Abstract

Characteristics of sea-breeze circulation over the tropical site of Jakarta, Indonesia, have been documented based on analyses of satellite images and data from long-term L-band boundary layer radar measurements carried out at Serpong (6.4°S, 106.7°E). Inspection of satellite imagery reveals that a sea-breeze front develops well along the northern coastal plain of West Java and propagates inland until its structure is deformed over complex topography. It is found that the sea-breeze signal detected by the boundary layer radar is most well defined during the dry season months of July–October. In all of these months, radar observations indicate a late afternoon intensification of sea-breeze flow in the 0.5–0.8-km height range between 1700 and 1800 LT, which is not elucidated upon by surface measurements. The effect of weather conditions on the sea-breeze pattern is investigated by using a cloudiness index derived from data of incoming solar radiation. The results show that sea-breeze intrusion over the radar site occurs earlier during more cloudy days, whereas the intensity of sea-breeze circulation weakens accordingly. In the rainy season months of January and February, diurnal wind variation is characterized by daytime onshore flow enhancement, which is not likely attributed to sea-breeze circulation.

1. Introduction

The sea-breeze circulation (SBC) is a mesoscale atmospheric phenomenon induced by differential heating between land and adjacent sea during the daytime. The existence of SBC has been recognized for centuries and extensively studied throughout the modern history of meteorology (see, e.g., Rotunno et al. 1992; Simpson 1994). However, it would not need long argument to note that our knowledge of the characteristics of SBC in the tropical&sol=uatorial region is still lacking observational data. Indeed, the effect of a small Coriolis factor in low latitudes on the horizontal extent and intensity of SBC has been a subject of theoretical and numerical studies for the past three decades with differences, not to say controversies, in the results. Readers may refer to the works by, among others, Walsh (1974), Rotunno (1983), Niino (1987), Yan and Anthes (1987), Dalu and Pielke (1989), and Aritt (1989) for more comprehensive discussions. It has been reported by Garrat (1985) that sea-breeze fronts can occasionally reach as far as 400 km inland in northern Australia. More observations in other tropical&sol=uatorial regions are, however, necessary to improve our understanding of the phenomenon.

The Indonesian archipelago and its enclosed seas constitute a unique climatological region in the Tropics known as the Maritime Continent where a large number of islands implies that SBC plays an important role in weather and climate dynamics. Jakarta is nowadays a densely populated metropolitan situated in the northern coast of the western part of Java, one of the five largest islands in the archipelago. Observations of tropical sea breeze using pilot balloons had historically been done in the city of Batavia (a former name of Jakarta) by van Bemmeln (1913) during the colonial era. However, evidence for the existence of well-defined SBC that can penetrate deeper inland has only been recently reported in a case study by Hadi et al. (2000), based on L-band boundary layer radar (BLR) observations at Serpong (6.4°S, 106.7°E). The radar site is situated about 35 km from the coastline near the southwestern suburb of Jakarta where unattended operation of BLR has been carried out since November 1992 (Hashiguchi et al. 1996). Such long-term BLR measurements provide a valuable dataset for investigating the climatological features of SBC in the region, which, to the authors' knowledge, have not been well documented.

This paper bears two main purposes: first, to investigate the horizontal extent of the tropical sea breeze, and second to identify seasonal as well as intraseasonal variabilities in the characteristic features of the associated circulation. Knowledge of the horizontal scale of the sea breeze is especially important because of the questions that have been raised in a number of theoretical studies as previously mentioned. Although available ground-based measurements are confined to a single point, there are cases where visible satellite images make it possible to identify inland spreading of the sea-breeze front (SBF). The results will also provide a means to assess the representativeness of data from the single-point observations at Serpong. Seasonal variation in the diurnal wind pattern observed with BLR can then be analyzed with regard to the characteristics of SBC over Jakarta. Among other things, the existence of SBC during the rainy season has yet to be clarified because it may have a profound implication in relation to diurnal convections over the Maritime Continent during the active monsoon period (e.g., Houze et al. 1981; Johnson 1992). On the other hand, it has been clearly shown by van Bemmeln (1913) and Hadi et al. (2000) that well-defined SBCs do occur over Jakarta in dry season months. However, it is recognized that development and evolution of SBC during the course of the day depends on several parameters such as atmospheric stability and the prevailing background winds (e.g., Estoque 1962; Walsh 1974), which are determined by synoptic meteorological condition. Additionally, interaction between inland propagating SBF and preexisting convective cells over land can modify the flow systems along the frontal zone (Wakimoto and Atkins 1994; Atkins et al. 1995; Dailey and Fovell 1999). In a simplified approach, cloudiness observed prior to a sea-breeze intrusion over the radar site is used here to represent factors that collectively affect the evolution of SBC. In this context, the relationship between cloudiness and the characteristics of SBC during the dry season will be investigated and discussed in more detail.

2. Datasets

a. BLR and collocated surface meteorological measurements

The main datasets used in this paper are those obtained from long-term BLR observations. Detailed descriptions of the wind profiler radar, which is especially designed for boundary layer observations, can be found in, among others, the work of Hashiguchi et al. (1996). Some important radar parameters can be summarized here such as 1) an operating frequency of 1357.5 MHz; 2) three parabolic antennas with 3.1 m2 aperture pointed in the north, east, and vertical directions; and 3) 1-kW peak power. It is probably important to note that the oblique beams are set to 15° zenith angle. Using this radar system, wind measurements can be made in the height range of 0.3–5.0 km with reasonable accuracy and a vertical resolution of about 100 m. Although the original dataset has a time resolution of about 107 s, 15-min-average profiles with the minimum height of 0.5 km are used in the present work.

Surface meteorological measurements collocated with BLR also provide important information concerning sea-breeze characteristics. Available instruments include 1) an Ogasawara WS-A54 anemometer, 2) an Ikeda RT-5 rain gauge, 3) an EKO MS-42 pyranometer, 4) an EKO MF-11 net pyrradiometer, and 5) Vaisala HMP-133Y temperature and humidity sensors. In this investigation, 10-min-average data are used.

b. Other data resources

In order to investigate the horizontal extent of SBC, visible imageries taken by Japan's Geostationary Meteorological Satellite (GMS) are incorporated. These images over Java are derived from two data resources:

  1. Geometrically corrected satellite images distributed via the Internet by the Department of Information Science, Kochi University (Kikuchi and Kitsuregawa 2001). Data are originally received at the Institute of Industrial Science (IIS), University of Tokyo. The dataset (hereafter referred to as IIS/Kochi-U) provides images with a spatial resolution of 1/20° or about 5 km per pixel.

  2. Satellite images for volcano activity monitoring provided via the Internet by the Space Science and Engineering Center (SSEC), University of Wisconsin—Madison. Although accurate geometrical correction is not applied, this dataset (hereafter SSEC UW—Madison) has good spatial resolution of about 1.5 km per pixel.

It would certainly be best to use high-resolution images of SSEC UW—Madison data for the purpose of this investigation but in some cases only those of the IIS/Kochi-U dataset are available.

Additionally, climatological data of sea surface temperature (SST) derived from Global Ocean Surface Temperature Atlas Plus (GOSTAplus) are also incorporated. The data were obtained from National Aeronautics and Space Administration (NASA) Physical Oceanography Distributed Active Archive Center at the Jet Propulsion Laboratory/California Institute of Technology. Finally, topographic data of Java derived from the GTOPO30 dataset are presented in this paper. This dataset is also provided by NASA Distributed Active Archive Center and freely available via the Internet.

3. Topography and low-level mean winds

In this section we briefly discuss topography and distinct seasonal change in the low-level mean winds as two a priori factors that are expected to affect the sea breezes observed with the BLR at Serpong. According to theoretical predictions, tropical sea breezes can penetrate at least 70 km inland (Niino 1987; also see, e.g., Simpson 1994) or further propagate as far as Rossby deformation radius (Dalu and Pielke 1989). However, as it can be seen from Fig. 1, Java is a mountainous landmass about 1000 km long and 100–200 km wide. The western part is characterized by a relatively wide plain in the north, mountain peaks rising more than 2000 m in the central region, and steep topography at the southern edge of the island. The sea breeze that traverses the coastal plain will reach an area of more complex topography at a distance of about 60 km from the north coast. Effects of topography are known, for example, to intensify the sea breeze through combination with mountain flows (Mahrer and Pielke 1977; Ookuchi et al. 1978). However, it is also pointed out by Ookuchi et al. (1978) that mountains may act as a barrier that prevents the sea breeze from propagating farther inland.

Fig. 1.

A map showing the topography of Java, derived from the GTOPO30 dataset. Contour lines are drawn at elevations of 100, 250, 500, 750, 1500, and 2500 m (thin lines), and 0, 1000, 2000, and 3000 m (thick lines). Approximate position of BLR is indicated

Fig. 1.

A map showing the topography of Java, derived from the GTOPO30 dataset. Contour lines are drawn at elevations of 100, 250, 500, 750, 1500, and 2500 m (thin lines), and 0, 1000, 2000, and 3000 m (thick lines). Approximate position of BLR is indicated

It is well known that ambient wind and land–sea temperature difference are two critical factors in the development of sea breezes (e.g., Walsh 1974; Simpson 1994). A numerical study by Aritt (1993) shows that weak to moderate offshore wind can intensify the sea breeze but a very strong one will prevent it from propagating inland. The low-level mean flow over Jakarta is mainly governed by monsoon circulation with a distinct reversal of zonal wind from predominantly easterly in the dry season to westerly in the rainy season. Such a gross feature of monsoon circulation over Java, which has long been documented (e.g., Braak 1929), is well illustrated by the wind rose diagrams plotted in Fig. 2. The diagrams, which consist of bars and circles, show distributions of wind speed and direction computed from height-averaged hourly mean wind profiles observed with BLR at Serpong during 1992–96. Wind direction is indicated by the orientation of the bars, in the sense that the wind comes from outside and goes into the center of the circles. The length of each segment of the bars represents the percentage of wind data whose speeds are indicated by the thickness (see legend). Dashed-line circles are drawn at 10% intervals to read both relative and cumulative data frequency in each direction, whereas the radius of the inner solid-line circle denotes the total percentage of data with wind speeds less than 1 m s−1. Here, we assume the months of July–August–September (JAS) and December–January–February (DJF) as the core periods of the dry and rainy seasons, respectively. Thus, Fig. 2 shows the aforementioned seasonal change in low-level (below 3 km) winds, and it can be said that wind speeds in dry seasons are typically less than 6 m s−1 with very weak meridional components, while more than 15% of those in rainy seasons are in the excess of 6 m s−1 with stronger northerlies. However, it should be noted that the presence of large-scale convections during an active monsoon period may cause more variabilities in the wind velocities observed with BLR. It has been reported by Hashiguchi et al. (1996) that the passage of a convection center over Serpong can induce horizontal wind variations of about 4 m s−1, whereas rainfall events due to local convection only cause wind perturbation of about 1 m s−1.

Fig. 2.

Wind rose diagrams (see text for explanation) showing distributions of low-level wind speed and direction observed with BLR in dry and rainy season months of JAS and DJF, respectively, during 1992–96. Hourly mean wind profiles averaged over 0.5–3.0-km-height range are used in the analysis

Fig. 2.

Wind rose diagrams (see text for explanation) showing distributions of low-level wind speed and direction observed with BLR in dry and rainy season months of JAS and DJF, respectively, during 1992–96. Hourly mean wind profiles averaged over 0.5–3.0-km-height range are used in the analysis

4. Horizontal extent of the sea-breeze circulation

Inland spreading of SBC can often be identified from cloud lines that develop near the frontal region where convergence occurs (e.g., Simpson 1994). Inspection of GMS visible images, especially those taken during August–October over Java, reveals that interpretation of cloud patterns makes it possible to investigate the horizontal extent and propagation speed of SBF. Figure 3 shows a satellite image form the IIS/Kochi-U dataset that was taken on 30 August 1999 at 1400 LT (UTC + 7). The low-resolution image suffers from smearing when resampled to fit the drawing area but low-level cloud forms are still discernible. Moreover, the absence of clouds at a higher level allows interpretation of the SBF position. By inspecting satellite images taken over the previous 5 h and judging from the inland expansion of cloud-free zones in Fig. 3, it can be noted that sea breezes may have developed over the majority of coastal plains circumventing the island but we are mainly concerned with those that occurred in the western part, where the BLR is situated. Here, the SBF can be identified as a distinct boundary between cloud-covered and cloud-free zones, which extends about 300 km along the northern coastal plain.

Fig. 3.

GMS visible image taken over Java on 30 Aug 1999 at 1400 LT (derived from IIS/Kochi-U dataset). Position of sea-breeze fronts along the northern and southern coastal regions of West Java are interpreted based on low-level cloud distribution. A white circle indicates BLR location (also see Fig. 1)

Fig. 3.

GMS visible image taken over Java on 30 Aug 1999 at 1400 LT (derived from IIS/Kochi-U dataset). Position of sea-breeze fronts along the northern and southern coastal regions of West Java are interpreted based on low-level cloud distribution. A white circle indicates BLR location (also see Fig. 1)

Tracking of SBF is more difficult over the southern coast because much of the cloud observed at 1400 LT seemed to have been generated due to the effect of topography. However, we could still identify a gradual expansion of the cloud-free zone over the relatively flat area of a small peninsula. We may then speculate that SBC was generated therein but that blocking by steep topography (also see Fig. 1) prevented it from propagating farther inland, which resulted in the development of a wider cloud band over the southern side of the mountains.

To investigate the front propagation, time–latitude sections of cropped images (of 20 pixels wide centered at the longitude of the BLR position) are plotted in Fig. 4. The averaged elevation profile in the right panel is derived from smoothed GTOPO30 topographic data. Hourly meridional wind velocities and relative humidity observed at the radar site are also depicted in the upper panel of Fig. 4. The images are again smeared out because of insufficient spatial resolution but the southward propagation of the front should be clearly identified from the expansion of the cloud-free region. On 29 August 1999, a higher-level cloud form appeared in the morning spreading above the northwestern sector of the radar position, which makes it difficult to see the SBF-related cloud movement. Front propagation is more obviously seen from the satellite images of 30 August 1999. It can be seen, however, that in both cases sea-breeze intrusions over the radar site are indicated by a marked increase in relative humidity (RH) and the enhancements of northerlies and southerlies in the lower (below 1 km) and upper (1.5–2.5 km) layers, respectively. Inspection of high time-resolution surface winds and relative humidity plotted in Fig. 5 suggests that SBF reached the radar site between 1300 and 1400 LT on 29 August 1999, while it occurred almost 1 h later on the following day.

Fig. 4.

Time–latitude sections of satellite images (from IIS/Kochi-U dataset) showing inland propagation of the sea-breeze front from the Jakarta coast north of the BLR site (white horizontal line) on 29 and 30 Aug 1999. Corresponding time series of hourly winds and RH are plotted in the upper panel. Vl and Vu denote BLR-derived meridional winds averaged in the 0.5–0.8- and 1.5–2.5-km-height ranges, respectively, while Vsfc is that obtained from an anemometer at 10-m height. Averaged elevation profile in the right panel is derived from smoothed GTOPO30 topographic data.

Fig. 4.

Time–latitude sections of satellite images (from IIS/Kochi-U dataset) showing inland propagation of the sea-breeze front from the Jakarta coast north of the BLR site (white horizontal line) on 29 and 30 Aug 1999. Corresponding time series of hourly winds and RH are plotted in the upper panel. Vl and Vu denote BLR-derived meridional winds averaged in the 0.5–0.8- and 1.5–2.5-km-height ranges, respectively, while Vsfc is that obtained from an anemometer at 10-m height. Averaged elevation profile in the right panel is derived from smoothed GTOPO30 topographic data.

Fig. 5.

Detailed time variations of zonal and meridional wind, and relative humidity corresponding to Fig. 4. Shadings are used to emphasize sea-breeze intrusion time over the radar site on 29 and 30 Aug 1999

Fig. 5.

Detailed time variations of zonal and meridional wind, and relative humidity corresponding to Fig. 4. Shadings are used to emphasize sea-breeze intrusion time over the radar site on 29 and 30 Aug 1999

It is probably important to remark that the RH data originally contained values that exceeded 100%, which is unrealistic. This seemed to happen because appropriate calibration was not performed in the measurements. We have made correction to the data by assuming that the daily maximum relative humidity observed during nightime, RHmax, does not exceed 100% and all values scale accordingly. If RHmax is found to be greater than 100%, the corrected values are calculated as RH × (100/RHmax). This correction has been applied to the RH data in Fig. 5 and throughout this paper. In this investigation, the errors in RH measurements do not actually affect our qualitative interpretation as we are only concerned with its relative time variation.

In most observed cases, cloud lines that can be associated with SBF start to emerge around 1000 LT near the coastline signifying the sea-breeze onset time, whereas the well-defined SBC is typically established over the radar site around 1400 LT. It implies a propagation speed of roughly 10 km h−1 or 2.8 m s−1 over flat topography, which is quite moderate considering that values of 1–5 m s−1 have been documented for front progression in the absence of a background wind (Stull 1988). However, as has been demonstrated by the case of 29–30 August 1999, the sea breeze may reach the radar site at different times from day to day. This feature can be examined more closely using higher-resolution satellite images (SSEC UW—Madison dataset) presented in Fig. 6, which were taken on 20 and 27 August 2000. Unfortunately, neither BLR nor surface meteorological data are available for comparison during this period. Nevertheless, it is obvious that on both days SBF had appeared near the coastline between 0932 and 1125 LT. It is also clear that SBF was progressing farther inland and passed over the radar site around 1432 LT before it reached the area of more complex topography at 1532 LT on 20 August 2000. Inland propagation of SBF is less easy to identify on 27 August 2000 because of more clouds that developed over land. However, inspection of the original satellite images (not shown) suggests that SBF reached the mountain base at 1432 LT, which is earlier than had occurred on 20 August 2000; leaving a patch of postfrontal clouds over the radar site (see Fig. 6). As it has also been shown in the case of 29–30 August 1999, there appears to be a tendency for SBF to arrive earlier at the radar site when more clouds are observed over land throughout the day.

Fig. 6.

Time–latitude sections of satellite images and average profile of elevation as in Fig. 4 (lower and right panels), except that higher-resolution imageries from the SSEC UW—Madison dataset have been used for observations on 20 and 27 Aug 2000

Fig. 6.

Time–latitude sections of satellite images and average profile of elevation as in Fig. 4 (lower and right panels), except that higher-resolution imageries from the SSEC UW—Madison dataset have been used for observations on 20 and 27 Aug 2000

It is difficult to discuss what happens after SBF encountered mountainous terrain without more detailed observations. Besides, satellite visible imagery is not usable beyond 1700 LT. Nonetheless, the last legible images obtained in the day on 20 and 27 August 2000 are worth examining to give an insight into possible situations that may evolve with time. For the case of 20 August 2000, Fig. 7a shows low-level cloud patterns over West Java at 1632 LT that suggest that deformation following irregular topography occurred to the structure of SBF propagating from the north. It is worth noting that formation of deep clouds over the entire island is apparently suppressed during this day, which probably indicates an effect of large-scale subsidence. It is suspected that, under such circumstances, blocking by certain parts of topography keeps SBF stationary for a period of time. Using radiosonde profiles, Hadi et al. (2000) showed that well-defined SBC still persisted over Serpong until 1800 LT on 11 October 1993, during which time a relatively strong inversion layer was identified at 3 km. Figure 7b, in contrast to the former case, shows enhanced cloud developments especially over sloping terrain on 27 August 2000 at 1533 LT. This means that more moist air from below was being transported deeper into the free troposphere. Another possible effect of orography can thus be considered in that venting of onshore flow over the mountain peaks may lead to reduction or even extinction of the upper-level return flow (e.g., Banta et al. 1993).

Fig. 7.

Satellite images showing low-level cloud patterns over West Java (a) at 1632 LT 20 Aug, and (b) at 1533 LT 27 Aug 2000 (also see Fig. 6)

Fig. 7.

Satellite images showing low-level cloud patterns over West Java (a) at 1632 LT 20 Aug, and (b) at 1533 LT 27 Aug 2000 (also see Fig. 6)

5. Seasonal variation in the sea-breeze signal

It was discussed in the previous section that a well-defined SBC can be identified from wind profiles observed with BLR at Serpong as enhancements in the northerly (sea breeze) flow below 1.0 km and southerly (return) flow in the 1.5–3.0-km height range (see Fig. 4). Assuming that there are only small changes in the total depth of sea-breeze circulation, seasonal variation in the sea-breeze signal is here investigated by calculating 24-h composite time series of meridional winds averaged over 0.5–0.8-km and 1.5–2.5-km height ranges after the daily mean has been subtracted. For this purpose, 15-min-average wind profiles observed during 1993, 1994, 1995, 1996, and 1999 are analyzed. Although long gaps exist in the time series, available data allow each climatological month to be represented by at least 2-yr observations. Only data with vertical velocity −1.0 ≤ w ≤ 1.5 m s−1 are used, so that effects of rain and other nonatmospheric echoes can be reduced. Composite diurnal variations of surface meridional wind data are also computed but without subtracting the daily mean. Figure 8 shows the resulted time series of 1000–2200 LT for June–November, while those for the rainy season are depicted in Fig. 9.

Fig. 8.

Composite time series of meridional winds observed with (a) BLR in the 0.5–0.8- (solid lines) and 1.5–2.5-km height ranges (dash–dotted lines), and (b) an anemometer at 10-m altitude, for the months of Jun–Nov. Vertical bars denote standard deviations, which are plotted at selected data points

Fig. 8.

Composite time series of meridional winds observed with (a) BLR in the 0.5–0.8- (solid lines) and 1.5–2.5-km height ranges (dash–dotted lines), and (b) an anemometer at 10-m altitude, for the months of Jun–Nov. Vertical bars denote standard deviations, which are plotted at selected data points

Fig. 9.

Composite time series of meridional winds observed with (upper panels) BLR and (lower panels) anemometer as in Fig. 8, except for rainy season months of Dec–Feb

Fig. 9.

Composite time series of meridional winds observed with (upper panels) BLR and (lower panels) anemometer as in Fig. 8, except for rainy season months of Dec–Feb

A gross feature that can be readily seen from Figs. 8a and 8b is the pronounced appearances of an SBC pattern in the dry season months of July–October. Closer examination reveals that wind maxima at the surface and upper layers occur at nearly the same time between 1500 and 1600 LT in July–August, while those of September–October are observed earlier between 1400 and 1500 LT. In contrast, the intensity of the sea-breeze flow in the 0.5–0.8-km layer appears to be strongest almost invariably between 1600 and 1800 LT in all of these months. It is also of interest to note that the return flow signal, that is, enhancement of southerlies in the 1.5–2.5-km height range, has the shortest lifetime in October. Outside July–October, sea-breeze-related flow patterns are in general quite diffuse and more difficult to assess. However, it can be seen from Fig. 9 that northerly enhancement in the lower layer is distinctly observed without a compensating reverse flow in the rainy season months of January and February. Furthermore, wind maxima at the surface and 0.5–0.8-km layers occur concurrently around midday.

It was mentioned earlier that the land–sea temperature difference is one of the critical parameters for sea-breeze development. As a consequence, annual variation of temperature over land and sea is an important factor that determines the seasonal variation in the strength of the sea-breeze signal. Figure 10 shows the average time series of mean daytime temperature and relative humidity from surface measurements at Serpong obtained during 1993–99. Mean temperature and relative humidity are calculated for each day from data at 0800–1400 LT. The curves plotted in Fig. 10 are the 30-point running mean of the resultant time series after averaging over available years of observation. In Fig. 10a, the annual variation of the monthly mean SST off the Jakarta coast averaged over 1986–94 is also presented. It can be said that the land–sea temperature difference can only be clearly seen in the months of July–October, while that of September is the most pronounced. A strong correlation between land–sea temperature difference and sea-breeze signal observed at Serpong (see Fig. 8) during the dry season is quite self-evident.

Fig. 10.

Composite annual variation of (a) temperature and (b) relative humidity observed in the daytime (averaged for 0800–1400 LT) at Serpong during 1993–99. Number of averaged days, i.e., available observations, is indicated with small square symbol. Solid, dotted, and dash–dotted lines represent mean, minimum, and maximum values, respectively. Dashed line in (a) denotes monthly mean SST off the Jakarta coast averaged over 1986–94

Fig. 10.

Composite annual variation of (a) temperature and (b) relative humidity observed in the daytime (averaged for 0800–1400 LT) at Serpong during 1993–99. Number of averaged days, i.e., available observations, is indicated with small square symbol. Solid, dotted, and dash–dotted lines represent mean, minimum, and maximum values, respectively. Dashed line in (a) denotes monthly mean SST off the Jakarta coast averaged over 1986–94

A diffuse land–sea temperature difference and subdued seaward return flow above the inflow current suggest that the daytime enhancement of low-level northerlies observed during the rainy season is not attributed to SBC as a thermally driven coastal meteorological phenomenon. It is likely that the diurnal wind pattern is related to deep cumulus cloud development over land (Houze et al. 1981). However, results of investigation into diurnal convective activities over the tropical western Pacific by Nitta and Sekine (1994) indicate that the maxima of deep cloud index over the northern part of West Java occur around 1800 LT, while the diurnal rainfall pattern recorded at Serpong (data not presented) also shows maximum precipitation between 1500 and 1800 LT. There appears to be a significant time lag between the onshore flow enhancement that starts before 1000 LT (see Fig. 9) and maximum cloud activity, which is difficult to clarify without more detailed investigation. Nevertheless, it should be clear that diurnal wind variations observed at Serpong during the rainy season exhibit distinct characteristics that are different from pronounced sea-breeze-induced flows detected in the months of July–October.

As previously described in Fig. 8, high time-resolution BLR and surface wind measurements give detailed characteristics of SBC over Jakarta that would be difficult to discern from conventional pilot balloon or radiosonde observations. A similar pattern of wind maxima in the surface and upper layers clearly indicates a response to the onset and decay of the SBC over the radar site during the dry season. Late afternoon enhancement of the sea-breeze flow in the 0.5–0.8-km layer is likely produced by the effects of mountains (e.g., Mahrer and Pielke 1977; Ookuchi et al. 1978; Lu and Turco 1994). Evening acceleration of SBF propagation, which may also induce such an enhancement, has actually been reported from various observations and numerical modeling (e.g., Simpson et al. 1977; Sha et al. 1991) but the influence of topography is obvious for SBC over Jakarta as discussed earlier. It is also of interest to note that the intensification in the sea-breeze current is not felt near the ground possibly because of intense turbulent mixing in the surface layer. It has been previously reported by Hadi et al. (2000) that, after SBC is established, the temperature structure in the lowest 300-m layer becomes superadiabatic, in contrast to the stabilized portion in the 0.5–0.8-km height range. An increase in the depth of sea-breeze current with time from about 0.8 km at 1500 LT to nearly 1.4 km at 1800 LT, which was evidenced by sounding wind profiles, also seems to account for the lag between sea-breeze intrusion and flow intensification above 0.5 km.

Another point of interest is the peculiar features of the sea-breeze signal in October. The temporal shift of the wind maxima (see Fig. 8) suggests that the sea breezes arrive earlier over the radar site during this month compared to July–August. It is possible that such a variability has been simply caused by changes in the direction of the prevailing background wind, since October may include a transition period from the dry to the rainy season. However, the diurnal wind pattern in the 1.5–2.5-km height range indicates that the lifetime of the return-flow signal is relatively shorter than its sea-breeze counterpart, which cannot be explained by the effect of background wind alone. A more plausible explanation for the obliteration of the upper-level countercurrent is venting of onshore flow into the free atmosphere over sloping topography or mountains. Such an orographic lifting is more effective under meteorological condition favorable for cloud development over land. It was also noted from the cases presented in section 4 that SBF tends to travel faster from the coast to the radar site during more cloudy days. Furthermore, a marked increase in maximum relative humidity from September to October (see Fig. 10b) is likely attributed to more enhanced cloudiness, considering that there is only a small difference between average land temperature in August and October. Hence, cloudiness is an important indicator to and one of the factors that can be associated with intraseasonal variability of SBC over Jakarta.

6. Relationship between cloudiness and sea-breeze profiles

In this section we investigate the characteristics of SBC in relation to meteorological conditions during the dry season, which is represented by the degree of cloudiness. Similar to the previous section, datasets obtained during 1993–99, excluding those of 1997 and 1998, are used. In order to perform the analysis, it is first necessary to express the degree of cloudiness observed during a given day in terms of an index that can be objectively determined.

a. Cloudiness index determined from incoming solar radiation

Instead of direct cloud observations, it is preferable to determine the cloudiness index using data of incoming solar radiation, which are available from long-term surface meteorological measurements collocated with BLR. A cloudiness index is defined here as the relative amount of incoming solar radiation observed during a given day from 0800 to 1400 LT. For each of the days in a specified month, examples of the required data are presented in Fig. 11, and the cloudiness index is evaluated as a ratio:

 
formula

where Robs(t) and Rm(t) denote observed diurnal and maximum incoming solar radiation. By putting Robs as the denominator in Eq. (1), the clearest day is thus indicated by CR = 1, which increases with higher degree of cloudiness.

Fig. 11.

Examples of incoming solar radiation data observed on a given day (solid line) with the maximum values for the specified month (dotted line), which are used to calculate cloudiness index CR

Fig. 11.

Examples of incoming solar radiation data observed on a given day (solid line) with the maximum values for the specified month (dotted line), which are used to calculate cloudiness index CR

From discussions in the previous section, we have presumed that SBC occurs very frequently over Jakarta during the dry season. Cloudiness in this case is regarded as a qualitative parameter that, in a sense, represents weather pattern. As a consequence, CR should be interpreted in the context of a meteorological condition rather than a specific physical quantity, although a close relation to temperature is implied by the utilization of insolation data. This is different from the work of Yoshikado (1981), for example, who used cumulative incoming solar radiation as a direct substitution for temperature to study the probability of sea-breeze occurrence at three locations in Japan.

b. Sea-breeze intrusion time

In relation to characteristic features of SBC in October, the fact that sea-breeze intrusion time over the radar site varies under different meteorological conditions has been previously brought forward. Figure 12 shows the correlation between the cloudiness index CR, defined in the above, and the times for intrusion and decay of SBC. These temporal parameters are estimated from surface meteorological measurements of meridional wind and relative humidity (see Fig. 5) by inspection. Although the method is subjective, results for sea-breeze intrusion time should be representative because only data with clear-cut changes in both wind and relative humidity are selected. With an intuitive approach, samples are divided into three groups for CR < 1.15, 1.15 < CR < 1.3, and CR > 1.3, which are marked by different symbols. It can be seen that sea-breeze intrusion tends to occur earlier with increasing cloudiness, which confirms previously mentioned satellite observations. We can also see this tendency by calculating Spearman's rank correlation statistics (see Table 1), which give a negative correlation coefficient ρ of −0.45, with more than 99% confidence, between CR and sea-breeze intrusion time. On the other hand, there is practically no correlation between CR and sea-breeze decay time as indicated by the small value of ρ, having less than 95% confidence. It should be noted, however, that the decay time is estimated with less accuracy because it is based on wind variation alone.

Fig. 12.

Correlation between cloudiness index with sea-breeze intrusion (asterisk, diamond, triangle) and decay (plus, cross, square) times determined from surface meteorological measurements at Serpong

Fig. 12.

Correlation between cloudiness index with sea-breeze intrusion (asterisk, diamond, triangle) and decay (plus, cross, square) times determined from surface meteorological measurements at Serpong

Table 1. 

Spearman's rank correlation coefficients, ρ, and two-sided tests, ρsig, with respect to null hypothesis calculated from N number of data in Fig. 12; tint and tdec denote intrusion and decay times, respectively

Spearman's rank correlation coefficients, ρ, and two-sided tests, ρsig, with respect to null hypothesis calculated from N number of data in Fig. 12; tint and tdec denote intrusion and decay times, respectively
Spearman's rank correlation coefficients, ρ, and two-sided tests, ρsig, with respect to null hypothesis calculated from N number of data in Fig. 12; tint and tdec denote intrusion and decay times, respectively

It is important to note that all samples used in this analysis are selected, by incorporating BLR observations of vertical wind velocity w, to exclude days with detected rainfall events. A wind profile is considered to be contaminated by precipitation if w < −1.0 m s−1 in more than 80% of the observed height ranges. A sample is rejected if any of the 15-min-averaged profiles observed in 24 h meets this condition.

c. Composites of wind and echo-power profiles

It has been shown from Fig. 8 that the sea-breeze flow intensity observed during the dry season is maximum between 1700 and 1800 LT. For a well-defined SBC, meridional wind profile averaged over the period of 1530–1730 LT (hereafter referred to as the sea-breeze profile) should be characterized by strong northerlies below 1.0 km and southerlies in 1.5–2.5-km heights. Figure 13a shows the correlation between CR and the strength of sea-breeze and return flows (averaged in 0.5–1.0- and 1.5–2.5-km height ranges, respectively). Despite much scatter in the data, the statistics presented in Table 2 show that correlation coefficients can be determined with more than 99% confidence for both the lower- and upper-level winds. Additionally, Fig. 13b depicts the composite profiles based on three ranges of CR values as previously discussed (also see Fig. 12) These profiles show that SBC is most well defined during clear days (CR < 1.15), which is consistent with the observed correlation in Fig. 13a. It is worth noting that, although the intensity of SBC decreases with a higher degree of cloudiness, there does not seem to be significant variation in the depth of sea-breeze flow, while the entire circulation appears to be confined below 3-km height.

Fig. 13.

(a) Correlation between cloudiness index and meridional wind velocities averaged in time over the period of 1530–1730 LT, and in height over 0.5–1.0-km (circle) and 1.5–2.5-km (asterisk) ranges. (b) Composite wind profiles of the same period as in (a) but grouped according to three different ranges of CR. Data are obtained from BLR measurements except those plotted near the bottom, which are observed with an anemometer. Standard deviations are indicated at selected points as horizontal bars, while N denotes number of samples

Fig. 13.

(a) Correlation between cloudiness index and meridional wind velocities averaged in time over the period of 1530–1730 LT, and in height over 0.5–1.0-km (circle) and 1.5–2.5-km (asterisk) ranges. (b) Composite wind profiles of the same period as in (a) but grouped according to three different ranges of CR. Data are obtained from BLR measurements except those plotted near the bottom, which are observed with an anemometer. Standard deviations are indicated at selected points as horizontal bars, while N denotes number of samples

Table 2. 

Similar to Table 1 except for data in Fig. 13 = a

Similar to Table 1 except for data in Fig. 13 = a
Similar to Table 1 except for data in Fig. 13 = a

Diurnal evolution of echo power profiles observed with BLR can provide additional information on the structure of the PBL, including the presence of clouds. The degree of cloudiness signified by CR can be confirmed by composite diurnal variations of reflectivity index depicted in Figs. 14a–c. For the days with CR < 1.15, marked development of a mixing layer is interrupted by a sea-breeze intrusion over the radar site and a weak echo region between 0.8- and 2.0-km heights has been observed around 1400–1800 LT. The weak echo region corresponds to the sea-breeze-induced shear layer and can be associated with the occurrence of Kelvin–Helmholtz (KH) instability, which has been discussed previously by Hadi et al. (2000). Although a similar pattern is also observed during more cloudy days, there is an indication that the decrease in echo power around 1.0-km height occurs earlier but with a reduced vertical dimension of the weak echo core. This is consistent with early sea-breeze intrusion for days with CR > 1.15. It is worth noting that echoes above 3.0 km are most likely related to the presence of clouds. Moreover, enhancement of the reflectivity index around 1200 LT (see Fig. 14c) corresponds to the highest range of CR values, which indicates the presence of deeper cumuli. Furthermore, Fig. 15 shows that the return flow pattern vanishes in the composite sea-breeze profile observed during the days with rainfall events detected between 1200 and 1500 LT.

Fig. 14.

Composite diurnal variation of reflectivity index observed with BLR, grouped according to three different ranges of CR corresponding to wind profiles of Fig. 13b 

Fig. 14.

Composite diurnal variation of reflectivity index observed with BLR, grouped according to three different ranges of CR corresponding to wind profiles of Fig. 13b 

Fig. 15.

Composite sea-breeze profile as in Fig. 13b except for days with rainfall events detected between 1200 and 1500 LT

Fig. 15.

Composite sea-breeze profile as in Fig. 13b except for days with rainfall events detected between 1200 and 1500 LT

Assuming a fixed time for sea-breeze onset near the coastline, early intrusion over the radar site means that SBF tends to travel faster inland during cloudy days. It is well known that offshore background winds may hinder SBF propagation but strengthen the circulation (e.g., Aritt 1993), and vice versa. Unless enhanced cloud development over land is correlated with onshore large-scale winds, however, there should be other factors that are responsible for the observed variations in propagation speed of SBF and sea-breeze intensity. Existence of large-scale subsidence, for example, will create a capping inversion layer whose relationship with a low degree of cloudiness is easier to understand. In this case, air motions in the PBL will be almost completely separated from the free atmosphere and a well-defined SBC may develop. Well-developed circulation will in turn create a strong shear layer, which induces intense KH instability and presumably larger KH billows. From the results of a numerical experiment, Sha et al. (1991) have shown that KH billows thus generated will produce frictional drag that can retard inland progression of SBF.

Although the phenomenon has been clearly identified, the mechanism that causes earlier sea-breeze intrusion under more cloudy conditions is rather complicated because the sea breeze itself could induce cloudiness in the humid tropical atmosphere by localized lifting along its leading edge. Dailey and Fovel (1999) pointed out that sea-breeze flows may be locally influenced by enhanced cloud development due to the interactions between preexisting convective rolls and an inland propagating SBF, in which case the direction of the background wind also plays an important role. In the present work, however, combined BLR and surface meteorological measurements alone do not provide sufficient information to separate out the effects of sea-breeze-induced cloudiness from those that existed prior to sea-breeze onset.

7. Summary

Three aspects of SBC over the tropical site of Jakarta, Indonesia, have been investigated. With regard to the horizontal extent, satellite images obtained during the dry season clearly show that SBF develops well along the northern coastal plain of West Java and propagates inland until its structure is modified over more complex topography at a distance of about 60–80 km from the coastline. The fact that cumulus convection may be suppressed over the entire region, or enhanced along sloping terrain, suggests that factors related to cloudiness are important for evolution of SBC later in the day. There is clearly a potential for the sea breezes to develop over other parts of Java where the topography is moderate, so that SBFs from northern and southern coasts may converge. Available data are not, however, sufficient to do further investigation. Offshore extent of the observed circulation has not been estimated in this study.

Long-term data from BLR and surface meteorological measurements have been analyzed to investigate the seasonal variability of SBC over Jakarta. It is found that the sea-breeze signal is most well defined during the dry season months of July–October, which appears to be correlated with the relatively large land–sea temperature difference. Other possible factors are the absence of large-scale convective activity and weak meridional component of synoptic-scale winds. A distinct daytime enhancement of low-level northerlies is observed during the rainy season months of January and February without being accompanied by a return flow pattern. This type of onshore flow is likely related to cloud convection over land but not directly driven by differential solar heating of land and ocean as in the case of sea breezes observed during the dry season.

Characteristic features of the sea-breeze signal observed in July–October provide additional information on the detailed vertical and temporal structure of SBC. It is found that wind maxima in the near-surface sea-breeze current and return flow aloft occur almost at the same time. These wind maxima can be associated with a direct response to sea-breeze intrusion over the BLR site, which mainly occurs between 1300 and 1500 LT. However, intensity of onshore flow in the 0.5–0.8-km height range becomes maximum late in the afternoon between 1700 and 1800 LT. Possible causes for such a vertical variation in the sea-breeze flow intensity are 1) an increase in the depth of the sea-breeze current with time, 2) a temperature profile that induces different turbulent layers, and 3) an interaction between SBF and flows over complex topography later in the day.

Near-surface and upper-level wind maxima in October exhibit a temporal shift from those of July and August, which indicates an intraseasonal variation in sea-breeze intrusion time. Satellite observations suggest a tendency for SBF to arrive earlier over the BLR site under meteorological conditions favorable for cloud development. Correlation between cloudiness and sea-breeze intrusion time is further investigated by utilizing data on incoming solar radiation. It can be confirmed that sea-breeze intrusion occurs earlier during more cloudy days, which, assuming a fixed time for sea-breeze onset near the coastline, implies that SBF propagates inland with a greater speed. Additionally, it is found that the sea-breeze profile observed in the period of 1530–1730 LT becomes less well defined, that is, the intensity of SBC is weaker, with increasing cloudiness. The possibility that retardation of SBF progression under clear weather is caused by intense KH instability in the sea-breeze-induced shear layer (Sha et al. 1991) is considered. However, explanations for an earlier sea-breeze intrusion under cloudy conditions have yet to be more thouroughly investigated.

Acknowledgments

Boundary Layer Radar and Serpong Observatory have been constructed, operated, and maintained in conjunction with the National Institute for Aeronautics and Space (LAPAN) and Agency for Assessment and Application of Technology (BPPT) of Indonesia. The first author is supported by the Overseas Economic Cooperation Fund fellowship program, which is jointly managed by the Japan–Indonesia Science and Technology Forum and ITB.

REFERENCES

REFERENCES
Aritt
,
R. W.
,
1989
:
Numerical modeling of the offshore extent of the sea breezes.
Quart. J. Roy. Meteor. Soc.
,
115
,
547
570
.
Aritt
,
R. W.
,
.
1993
:
Effects of the large-scale flow on the characteristic features of the sea breeze.
J. Appl. Meteor.
,
32
,
116
125
.
Atkins
,
N. T.
,
R. M.
Wakimoto
, and
T. M.
Weckwerth
,
1995
:
Observations of the sea-breeze front during CaPE. Part II: Dual-Doppler and aircraft analysis.
Mon. Wea. Rev.
,
123
,
944
968
.
Banta
,
R. M.
,
R. D.
Olivier
, and
D. H.
Levinson
,
1993
:
Evolution of the Monterey Bay sea-breeze layer as observed by pulsed Doppler lidar.
J. Atmos. Sci.
,
50
,
3959
3982
.
Braak
,
C.
,
1929
:
The Climate of The Netherland Indies.
Vols. I and II. Verhandelingen No. 8, Koninklijk Magnetisch en Meteorologisch Observatiorium te Batavia, 1602 pp
.
Dailey
,
P. S.
, and
R. G.
Fovell
,
1999
:
Numerical simulation of the interaction between the sea-breeze front and horizontal convective rolls. Part I: Offshore ambient flow.
Mon. Wea. Rev.
,
127
,
859
878
.
Dalu
,
G. A.
, and
R. A.
Pielke
,
1989
:
An analytical study of sea breeze.
J. Atmos. Sci.
,
46
,
1815
1825
.
Estoque
,
M. A.
,
1962
:
The sea breeze as a function of the prevailing situation.
J. Atmos. Sci.
,
19
,
244
250
.
Garrat
,
J. R.
,
1985
:
The inland boundary layer at low latitudes. I. The nocturnal jet.
Bound.-Layer Meteor.
,
32
,
307
327
.
Hadi
,
T. W.
,
T.
Tsuda
,
H.
Hashiguchi
, and
S.
Fukao
,
2000
:
Tropical sea-breeze circulation and related atmospheric phenomena observed with L-band boundary layer radar in Indonesia.
J. Meteor. Soc. Japan
,
78
,
123
140
.
Hashiguchi
,
H.
,
S.
Fukao
,
T.
Tsuda
,
M. D.
Yamanaka
,
S. W. B.
Harijono
, and
H.
Wiryosumarto
,
1996
:
An overview of the planetary boundary layer observation over equatorial Indonesia with an L-band clear air Doppler radar.
Beitr. Phys. Atmos.
,
69
,
13
25
.
Houze
,
R. A.
,
S. G.
Geotis
,
F. D.
Marks
, and
A. K.
West
,
1981
:
Winter monsoon convection in the vicinity of North Borneo. Part I: Structure and time variation of the clouds and precipitation.
Mon. Wea. Rev.
,
109
,
1595
1614
.
Johnson
,
R. H.
,
1992
:
Heat and moisture sources and sinks of Asian monsoon precipitating systems.
J. Meteor. Soc. Japan
,
70
,
223
371
.
Kikuchi
,
T.
, and
M.
Kitsuregawa
,
2001
:
GMS-5 Meteorological Satellite Image Database and Integrated Visualization System.
(in Japanese with English abstract). IPSJ Rep. SIG8 (TOD10)
.
Lu
,
R.
, and
R. P.
Turco
,
1994
:
Air pollutant transport in a coastal environment. Part I: Two-dimensional simulations of sea-breeze and mountain effects.
J. Atmos. Sci.
,
51
,
2285
2308
.
Mahrer
,
Y.
, and
R. A.
Pielke
,
1977
:
The effects of topography on sea and land breezes in a two-dimensional numerical model.
Mon. Wea. Rev.
,
105
,
1151
1162
.
Niino
,
H.
,
1987
:
The linear theory of land and sea-breeze circulation.
J. Meteor. Soc. Japan
,
65
,
901
921
.
Nitta
,
T.
, and
S.
Sekine
,
1994
:
Diurnal variation of convective activity over the tropical western Pacific.
J. Meteor. Soc. Japan
,
72
,
627
641
.
Ookuchi
,
Y.
,
M.
Uryu
, and
R.
Sawada
,
1978
:
A numerical study on the effects of mountain on the land and sea breezes.
J. Meteor. Soc. Japan
,
56
,
368
385
.
Rotunno
,
R.
,
1983
:
On the linear theory of the land and sea breeze.
J. Atmos. Sci.
,
40
,
1999
2009
.
Rotunno
,
R.
, and
Coauthors
.
1992
:
Coastal Meteorology, A Review of the State of the Science.
National Academy Press, 99 pp
.
Sha
,
W.
,
T.
Kawamura
, and
H.
Ueda
,
1991
:
A numerical study on sea/land breeze as a gravity current: Kelvin–Helmholtz billows and inland penetration of sea-breeze front.
J. Atmos. Sci.
,
48
,
1649
1665
.
Simpson
,
J. E.
,
1994
:
Sea Breeze and Local Wind.
Cambridge University Press, 234 pp
.
Simpson
,
J. E.
,
D. A.
Mansfield
, and
J. R.
Milford
,
1977
:
Inland penetration of sea-breeze fronts.
Quart. J. Roy. Meteor. Soc.
,
103
,
47
76
.
Stull
,
R. B.
,
1988
:
An Introduction to Boundary-Layer Meteorology.
Kluwer Academic, 666 pp
.
van Bemmeln
,
W.
,
1913
:
Die Erforschung des tropischen Luftozeans.
Niederländisch-Ost-Indien (The Study of the Tropical Air–Ocean in Netherlands East India), Springer-Verlag, 55 pp
.
Wakimoto
,
R. M.
, and
N. T.
Atkins
,
1994
:
Observations the sea-breeze front during CaPE. Part I: Single-Doppler, satellite, and cloud photogrammetry analysis.
Mon. Wea. Rev.
,
122
,
1092
1114
.
Walsh
,
J. E.
,
1974
:
Sea breeze theory and application.
J. Atmos. Sci.
,
31
,
2012
2026
.
Yan
,
H.
, and
R. A.
Anthes
,
1987
:
Effect of latitude on the sea-breeze.
Mon. Wea. Rev.
,
115
,
936
956
.
Yoshikado
,
H.
,
1981
:
Statistical analysis of the sea breeze pattern in relation to general weather condition.
J. Meteor. Soc. Japan
,
59
,
98
107
.

Footnotes

*

Permanent affiliation: Department of Geophysics and Meteorology, Institut Teknologi Bandung, Bandung, Indonesia

Corresponding author address: Dr. Tri W. Hadi, Radio Science Center for Space and Atmosphere, Kyoto University, Gokasho, Uji, Kyoto 611-0011, Japan. Email: tri@kurasc.kyoto-u.ac.jp