase was located slightly south and extended farther southeastward than its counterpart in the MaxV years. The convergence zone and TC genesis positions also reached an area farther to the east. Similarly, in the westerly low phase, the monsoon trough and concentration of the TC genesis positions also shifted eastward. Compared with their counterparts in the MaxV years, the mean-state monsoon troughs in the westerly phases of the MinV years were considerably weaker, and fewer significant winds were observed. By contrast, the subtropical anticyclones in the easterly phases of the MinV years were much weaker than those in the MaxV years. In the easterly high phase, a weak monsoon trough extended from the South China Sea to the east of the Philippines, and the area southwest of the subtropical anticyclone was occupied by a maximal convergence area and a clustering area of TC geneses. The easterly low phase exhibited a similar pattern except that the monsoon trough protruded farther east and generated significant southerly winds near the confluence zone between the monsoon trough and the subtropical anticyclone. Overall, the monsoon troughs in the easterly phases of the MinV years expanded farther east and the subtropical anticyclones were weaker than their counterparts in the MaxV years. The TC genesis positions also shifted southeastward in the MinV years, in agreement with the results reported by Chia and Ropelewski (2002).
As in Fig. 5, but for the MinV years.
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
One objective of this study was to examine the interannual variability of the ISO. Figures 7 and 8 illustrate the evolution of the ISO westerly and easterly events for the MaxV and MinV years that was derived by compositing the four types of ISO events selected in Table 1. At day −6 in the ISO westerly phase of the MaxV years, a large ISO cyclonic anomaly spanned from central Vietnam eastward through the western portion of the Philippine Sea, extending southeastward to 10°N, 150°E. The negative OLR anomaly area was located over the southern flank of the ISO cyclonic anomaly, and two negative OLR anomaly centers were present: one west and the other east of the Philippines. The ISO cyclonic anomaly at day 0 intensified and moved slightly northward, expanding northeastward toward southern Japan. The associated negative OLR anomaly east of the Philippines grew and moved northward faster than did its counterpart west of the Philippines. The ISO cyclonic anomaly moved farther northward and decayed rapidly at day 6. The orientation of the cyclonic anomaly then followed a northeast–southwest tilt with rapidly decaying negative OLR anomalies. The circulation and OLR anomaly patterns in the easterly phase resembled those of the westerly phase except that the ISO cyclonic anomaly was replaced by an ISO anticyclonic anomaly; barely any negative OLR anomaly was observed.
Composite 30–80-day filtered anomalies (from the 35-yr climatological JASO mean) of 850-hPa streamfunction (contours) and winds (vectors) in the MaxV years for the ISO westerly events at (a) day −6, (c) day 0, and (e) day +6, and for the easterly events at (b) day −6, (d) day 0, and (f) day +6. The contour interval for the streamfunction anomalies is 5 × 105 m2 s−1. Also shown are the 30–80-day filtered OLR (shaded) anomalies from the 35-yr climatological JASO mean.
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
As in Fig. 7, but for the MinV years.
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
The circulation and OLR anomalies in the MinV years revealed weaker patterns, as shown in Fig. 8. A weaker ISO cyclonic anomaly at day −6 in the westerly phase was located in a similar area to that in the MaxV years. The negative OLR anomalies, although weaker, also exhibited two centers: one in the South China Sea and the other near 10°N, 140°E. The ISO cyclonic anomaly center gradually moved westward at day 0, forming a more solid structure. The center of the eastern negative OLR anomaly intensified and moved northwestward to merge with that from the South China Sea. Although the center of the circulation and negative OLR anomaly pattern moved westward, negative OLR anomalies and significant anomalous winds remained in the eastern end of the system. After 6 days, the ISO cyclonic anomaly and negative OLR anomaly decayed and moved slightly northwestward. However, the composite ISO anticyclonic anomaly in the easterly phase of the MinV years exhibited a clear westward propagation from day −6 to day 0. Subsequently, the ISO anticyclonic anomaly decayed and became stationary at day 6. To summarize the aforementioned ISO evolution, the ISO anomalies tended to move westward in the MinV years but northward in the MaxV years.
The composite 5–16-day filtered streamfunction and wind anomalies at day 0 for those eight categories, as listed in Table 2, are shown in Fig. 9 (MaxV) and Fig. 10 (MinV). Also shown are the corresponding ISO streamfunction anomalies that were concurrently observed at day 0 of the 5–16-day submonthly cases. In the westerly high phase of the MaxV years, a cyclonic anomaly was centered in the South China Sea, and an anticyclonic anomaly was located to the south of Japan. Additionally, a secondary cyclonic anomaly east of Luzon was followed by another weaker anticyclonic anomaly centered near 5°N, 150°E. These circulation anomalies formed a wave pattern that was situated in an ISO cyclonic anomaly. According to Ko and Liu (2016a,b), the anticyclonic anomaly near Japan tended to be stationary and prevent the submonthly wave pattern from propagating northward in the JSCS high phases. Consequently, the cyclonic anomalies were forced to move westward. However, a weak and less organized submonthly pattern was observed in the easterly high phase. Although a submonthly anticyclonic anomaly was centered just south of Japan, only a weak submonthly cyclonic anomaly emerged over southern China, and the whole submonthly pattern was located in an intense ISO anticyclonic anomaly. In the westerly low phase, a solid and wavy structure of the 5–16-day streamfunction anomalies was present from northeast of Taiwan to 5°N, 150°E. The submonthly cyclonic anomaly center to the northeast of Taiwan was located in an ISO cyclonic anomaly whose center was slightly shifted eastward compared with that in the westerly high phase. By contrast, in the easterly low phase, a submonthly cyclonic center near southern Japan and a weak anticyclonic anomaly southeast of the Philippines were located in the eastern portion of an ISO anticyclonic anomaly, the eastern portion of which was smaller than that observed during the easterly high phase. The submonthly wave patterns in the JSCS low phases tended to move northward and recurve (Ko and Liu 2016a,b).
Composite 5–16-day filtered anomalies from the 35-yr climatological JASO mean for the 850-hPa streamfunction (thick contours), winds (vectors) at day 0 for (a) westerly high index cases, (b) easterly high index cases, (c) westerly low index cases, and (d) easterly low index cases, in the MaxV years. The contour interval for the streamfunction anomalies is 5 × 105 m2 s−1. Only the anomalous winds exceeding the 95% confidence levels are shown. Also shown are the corresponding 30–80-day filtered streamfunction anomalies (thin contours with shading).
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
As in Fig. 9, but for the MinV years.
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
The 850-hPa submonthly circulation anomalies at day 0 in the westerly high phase of the MinV years exhibited a similar pattern to that in the MaxV years except that an anticyclonic anomaly near the equatorial western Pacific was not present. Additionally, the associated ISO cyclonic anomaly was much weaker and smaller, centered near a similar location as that in the MaxV years. The amplitude of the submonthly cyclonic anomaly in the South China Sea, however, was slightly stronger than that in the MaxV years. In contrast to the pattern of the MaxV years, the submonthly circulation in the easterly high phase of the MinV years featured a more solid and intense structure and was located in an ISO anticyclonic anomaly protruding northeastward from the South China Sea to 35°N, 150°E. A nearly north–south-oriented cyclonic–anticyclonic pattern was observed in the westerly low phase, and the submonthly cyclonic anomaly was located in an ISO cyclonic anomaly that expanded eastward to 20°N, 155°E. The center of the ISO cyclonic anomaly also shifted eastward to the east of Taiwan. Similarly, in the easterly low phase, a larger north–south-oriented pattern was present near the same area as in the westerly low phase, although it was located near the eastern rim of a much weaker and westward-shifted ISO anticyclonic anomaly. Overall, the east–west expansion and shrinking of the ISO anomaly features were more pronounced in the MinV years; additionally, the submonthly features were stronger than those in the MaxV years.
4. Relationship with TCs
The concurrent links between the day-0 submonthly wave patterns at 850 hPa and the TCs are illustrated in Figs. 11 and 12. The TCs were located near the cyclonic anomalies for most of the phases, which accords with the findings of Ko and Liu (2016a,b). In particular, a TC clustering area was located near the western rim of the submonthly anticyclonic anomaly at day 0 in the westerly high phase of the MaxV years. This is also an area where the anomalous southwesterly flow generated by the submonthly cyclonic anomaly centered near southern China usually drives abundant moisture that follows TCs when they reach Taiwan and southeastern China. Few TCs were observed in the easterly high phase because East Asia was dominated by a large subtropical anticyclone, as shown in Fig. 5b. For the westerly phase, TCs mostly clustered over the major submonthly cyclonic anomalies to the south of Japan. A similar pattern was observed in the easterly low phase, except that fewer TCs occurred.
Composite 5–16-day filtered anomalies (from the 35-yr climatological JASO mean) of 850-hPa streamfunction (thick contours) at day 0 for (a) westerly high index cases, (b) easterly high index cases, (c) westerly low index cases, and (d) easterly low index cases, in the MaxV years. The contour interval for the streamfunction anomalies is 5 × 105 m2 s−1. Also shown are streamfunction tendency of the composite anomalies (shaded, m2 s−2) and the corresponding TC positions along with the past 1-day tracks (thin lines).
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
As in Fig. 11, but for the MinV years.
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
In contrast to the clustering of TCs in the MaxV years, TCs in the westerly high phase of the MinV years were more evenly distributed over the submonthly cyclonic anomalies. More TCs were present in the westerly high phase as a result of the cyclonic anomalies of the wave pattern, which were more evident because of the weakening of the subtropical anticyclonic anomaly. Most of the TCs clustered near the major submonthly cyclonic anomalies in the westerly and easterly low phases, and the submonthly cyclonic anomalies of the wave patterns were stronger than their counterparts in the MaxV years.
5. Submonthly cyclonic anomaly center tracks and large-scale circulation changes
Track analysis of the submonthly cyclonic anomaly centers has proved useful in studying the effects of large-scale circulation patterns on smaller-scale phenomena (Ko and Liu 2016b). The submonthly cyclonic anomaly center tracks between day −2 and day +2 and the composite 5–16-day filtered 850-hPa streamfunction at day 0 are illustrated in Fig. 13. The movement of the submonthly cyclonic anomaly center in the westerly high phase was faster in the MaxV years than in the MinV years. Before day 0, the track of the MaxV year cyclonic anomaly center moved from southern Taiwan to Hong Kong, and it was located farther north than in the MinV years. The Max year cyclonic anomaly center then moved southwestward to central Vietnam around day 0 and turned westward afterward. In the MinV years, the submonthly cyclonic anomaly center took a path across the northern portion of the South China Sea. Although the cyclonic anomaly of the submonthly wave pattern was weaker in the easterly high phase of the MaxV years, the cyclonic anomaly center followed a route from northern Luzon to southern China. This track was farther northeast than that in the MinV years. The tracks of the submonthly cyclonic anomaly centers in the westerly low phases, however, are controversial. The submonthly cyclonic anomaly center in the MaxV years exhibited a northwestward propagation route to the oceanic area between Taiwan and Japan before day 0, but this became a westward propagation route after day 0; that is, the MaxV year cyclonic anomaly center in the westerly low phase did not follow a recurving path between day −2 and day +2. However, the submonthly cyclonic anomaly center in the MinV years followed a similar path as in the MaxV years before day 0, but it subsequently recurved and propagated northward. In other words, the cyclonic anomaly of the submonthly wave pattern and the associated TCs would approach eastern China and Taiwan in the MaxV years, but they were more likely to reach Korea and Japan in the MinV years. In the easterly low phase, the submonthly cyclonic anomaly center in the MaxV years followed a northward path through the Japan Sea before turning northeastward to higher latitudes. Furthermore, this track was located to the northwest of that in the MinV years. Therefore, the cyclonic anomaly of the wave pattern and TCs were more likely to hit Korea in the MaxV years, whereas they might influence Japan in the MinV years. The impact areas of TCs could differ between the JSCS low phases of the MaxV and MinV years.
Composite 5–16-day filtered anomalies (from the 35-yr climatological JASO mean) 850-hPa streamfunction (thin contours) at day 0 for (a) westerly high index cases, (b) easterly high index cases, (c) westerly low index cases, and (d) easterly low index cases, in the MaxV and MinV years. The contour interval for the streamfunction is 5 × 105 m2 s−1. The blue lines show the day-0 streamfunction anomalies for the MaxV years and red lines are for the MinV years, respectively. Also shown are the tracks of the cyclonic anomaly centers (thick lines) from day −2 to +2.
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
The aforementioned tracks of the submonthly cyclonic anomaly center raise a question: What mechanism is responsible for the different tracks? The mean flow that was averaged between day −5 and +5 for each category of the submonthly cases might clarify this issue. Ko and Liu (2016b) used mean-state difference maps between the westerly and easterly phases to account for the shifting of the cyclonic anomaly center tracks. In the present study, the mean-state differences between the MaxV and MinV years were examined (MaxV − MinV), and the results are shown in Fig. 14. After the mean-state streamfunction in the westerly high phase of the MinV years was subtracted from that in the MaxV years, a trough was found extending from southern China eastward to 20°N, 140°E. Strong southwesterly difference winds from the eastern portion of the South China Sea to the oceanic area east of Taiwan, and difference southeasterlies from north of the trough to eastern China, Korea, and southern Japan were observed. In other words, compared with the mean flow in the MinV years, that in the aforementioned areas in the westerly high phase of the MaxV years was considerably stronger. However, nothing was found to be responsible for driving the cyclonic anomaly center of the wave pattern farther westward in the northern portion of the South China Sea. Two difference anticyclones were observed in the easterly high phase: one was centered near 20°N, 140°E, and the other smaller one was centered south of Hong Kong. These two difference anticyclones both generated difference southwesterlies on their northern flanks. The difference anticyclone near Hong Kong and the associated difference southwesterlies could be a key factor in driving the submonthly cyclonic anomaly center to drift farther northeastward, as illustrated in Fig. 13b. By contrast, the mean-state difference maps of the westerly low phase reveal a pattern similar to that in the westerly high phase. The difference southeasterly winds to the north of the trough were responsible for the northwestward movement of the cyclonic anomaly center tracks for the MaxV and MinV years before day 0, but those two tracks eventually followed different paths: westward for the MaxV years and northward for the MinV years. The mean-state difference pattern lacked a key feature for it to explain how the cyclonic anomaly center was driven westward in the westerly low phase of the MaxV years. However, in the easterly low phase, a difference anticyclone was centered near 20°N, 140°E. This difference anticyclone generated difference southwesterlies on its northwestern flank, and these difference winds continued downstream, forming difference southeasterlies near southern Japan and Korea. The overall pattern of these difference southerlies was located in an area where those two tracks of the cyclonic anomaly centers were collocated. Therefore, the difference southerlies could be responsible for the northward shift of the cyclonic anomaly center in the MaxV years compared with that in the MinV years. To summarize the aforementioned results, the mean-state difference maps between the MaxV and MinV years could be used to account for the submonthly cyclonic anomaly center track changes for the easterly phases but not for the westerly phases.
Composite mean state differences of the 850-hPa streamfunction, winds averaged for 10-day periods centered at day 0 for (a) MaxV − MinV westerly high index cases, (b) MaxV − MinV easterly high index cases, (c) MaxV − MinV westerly low index cases, and (d) MaxV − MinV easterly low index cases. The contour interval for the difference streamfunction is 5 × 105 m2 s−1. The mean state wind speed differences are shown as orange contours (contour interval 1.5 m s−1) with yellow shading ≥1.5 m s−1.
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
To identify the key features that explain the track changes of the submonthly cyclonic anomaly centers between the MaxV and MinV years for the westerly phases, ISO mean-state difference maps for the streamfunction and winds were plotted in the manner as those shown in Fig. 14, except for the 30–80-day filtered data; the results are shown in Fig. 15. In the westerly high phase, the ISO mean-state difference map exhibited an ISO difference cyclone from northern Vietnam to the oceanic area southeast of Taiwan. This ISO difference cyclone would generate difference northeasterlies near southern China, and these difference winds could contribute to the cyclonic anomaly center of the wave pattern moving faster and following a southwestward route in the MaxV years. Similarly, in the westerly low phase, an ISO difference cyclone occurred in an area similar to that in the westerly high phase, generating strong difference southeasterlies northeast of Taiwan and difference easterlies downstream near the north of Taiwan where the westward-propagating submonthly cyclonic anomaly was collocated. Therefore, the ISO mean-state difference circulation could be responsible for the track changes of the cyclonic anomaly centers between the MaxV and MinV years in the westerly phases.
Composite mean state differences of the 30–80-day filtered anomalies (from the 35-yr climatological JASO mean) for the 850-hPa streamfunction, winds averaged over 10-day periods centered at day 0 for (a) MaxV − MinV westerly high index cases and (b) MaxV − MinV westerly low index cases. The contour interval for the difference streamfunction is 10 × 105 m2 s−1. The mean state difference wind speed is shown as orange contours (contour interval 0.5 m s−1) with yellow shading ≥0.5 m s−1.
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
Because changes in large-scale circulation could affect the propagation routes of the cyclonic anomalies for the submonthly wave patterns, the weather associated with those wave patterns and TCs merit further investigation. Figure 16 shows the TCCIP rainfall averaged between day −2 and day +2 for all phases in the MaxV years. The entirety of Taiwan experienced heavy rainfall in the westerly high phase because the submonthly cyclonic anomaly at day −2 shown in Fig. 13a was near Taiwan, where a TC clustering area was also collocated, as shown in Fig. 11a. In the easterly high phase, as shown in Fig. 16b, maximal rainfall areas were observed over northeastern and eastern Taiwan, with scattered rainfall areas over the central mountain range. According to Ko and Liu (2016a), rainfall areas in eastern Taiwan were detected when the JSCS high wave patterns pattern and TCs moved westward to the northern portion of the South China Sea. Heavy rainfall areas over northern Taiwan and most of central Taiwan were observed in the westerly low phase as the northwestward-propagating wave pattern approached Taiwan. Two maximal rainfall areas were observed in the easterly low phase, as shown in Fig. 16d: one was over northern Taiwan and the other near southwestern Taiwan. The observation of the rainfall area near northern Taiwan agrees with the findings reported by Ko and Liu (2016a), indicating that the rainfall could be caused by the northerly winds created by the northward passage of the submonthly cyclonic anomaly to the east of Taiwan. However, after the passage of the submonthly cyclonic anomaly and TCs, anomalous southwesterlies could follow the tail of the TCs and bring rainfall to southwestern Taiwan. By contrast, the rainfall in the MinV years, as presented in Fig. 17, exhibited similar but weaker patterns because the propagation routes of the submonthly cyclonic anomalies were farther from Taiwan, as shown in Fig. 13. Therefore, changes in large-scale circulation could influence the submonthly wave patterns and affect local weather in regions such as Taiwan.
Composite TCCIP rainfall (shaded) averaged between day −2 and day +2 for (a) westerly high index cases, (b) easterly high index cases, (c) westerly low index cases, and (d) easterly low index cases, in the MaxV years. Also shown is the corresponding composite 5–16-day filtered streamfunction at 850 hPa. The contour interval is 3 × 105 m2 s−1.
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
As in Fig. 16, but for the MinV years.
Citation: Monthly Weather Review 145, 9; 10.1175/MWR-D-16-0482.1
6. Summary and discussion
The interannual ISO variability and associated submonthly wave patterns at 850 hPa for JASO during the 1979–2013 period were investigated by classifying the years into MaxV years and MinV years. The selected ISO events and submonthly wave pattern cases were then further divided into categories on the basis of the MaxV and MinV years. The 850-hPa streamfunction for the mean-state patterns of those eight categories suggested that the monsoon troughs and TC genesis positions extended farther southeastward in the MinV years. Moreover, the SST anomalies showed an ENSO pattern in the MinV years, and the southeastward displacement of TC genesis positions agrees with the results of Chia and Ropelewski (2002). However, their study determined that ENSO signals were not evident in every year (e.g., 1988).
The ISOs in the MaxV and MinV years exhibited distinct propagating tendencies. Stronger ISOs were observed in the MaxV years, moving slowly northward toward southern China. By contrast, in the MinV years, the ISO center propagated westward from the northeast of Luzon to the South China Sea. By overlaying the composite 5–16-day filtered streamfunction and the corresponding 30–80-day filtered streamfunction, we connected these two anomalies. Apparently, the ISO cyclonic anomalies in the westerly phase provided favorable conditions for the submonthly wave patterns and TCs to grow. However, the wave patterns were less intense in the MaxV years, possibly because of the stronger subtropical anticyclonic anomaly of the climatological mean JASO pattern in the MaxV years. This subtropical anticyclonic anomaly would prevent the submonthly wave patterns and TCs from growing. Additionally, the east–west expansion and contraction of the ISO anomalies were more evident in the MinV years, resulting in more activity to the east of the monsoon trough. Chia and Ropelewski (2002) found that TCs in the ENSO years tended to be stronger because they formed in areas farther southeast than the climatological mean-genesis region and thus had a longer lifetime.
Track analyses of the submonthly cyclonic anomaly centers suggested by Ko and Liu (2016b) provided a means for examining how the large-scale circulation affected the propagation routes of smaller-scale phenomena. In that study, the propagation routes of the cyclonic anomaly centers for the submonthly wave patterns in the westerly phases were forced to move farther west relative to those in the easterly phases. Hence, the wave patterns in the easterly phases were located farther east. The mean-state difference obtained by subtracting the streamfunction in the easterly phase of the MinV years from that of the MaxV years revealed a difference anticyclone that generated difference southwesterly or southerly winds to push the submonthly cyclonic anomaly centers farther northeast or north. In addition, the propagation routes of the easterly cyclonic anomaly centers in the MaxV years were located near the western rim of the difference climatological subtropical anticyclone that generated the difference southerly flow forcing the wave pattern to move farther north. However, the propagation routes of the westerly submonthly cyclonic anomaly centers were located farther west of the difference subtropical anticyclones, and thus the mean-state differences of the winds could barely account for their movement. The difference easterly flow generated by the difference ISO cyclone in the MaxV years was located in the areas where the westerly submonthly cyclonic anomaly centers followed. Thus, the difference ISO easterly flow would push the submonthly cyclonic anomaly center in the westerly high phase to move quickly westward and shift the submonthly cyclonic anomaly center so that it turned westward in the westerly low phase. In other words, the propagation of the cyclonic anomaly centers of the submonthly wave patterns in the easterly phases could be dominated by the mean-state subtropical anticyclones, whereas the westerly wave patterns would follow tracks controlled by the ISO cyclonic anomalies. Ko and Hsu (2009) also found that TC geneses were related to the moisture convergence by the ISO westerly flow, whereas in the easterly phase, they were related to moisture convergence caused by the climatological mean flow.
One objective of this study was to explore the impact of large-scale circulation changes on Taiwan’s weather. Because the submonthly cyclonic anomalies and the associated TCs in the MaxV years were closer to Taiwan, heavier rainfall was observed in those years. Therefore, forecasters in MaxV years should be more careful in assessing the rainfall amount. The results in the current study may be useful in improving forecasting skills. Further research on the interdecadal variability is under way to investigate this complex multiscale interaction.
Acknowledgments
The authors thank the TCCIP Project Office (NSC 100-2621-M-492-001) for providing the precipitation data used in this study. The authors also thank NCEP for providing the global analyses and NOAA for the OLR data. Comments from two anonymous reviewers and Editor Dr. Paul Roundy considerably improved the communication of the results, and the authors are grateful to them for their suggestions. This manuscript was edited by Wallace Academic Editing. The support by the Ministry of Science and Technology, Taiwan (Grant MOST 106-2111-M-017-001 issued to Dr. Ken-Chung Ko) is also acknowledged.
REFERENCES
Chang, C.-P., J. M. Chen, P. A. Harr, and L. E. Carr, 1996: Northwestward-propagating wave patterns over the tropical western North Pacific during summer. Mon. Wea. Rev., 124, 2245–2266, doi:10.1175/1520-0493(1996)124<2245:NPWPOT>2.0.CO;2.
Chia, H.-H., and C. F. Ropelewski, 2002: The interannual variability in the genesis location of tropical cyclones in the Northwest Pacific. J. Climate, 15, 2934–2944, doi:10.1175/1520-0442(2002)015<2934:TIVITG>2.0.CO;2.
Glasbey, C. A., and I. M. Nevison, 1997: Rainfall modeling using a latent Gaussian variable. Modelling Longitudinal and Spatially Correlated Data: Methods, Applications, and Future Directions, T. G. Gregoire et al., Eds., Lecture Notes in Statistics, Vol. 122, Springer, 233–242.
Hendon, H. H., M. C. Wheeler, and C. Zhang, 2007: Seasonal dependence of the MJO–ENSO relationship. J. Climate, 20, 531–543, doi:10.1175/JCLI4003.1.
Hsu, H.-H., 2005: Intraseasonal variability of the atmosphere–ocean–climate system: East Asian monsoon. Intraseasonal Variability in the Atmosphere–Ocean Climate System, K.-M. Lau and D. Waliser, Eds., Springer-Praxis, 65–98.
Hsu, H.-H., and C.-H. Weng, 2001: Northwestward propagation of the intraseasonal oscillation in the western North Pacific during the boreal summer: Structure and mechanism. J. Climate, 14, 3834–3850, doi:10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2.
Hsu, P.-C., T. Li, and C.-H. Tsou, 2011: Interactions between boreal summer intraseasonal oscillations and synoptic-scale disturbances over the western North Pacific. Part I: Energetics diagnosis. J. Climate, 24, 927–941, doi:10.1175/2010JCLI3833.1.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471, doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.
Kaylor, R. E., 1977: Filtering and decimation of digital time series. Tech. Note BN 850, Institute of Physical Science Technology, University of Maryland, College Park, 42 pp.
Ko, K.-C., and H.-H. Hsu, 2006: Sub-monthly circulation features associated with tropical cyclone tracks over the East Asian monsoon area during July-August season. J. Meteor. Soc. Japan, 84, 871–889, doi:10.2151/jmsj.84.871.
Ko, K.-C., and H.-H. Hsu, 2009: ISO modulation on the submonthly wave pattern and the recurving tropical cyclones in the tropical western North Pacific. J. Climate, 22, 582–599, doi:10.1175/2008JCLI2282.1.
Ko, K.-C., and P.-S. Chiu, 2014: ISO-modulating effects on the East Asian summer monsoon circulation patterns associated with southern Taiwan’s monsoon rainfall. Mon. Wea. Rev., 142, 3163–3177, doi:10.1175/MWR-D-13-00327.1.
Ko, K.-C., and J.-H. Liu, 2016a: Quasi-periodic behavior of the Pacific–Japan pattern affecting propagation routes of summertime wave patterns and the associated tropical cyclone tracks over the western North Pacific. Mon. Wea. Rev., 144, 393–408, doi:10.1175/MWR-D-15-0080.1.
Ko, K.-C., and J.-H. Liu, 2016b: Relationship between summertime intraseasonal oscillations and submonthly wave patterns under meridional periodic fluctuations. J. Climate, 29, 6151–6166, doi:10.1175/JCLI-D-16-0217.1.
Ko, K.-C., H.-H. Hsu, and C. Chou, 2012: Propagation and maintenance mechanism of the TC/submonthly wave pattern and TC feedback in the western North Pacific. J. Climate, 25, 8591–8610, doi:10.1175/JCLI-D-11-00643.1.
Lau, K.-H., and N.-C. Lau, 1990: Observed structure and propagation characteristics of tropical summertime synoptic scale disturbances. Mon. Wea. Rev., 118, 1888–1913, doi:10.1175/1520-0493(1990)118<1888:OSAPCO>2.0.CO;2.
Maloney, E. D., and M. J. Dickinson, 2003: The intraseasonal oscillation and the energetics of summertime tropical western North Pacific synoptic-scale disturbances. J. Atmos. Sci., 60, 2153–2168, doi:10.1175/1520-0469(2003)060<2153:TIOATE>2.0.CO;2.
Nitta, T., 1987: Convective activities in the tropical western Pacific and their impact on the northern hemisphere summer circulation. J. Meteor. Soc. Japan, 65, 373–390, doi:10.2151/jmsj1965.65.3_373.
Tam, C.-Y., and N.-C. Lau, 2005: Modulation of the Madden–Julian oscillation by ENSO: Inferences from observations and GCM simulations. J. Meteor. Soc. Japan, 83, 727–743, doi:10.2151/jmsj.83.727.
Wang, B., and H. Rui, 1990: Synoptic climatology of transient tropical intraseasonal convection anomalies: 1975–1985. Meteor. Atmos. Phys., 44, 43–61, doi:10.1007/BF01026810.
