1. Introduction
Temperature inversion layers (hereafter inversion layers) frequently develop at a height of ∼4 km over the Indochina Peninsula in the month prior to the onset of the rainy season, but the inversion layers have not been systematically studied. These inversion layers are considerably higher than those of the trade inversions that typically occur over the tropical ocean at ∼2 km height (e.g., Gutnick 1958). The Indochina Peninsula has the earliest onset of the rainy season of any region in Asia of comparable latitude (He et al. 1987; Hsu et al. 1999). Matsumoto (1995) documented that significant rainfall occurs over inland parts of the Indochina Peninsula prior to the onset of the rainy season. There is a possibility that the inversion layers are related to the rainy season onset through thermodynamical processes.
The final goal of our study is to clarify the processes of interaction between inversion layers and seasonal changes in convective activity during the dry season to rainy season transition in the Indochina Peninsula region. As the first step, we first provide a climatological description of the seasonal changes in inversion layers.
An inversion layer is an extremely stable layer. Stable layers including inversion layers are known to affect convective activity (e.g., Chen and Feng 1995; Johnson et al. 1996; Chen and Feng 2001). Three kinds of relationships can exist between inversion layers and convective activity: 1) inversions influencing convection, 2) convection influencing inversions, and 3) equal interaction between inversions and convection. In cases 1 and 3 above, inversions may play a key role in the transition from the dry to rainy seasons in the Indochina region. For case 2, an inversion is simply a by-product of convective activity. Although the physical relationship between inversion and convective activity may prove important in our understanding of the onset of the rainy season, no previous studies have described the climatological features of the seasonal transition of inversion layers.
Liu (1990) investigated the horizontal distribution of inversion layer frequency within the Northern Hemisphere and documented the common occurrence of inversions at a height of ∼2 km over Southeast Asia during January; the regions that have inversion layers during January also have the earliest starts to the rainy season. The descriptions of Liu (1990) are based on data from January and July only; seasonal variation in inversion layers at the onset of the rainy season has not yet been studied. Continuous inversion layer data throughout the year would be helpful in understanding the relationship between inversion layers and the onset of the rainy season. Inversion layer climatology has the potential to provide us with a new perspective on monsoon climatology in Asia and other monsoon regions such as North America, the Amazon, and West Africa.
2. Data processing
We analyzed operational rawinsonde data at mandatory and significant levels from 14 stations in the Indochina Peninsula region (southern China, Vietnam, Thailand, the Malay Peninsula, and Singapore) collected over 29 years from 1973 to 2001. The stations used in this study are listed in Table 1 and their locations are shown in Fig. 1.
In this study we used only morning data obtained at 0000 UTC (all times stated as hours only), which corresponds to 0700 or 0800 local time (LT) at the 14 stations. As 1200 UTC observations were commonly not recorded at most of the 14 stations, a statistical analysis of a varying mixture of 0000 and 1200 UTC data may result in misleading description of climatological features. We therefore present only the 0000 UTC data, except for a brief comparison between the 0000 and the 1200 UTC data in section 3.
A typical sounding profile is shown in Fig. 2. A remarkable temperature inversion layer is seen at a height of ∼4 km. We focus on this type of lower-tropospheric inversion layer in this study. Although surface inversion layers are also frequent at 0000 UTC below a height of ∼1 km, as seen in Fig. 2, they are not of interest for this study.
Erroneous temperature data were excluded from analysis according to the following procedure. First, temperature data that differed from the mean by more than two standard deviations were identified as erroneous. The standard deviations and means were calculated at every pentad1 and at every 600-m height range. Second, pre-1990 temperature data for Hanoi, Da Nang, and Ho Chi Minh, Vietnam, occasionally have exactly the same values at several successive significant levels. We considered such data to be erroneous and did not use them in this study.
After excluding erroneous data, the data at significant levels were linearly interpolated into constant height intervals of 100 m in order to simplify statistical analysis. In Fig. 2, linearly interpolated data are shown by asterisks. It is found that the structure of the inversion at a height of ∼4 km is well reproduced by the linear interpolation procedure.
In sections 3 and 4 we investigate seasonal variations in the height and appearance frequencies of the stable and unstable layers. A climatological value of the frequency distribution was evaluated at temporal intervals of every half-month and at 100-m height intervals.
3. Seasonal variation at Ubon Ratchathani
In this section we describe seasonal variations in the height and appearance frequencies of the stable and unstable layers at Ubon Ratchathani on the central Indochina Peninsula (see Fig. 1).
Figure 4 shows a time–height cross section of the frequency of stable layers at Ubon Ratchathani. Stable layers commonly occur during the dry season (November–April) but are rare during the rainy season (May–October). This feature is consistent with the results shown by Liu (1990). The dry and rainy seasons roughly correspond to the boreal winter and summer, respectively (Fig. 5a). During the dry season, stable layers develop at two distinct heights: near the land surface and at a height of 2–5 km. The upper stable layers are the focus of this study, while the lower stable layers represent surface inversion layers. Figure 6 shows a time–height cross section of the frequency of evening-time stable layers at Ubon Ratchathani. A comparison of Figs. 4 and 6 demonstrates that surface inversions appear dominantly during the morning and only rarely during the evening. In contrast, the upper-level stable layers develop during both morning and evening, at a similar height regardless of the time of day, although they occur slightly more frequently during the morning than in the evening. These observations concerning the slight difference in occurrence of upper-level stable layers may provide insight into the generation mechanism of the inversion layers. However, we will not discuss this data further in this paper because diurnal variations are not sufficiently documented by the twice-daily radiosonde data.
The characteristics of the stable layers at 2–5-km height changes over the course of the dry season. The stable layers commonly occur at ∼2 km height during late November, which is the beginning of the dry season. The frequency of stable layers gradually increases from November to December and reaches a maximum during early January. The height at which stable layers develop increases steadily from January to April, reaching a maximum of 4–5 km in April immediately prior to the onset of the rainy season. This variation in the height of the stable layers is much greater than the variation in height of trade inversion layers over the ocean (e.g., Gutnick 1958; Schubert et al. 1995, and references therein). The occurrence of stable layers gradually decreases from January to April.
Figure 7 shows a time–height cross section of the appearance of unstable layers. Unstable layers frequently occur below the heights at which stable layers develop during the dry season (see Figs. 4 and 7). From December to April, unstable layers generally develop at ∼2 km height, with the maximum observed frequency of unstable layer development occurring during January at a height of ∼1.5 km. During February and March, the region with an unstable layer frequency of >5% extends up to 2.5 km in height (Fig. 7); unstable layers occur only rarely during the rainy season.
The seasonal variations in stability for the 29 years of data summarized above shows gradual changes in terms of time and height; however, there is significant interannual and intraseasonal variability in the appearance of stable layers. For example, Fig. 8a shows a time–height cross section of stable layer appearance for the year from July 1999 to June 2000. Stable layers appear intermittently: during early November they occur dominantly at a height of 3–4 km, during mid-December below 2 km, during late January at a height of ∼3 km, and during March at a height of ∼4 km. Similarly intermittent occurrences with an intraseasonal temporal scale were also observed in most of the other years for which data were recorded, although the temporal scales vary with each year. In only a few years is the intermittent nature of stable layer occurrence poorly defined, such as the winter from 1996 to 1997 (Fig. 8b). Despite the observed intermittent nature of occurrence, stable layers tend to develop at heights where the high-frequency appearances are found in the climatological features as shown in Fig. 4. In summary, the climatological feature of seasonal variation in stability is a general feature that consists of the intermittent appearance of stable and unstable layers and that stable layers develop at preferred seasonal heights.
4. Regional differences in seasonal stability variation over Southeast Asia
In the previous section, we described the frequency of stable and unstable layers in the central region of the Indochina Peninsula. In this section, we describe results from other stations over the Indochina Peninsula region. Characteristics of seasonal variations are classified into three types (I, II, and III), as shown below.
The type-I variations are characterized by an increase in the height of stable layers from midwinter through the end of winter and by the appearance of unstable layers below the stable layers. Stations that record type-I variations are located in the inland regions of the Indochina Peninsula and southern China. Ubon Ratchathani is representative of type-I stations, while the other type-I stations are Kunming, China, and Chiang Mai, Thailand (see Fig. 1). Seasonal variations recorded at type-I stations, except for Ubon Ratchathani, are shown in Fig. 9. For these stations, stable and unstable layers mainly developed during boreal winter; winter roughly corresponds to the dry season (see Fig. 5a). In each panel of Fig. 9, the same characteristics are observed as those described in the previous section for Ubon Ratchathani.
There are some differences among the type-I stations. The straight lines drawn upon Figs. 4, 9a and 9c show the estimated increase in stable layer heights as determined by applying a linear least squares fit to the most frequently occurring stable layer heights at half-month bins from late November until late April. The increase rates were estimated to be 0.4 km month−1 at Ubon Ratchathani, 0.3 km month−1 at Kunming, and 0.4 km month−1 at Chiang Mai. The increase rates vary by ∼0.1 km month−1 between the three stations, although the seasonal increase of stable layer heights is a common feature of all type-I stations. The heights at which stable layers most frequently appear are similar for all three stations, with the difference being <1 km; however, the difference in elevation at the Kunming and Ubon Ratchathani stations is close to 2 km (see Table 1). This indicates that the height of the stable layer does not depend on the elevation of the underlying topography. Differences between the three type-I stations are also evident in terms of the appearance of unstable layers: unstable layers occur only rarely at Kunming compared with the other two stations.
Type-II variations are similar to type I, with the additional appearance of stable layers in the height range of 0.5–2 km above the land surface. Type-II stations are located in the coastal regions of the Indochina Peninsula, and include the stations at Nanning, China; Hanoi; Hong Kong; Bangkok, Thailand; and Da Nang (see Fig. 1). The nature of seasonal variations at these stations is shown in Fig. 10. Stable layers begin to appear frequently from October to November at ∼2 km height. The behavior of stable layers in the height range from 2 to 5 km is similar to that of type-I layers. The height of the upper stable layers increases, and frequency decreases, from December to May and reaches over 4 km in height, similar to type-I layers. The lower stable layers frequently occur at 0.5–2-km height until April or May and occur only rarely during June. In this region, the season of frequent development of stable layers corresponds to the dry season (Fig. 5b). Unstable layers commonly occur at a height of 2–3 km, between the heights of the upper and lower stable layers. The unstable layers first appear in December and reach a maximum height of ∼3 km in March before decreasing in frequency, with only rare occurrences during April.
A characteristic feature of type-III variations is that the stable layers generally occur at nearly constant altitude through the boreal winter. Type-III variations are recorded in the southern part of Southeast Asia, including the southern edge of the Indochina Peninsula and the Malay Peninsula, and include the stations at Ho Chi Minh on the Indochina Peninsula, and Songkhla, Thailand; Kota Bharu, Malaysia; Kuantan, Malaysia; Penang, Malaysia; and Singapore on the Malay Peninsula (see Fig. 1). Seasonal variations recorded by these stations are shown in Fig. 11. Stable layers frequently develop at ∼2 km height during boreal winter and maintain a constant height over time. Here, we should also note the relationship between seasonal variation in inversion layers and that of precipitation. At type-I, -II, and -III stations, the inversion layers frequently appear during the boreal winter; however, at type-III stations the boreal winter does not necessarily correspond to the dry season (see Figs. 5c and 5d).
There are some differences among the type-III stations. Additional stable layers appear at 1–2 km above the surface during February and March over Songkhla and Kota Bharu, located on the east coast of the Malay Peninsula (Figs. 11c and 11e, respectively). This feature is not evident in data from the evening observations (figures not shown). The frequency of both stable and unstable layers recorded at type-III stations generally increases with increasing latitude. At Singapore (Fig. 11k), stable layers at ∼2 km height occur less frequently than at other stations, while the frequency of layers at heights >6.5 km is high throughout the year. Unstable layers occur only seldomly at the three southernmost stations, Penang, Kuantan, and Singapore (Figs. 11h, 11j and 11l, respectively).
5. Discussion of inversion generation mechanisms
In sections 3 and 4, we described climatological features of seasonal variation in stable layers. For type-I and type-II stations, the peak height of stable layers increases from ∼2 km during December to ∼5 km during April. Unstable layers commonly develop beneath the stable layers. Within individual years, the stable layers have an intermittent occurrence at intraseasonal time scales. These observations indicate that stable layers are possibly related to the five types of well-known inversions: 1) capping inversions, 2) advective inversions, 3) 0°C inversions, 4) radiatively generated inversions, and 5) trade wind inversions. We now discuss the classification of the observed stable layers in terms of these five types of inversions.
a. Capping inversions
Type-I inversions that occur from midwinter through March may be capping inversions. We showed earlier that unstable layers frequently develop at a height of ∼2 km at Ubon Ratchathani (Fig. 7) and that stable layers commonly develop above unstable layers (Fig. 4). This type of vertical stability structure is consistent with the well-known occurrence of a mixed boundary layer accompanied by a capping inversion layer.
Seasonal variation in the stable layer height is also consistent with the nature of capping inversions. We showed that stable layers at Ubon Ratchathani increase their peak height from ∼2 km during January to 4–5 km during March. It is likely that the depth of mixed boundary layers increases with the seasonal increase in solar heating of the ground surface after midwinter. Capping inversions may therefore progressively increase in height from midwinter until the end of dry season; this trend is consistent with the seasonal evolution of stable layer heights documented in this study. Although the 4–5-km inversion heights documented for type-I and type-II stations during March and April are higher than the inversion top height of known trade wind boundary layers, such a deep boundary layer is sometimes observed in tropical inland regions (e.g., Hashiguchi et al. 1996).
b. Advective inversions
Some stable layers that occur during boreal winter may be considered advective inversions. For most of the years analyzed in the present study, stable layers occur intermittently at an intraseasonal time scale (Fig. 8). Intermittent cold air intrusions into the lowest layer over the South China Sea to the east of the Indochina Peninsula have been observed in winter (e.g., Murakami 1979). At the top of a cold intrusion, a stable layer can be generated. The time scales of the cold air intrusion reported by Murakami (1979) are similar to those of the intermittent appearance of stable layers observed in our study. It is therefore possible that some of the inversions documented in this study are caused by cold air advection.
c. 0°C inversions
It is possible that some of the type-I and type-II stable layers near the 0°C level (at ∼5 km height) that occur later than March are 0°C inversions, as Johnson et al. (1996) documented that 0°C inversion layers are generated near the melting region in tropical convective systems. The fact that rainfall occurs over the Indochina Peninsula after March and prior to the onset of the rainy season (Matsumoto 1997) strengthens the possibility of the occurrence of 0°C inversions. As the 0°C levels in this region exist at ∼4–5 km height, without significant seasonal variation, the midwinter stable layers that develop at ∼2–3 km height cannot be considered 0°C inversions.
d. Radiatively generated inversions
The lower-level stable layers that develop in the northwestern areas of the South China Sea (Fig. 10) at 1–2-km height from February to April, which distinguish type-II stations from type I, may be considered radiatively generated inversions. Klein and Hartman (1993) showed that, during the boreal winter, low-level clouds commonly occur over the regions of type-II stations. These geographical and seasonal coincidences indicate that the lower stable layers recorded at the type-II stations have a close relationship with low-level clouds. This interpretation is physically reasonable because the top of a cloud is a likely location for the development of a stable layer resulting from cooling by infrared radiation.
e. Trade wind inversions
Trade inversion layers are historically well documented at a height of 1–2 km in the Hadley circulation tropical subsidence zone over oceanic areas. We found that stable layers commonly developed during the boreal winter at a height of 1–2 km at type-III stations, which are surrounded by ocean. It is therefore possible that the stable layers recorded at the type-III stations are trade inversions.
6. Summary
In this study we used rawinsonde data for the 29 years from 1973 to 2001 to describe seasonal variations in stable (Δθ/Δz > 10 K km−1) and unstable (Δθ/Δz < 1 K km−1) layers in the lower troposphere over the Indochina Peninsula region.
As a general feature, the stable and unstable layers commonly appear during boreal winter. Analysis of seasonal height variations during the boreal winter helped to identify three types of seasonal variations. Type-I variations are characterized by stable layers with a single frequency maximum at ∼2 km height during midwinter over the inland Indochina Peninsula and southern China. As spring approaches, the frequency of stable layers decreases while their elevation above the surface increases by 2–3 km. It is an interesting fact that the height of stable layers does not vary with station elevation. Type-II variations are similar to those of type I, with an additional stable layer at 0.5–2 km in elevation near the coast of the Indochina Peninsula. Type-III variations are identified on the Malay Peninsula and are characterized by poorly defined seasonal changes in the vertical distribution of stable layers throughout the boreal winter.
The frequent occurrence of temperature inversions during boreal winter has previously only been described from incomplete seasonal data (see section 1). In the present study, we provided a detailed analysis of seasonal stability variations and presented many new observations as summarized above. We proposed that the five types of known inversions may explain some aspects of the seasonal, vertical, and geographical variations in stable layers. The next step is to clarify the physical mechanisms of stability variations, as this will lead to a better understanding of the seasonal transition of convective activity from dry to rainy seasons over the Indochina Peninsula and other monsoon regions.
Acknowledgments
The authors thank Drs. T. Yasunari, J. Matsumoto, T. Satomura, K. Kita, and R. Kawamura for their useful comments.
REFERENCES
Chen, Y-L., and J. Feng, 1995: The influences of inversion height on precipitation and airflow over the island of Hawaii. Mon. Wea. Rev., 123 , 1660–1676.
Chen, Y-L., and J. Feng, 2001: Numerical simulations of airflow and cloud distributions over the windward side of the island of Hawaii. Part I: The effects of trade wind inversion. Mon. Wea. Rev., 129 , 1117–1134.
Gutnick, M., 1958: Climatology of the trade-wind inversion in the Caribbean. Bull. Amer. Meteor. Soc., 39 , 410–420.
Hashiguchi, H., S. Fukao, T. Tsuda, M. D. Yamanaka, S. W. B. Harijono, and H. Wiryosumarto, 1996: An overview of the planetary boundary layer observations over equatorial Indonesia with an L-band clear-air Doppler radar. Beitr. Phys. Atmos., 69 , 13–25.
He, H., J. W. McGinnis, Z. Song, and M. Yanai, 1987: Onset of the Asian monsoon in 1979 and the effect of the Tibetan Plateau. Mon. Wea. Rev., 115 , 1966–1995.
Hsu, H-H., C-T. Terng, and C-T. Chen, 1999: Evolution of a large-scale circulation and heating during the first transition of Asian summer monsoon. J. Climate, 12 , 793–810.
Johnson, R. H., P. E. Ciesielski, and K. A. Hart, 1996: Tropical inversions near the 0°C level. J. Atmos. Sci., 53 , 1838–1855.
Klein, S. A., and D. L. Hartman, 1993: The seasonal cycle of low stratiform clouds. J. Climate, 6 , 1587–1606.
Liu, F. H., 1990: Climatological study of temperature inversion layers in northern hemispheric troposphere. Climatological Note 39, Institute of Geoscience, University of Tsukuba, 80 pp.
Matsumoto, J., 1995: Rainfall climatology over Asian monsoon region. Toward Global Planning of Sustainable Use of the Earth, S. Murai, Ed., Elsvier, 419–422.
Matsumoto, J., 1997: Seasonal transition of summer rainy season over Indochina and adjacent monsoon region. Adv. Atmos. Sci., 14 , 231–245.
Murakami, T., 1979: Winter monsoonal surges over East and Southeast Asia. J. Meteor. Soc. Japan, 57 , 133–158.
Schubert, W. H., P. E. Ciesielski, C. Lu, and R. H. Johnson, 1995: Dynamical adjustment of the trade wind inversion layer. J. Atmos. Sci., 52 , 2941–2952.
Locations of operational rawinsonde stations. Open squares, open triangles, and plus symbols indicate type-I, type-II, and type-III stations, respectively (see section 4).
Citation: Journal of Climate 19, 13; 10.1175/JCLI3792.1
Vertical profile of temperature observed at 0000 UTC 2 March 1997 at Ubon Ratchathani station. Squares indicate data reported at mandatory and significant levels. Asterisks indicate linearly interpolated data at 100-m height intervals.
Citation: Journal of Climate 19, 13; 10.1175/JCLI3792.1
Histogram of Δθ/Δz frequency over 29 years of data for November to April at a height of 2–6 km at Ubon Ratchathani. The frequency bins for [−5, −4] K km−1 and [20, 21] K km−1 include contributions of layers with Δθ/Δz < −5 K km−1 and with Δθ/Δz > 21 K km−1, respectively. Stable (Δθ/Δz > 10 K km−1) and unstable (Δθ/Δz < 1 K km−1) layers account for 4.9% and 1.3%, respectively, of all the layers. The mean value and the standard deviation of the distribution are 5.1 and 2.7 K km−1, respectively. The typical value of the atmospheric moist adiabatic lapse rate at 2–6-km height over the subtropics is 3–7 K km−1.
Citation: Journal of Climate 19, 13; 10.1175/JCLI3792.1
Time–height cross section of stable layer frequency (%) for morning data at Ubon Ratchathani (15°15′N, 104°52′E). Ubon Ratchathani is a type-I station (see section 4). Data is from July to June to cover the entire dry season. Asterisks indicate the levels with highest frequency of stable layer development (estimated by excluding surface stable layers) for half-month bins from the second half of November to the second half of April. The straight line is a regression line of these highest frequency levels obtained by a least squares fitting.
Citation: Journal of Climate 19, 13; 10.1175/JCLI3792.1
Mean annual variation in precipitation averaged from 1997 to 2003 at (a) Ubon Ratchathani, (b) Hanoi, (c) Songkhla, and (d) Singapore.
Citation: Journal of Climate 19, 13; 10.1175/JCLI3792.1
Time–height cross section of stable layer frequency (%) for evening data at Ubon Ratchathani. Data is from July to June to cover the entire dry season.
Citation: Journal of Climate 19, 13; 10.1175/JCLI3792.1
As in Fig. 6 but for unstable layer frequency.
Citation: Journal of Climate 19, 13; 10.1175/JCLI3792.1
Time–height cross section of stable layer frequency (%) for morning data at Ubon Ratchathani: (a) data from July 1999 to June 2000 and (b) data from July 1996 to June 1997. Frequencies were evaluated at every pentad and at 100-m height intervals. Data is from July to June to cover the entire dry season.
Citation: Journal of Climate 19, 13; 10.1175/JCLI3792.1
Time–height cross section of stable layer frequency (%) for morning data at type-I stations (a) Kunming and (c) Chiang Mai; and unstable layers at (b) Kunming and (d) Chiang Mai.
Citation: Journal of Climate 19, 13; 10.1175/JCLI3792.1
As in Fig. 9 but for stable layer frequency at type-II stations (a) Nanning, (c) Hanoi, (e) Hongkong, (g) Da Nang, and (i) Bangkok; and for unstable layer frequency at type-II stations (b) Nanning, (d) Hanoi, (f) Hongkong, (h) Da Nang, and (j) Bangkok.
Citation: Journal of Climate 19, 13; 10.1175/JCLI3792.1
As in Fig. 9 but for stable layer frequency at type-III stations (a) Ho Chi Minh, (c) Songkhla, (e) Kota Bharu, (g) Penang, (i) Kuantan, and (k) Singapore; and for unstable layer frequency at type-III stations (b) Ho Chi Minh, (d) Songkhla, (f) Kota Bharu, (h) Penang, (j) Kuantan, and (l) Singapore.
Citation: Journal of Climate 19, 13; 10.1175/JCLI3792.1
Operational rawinsonde stations used in this study.
A pentad is seasonal division by 5 days; one year consists of 73 pentads.
