1. Introduction
Eastward propagation of the intraseasonal oscillation along the equator, especially during the boreal winter, has been widely explored. Many mechanisms, mostly based on equatorial wave dynamics, have been proposed. It is also known that the intraseasonal oscillation propagates not only eastward but also poleward (Lau and Chan 1985, 1986; Wang and Rui 1990; Hsu 1996). The poleward propagation is particularly evident in south and Southeast Asia during the boreal summer (e.g., Yasunari 1979, 1981; Krishnamurti and Sabrahmanyan 1982; Murakami and Nakazawa 1985). In addition, northward and/or northwestward propagation of a large-scale convection system is also observed in the western North Pacific (e.g., Lau and Chan 1986; Knutson and Weickmann 1987; Nitta 1987; Chen and Murakami 1988; Kawamura et al. 1996).
While many mechanisms have been proposed to explain the eastward propagation along the equator, northward propagation in south Asia and northwestward propagation in the western North Pacific has received less attention. Nitta (1987) documented the northwestward propagation of intraseasonal cloud variation, which propagates at a speed corresponding to the mean wind speed in the lower troposphere. This study seemed to suggest that the northwestward propagation is a result of the mean flow advection. Kawamura et al. (1996) found that the dominant empirical orthogonal function of the infrared equivalent blackbody temperature over the western Pacific propagates northward with time. Their calculations revealed that the moisture convergence, which is mainly contributed by the low-level anomalous southeasterly, at the northern part of the cyclonic circulation is induced by convection. It was suggested that the supply of moisture favored the northward propagation of the tropical intraseasonal oscillation in the western North Pacific.
Wang and Xie (1997) recently proposed a model for the boreal summer intraseasonal oscillation based on the results of a numerical study. On an equatorial beta plane, a coupled Kelvin–Rossby wave packet, which is initiated in the western Indian Ocean, propagates eastward into the western Pacific and disintegrates when it approaches the date line where the moisture supply is significantly reduced. A moist Rossby wave, which is sustained by the instability due to the easterly vertical shear, then emanates from the western Pacific and propagates westward. The westward-propagating Rossby wave drifts northward into the western North Pacific to the latitude near 20°N. It then continues to move westward to south Asia along 20°N. The model results indicate that the emanation of Rossby waves is the mechanism responsible for the northwestward propagation in the western North Pacific. While the proposed mechanisms seem to work quite well in the model, more evidence seems needed to confirm some simulated features such as the disintegration of the Kelvin–Rossby wave packet.
Although several studies have investigated the northwestward propagation of the intraseasonal oscillation in the western North Pacific, the nature of the phenomenon is not well documented and understood. This diagnostic study attempts to unravel the mystery of this phenomenon and propose a mechanism responsible for the northwestward propagation. The data and analysis procedures are described in section 2. In sections 3 and 4 the structure and temporal evolution of the northwestward-propagating intraseasonal oscillation in the western North Pacific are presented. The mechanisms of the observed features are explained in the discussions and conclusions presented in section 5.
2. Data and analysis procedure
The data used in this study include the following: 1) the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis (ERA; Gibson et al. 1997), 2) the optimally interpolated sea surface temperature (SST) produced by the National Centers for Environmental Prediction (Reynolds and Smith 1994), and 3) the interpolated outgoing longwave radiation (OLR; Liebmann and Smith 1996). The ERA and OLR data covers only 15 summers, defined as June–August, from 1979 to 1993, while the SST data covers only 12 summers from 1982 to 1993. The grid spacing of the ERA and OLR is 2.5° × 2.5°, while the SST is on a 1° × 1° grid. The SST data, originally available weekly, were linearly interpolated into daily data. This procedure did not produce higher-frequency signals. It however created a dataset that is convenient for the bandpass-filtering procedure described below.
In order to isolate the intraseasonal signals, all data were filtered through a 30–60-day bandpass Butterworth filter (Hamming 1989). Zonal means were then removed to emphasize the eddy components of the circulation. To show the large-scale structure clearly, all data except the SST were spatially smoothed using a T24 spectral smoother as proposed by Sardehsmukh and Hoskins (1984).
3. Characteristics of convection and circulation
The western North Pacific, that is, from the South China Sea to the date line and from the equator to 30°N, is one of the major regions of strong intraseasonal variability. Figures 1a and 1b present the ratio of the 30–60-day variance to the total variance for OLR and SST, respectively. The total variance is computed from the unfiltered daily data and therefore includes the variability from the subsynoptic (except the SST) to the interannual timescales. The largest variance ratio is observed in a large area covering the South China Sea and the Philippine Sea for both the OLR and SST. Our calculation (not shown) indicates that other variables such as surface heat fluxes and vorticity also exhibit high percentages in the same region. The results suggest that the western North Pacific is a region of vigorous intraseasonal variability in both the atmosphere and ocean. It follows that the ocean–atmospheric interaction occurring in the western North Pacific may affect the intraseasonal variability in the region.
The vectors shown in Figs. 1c and 1d indicate the propagation of the intraseasonal OLR and 850-hPa vorticity anomalies, respectively. Five-day-lagged correlations between the variable at each grid point (i.e., base point) and the variable at all points five days later were calculated. A vector was drawn from the base point to the point where the lagged correlation was at maximum. This vector represents the preferred propagation direction of the intraseasonal anomaly at each point. Results shown in Fig. 1c resemble those presented in Nitta (1987) and Lau and Chan (1986), although the length and the data period used was different. For OLR, major regions where the lagged correlation was greater than 0.75 appear in the Philippine Sea, the southern part of the South China Sea, and the Bay of Bengal. As revealed in Fig. 1c, the intraseasonal convective perturbations in the Philippine Sea tend to propagate northwestward from the Tropics to 20°N, while their counterparts in the southern South China Sea tend to propagate northward. To the east of the northwestward-propagating region where the lagged correlation is slightly weaker, westward and southwestward propagation is also evident. Near the equator in the western Pacific, the favored propagation route of the intraseasonal convection activity is, however, eastward or northeastward. The complicated distribution of the propagation routes reveals the complex nature of the phenomenon examined in this study. It will be shown later that the eastward propagation near the equator and the westward propagation in the subtropics are all parts of the evolution of the convection and circulation that propagate northwestward in the later stage.
The propagation of the intraseasonal vorticity anomaly at 850 hPa is most evident in the western North Pacific (Fig. 1d) and exhibits less geographical dependence than its OLR counterpart. While the westward propagation is the dominant feature in the region east of 135°E, the anomalies in the region west of 135°E propagate northwestward. As a whole, the major region of propagation for the OLR is located to the southeast of the region for the 850-hPa vorticity. This spatial relationship reflects the coupling between the tropical convection and circulation that often occurs in the low latitudes.
In order to isolate the dominant intraseasonal signals in the western North Pacific, an index defined as the area mean of the filtered OLR in the domain (0°–20°N, 120°–160°E) is calculated to represent the convective activity in the region. One may wonder whether a mean value in such a large domain can represent the circulation and convection activity in the western North Pacific. To ensure the adequacy of this index, an empirical orthogonal function (EOF) analysis was applied to the filtered OLR field in various domains covering the above area. The results indicate that the regressed evolution of the OLR based on this index (shown in Fig. 2) can be represented by the first two leading EOF that together form a propagating pattern. The comparison suggests that this index is a good representative of the major intraseasonal OLR variability in the region. Our calculation also indicates that this index reveals more structure than using the EOF results. In the following sections, the lagged regression coefficients between the index and various variables are presented to reveal the structure and temporal evolution of the northwestward-propagating intraseasonal perturbation. In order to focus on the statistically meaningful features, only those regression coefficients significant at the 0.05 level are plotted in the following figures. The regression coefficients were multiplied by one standard deviation of the index to reconstruct the corresponding variables such as OLR and vorticity. The value shown in each figure can be interpreted as the fluctuation of the plotted variables corresponding to one standard deviation of the index.
4. Spatial structure and temporal evolution
a. OLR and diabatic heating
Lagged regression maps representing the temporal evolution of the OLR and diabatic heating are shown in Fig. 2. The diabatic heating is calculated as the residual of the thermodynamic equation and is vertically integrated from the surface to 10 hPa. The evolution shown in Fig. 2 covers the period from day −15 to day 10, which is about half a cycle. As a whole, the diabatic-heating pattern corresponds well to the OLR pattern and is nearly 180° out of phase with the OLR. Since the diabatic heating in the low latitudes is dominated by the convective heating, it is hence reflected in the OLR.
At day −15, a region of positive OLR (negative heating) anomaly is found located over the Philippine Sea, while a negative OLR (positive heating) anomaly is located to the south near the equator. In the next five days, the positive OLR anomaly over the Philippine Sea moves slightly northwestward toward Taiwan and begins weakening, while the negative OLR anomaly in the south strengthens and expands both northwestward and southeastward. It is also worthy to note that at day −10 (Fig. 2b) a wavelike OLR and diabatic-heating structure is evident to the east of the positive anomaly between 15° and 30°N. At day −5, the positive OLR anomaly north of the Philippines disappears, while the negative OLR anomalies in the south and in the east merge and form a large negative anomaly in the western North Pacific. During this merging period, four maximum diabatic-heating centers are evident near (17.5°N, 145°E), (5°S, 147.5°E), (5°N, 130°E), and (10°N, 115°E). The newly formed negative anomaly becomes more compact in the next five days and moves northwestward and reaches Taiwan at day 10 when it begins weakening. At this time, the positive OLR anomalies near Indonesia and (20°N, 155°E) are well developed and are ready to repeat the evolution of its predecessor. The temporal evolution shown above evidently reflects the complex distribution of the preferred propagation routes shown in Fig. 1c, namely, the northeastward movement near the equator, the westward propagation near 20°N, and the following northwestward propagation in the western North Pacific.
The convective activities in the western North Pacific are negatively correlated with those in the Indian Ocean. During the development and movement of the negative OLR anomaly in the Philippine Sea, a positive OLR anomaly appears near the equator in the Indian Ocean and propagates northward almost simultaneously with its counterpart in the western North Pacific. This seesaw relationship has been well documented and discussed in previous studies, for example, Lau and Chan (1986) and Zhu and Wang (1993). It is interesting to note that both systems in the Indian Ocean and the western North Pacific propagate toward the Asian continent, although northward for the former and northwestward for the latter. Northward propagation in the Indian Ocean will be explored in another paper. In this study, we shall focus on the development and movement of the component in the western North Pacific.
b. Circulation
The temporal evolution of the corresponding vorticity pattern at 850 hPa is shown in Fig. 3. At day −15, a dipole of vorticity anomaly crosses the equator in the western Pacific. The negative vorticity anomaly in the Northern Hemisphere is located to the northwest of the positive OLR anomaly in the western North Pacific. This vorticity–OLR relationship is consistent with the relative position of the heating-induced circulation and an off-equator heating as shown in the theoretical study by Gill (1980). This vorticity dipole and the positive OLR anomaly move westward slightly and weaken over the next five days. During this period, a wavelike structure that extends from the northern South China Sea into the extratropical central Pacific moves westward simultaneously. Interestingly, the wavelike vorticity and OLR patterns tend to be in quadrature in a manner that the positive (negative) OLR anomalies are located to the east or southeast of the negative (positive) vorticity anomalies. Such a vorticity–OLR spatial relationship resembles that of a divergent Rossby wave on a sphere. The wavelike pattern also bears a similarity to a Rossby wave train forced by tropical heating (Hoskins and Karoly 1981).
When the negative OLR anomaly in the western North Pacific begins developing and strengthens at day −5, the positive vorticity anomaly that was at 150°E moves westward to 135°E along 20°N and grows in both spatial coverage and strength. Five days later at day 0, the positive vorticity anomaly strengthens further and moves westward to 125°E while the negative OLR anomaly located to its southeast propagates northwestward. The simultaneous strengthening of the circulation and convection suggests the occurrence of a positive feedback through convection–circulation interaction. The vorticity dipole and the negative OLR anomaly continue moving westward over the next 10 days and begin to decay at day 10. A Rossby wave–like structure of opposite signs in both vorticity and OLR is again evident in the subtropical western North Pacific.
Climatologically there is a monsoon trough, characterized by the active convection and the southwesterly penetration from the Indian Ocean, in the western North Pacific during the boreal summer. During the active convection phase, one can infer from Figs. 2c–f and Figs. 3c–f that the monsoon trough deepens and the accompanying southwesterly is strengthened all the way from the Philippine Sea to south Asia. The intraseasonal OLR and 850-hPa variations documented here apparently correspond to the intraseasonal fluctuations of the monsoon trough.
The westward-propagating circulation exhibits characteristics similar to a westward-propagating Rossby wave. However, the near-surface convergence, moisture, and stability in the lower troposphere also contribute to the westward propagation. The 10-m divergence field shown in Fig. 4 clearly exhibits the wavelike structure in the subtropics at day −15 (e.g., Fig. 4a) and the merging of the eastward-moving convergence anomaly in the equator and the westward-moving convergence anomaly in the western North Pacific (e.g., Figs. 4a–c). Both convergence anomalies in the equatorial and subtropical regions tend to lead the corresponding OLR anomaly in its propagation direction, indicating the leading phase of the near-surface convergence relative to the deep convection (inferred from OLR). A similar phase relationship can be seen even more clearly when the negative OLR anomaly in the western North Pacific propagates northwestward. For example, a region of 10-m convergence anomalies is always located to the northwest or north of the negative OLR anomaly in the western North Pacific from day −5 to day 10 (Figs. 4c–f).
Figure 5 presents the moisture transport divergence [i.e., ∇·(VQ), where V is velocity and Q is specific humidity] and the divergent component of moisture transport [i.e., (VQ)χ, where χ represents the divergent component], at 925 hPa from day −5 to day 5. The evolution of ∇·(VQ) bears a similarity to that of the 10-m divergence. The center of maximum moisture convergence is however located slightly to the southwest of the maximum 10-m convergence in the Philippine Sea (Figs. 4c and 5a). Moisture is transported from the eastern equatorial Indian Ocean into the Philippine Sea at day −5. Five days later, while the moisture convergence region (i.e., negative anomaly) moves to Taiwan and the Philippines, a moisture divergence region (i.e., positive anomaly) develops near the Maritime Continent. The vectors shown in Fig. 5b indicate that the moisture is transported into the region mainly from the Indian Ocean and the Maritime Continent. While the moisture convergence region continues moving westward, the divergence region in the Maritime Continent strengthens and the one in the Indian Ocean weakens. Although there are indications of moisture transport from other directions, the transport from the southwest and south are clearly the dominant features. Climatologically, the prevailing southwesterly in the boreal summer transports the moisture into the South China Sea and the Philippine Sea. The southwesterly anomaly shown in Fig. 5 (and also in Fig. 7) indicates an enhanced moisture transport by the strengthened southwesterly when a negative OLR anomaly appears in the western North Pacific.
While the near-surface convergence and moisture transport favors the northwestward/westward propagation of the circulation–convection system, one can also find the signals in the thermodynamic structure. Shown in Fig. 6 are the cross sections of the potential instability, vertical circulation, and diabatic heating along 17.5°N, where the convection and near-surface convergence are most significant, from day −10 to day 10. The potential instability is defined as −∂θe/∂p, where θe is the equivalent potential temperature calculated as suggested by Bolton (1980). The computation of −∂θe/∂p follows the finite difference scheme used in Brunet et al. (1995). At day −10, a region of positive heating anomaly with a maximum at 400 hPa is located between 150° and 160°E. This positive anomaly of about 20° longitude in width can be traced back near the date line, although very weak, at day −20 (not shown). It propagates westward to 145°E at day −5 and then suddenly flares in a region covering from 155° to 110°E at day 0. This sudden development of the anomalous heating coincides with the merging of the two negative OLR anomalies in the western North Pacific, as shown in Fig. 2. The center of positive heating continues moving westward in the next 10 days to 125°E.
During the course of westward propagation, an anomalous region of negative potential instability in the lower troposphere is always located to the west of the positive heating anomaly. In addition, an anomalous upward motion is also evidently stronger to the west of the positive heating anomaly. One can even see a strong upward motion between 600 and 900 hPa, where the low-level negative potential instability is most significant. The region west of the enhanced convection in the western North Pacific is potentially less stable and corresponds to the near-surface convergence anomaly, which provides a lifting effect. The collocation of this convergence, anomalous upward motion, and positive instability anomalies suggests a greater opportunity for the region to become less stable and even unstable. The zonally asymmetrical distribution of both the dynamic and thermodynamic structures apparently favors the westward propagation of the convection system, which happens to couple with a westward-propagating circulation with characteristics similar to a Rossby wave. Such a coupling system can sustain itself through the convection–circulation interaction and continues propagating westward.
c. Surface heat flux and sea surface temperature
Many recent studies have suggested the possible effect of the ocean–atmosphere interaction on the intraseasonal oscillation (e.g., Hendon and Glick 1997; Jones and Weare 1996). Whether such an ocean–atmosphere interaction also exists in the phenomenon examined here is an interesting question. Shown in Fig. 7 is the temporal evolution of the latent heat flux and the 925-hPa wind from day −5 to day 5. A cyclonic circulation is evident in the western North Pacific at day −5, while negative latent heat flux anomalies (indicating more flux into the atmosphere) are located at its southwestern and northeastern corners. At day 0, a negative latent heat flux anomaly appears in a large area all the way from the eastern Arabian Sea to the Philippine Sea and even extending northeastward to Japan, where the westerly and southwesterly anomalies in the same area strengthen. The centers of these anomalies are located mainly over the oceans, for example, the Bay of Bengal, the South China Sea, the Philippine Sea, and in the Pacific south of Japan. A region with a negative anomaly can also be seen in the eastern Indian Ocean in the Southern Hemisphere.
In general, the negative latent heat flux anomalies are located over the oceans at the southern and eastern flanks of the anomalous cyclonic circulation. In the northwestern corner of this circulation, although the circulation is only slightly weaker than its counterpart in the southeastern corner, there is no obvious variation in latent heat flux presumably because it is over or near the land. The relationship between the circulation and the latent heat flux remains similar for the next 10 days when the system moves toward the landmass of east Asia and begins weakening. By comparing (VQ)χ and the latent heat flux, a southwesterly and southerly moisture transport occurs mainly over the negative latent heat flux anomaly. This collocation suggests that through evaporation the oceans provide moisture, which is then transported to the Philippine Sea to converge into the region of deep convection.
The SST and net surface radiation flux at day −10, day 0, and day 10 are shown in Fig. 8. At day −10, before the deep convection moves in, there is a positive SST anomaly in the Philippine Sea, while a positive anomaly of net surface radiation flux (indicating more radiation into the ocean surface) is located to the northwest of the positive SST anomaly. At day 0, the positive SST anomaly almost disappears, while a negative anomaly of net surface radiation flux, which corresponds to the negative OLR anomaly shown in Fig. 2d, covers the Philippine Sea and the South China Sea. At the same time, negative SST anomalies can be found in the southern South China Sea and the subtropical western Pacific southeast of Japan. By day 10, a region with a negative SST anomaly appears in the western North Pacific, while a region with a negative radiation anomaly is located to the west. This pattern resembles the feature shown in Fig. 8a except that the signs are reversed. Coincidentally, the negative SST anomaly happens to occur in the regions where both the latent heat flux anomaly and the net surface radiation anomaly have been strongly negative for the past 10–15 days.
The above results reveal a close spatial and temporal relationship between the SST, latent heat flux, and net surface radiation flux anomalies. The SST fluctuations seem to be affected by the atmospheric conditions. When there is stronger (weaker) convection, the corresponding increased (decreased) cloudiness and enhanced (weakened) southwesterly leads to less (more) downwelling shortwave radiation into the ocean surface and more (less) latent heat flux leaving the ocean surface. These processes can result in the rise or fall of the SST. A rough estimate can be done based on the National Oceanographic Data Center World Ocean Atlas 1994 (Levitus et al. 1994) to check whether the above hypothesis is probable. According to the atlas, the climatological depth of the mixed layer in the Philippine Sea is about 20 m in summer. Such a mixed layer can be cooled (or warmed) by about 0.1K in about 15 days by a net heat flux of 8 Wm−2 by neglecting the effects of ocean dynamics. This rough estimate is consistent with the amplitudes of the documented surface heat flux (Figs. 7 and 8) and with the intraseasonal timescale considered in this study. This relationship is similar to the ocean–atmosphere interaction in a Madden–Julian oscillation, which propagates eastward along the equator, as documented in Hendon and Glick (1997) and Jones and Weare (1996).
5. Discussions and conclusions
This diagnostic study examined the spatial and temporal characteristics associated with the northwestward propagation of the convection–circulation system in the western North Pacific at the intraseasonal timescale. The evolution of this phenomenon can be divided into the following three stages as illustrated in the schematic diagrams shown in Fig. 9. The following discussion is based on the situation when the convection in the western North Pacific is enhanced. Similar discussions considering the anomalies in reversed signs can also be applied to the situation when the convection is suppressed.
Stage 1: A convective region develops in the equatorial western Pacific near Indonesia and propagates eastward along the equator, while another convective region propagates westward near 20°N from the central North Pacific to the Philippine Sea. The latter is located to the east of a positive vorticity anomaly, which is one of the centers of a wavelike structure extending from the Philippine Sea northeastward to the central North Pacific. The merging of these two convective regions enhances the convection in the western North Pacific, which in turn enhances the low-level cyclonic anomaly located to the northwest. At the same time, the convection near the Philippines and Taiwan is relatively inactive and the SST anomaly is positive.
Stage 2: A deep convection moves northwestward in the western North Pacific to the southwestern corner of the well-developed low-level cyclonic anomaly. This coupled circulation–convection system then propagates westward. During the earlier northwestward or later westward propagation of the deep convection, a region of near-surface convergence is always located to the northwest or west of the deep convection. At the same time, the southwesterly is strengthened all the way from the Bay of Bengal to the Philippine Sea. Latent heat flux into the atmosphere is significantly enhanced underneath the strengthened southwesterly. Moisture is transported northeastward from the oceans located in the southwest of the cyclonic anomaly and converges at the northwestern corner of the deep convection where the lower troposphere becomes potentially less stable.
Stage 3: The system consequently propagates westward toward Taiwan and southern China and starts weakening when it approaches the landmass of east Asia. Quite often, a low-level anticyclonic circulation in the Philippine Sea and suppressed convection in the Philippine Sea and near Indonesia begin developing. These systems, which can be inferred from Fig. 3a but with the signs reversed, are ready to follow the path of the previous convection-enhanced system to propagate toward marine east Asia.
The results presented here suggest an interesting interaction between circulation and convection and between the atmosphere and oceans. Mechanisms are proposed in the following discussions to explain the observed interesting phenomenon.
a. Merging of the convection
The western North Pacific is one of the places where the SST is highest in the Pacific and the low-level confluence is strong during the boreal summer. The atmosphere and ocean in this region also exhibit strong intraseasonal variability for reasons not yet understood. Before an enhanced convection event, there is often a preceding suppressed convection event, which is characterized by the northeasterly anomaly in the South China Sea and the Philippine Sea. This feature is an indication of a weakened southwesterly that exists climatologically in the region. The moisture transport at this stage is similar to the pattern shown in Figs. 5a and 5b but with the signs reversed. The results are the anomalous moisture converging near the equator and the anomalous moisture divergence in marine east Asia. This could lead to a higher probability for the convection to develop near the equator, instead of farther north as usual. This convection then propagates eastward and northeastward near the equator. While it is tempting to explain this eastward-propagating feature as the equatorial Kelvin wave, the corresponding vorticity and circulation anomalies (e.g., Figs. 3 and 7, respectively) do not appear as a dipole straddling the equator like a Kelvin wave (Hendon 1986). Instead, there is a positive vorticity anomaly and cyclonic vortex straddling the equator to the east of the equatorial convection (e.g., Figs. 3c and 7b). Since the focus of this study is the northwestward propagation, the exact mechanism leading to the eastward propagation is not further explored here.
Theoretically, the suppressed convection that has been prevailing in the western North Pacific could force a divergent Rossby wave emanating from the Philippine Sea into the subtropics following a route similar to a great circle (e.g., Hoskins and Karoly 1981). East of the anticyclonic circulation anomaly in the Philippine Sea there is a cyclonic circulation anomaly with a low-level convergent anomaly. While the energy disperses downstream following the great-circle-like route, the anticyclonic and cyclonic circulation anomalies propagate westward along the latitudes near 20°N like a Rossby wave often does. At the same time, the equatorial convection can force and maintain a Rossby wave–like cyclonic circulation to its north in the Philippine Sea. The combination of these two Rossby wave–like circulation leads to the rapid development of the cyclonic circulation in the Philippine Sea (e.g., Figs. 3b and 3c). The southwesterly anomaly embedded in the cyclonic circulation then intensifies in the region from the Indian Ocean to the Philippine Sea. The enhanced southwesterly transports more moisture into the Philippine Sea, where it becomes more susceptible to the development of a deep convection. It is proposed that, through the above processes, the eastward-propagating convection along the equator and the westward-propagating low-level convergence along 20°N to the Philippine Sea creates a condition favoring the flaring of a deep convection in the Philippine Sea and resulting in the merging of these convections.
b. Northwestward propagation
After the merging of the two convection regions, the deep convection in the Philippine Sea begins developing and couples with the cyclonic circulation. The latent heat released through the deep convection is then used to strengthen the low-level cyclonic circulation in a manner similar to the forcing mechanism proposed by Gill (1980). The enhanced southwesterly associated with the strengthened cyclonic circulation extracts moisture from the Indian Ocean, the Maritime Continent, the South China Sea and transports this moisture to the northwest corner of the deep convection where the moisture converges. The region to the northwest of the deep convection becomes potentially unstable and the convection can continue to flare with the help of near-surface convergence and upward motion. The deep convection consequently propagates northwestward.
Another possible mechanism is the mean flow advection as suggested by Nitta (1987). The speed of the northwestward propagation shown here is around 1–1.5 m s−1, which is consistent with the results of Wang and Rui (1990) but is much less than the 5–10 m s−1 documented by Nitta (1987). Moreover, the climatological mean flow at 850 hPa in the concerned region is around 3–4 m s−1 and is mainly southerly in the ocean to the east of the Philippines and Taiwan. Therefore, Nitta's conclusion does not seem applicable for this study.
Why does the moisture convergence occur in the northwest of the deep convection instead of in the center? Friction may play an important role in triggering the northwestward propagation as it does in the eastward propagation of the equatorial Kelvin wave (e.g., Salby et al. 1994). Hypothetically, the surface friction can cause the near-surface flow to converge near the center of the cyclonic circulation in the Philippine Sea. The deep convection that forces or maintains a Rossby wave is usually located to the southeast of the cyclonic circulation. Therefore, the near-surface moisture convergence resulted from this friction is located to the northwest of the deep convection and results in the northwestward propagation. This phase–shift relationship can be seen clearly by comparing the 10-m divergence (Fig. 4), the 850 vorticity (Fig. 3), and the moisture convergence (Fig. 5).
c. Tropical versus subtropical features
Throughout this evolution, the low-level cyclonic circulation propagates mainly westward while the deep convection in the western North Pacific propagates northwestward. This westward-propagating circulation is likely to be a Rossby wave, which is sustained by the latent heat release through the coupling with the deep convection. While a Rossby wave itself can propagate westward, the moisture convergence and potentially unstable condition that are always located to the west of the deep convection can also contribute to the westward propagation. This westward-propagating feature exhibits characteristics similar to the moist Rossby wave sustained by the instability in the easterly vertical shear found by Xie and Wang (1996).
While the westward propagation seems to be consistent with the emanation of a Rossby wave from the western Pacific as simulated by Wang and Xie (1997), there are several features that do not agree with their model results. The major discrepancies are listed as follows.
While the circulation bears the characteristics of the Rossby wave, the characteristics similar to the Kelvin wave do not seem to exist. For example, there should be a vorticity dipole to the east of 150°E that straddles the equator and is meridionally confined to the equator (e.g., Fig. 3c) as shown in Hendon (1986), if there were a Kelvin wave component. Instead, there is a positive anomaly, not a dipole, straddling the equator (e.g., Fig. 3d). The 925-hPa wind shown in Fig. 7a does not resemble a Kelvin–Rossby wave packet either. Furthermore, the orientation of the convection in the Philippine Sea at day −5 does not appear in the horseshoe shape as it is supposed to be in a Kelvin–Rossby wave packet (Wang and Rui 1990; Wang and Xie 1997). It seems more likely that the westward-propagating vorticity anomaly originates from a perturbation in the subtropics and then amplifies in the Philippine Sea through the coupling with the convection.
The positive vorticity anomaly in the Philippine Sea shown in Fig. 3c can be traced back to the vorticity anomaly that is located to the east 10 days early. This anomaly is clearly not related to the Kelvin–Rossby wave packet. The subtropical feature, which is statistically significant in various variables and evolves systematically through the evolution, cannot be neglected.
When the system approaches the landmass in east Asia, the circulation begins weakening and stalling presumably due to the increasing friction. The moisture supply from the ocean therefore begins declining and leads to the further weakening of the circulation. The stalling of this system is apparently reflected in the slowing down of the moisture convergence center from day 0 to day 5. This may explain why the system does not continue moving to south Asia to trigger another round of intraseasonal oscillation as discussed in Wang and Xie (1997), because there is no land–sea contrast and topography in their model.
The existence of these subtropical features and their merging with the tropical features documented in this study do not exclude the possibility that the northwestward propagation can occur even without the presence of these subtropical features. In the study by Lau and Chan (1986), the convection in the western North Pacific did not merge with any subtropical features and still propagated northwestward. This discrepancy is probably due to the different methodology. Lau and Chan (1986) applied the extended EOF (EEOF) to the OLR in a domain covering the Asian monsoon region and the Pacific. Since their EEOF is essentially a large-scale tropical feature, the more regional subtropical structures shown here may not be seen in their study. Our index was designed to emphasize the convection activity in the Philippine Sea and therefore captures more subtropical structures.
d. Ocean–atmosphere interaction
The influence of the atmosphere on the ocean is clearly seen in the areas where the variations in convection and near-surface circulation are significant. The SST variation is qualitatively consistent with the variation in the net surface radiation flux and latent heat flux, which affects the energy flux budget at the ocean surface. The oceans also play an important role in supplying moisture, which is later transformed into latent heating in the convection to enhance the circulation. However, the major supply of moisture comes from the oceans located to the southwest of the convection through a long-range transport. The ocean located to the northwest of the convection does not contribute much moisture. The positive SST anomaly located to the northwest of the convection contributes negligible sensible and latent heat fluxes. Its influence along the propagation track seems to be much less than the moisture convergence and low-level lifting. Overall speaking, the atmosphere seems to play a dominant role during the ocean–atmosphere interaction processes, while the ocean plays a more passive role in response to the atmospheric forcing.
Acknowledgments
The authors appreciate the valuable comments of two anonymous reviewers and Harry Hendon. Their comments have led to a significant improvement of this paper. This study is supported by the National Science Council, Taiwan, under Grants NSC 89-2111-M-002-012-AGT and NSC 89-2111-M-002-032.
REFERENCES
Bolton, D., 1980: The computation of equivalent potential temperature. Mon. Wea. Rev, 108 , 1046–1053.
Brunet, G., R. Vautard, B. Legras, and S. Edouard, 1995: Potential vorticity on isentropic surfaces: Climatology and diagnostics. Mon. Wea. Rev, 123 , 1037–1058.
Chen, T-C., and M. Murakami, 1988: The 30–50 day variation of convective activity over the western Pacific Ocean with the emphasis on the northwestern region. Mon. Wea. Rev, 116 , 892–906.
Gibson, J. K., P. Kallberg, S. Uppala, A. Hernandez, A. Nomura, and E. Serrano, 1997: ECMWF Re-analysis Project Report Series. 1. ERA description. European Centre for Medium-Range Weather Forecasts, 72 pp.
Gill, A., 1980: Some simple solutions for heat-induced tropical circulation. Quart. J. Roy. Meteor. Soc, 106 , 447–462.
Hamming, R. W., 1989: Digital Filters. Prentice-Hall, 284 pp.
Hendon, H. H., 1986: Streamfunction and velocity potential representation of equatorial trapped wave. J. Atmos. Sci, 43 , 3038–3042.
Hendon, H. H., and J. Glick, 1997: Intraseasonal air–sea interaction in the tropical Indian and Pacific Oceans. J. Climate, 10 , 647–661.
Hoskins, B. J., and D. J. Karoly, 1981: The steady linear response of a spherical atmosphere to thermal and orographic forcing. J. Atmos. Sci, 38 , 1179–1196.
Hsu, H-H., 1996: Global view of the intraseasonal oscillation during northern winter. J. Climate, 9 , 2386–2406.
Jones, C., and B. C. Weare, 1996: The role of low-level moisture convergence and ocean latent heat fluxes in the Madden and Julian oscillation: An observational analysis using ISCCP data and ECMWF analyses. J. Climate, 9 , 3086–3104.
Kawamura, R., T. Murakami, and B. Wang, 1996: Tropical and mid-latitude 45-day perturbations over the western Pacific during the northern summer. J. Meteor. Soc. Japan, 74 , 867–890.
Knutson, T. R., and K. M. Weickmann, 1987: 30–60-day atmospheric oscillations: Composite life cycles of convection and circulation anomalies. Mon. Wea. Rev, 115 , 1407–1436.
Krishnamurti, T. N., and D. Subrahmanyam, 1982: The 30–50 day mode at 850 mb during MONEX. J. Atmos. Sci, 39 , 2088–2095.
Lau, K. M., and P. H. Chan, 1985: Aspects of the 40–50 day oscillation during the northern winter as inferred from outgoing longwave radiation. Mon. Wea. Rev, 113 , 1889–1909.
Lau, K. M., and P. H. Chan, 1986: Aspects of the 40–50 day oscillation during the northern summer as inferred from outgoing longwave radiation. Mon. Wea. Rev, 114 , 1354–1367.
Levitus, S., J. Antonov, and T. P. Boyer, 1994: International Variability of Upper Ocean Thermal Structure. Vol. 5, World Ocean Atlas, NOAA Atlas NESDIS, 176 pp.
Liebmann, B., and C. A. Smith, 1996: Description of a complete (interpolated) outgoing longwave radiation dataset. Bull. Amer. Meteor. Soc, 77 , 1275–1277.
Murakami, T., and T. Nakazawa, 1985: Tropical 45 day oscillations during the 1979 northern summer. J. Atmos. Sci, 42 , 1107–1122.
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.
Reynolds, R. W., and T. M. Smith, 1994: Improved global sea surface temperature analyses using optimum interpolation. J. Climate, 7 , 929–948.
Salby, M., R. Garcia, and H. Hendon, 1994: Planetary circulations in the presence of climatological and wave-induced heating. J. Atmos. Sci, 51 , 2344–2367.
Sardeshmukh, P. D., and B. J. Hoskins, 1984: Spatial smoothing on the sphere. Mon. Wea. Rev, 112 , 2524–2529.
Wang, B., and H. Rui, 1990: Synoptic climatology of transient tropical intraseasonal convection anomalies. Meteor. Atmos. Phys, 44 , 43–61.
Wang, B., and X. Xie, 1997: A model for the boreal summer intraseasonal oscillation. J. Atmos. Sci, 54 , 72–86.
Xie, X., and B. Wang, 1996: Low-frequency equatorial waves in vertically sheared zonal flow. Part II: Unstable waves. J. Atmos. Sci, 53 , 3589–3605.
Yasunari, T., 1979: Cloudiness fluctuations associated with the Northern Hemisphere summer monsoon. J. Meteor. Soc. Japan, 57 , 227–242.
Yasunari, T., 1981: Structure of an Indian summer monsoon system with around 40-day period. J. Meteor. Soc. Japan, 59 , 336–354.
Zhu, B., and B. Wang, 1993: The 30–60-day convection seesaw between the tropical Indian and western Pacific Oceans. J. Atmos. Sci, 50 , 184–199.
Ratio of the 30–60-day band variance to the total variance for (a) OLR and (b) SST. Arrows indicates the distance and the direction of a (c) OLR and (d) 850-hPa vorticity anomaly that moves from day 0 to day 5. The shading in (c) and (d) present the lagged correlation distribution. Only those greater than 0.75 are plotted. Contour intervals are 0.02 and 0.05 for (a) and (b), respectively, and are 0.05 for (c) and (d)
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Ratio of the 30–60-day band variance to the total variance for (a) OLR and (b) SST. Arrows indicates the distance and the direction of a (c) OLR and (d) 850-hPa vorticity anomaly that moves from day 0 to day 5. The shading in (c) and (d) present the lagged correlation distribution. Only those greater than 0.75 are plotted. Contour intervals are 0.02 and 0.05 for (a) and (b), respectively, and are 0.05 for (c) and (d)
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Ratio of the 30–60-day band variance to the total variance for (a) OLR and (b) SST. Arrows indicates the distance and the direction of a (c) OLR and (d) 850-hPa vorticity anomaly that moves from day 0 to day 5. The shading in (c) and (d) present the lagged correlation distribution. Only those greater than 0.75 are plotted. Contour intervals are 0.02 and 0.05 for (a) and (b), respectively, and are 0.05 for (c) and (d)
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the OLR (contoured) and diabatic heating (shaded) at (a) day −15, (b) day −10, (c) day −5, (d) day 0, (e) day 5, (f) day 10. Contour intervals are 2 and 15 W m−2 for the OLR and diabatic heating, respectively. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the OLR (contoured) and diabatic heating (shaded) at (a) day −15, (b) day −10, (c) day −5, (d) day 0, (e) day 5, (f) day 10. Contour intervals are 2 and 15 W m−2 for the OLR and diabatic heating, respectively. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the OLR (contoured) and diabatic heating (shaded) at (a) day −15, (b) day −10, (c) day −5, (d) day 0, (e) day 5, (f) day 10. Contour intervals are 2 and 15 W m−2 for the OLR and diabatic heating, respectively. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the OLR (shaded) and 850-hPa vorticity (contoured) at (a) day −15, (b) day −10, (c) day −5, (d) day 0, (e) day 5, (f) day 10. Contour intervals are 2 W m−2 and 1 × 10−7 s−1 for the OLR and vorticity, respectively. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the OLR (shaded) and 850-hPa vorticity (contoured) at (a) day −15, (b) day −10, (c) day −5, (d) day 0, (e) day 5, (f) day 10. Contour intervals are 2 W m−2 and 1 × 10−7 s−1 for the OLR and vorticity, respectively. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the OLR (shaded) and 850-hPa vorticity (contoured) at (a) day −15, (b) day −10, (c) day −5, (d) day 0, (e) day 5, (f) day 10. Contour intervals are 2 W m−2 and 1 × 10−7 s−1 for the OLR and vorticity, respectively. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the OLR (shaded) and 10-m divergence (contoured) at (a) day −15, (b) day −10, (c) day −5, (d) day 0, (e) day 5, (f) day 10. Contour intervals are 2 W m−2 and 0.4 × 10−6 s−1 for the OLR and 10-m divergence, respectively. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the OLR (shaded) and 10-m divergence (contoured) at (a) day −15, (b) day −10, (c) day −5, (d) day 0, (e) day 5, (f) day 10. Contour intervals are 2 W m−2 and 0.4 × 10−6 s−1 for the OLR and 10-m divergence, respectively. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the OLR (shaded) and 10-m divergence (contoured) at (a) day −15, (b) day −10, (c) day −5, (d) day 0, (e) day 5, (f) day 10. Contour intervals are 2 W m−2 and 0.4 × 10−6 s−1 for the OLR and 10-m divergence, respectively. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the ∇·(VQ) (contoured) and (VQ)χ (arrows) at (a) day −5, (b) day 0, (c) day 5. Contour interval is 1 × 10−9 g K g−1 s−1. The length of the reference arrow is equivalent to 0.01 g K g−1 m s−1. Solid and dashed lines indicate divergence and convergence, respectively. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the ∇·(VQ) (contoured) and (VQ)χ (arrows) at (a) day −5, (b) day 0, (c) day 5. Contour interval is 1 × 10−9 g K g−1 s−1. The length of the reference arrow is equivalent to 0.01 g K g−1 m s−1. Solid and dashed lines indicate divergence and convergence, respectively. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the ∇·(VQ) (contoured) and (VQ)χ (arrows) at (a) day −5, (b) day 0, (c) day 5. Contour interval is 1 × 10−9 g K g−1 s−1. The length of the reference arrow is equivalent to 0.01 g K g−1 m s−1. Solid and dashed lines indicate divergence and convergence, respectively. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Cross sections of lagged regression coefficients between the convection index and diabatic heating (contoured), zonal wind, vertical velocity, and potential instability (−∂θe/∂p, shaded) at 17.5°N at (a) day −10, (b) day −5, (c) day 0, (d) day 5, and (e) day 10. Contour intervals are 2 × 10−6 K s−1 and 4 × 10−6 K Pa−1 for the diabatic heating and potential instability, respectively. Vertical velocity (ω) has been multiplied by 100. Length of the reference arrow is equivalent to 1 m s−1 or 0.01 hPa s−1. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Cross sections of lagged regression coefficients between the convection index and diabatic heating (contoured), zonal wind, vertical velocity, and potential instability (−∂θe/∂p, shaded) at 17.5°N at (a) day −10, (b) day −5, (c) day 0, (d) day 5, and (e) day 10. Contour intervals are 2 × 10−6 K s−1 and 4 × 10−6 K Pa−1 for the diabatic heating and potential instability, respectively. Vertical velocity (ω) has been multiplied by 100. Length of the reference arrow is equivalent to 1 m s−1 or 0.01 hPa s−1. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Cross sections of lagged regression coefficients between the convection index and diabatic heating (contoured), zonal wind, vertical velocity, and potential instability (−∂θe/∂p, shaded) at 17.5°N at (a) day −10, (b) day −5, (c) day 0, (d) day 5, and (e) day 10. Contour intervals are 2 × 10−6 K s−1 and 4 × 10−6 K Pa−1 for the diabatic heating and potential instability, respectively. Vertical velocity (ω) has been multiplied by 100. Length of the reference arrow is equivalent to 1 m s−1 or 0.01 hPa s−1. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. The regression coefficients have been multiplied by one std dev of the convection index and only those that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the surface latent heat flux (shaded) and 925-hPa winds at (a) day −5, (b) day 0, and (c) day 5. Contour interval is 2 W m−2. Length of the reference arrow is equivalent to 1 m s−1. Dark shading and solid lines indicate downward flux, while light shading and dashed lines indicate upward flux. Regression coefficients have been multiplied by one std dev of the convection index and only those regressions that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the surface latent heat flux (shaded) and 925-hPa winds at (a) day −5, (b) day 0, and (c) day 5. Contour interval is 2 W m−2. Length of the reference arrow is equivalent to 1 m s−1. Dark shading and solid lines indicate downward flux, while light shading and dashed lines indicate upward flux. Regression coefficients have been multiplied by one std dev of the convection index and only those regressions that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the surface latent heat flux (shaded) and 925-hPa winds at (a) day −5, (b) day 0, and (c) day 5. Contour interval is 2 W m−2. Length of the reference arrow is equivalent to 1 m s−1. Dark shading and solid lines indicate downward flux, while light shading and dashed lines indicate upward flux. Regression coefficients have been multiplied by one std dev of the convection index and only those regressions that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the SST (contoured) and net surface radiation flux (shaded) at (a) day −10, (b) day 0, (c) day 10. Contour intervals are 2 W m−2 and 0.03 K for net surface radiation flux and SST, respectively. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. Regression coefficients have been multiplied by one std dev of the convection index and only those regressions that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the SST (contoured) and net surface radiation flux (shaded) at (a) day −10, (b) day 0, (c) day 10. Contour intervals are 2 W m−2 and 0.03 K for net surface radiation flux and SST, respectively. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. Regression coefficients have been multiplied by one std dev of the convection index and only those regressions that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Lagged regression coefficients between the convection index and the SST (contoured) and net surface radiation flux (shaded) at (a) day −10, (b) day 0, (c) day 10. Contour intervals are 2 W m−2 and 0.03 K for net surface radiation flux and SST, respectively. Dark shading and solid lines indicate positive values, while light shading and dashed (and dotted) lines indicate negative values. Regression coefficients have been multiplied by one std dev of the convection index and only those regressions that are significant at the 0.05 level are plotted
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Schematic diagrams illustrate the three stages of this evolution. Circles with arrowheads indicate the 850-hPa anomalous circulation. The large arrow with broad shaft in (b) indicates the major moisture transport, while the dashed arrow indicates the propagation of the convection. Shaded regions marked with LHF represent the regions where anomalous latent heat flux was released into the atmosphere. The cloudlike symbol indicates the convection-active region
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Schematic diagrams illustrate the three stages of this evolution. Circles with arrowheads indicate the 850-hPa anomalous circulation. The large arrow with broad shaft in (b) indicates the major moisture transport, while the dashed arrow indicates the propagation of the convection. Shaded regions marked with LHF represent the regions where anomalous latent heat flux was released into the atmosphere. The cloudlike symbol indicates the convection-active region
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
Schematic diagrams illustrate the three stages of this evolution. Circles with arrowheads indicate the 850-hPa anomalous circulation. The large arrow with broad shaft in (b) indicates the major moisture transport, while the dashed arrow indicates the propagation of the convection. Shaded regions marked with LHF represent the regions where anomalous latent heat flux was released into the atmosphere. The cloudlike symbol indicates the convection-active region
Citation: Journal of Climate 14, 18; 10.1175/1520-0442(2001)014<3834:NPOTIO>2.0.CO;2
