Intraseasonal Variation of Visibility in Hong Kong

Visibility impairment occurs as a result of the scattering and absorption of light by particles and gases in the atmosphere. Visibility impairment or haze is an important environmental issue associated with public health, hospital admissions, and mortality that receives a lot of attention throughout the world. A huge increase in ambient pollutant emissions has occurred during China's rapid development, such as construction, transportation, and energy consumption, causing deterioration in air quality. Hong Kong is located on the Pearl River Delta (PRD), which is one of the most industrial and GDP-contributing regions in Southeast China. Local emissions of pollutants and transboundary transportation of pollutants from mainland China are the major sources of ambient pollutants in Hong Kong (Wang, 2003; Leung and Lam, 2008; Leung et al., 2008).

Air pollution levels can be quantified not only by air pollutant concentration and general air quality indices [e.g., air pollution index (API) and air quality health index], but also in terms of visibility (Huang et al., 2009), which can act as a surrogate for air pollution levels when long-term routine air pollutant concentration data are not available. In a study of the relationship between mortality and fine particulate matter (PM_2.5) exposure, visibility data were used to estimate PM_2.5 levels (McDonnell et al., 2000). Visibility degradation occurs mainly as a result of the absorption and scattering of light by suspended particulates such as primary and secondary salts, inorganic carbon, and crustal minerals. The relationship between visibility and air particles has been widely discussed (Malm et al., 1994), and this information can be useful in Hong Kong. In terms of public impact, visibility impairment badly hampers local traffic and aviation, causes adverse health effects such as cardiovascular and respiratory disease (Pope III and Dockery, 2006), and even increases the mortality risk of the general public in Hong Kong (Thach et al., 2010).

Figure 1.  (a) Monthly variation of visibility in Hong Kong. (b) Power spectrum of daily reduced-visibility hours in Hong Kong during summer. (c) Power spectrum of daily reduced-visibility hours in Hong Kong during winter. The green dashed lines denote the Markov red noise spectrum, while the red and blue dashed lines represent the 90% and 5% confidence levels, respectively.

Visibility is associated not only with an increase in pollutant emissions, but also with direct and indirect atmospheric conditions. Direct atmospheric conditions are associated mostly with ambient water content and include fog, mist, rain, and their interaction with particulates. Indirect conditions include pollutant transport and accumulation. A case study conducted by (Wu et al., 2005) showed that a hurricane event on 2 November 2003 caused strong descending motion and weak surface wind that increased aerosol concentrations as a result of extreme low visibility. In long-term air pollution studies, interannual variation in air pollution levels has been noted (Kim et al., 2013).

In this study, we aim to further investigate the intraseasonal variation of visibility and the associated modulating factors during summer (July-September) and winter (January-March) in Hong Kong in order to reveal the influence of atmospheric conditions. Moreover, an understanding of the impacts of the predictors that give rise to the observed variability in Hong Kong visibility is central for future success in subseasonal prediction. Successful prediction of the response of visibility to the tropical intraseasonal oscillation (ISO) can have profound social and economic consequences. The failure or even delay of a prediction can make all the difference between risk and health for Hong Kong. Improved knowledge of the impacts of climate variability on Hong Kong visibility will be used to develop more effective management strategies for public health units and will enable air quality managers to provide people living in cities and towns with safe and healthy environments.

Daily meteorological data, including RH, specific humidity (q), omega, geopotential height (GPH), zonal wind (u), meridional wind (v), and temperature, were acquired online (http://www.esrl.noaa.gov/psd/data/gridded/data.ncep.reanalysis.html) from the NCEP-NCAR Reanalysis-1 dataset (Kalnay et al., 1996). Hourly visibility data during 2000-07 were collected from the Hong Kong Observatory (HKO) Headquarters. Since visibility is affected by atmospheric conditions such as rain, fog, and hygroscopic growth interacting with RH, daily reduced visibility here is defined as visibility below 8 km along with RH <95%, following previous studies (Chang and Koo, 1986; Leung and Lam, 2008).

Figure 2.  Composites of anomalies of OLR (units: W m^-2), 850 hPa omega (units: 10^-2 s^-1), 850 hPa RH (units: %), 850 hPa wind (units: m s^-1) and geopotential height (gph; units: m), and 850 hPa divergent wind (units: m s^-1) and divergence (units: 10^-6 s^-1) associated with visibility impairment in Hong Kong during summer. Colored shading and colored vectors denote anomalies that are over the 90% confidence level, and the crosses denote the position of Hong Kong.

Visibility at the HKO can be used as a measure for local air quality, and visibility in Hong Kong demonstrates clear seasonal variation (Fig. 1a), with better (poor) visibility being observed in summer (winter), which is consistent with winter in general having higher pollution levels than summer. The average and the 90th percentile of the daily reduced-visibility hours during summer are 1.59 h and 7 h, respectively, compared to values of 6.48 h and 17 h in winter, as shown in Table 1. In this study, we consider a day to suffer from visibility impairment when the number of hours of reduced visibility of that particular day exceeds a threshold of the 90th percentile (7 hours during summer and 17 hours during winter).

To further identify the dominant period of variation, Figs. 1b and c show the power spectra of daily reduced-visibility hours during summer and winter. Both spectra reveal dominant peaks on the synoptic time scale at 3-10 days, as well as on the intraseasonal time scale at 20-30 days in summer and 20-40 days in winter. This suggests that visibility in Hong Kong is affected not only by synoptic systems, but perhaps also by components of the ISO. Therefore, we want to know whether the phase changes of the ISO could have significant effects on local visibility impairment and air quality. We can then identify the coupling of the synoptic scale, ISO type, local meteorological conditions, and air quality conditions in Hong Kong.

The circulation characteristics associated with visibility impairment in summer and winter are first investigated using composite analysis in this section. The composite maps of different environmental variables are constructed based on the days with visibility impairment during summer and winter. There are a total of 78 (65) days of visibility impairment in summer (winter) during the study period. The Student's t-test is then used to determine the level of significance associated with the circulation anomalies.

As shown in Fig. 2, visibility impairment in Hong Kong during summer is characterized by the following features:

In summer, the first noticeable feature is the alternating circulation anomalies over Hong Kong and east of Taiwan. Enhanced convection in Taiwan is associated with anomalous convergence and rising motion. This in turn results in anomalous subsidence and divergence over Hong Kong, which suppresses dispersion and promotes stagnation and accumulation of pollutants in Hong Kong.

Figure 3.  Composites of anomalies of OLR (units: W m^-2), 1000 hPa temperature (units: K), 850 hPa RH (units: %), and 500 hPa wind (units: m s^-1) and geopotential height (gph; units: m) associated with visibility impairment in Hong Kong during winter. Colored shading and colored vectors denote anomalies that are over the 90% confidence level, and the crosses denote the position of Hong Kong.

On the other hand, visibility impairment in winter is found to be associated with positive outgoing longwave radiation (OLR) anomalies, geopotential height anomalies, and temperature anomalies over Southeast China in the vicinity of Hong Kong (Fig. 3). In other words, Hong Kong is under the control of suppressed convection during visibility impairment in winter. Compared to the alternating circulation anomalies in summer (Fig. 2), the spatial scale of the suppressed convection is much larger in winter, covering broader areas of China. The suppressed convection favors the accumulation of local pollutants, impairing the visibility and air quality in Hong Kong.

Apart from anomalous sinking motion, suppressed convection associated with reduced moisture in summer in Hong Kong inhibits the wet deposition of pollutants and degrades the visibility and air quality in Hong Kong.

Similar to cases in summer (Fig. 2), visibility impairment in winter is also characterized by a significant reduction in moisture in Hong Kong (Fig. 3), which suppresses the wet deposition of pollutants and worsens the visibility and air quality in Hong Kong.

In summer, the cyclonic anomalies associated with enhanced convection in Taiwan also induce anomalous northeasterlies in the vicinity of Hong Kong, which favors the transport of regional pollutants from the PRD, further worsening the visibility and air quality in Hong Kong.

In winter, enhanced northeasterlies associated with anticyclonic anomalies over Southeast China facilitate the transport of regional pollutants from the PRD and further worsen the visibility and air quality in Hong Kong (Fig. 3).

Figure 4.  Composites of the synoptic (left panels), intraseasonal (middle panels), and LFBS (right panels) components of OLR (units: W m^-2), 850 hPa omega (units: 10^-2 s^-1), 850 hPa RH (units: %), and 850 hPa divergence (10^-6 s^-1) associated with visibility impairment in Hong Kong during summer. Colored shading denotes anomalies that are over the 90% confidence level, and the crosses denote the position of Hong Kong.

Figure 5.  Composites of the synoptic (left panels), intraseasonal (middle panels), and LFBS (right panels) components of OLR (W m^-2), 1000 hPa temperature (units: K), and 850 hPa RH (units: %) associated with visibility impairment in Hong Kong during winter. Colored shading denotes anomalies that are over the 90% confidence level, and the crosses denote the position of Hong Kong.

To further clarify the changes in atmospheric circulations at different time scales, the total field of each environmental variable is decomposed into three components; namely, the synoptic component (3-10 days), the intraseasonal component (10-90 days), and the low-frequency background state (LFBS) component (>90 days), following (Li and Zhou, 2015). Through such decomposition, the relative contributions of each of these components to the overall circulation changes can be evaluated. Figure 4 shows the decomposition results, while Table 2 summarizes the contributions of each individual component to the overall circulation anomalies in Hong Kong during summer. Interestingly, among the three terms, the intraseasonal components dominate, accounting for 50%-70% of the overall anomalies in Hong Kong during summer, as shown in Fig. 4 and Table 2. The synoptic components rank second, making up 15%-30% of the overall anomalies. The LFBS components, on the other hand, generally have the smallest contribution. The results here again stress the importance of the intraseasonal component in controlling the visibility variation in Hong Kong. Therefore, apart from synoptic systems, the relationship between the ISO and visibility variation in Hong Kong should definitely be considered and will be discussed in detail in the next section.

The total field of each environmental variable is further decomposed into synoptic, intraseasonal, and LFBS components to reveal their relative contributions to the overall circulation changes in Hong Kong during winter. The decomposition results and the contributions of each individual component are depicted in Fig. 5 and Table 3, respectively. Again, the intraseasonal components are the largest contributor among the three terms, followed by the LFBS and the synoptic components (Fig. 5 and Table 3). As in summer, the intraseasonal component plays a dominant role in controlling visibility impairment in Hong Kong during winter.

Figure 6.  The first two leading EOF modes of (a, b) 30-60-day filtered and (d, e) 10-30-day filtered OLR anomalies in summer. Panels (c, f) show the correlation coefficients between the associated time series of PC1 and PC2 at different lags, respectively.

The previous section highlighted the importance of intraseasonal components in controlling visibility impairment in Hong Kong during summer. As pointed out in previous studies (Zhou and Chan, 2005; Li and Zhou, 2013a, b ), the tropical ISO consists of the pronounced 30-60-day MJO and the 10-30-day ISO. In this section, how these two intraseasonal components modulate local visibility in Hong Kong will be examined and the associated modulating mechanisms will be discussed.

Following Li and Zhou (2013a, b), an EOF analysis is applied to 30-60-day filtered and 10-30-day filtered OLR anomalies, respectively, to extract the dominant convective signals associated with the MJO and the 10-30-day ISO (Fig. 6). The phase and amplitude of the ISO can then be expressed in terms of the two leading principal components (PCs) in the same way as Li and Zhou (2013a, b):

For 10-30-day ISO: = Phase _10-30-ISO = tan^-1 [PC2 _10-30-ISO/PC1 _10-30-ISO]> Amplitude _10-30-ISO = [PC1 _10-30-ISO² + PC2 _10-30-ISO²]^1/2[0.5mm] For MJO: > Phase _MJO = tan^-1 [PC2 _MJO/PC1 _MJO]> Amplitude _MJO = [PC1 _MJO² + PC2 _MJO²]^1/2

5.1.1. In summer

Figure 7.  Composites of 30-60-day filtered OLR anomalies (shading; units: W m^-2) and 850 hPa wind anomalies (vectors; units: m s^-1) for different MJO phases in summer. Only anomalies exceeding the 90% confidence level based on the Student’s t-test are shown. Triangles denote the position of Hong Kong.

Consistent with Li and Zhou (2013a, b), the leading EOF modes of the 30-60-day filtered OLR describe the northeastward propagation of the MJO-related convection during boreal summer (Figs. 6a and b), while those of the 10-30-day filtered case reveal northwestward-propagating convection associated with the 10-30-day ISO (Figs. 6d and e). Figure 7 shows the circulation anomalies for different MJO phases, while Table 4 depicts the corresponding changes in local visibility and air quality associated with each of the MJO phases during summer. Two of the MJO phases, phase 3+4 and phase 7+8, reveal significant changes in local visibility (Table 4). In phase 3+4, local visibility is observed to be much better than that of the climatology. The daily reduced-visibility hours drop significantly to 0.87, compared to the climatological value of 1.59. During this phase, the MJO-related convection is oriented in a north-south direction in a way similar to that of EOF2 (Fig. 6b), with enhanced convection and cyclonic circulation dominating over Southeast China (Fig. 7b). Examination of the vertical profile of different environmental variables shows that Hong Kong is actually subjected to enhanced rising motion, a richer moisture supply, and strengthened low-level southwesterly wind during this period (Fig. 8), all of which are very favorable for the dispersion and wet deposition of local pollutants, leading to general improvement in local visibility as well as air quality in Hong Kong. In addition, (Li and Zhou, 2013a) found a significant reduction in the frequency of tropical cyclones (TCs) in the western North Pacific during this phase. The suppressed synoptic-scale TC activity might be another reason for the general improvement in visibility and air quality in Hong Kong during this period.

Figure 8.  Vertical profiles of 30-60-day filtered anomalies [averaged over (21^°-24^°N, 112^°-115^°E)] of RH, specific humidity (q), omega, geopotential height (GPH), zonal wind (u), meridional wind (v), temperature (Temp), and relative vorticity (Vort) for MJO phase 3+4 (red lines) and phase 7+8 (blue lines) in summer.

Figure 9.  Composites of 10-30-day filtered OLR anomalies (shading; units: W m^-2) and 850 hPa wind anomalies (vectors; units: m s^-1) for different phases associated with the 10-30-day ISO in summer. Only anomalies exceeding the 90% confidence level based on the Student's t-test are shown. Triangles denote the position of Hong Kong.

Figure 10.  Vertical profiles of 10-30-day filtered anomalies [averaged over (21^°-24^°N, 112^°-115^°E)] of RH, specific humidity (q), omega, geopotential height (GPH), zonal wind (u), meridional wind (v), temperature (Temp), and relative vorticity (Vort) for phase 1+2 (red lines) and phase 5+6 (blue lines) associated with the 10-30-day ISO in summer.

Figure 11.  The first two leading EOF modes of (a, b) 30-60-day filtered and (d, e) 10-30-day filtered OLR anomalies in winter. Panels (c, f) show the correlation coefficients between the associated time series of PC1 and PC2 at different lags, respectively.

In contrast, in phase 7+8, the situation is reversed. Degradation in visibility and the API can be found during this period (Table 4). The number of hours of reduced visibility increases to 2.38 d^-1, compared to the climatological value of 1.59. The suppressed MJO-related convection over Southeast China induces stronger descending motion, reduced moisture, and strengthened northeasterlies in the vicinity of Hong Kong (Fig. 8), which favors the accumulation of local pollutants as well as the transport of remote pollutants from the PRD. Apart from this, statistically enhanced TC genesis during this phase, as noted previously by (Li and Zhou, 2013a), might also contribute to the local deterioration of visibility and air quality in Hong Kong during summer.

We next move on to look into the effect of the 10-30-day ISO. Similar to the MJO cases, significant changes in local visibility and the API can be observed during different phases of the 10-30-day ISO. Specifically, there is significant improvement (deterioration) in local visibility and the API in phase 1+2 (phase 5+6) associated with the 10-30-day ISO (Table 5). During these two phases, alternating circulation anomalies can be observed locally in Hong Kong and the region east of Taiwan (Figs. 9a and c), which resemble the general circulation patterns associated with visibility impairment, as shown previously in Fig. 2. In phase 1+2, Hong Kong is under the control of the enhanced convection, while a suppressed convective center is found in the region east of Taiwan (Fig. 9a). Such an orientation in convection leads to stronger ascending motion, richer moisture, and strengthened southwesterlies in the vicinity of Hong Kong (Fig. 10), which favors the dispersion and wet deposition of local pollutants, leading to better visibility and air quality in Hong Kong (Table 5). In phase 5+6, however, the circulation pattern is reversed (Fig. 9c). Hong Kong suffers from suppressed convection with enhanced descending motion, reduced moisture, and strengthened northeasterlies (Fig. 10), resulting in much poorer visibility and air quality during this period.

Figure 12.  Composites of 30-60-day filtered OLR anomalies (shading; units: W m^-2) and 850 hPa wind anomalies (vectors; units: m s^-1) for different MJO phases in winter. Only anomalies exceeding the 90% confidence level based on the Student’s t-test are shown. Triangles denote the position of Hong Kong.

5.1.2. In winter

For the 30-60-day MJO in winter, shown in Fig. 11, the leading EOF modes of filtered OLR anomalies reveal an east-west dipole pattern. Compared to the leading EOF modes in summer (Fig. 6), the northward-propagating component of the MJO-related convection is much weaker in winter and the convective center is confined mainly to the tropical region south of 20^°N. The derived pattern here is consistent with that of previous studies (Wheeler and Hendon, 2004; Huang et al., 2011), which suggests that the MJO propagates mainly eastward, instead of northeastward, during boreal winter. Likewise, for the 10-30-day ISO, the convective center shifts southward during winter, with a much weakened northward-propagating component compared to that in summer (Fig. 11).

Figure 12 shows the circulation anomalies for different MJO phases, while Table 6 depicts the corresponding changes in local visibility and air quality associated with each of the MJO phases during winter. Compared to the significant changes in local visibility during different MJO phases in summer, the influence of the MJO is less prominent in winter (Table 6). The reduced-visibility hours do not reveal significant differences among different MJO phases, though phase 5+6 (phase 1+2) generally shows some improvement (deterioration) in local air quality. Compared to the circulation anomalies in summer (Fig. 9), both the convection and wind anomalies are significantly weakened over the region north of 20^°N in winter (Fig. 12). The insignificant MJO modulation in winter can be attributed primarily to the southward shift in the MJO-related convection during boreal winter. The weakening of the northward-propagating component of the MJO during boreal winter tends to weaken its modulation effect locally in Hong Kong.

Figure 13.  Composites of 10-30-day filtered OLR anomalies (shading; units: W m^-2) and 850 hPa wind anomalies (vectors; units: m s^-1) for different phases associated with the 10-30-day ISO in winter. Only anomalies exceeding the 90% confidence level based on the Student’s t-test are shown. Triangles denote the position of Hong Kong.

Similarly, the influences of the 10-30-day ISO on local visibility are found to be weaker in winter compared to those in summer. No significant changes in local visibility can be observed for different ISO phases (Table 7), though the local API does show some improvement (deterioration) in phase 7+8 (phase 3+4). As shown in Fig. 13, the convective centers associated with the 10-30-day ISO are confined mainly to the tropical region south of 20^°N in winter, with the circulation anomalies being greatly weakened in the vicinity of Hong Kong. As a result, the impact of the 10-30-day ISO can not extend to Hong Kong (at 22^°15'N/114^°10'E), which explains why the modulation of local visibility by the 10-30-day ISO appears to be weaker in winter compared to that in summer (Table 7).

In this study, visibility variation in Hong Kong during summer and winter is investigated. Visibility in Hong Kong has clear intraseasonal variation. Examination of different environmental parameters suggests that the intraseasonal component dominates the overall circulation anomalies in both summer and winter. Owing to the dominance of the intraseasonal variation, we further examine the respective impacts of the dominant modes of the tropical ISO, the 30-60-day MJO, and the 10-30-day ISO on visibility variation in Hong Kong in summer and winter. An EOF analysis is first applied to 30-60-day filtered and 10-30-day filtered OLR anomalies to extract the dominant convective signals associated with the MJO and 10-30-day ISO, respectively. It is found that the leading EOF modes of the 30-60-day filtered OLR describe the northeastward propagation of the MJO-related convection, while those of the 10-30-day filtered case capture the northwestward-propagating convection associated with the 10-30-day ISO in summer. On the other hand, the convection associated with the MJO and 10-30-day ISO in winter seems to be weaker and southward-shifted, and the convective center is confined mainly to the tropical region south of 20^°N.

Consistent with the change in the confinement of the convective center, our results also suggest that the two dominant modes of the ISO contribute differently to visibility variation in Hong Kong in summer and winter. In summer, local visibility and air quality are found to be significantly affected by the MJO and 10-30-day ISO through modulating the associated atmospheric circulations, while the southward shift in the convective center of the MJO and 10-30-day ISO tend to weaken their impacts on local visibility in winter. (Kim et al., 2013) previously suggested that the East Asian winter monsoon can transport regional pollutants from the north, causing deterioration in local visibility in winter. We thus suspect that midlatitude intraseasonal signals might be the key factor in visibility modulation in winter, which we plan to explore further in future studies.

This study is only a preliminary investigation of the environmental impacts on local visibility and air pollution in Hong Kong. The results highlight the importance of the tropical ISO in contributing to the overall visibility variation in Hong Kong. A follow-up study using higher-resolution data and model simulations is currently underway to further verify our results and look into the details of the modulation mechanisms associated with the ISO. Nevertheless, this study lays the foundation for the development of possible mitigation strategies for any related detrimental effects and will be beneficial to regional air quality management, public health risk control, and sustainable city planning. We know that Hong Kong is experiencing increasing emissions of gaseous pollutants and particulates due to the increase in population, as well as per capita energy consumption. In recent years, air quality in Hong Kong has become an issue of great public concern. There is ample evidence for the adverse effects of air pollution on human health, materials, and natural ecosystems in this region. Under these pressures, identifying and understanding the atmospheric factors behind the degradation of air quality in Hong Kong has become an important scientific research field. Visibility impairment in Hong Kong is related largely to meteorological conditions. Successful prediction of the response of visibility to the ISO can have profound social and economic consequences. The failure or even delay of a prediction can make all the difference between risk and health for Hong Kong. Improved knowledge of the impacts of climate variability on Hong Kong visibility will be useful for the development of more effective management strategies for public health units and for safeguarding the wellbeing of citizens.