Some Interesting Observations of Cross-Mountain East-to-Southeasterly Flow at Hong Kong International Airport and Their Numerical Simulations

Abstract

With the availability of more ground-based remote-sensing meteorological equipment at Hong Kong International Airport, many more interesting features of terrain-disrupted airflow have been observed, such as the applications of short-range Doppler LIDAR. This paper documents a number of new features observed at the airport area, such as the hydraulic jump-like feature, vortex, and extensive mountain wake/reverse flow. The technical feasibility of using a numerical resolution weather prediction model to simulate such features is also explored. It is found that the presently available input data and numerical model may not be able to capture the fine features of the atmospheric boundary layer, and thus they are not very successful in reproducing many small-scale terrain-disrupted airflow features downstream of an isolated hill. On the other hand, more larger-scale terrain-disrupted flow features may be better captured, but there are still limitations with the available turbulence parameterization schemes. This paper aims at documenting the newly observed flow features at the Hong Kong International Airport, enhancing the understanding of low-level windshear, and evaluating the outputs of numerical resolution simulations for reproducing such observed features and its technical feasibility on short-term forecasting.

1. Introduction

Hong Kong International Airport (HKIA) is situated in an area of complex terrain. To the south of it is the mountainous Lantau Island, with peaks reaching about 1000 m above mean sea level, with valleys of around 400 m in between. HKIA, located in Hong Kong, experiences a humid subtropical climate. During autumn and winter, the predominantly winds are easterlies under the effect of northeast monsoon. Temperatures are mild, with the passage of cold fronts bringing down the temperatures and causing rapid wind changes. Meanwhile, in spring and summer, the prevailing wind is southeast-to-southwesterly. Due to the abundant moisture transfer from the South China Sea, the weather is hot and humid, occasionally with thunderstorms. In particular, when east-to-southeasterly winds cross over the Lantau terrain, various kinds of terrain-disrupted airflow occur inside and around HKIA. Such airflow disturbances may take the form of mountain waves, mountain wakes, and reverse flow, which may be observed in clear air conditions using the Doppler Light Detection And Ranging (LIDAR) systems inside HKIA [1] using plan position indicator (PPI) scans.

Recently, short-range LIDAR has also been set up around HKIA [2], and range height indicator (RHI) scans are performed to reveal the two-dimensional airflow features perpendicular to the mountains of Lantau Island. With more scanning data collected in the recent time, interesting flow patterns, such as hydraulic jump-like features and horizontally oriented vortex, are also observed for the first time around HKIA. Such features are documented in the present paper, for the first time in Hong Kong.

The above-mentioned airflow disturbances may be hazardous to the aircraft operating at HKIA, due to the occurrence of low-level windshear and turbulence. As such, simulation of these features would be useful for early alerting of such hazardous weather. This paper will discuss the technical feasibility of forecasting such flow features using large eddy simulations (LES) with high spatial resolution, namely, horizontal resolution in the order of 40 m. Some windshear features are reproduced successfully, but there are also limitations of the LES model, so certain features are not simulated very well.

Mountain-disturbed airflow has been studied extensively in other mountain ranges, such as the Rocky Mountains [3] or the Alps [4]. Clark et al. [5] studied the famous Colorado downslope flow as well, which has important practical application values. Smith and Skyllingstad [6] studied waves and rotors arising from terrain. Moreover, Ehard et al. [7] analysed LIDAR observations covering 18 years to study the wintertime gravity wave activity near Scandinavian mountain ranges. There are also a lot of similar studies that evaluate gravity wave activity, but mostly in the stratosphere [8,9,10,11]. The present paper will focus on smaller-scale mountains with a spatial dimension of around 10 km only. The elaboration will be on the low-level waves observed in the vicinity of an international airport, which may affect aircraft operations.

2. Meteorological Instruments at HKIA

The Hong Kong Observatory (HKO) operates a suite of meteorological equipment inside and around HKIA to monitor and warn against weather that may be hazardous to aircraft operation. In particular, because the airflow around HKIA could be complex due to terrain-disrupted airflow and thunderstorms, different kinds of in-situ and remote-sensing wind measuring equipment are operated. The major equipment used in HKIA is shown in Figure 1a. Of particular concern to this paper is the long-range Doppler LIDAR (range up to 10 km with a spatial resolution of around 100 m) operated at the centre of the Centre Runway of HKIA and the Terminal Doppler Weather Radar (TDWR) operated to the northeast of HKIA for monitoring wind conditions in the airport region in rainy weather. The long-range Doppler LIDAR has a wavelength of 1550 nm and a range resolution of 100 m, while the TDWR is in C band (5.625 GHz), with a range resolution of 150 m.

Short-range LIDARs, namely those with a measuring range of around 6 km but with higher spatial resolution (30 m or so), have been set up gradually inside the airport. At first, they were mainly used to monitor airflow disturbances arising from man-made structures and buildings. However, due to their relatively small size, though with limited measurement range, they were also useful for monitoring airflow disturbances over localized areas, such as performing RHI scans downwind of isolated hills on Lantau Island (Figure 1c), with rather frequent updates (every couple of minutes). The short-range LIDAR at Siu Ho Wan (scanning pattern in Figure 1b) is considered in this paper, mainly to monitor the airflow disturbances arising from Lo Fu Tau (location in Figure 1a) when there is mountain-crossing airflow.

3. Numerical Simulation

The Regional Atmospheric Modelling System (RAMS) version 6.3 is used in this paper. It is downscaled from the re-analysis of the European Centre of Medium-Range Weather Forecasts (ERA5 from ECMWF) as a boundary condition. The spatial resolution of the simulation is increased by a factor of five, nested to focus in the HKIA area. The resolution is 25 km, 5 km, 1 km, and 200 m, and the innermost domain has a high spatial resolution of 40 m. LES mode simulation is performed. Technical details can be found in [12]. No additional data are assimilated in the downscaling. Boundary conditions are updated every hour from ECMWF.

To perform the LES simulation, the choice of turbulence parameterization scheme is an important factor. In this work, the Deardorff scheme [13] was used. It is found to perform very well in simulating the terrain-disrupted airflow at HKIA by comparison with LIDAR and weather station observations [12], as well as elsewhere, such as the airports in the Qinghai–Tibet Plateau in China (the relevant papers are under review). Moreover, the simulation used a rather fine vertical grid. The vertical grid spacing at the lowest level is about 10 m, and the stretching ratio is 1.08 (i.e., the ratio of the model level heights between the adjacent model levels). The Deardorff closure scheme [13] is the most commonly used formulations for LES applications [14]. It is found to capture the main features of turbulent flow without introducing complicated mathematical formulations. The scheme was found to rather successfully simulate the same-scale flow features at HKIA, such as Foehn wind and mountain waves.

4. Hydraulic Jump-like Feature

Between around 10 and 11 UTC of 20 March 2024 (UTC = Hong Kong time + 8 h), jump-like features are observed to persist for several tens of minutes in the RHI scan of the short-range LIDAR at Siu Ho Wan. Some examples of such features are shown in Figure 2a,b. In particular, there is a “jump” of the outbound flow at around 1.3 to 1.5 km away from the LIDAR, with re-circulating flow aloft (slight inbound flow) and below (much smaller outbound flow). Such features have never been observed before in the mountains of Hong Kong, and this is the first time that a LIDAR at HKIA captures such flow features.

Synoptically, the northeast monsoon dominated over the whole of China on that day based on the surface isobaric pattern, with east-to-northeasterly winds prevailing over the south China coast. The upper air conditions may be revealed by radiosonde ascent in Hong Kong (at King’s Park, about 18 km east of Siu Ho Wan) at the nearest time, namely, 13 UTC on 20 March 2024. Figure 2c,d show the vertical profiles of temperature, dew point, wind speed, and wind direction in the atmospheric boundary layer and the middle troposphere. Lo Fu Tau has a height of about 460 m above mean sea level. No apparent temperature inversion is seen below 1 km or so above mean sea level. Also, no low-level jet is observed.

The jump-like structure is studied further using a two-layer model. The depth of the atmospheric boundary layer, h, can be of the order of the terrain elevation, i.e., 460 m, and the nature of the flow depends upon the Froude number (Fr),

$$Fr={\displaystyle \frac{{〈U〉}_h}{\sqrt{g_{r}h}}}$$

(1)

where $〈X〉$ indicates a vertical/depth average over the extent of X, U is the wind speed, and $g_r$ the reduced gravity,

$$g_r={\displaystyle \frac{g\left({\theta }_T-{〈\theta 〉}_h\right)}{{〈\theta 〉}_h}}$$

(2)

where g is the gravitational constant, ${\theta }_T$ the free troposphere potential temperature, and $\theta$ the potential temperature. Based on the radiosonde data, ${\theta }_T$ is about 295 K, and ${〈\theta 〉}_h$ is about 290 K. The wind speed U is about 10 m/s at the mountain top. As such, the Froude number is more than 1, i.e., it is supercritical, and this supports the occurrence of the jump-like feature.

A computer simulation initialized at 06 UTC, 20 March 2024, was performed. In order to update the boundary conditions frequently enough (as the boundary layer of the troposphere is expected to change every few hours in the spring time in Hong Kong), the ECMWF ERA5 data were used as the boundary condition, updated every hour. Around the observed period of the event (between 10 and 11 UTC of 20 March 2024), there is no apparent feature of hydraulic jump. However, at about 00 UTC of 21 March 2024, slight features of hydraulic jump are occasionally observed in the generally non-stationary flow. An example is given in Figure 3a, showing the simulated RHI image of Doppler velocity up to a range of 6 km away from the LIDAR. A zoom-in of the first 2500 m or so is given in Figure 3b. A slight jump feature is observed. Re-circulating flow can be seen above the jump and is barely visible below it. The reproduction of jump-like features is not straight-forward, even with the use of the rapidly updated boundary condition. The ERA5 data may not be able to capture the rather rapidly changing atmospheric boundary layer in the spring time over southern China. The vertical resolution of the boundary layer in ERA5 may also not be fine enough to analysis these low-level features. As such, the jump-like feature is just barely reproducible, not exactly at the observation time and the jump in the simulation is less prominent compared with actual observations.

5. Mountain Wave

A mountain wave case is observed at about 00 UTC of 21 March 2024 from the RHI scan of Siu Ho Wan short-range LIDAR (Figure 4a). The wave occurs at about the height of Lo Fu Tau. The wave length was about 2 km (the “trough” of the wave is found at about 3.5 km and 5.5 km away from the LIDAR). This is in fact the lower boundary of the vertical wavelengths analysed by [7] which identified dominating vertical wave lengths of a range from 2 to 13 km. This can also be attributed to the difference in the height of the mountain ranges and the vertical extent of waves discussed.

Based on the surface isobaric pattern, following the dominance of the northeast monsoon over China, the high pressure area moved eastwards, with a ridge of high pressure building up over southeastern coast of China, bringing easterly winds to Hong Kong. The radiosonde ascent data (launched in King’s Park, about 18 km east of Siu Ho Wan) at around that time (01 UTC of 21 March 2024) are shown in Figure 4b,c. There is a temperature inversion of a couple of degrees Celsius between about 700 to 800 m above mean sea level. Slightly higher wind speed in the easterly wind is observed at this height range.

Mountain waves have been observed at around HKIA [15] but over a much larger area. This is the first time that a rather localized mountain wave is observed in this airport area, with the use of a short-range LIDAR. To analyse the wave features, the scorer parameter is utilized, which is commonly used to describe whether gravity wave will develop or not. It combines the Brunt–Väisälä frequency and the wind along the vertical profile and defined in Equation (3) below, where $l^2$ is the scorer parameter, $N\left(z\right)$ is the Brunt-Väisälä frequency and $U\left(z\right)$ is the horizontal wind speed for different $z$ altitude.

$$l^2\left(z\right)={\displaystyle \frac{N^2\left(z\right)}{U^2\left(z\right)}}-{\displaystyle \frac{1}{U\left(z\right)}}{\displaystyle \frac{d^{2}U}{d{z}^2}}$$

(3)

The Scorer parameter is calculated from the radiosonde data and the result is shown in Figure 4d. The parameter has a peak at about 400 m and decreases with altitude upwards. This may suggest that a standing mountain wave could occur downstream of Lo Fu Tau (the LIDAR observed mountain wave is also stationary for a few tens of minutes).

For numerical simulation, the same simulation set as described in the hydraulic jump-like case above is used, i.e., initialized at 06 UTC, 20 March 2024. At the time of the event (around 00 UTC, 21 March 2024), mountain wave is not observed. However, wave-like feature is found at an earlier time, namely, at about 13 UTC, 20 March 2024, as shown in Figure 5a for the full range of the LIDAR and in Figure 5b for a zoom-in. Wave “trough” is found at about 1000 m, 2500 m and 4000 m around the LIDAR, i.e., a wavelength of about 1500 m. The locations of the wave troughs and the wavelength in the simulation are not exactly the same in the actual LIDAR observations. Also, the simulated wave is not as apparent as in the real LIDAR image. Similar to the hydraulic jump case, the mountain wave may be related to the rapidly changing atmospheric boundary layer which cannot be fully resolved by the limited vertical resolution of the ECMWF re-analysis data.

HKO has operated radar wind profiles (providing vertical profile of horizontal wind) and microwave radiometers (providing vertical profile of temperature and humidity) in Hong Kong. Such profiles at the times of the observed jump-like feature and mountain wave have been used in a computational fluid dynamic (CFD) model with built-in terrain of Lantau Island, especially around Lo Fu Tau. However, the observed features in the RHI scans of the LIDAR also do not show up well in the simulation. Basically, the RAMS simulations presented in this paper appear to be even better than the results of CFD simulations. Further simulation efforts would be made in order to reproduce such features more correctly.

6. Vortex

Vortex downstream of Lantau Island has not been observed before because there are no wind-measuring equipment performing RHI scans downstream of mountains. With the use of the short-range LIDAR at Siu Ho Wan, a vortex was observed on 24 March 2024. From the surface isobaric pattern, the high pressure area continued to move east and was then situated over the seas south of Japan, with a ridge of high pressure extending from the northwest Pacific to Taiwan and then southern China. Easterly winds continued to prevail over the south China coast. The surface wind observations by that time are shown in Figure 6a. Downwind of Lo Fu Tau, there are basically moderate east-to-southeasterly winds. An RHI velocity image is shown in Figure 6b. Reverse flow (inbound flow) is observed directly downstream of the LIDAR nearer the ground. Higher up, outbound flow is observed at higher altitudes, and it is found to “descend” with distance downstream of the LIDAR, reaching the ground at about 3000 m away from the LIDAR.

RAMS simulation is again performed using ERA5 as boundary condition. It is initialized at 06 UTC, 24 March 2024. Elevated outbound flow is barely observed in the simulated LIDAR picture, with the majority outbound flow originated near the ground in the simulation. When elevated outbound flow is observed, reserve flow is not apparent downwind of Lo Fu Tau. An example is given in Figure 7. In general, as in the simulation described above, just non-stationary flow is given in the simulation. However, in the real LIDAR images, the reserve flow is quite persistent, remaining basically stationary in a period of a couple tens of minutes. Further simulation effort would be required in order to capture the stationary vortex pattern.

7. Mountain Wake

As in [1] and documented elsewhere, in clear-air conditions, a mountain wave is frequently observed at HKIA in non-rainy weather. Before, there was no documentation of mountain waves observed in rainy weather at this airport. A case was found on 10 March 2024, with a Doppler velocity image from the TDWR (Figure 8a). A wake (weak winds in grey) is found to the west of the southern part of Lantau Island. On the other hand, there are multiple jets (strong winds in yellow) in the northern part of this island.

Simulation was performed as initialized at 00 UTC, 10 March 2024. On that day, based on a surface isobaric pattern, a ridge of high pressure dominated over the eastern and southeastern coast of China. This is the typical pattern for easterly winds prevailing over the south China coast. The result of domain 4 at the observed time of the event is given in Figure 8b. The pattern of the wind is found to be reproduced reasonably well in the simulation. The mountain wave and the multiple jet features are well forecast. With the current setup of RAMS, maybe such “larger-scale” terrain-disrupted airflow features could have a better chance of successful reproduction, as compared to the reproduction of localized airflow features just downstream of a small and isolated hill on Lantau Island, as described earlier in this paper.

8. Reverse Flow Leading to Low-Level Windshear

Another case of extensive reverse flow is found on the rainy day of 21 April 2024. Synoptically, a trough of low pressure existed over inland areas of southern China, leading to generally southerly winds over the south China coast based on a surface isobaric pattern. The surface observations (Figure 9a) show that strong to gale force south-to-southeasterly winds prevail in the airport area. From the long-range LIDAR data (Figure 9b), extensive reverse flow (in green) could be found both to the west and to the east of the long-range LIDAR at the centre of the Centre Runway of HKIA. Such features could be hazardous to the aircraft operating at the airport because of low-level windshear and turbulence. In fact, at around the time of the above observation, two aircraft landing at the north runway of HKIA from the west (07LA runway corridor) reported the encountering of significant windshear.

Additional observations of this case are provided in Figure 10. Figure 10a shows the virtual temperature from the radio acoustic sounding system (RASS) at Sha Lo Wan (location in Figure 1a). A warming up of the lower atmosphere (in red) occurs occasionally due to the Foehn effect of cross-mountain flow. Over the 07LA runway corridor, at the time of the windshear report from the pilot, the LIDAR-measured headwind profile (Figure 10b) shows a significant jump of headwind at about the runway end (0 nautical mile point), with tailwind while the aircraft is airborne (negative headwind) and positive headwind over the runway (from 0 to −2 nautical miles from the runway end).

Simulation was performed as initialized at 00 UTC, 21 April 2024. Two sample images of the simulated LIDAR Doppler velocity fields are given in Figure 11. Downstream of the mountains of Lantau Island, there are “patches” of reverse flow (in grey and green) both to the west and to the east of the location of the centre LIDAR of the Centre Runway. However, compared to the actual observations (Figure 9b), the reverse-flow areas are less extensive.

The simulated virtual temperature profiles of RASS are shown in Figure 12a. Again, a warming up of the lower atmosphere (in red) is successfully reproduced. In the simulated 07LA headwind profile (Figure 12b), the increase in headwind near the runway threshold from 1 to 0 nautical mile is reproduced, consistent with actual observations (Figure 10b).

In general, the “larger-scale” airflow disruption at HKIA is quite successfully captured by numerical simulation. The headwind profiles are correctly simulated, which would be useful for early alerting of low-level windshear. However, in the simulation, probably because of the available choice of turbulence parameterization schemes, the reverse flow tends to be patchier, and the rather extensive reverse-flow areas in the actual LIDAR PPI observation are not very successfully reproduced.

9. Conclusions

In earlier attempts [12], high-resolution numerical simulations could successfully reproduce many terrain-disrupted airflow features at HKIA. As such, early alerting of low-level windshear and turbulence is found to be technically feasible. With more cases accumulated, and with localized observations such as short-range LIDAR measurements, it is found that there are limitations in such simulations. This paper is novel in two ways: first, many new features of terrain-disrupted airflow have been documented, such as hydraulic jump-like features and vortices; second, the technical feasibility of reproducing terrain-disrupted airflow features, especially the new ones, is explored. The first point would be of use for reference by aviation weather forecasters working at airports with similarly complex terrain. The second point is to stimulate further development of fine-scale numerical weather prediction models, as with the use of the existing parameterization schemes, there are still limitations as to the successful reproduction of some features, especially the small-scale ones associated with isolated hills.

More short-range LIDARs are being deployed at HKIA. It is expected that more interesting observations could come out with the availability of more observational data. Such cases will be reported in future papers, and the uses and limitations of high-resolution numerical simulations will be further studied.