Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Aviation Model: A Fine-Scale Numerical Weather Prediction System for Aviation Applications at the Hong Kong International Airport

Aviation Model: A Fine-Scale Numerical Weather Prediction System for Aviation Applications at the... Hindawi Publishing Corporation Advances in Meteorology Volume 2013, Article ID 532475, 11 pages http://dx.doi.org/10.1155/2013/532475 Research Article Aviation Model: A Fine-Scale Numerical Weather Prediction System for Aviation Applications at the Hong Kong International Airport 1 2 1 Wai-Kin Wong, Cheong-Shing Lau, and Pak-Wai Chan Hong Kong Observatory, Hong Kong Department of Computing, Polytechnic University of Hong Kong, Hong Kong Correspondence should be addressed to Pak-Wai Chan; pwchan@hko.gov.hk Received 29 December 2012; Revised 16 April 2013; Accepted 22 May 2013 Academic Editor: Richard Leaitch Copyright © 2013 Wai-Kin Wong et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The Hong Kong Observatory (HKO) is planning to implement a fine-resolution Numerical Weather Prediction (NWP) model for supporting the aviation weather applications at the Hong Kong International Airport (HKIA). This new NWP model system, called Aviation Model (AVM), is configured at a horizontal grid spacing of 600 m and 200 m. It is based on the WRF-ARW (Advance Research WRF) model that can have sufficient computation efficiency in order to produce hourly updated forecasts up to 9 hours ahead on a future high performance computer system with theoretical peak performance of around 10 TFLOPS. AVM will be nested inside the operational mesoscale NWP model of HKO with horizontal resolution of 2 km. In this paper, initial numerical experiment results in forecast of windshear events due to seabreeze and terrain eeff ct are discussed. eTh simulation of sea-breeze-related windshear is quite successful, and the headwind change observed from flight data could be reproduced in the model forecast. Some impacts of physical processes on generating the fine-scale wind circu lation and development of significant convection are illustrated. The paper also discusses the limitations in the current model setup and proposes methods for the future development of AVM. 1. Introduction air traffic for in-bound and out-bound flights. Moreover, the airel fi d operation will be greatly affected by widespread or The Hong Kong International Airport [ 1]islocated near prolonged thunderstorms and lightning. Hence, it is partic- the Lantau Island where airflow disturbances are generated ularly essential to provide accurate forecast of the timing and due to the complex mountainous orography. Mountains with severity of the convective weather to the aviation users such as summits of close to 1000 m above ground level (AGL) and the Air Traffic Control to assure ecffi iency and safety of flights valleys of altitudes at around 400 m are found about 4 km and airfield operations, as well as to reduce flight delays and to the southeast of HKIA. Terrain-disrupted airflow could diversions, and to maximize capacity and optimize the flow occuroverand around HKIA when theprevailingwinds of air traffic within the HKFIR. arefromeasttosouthwest,inparticularwhenstrong-east- Currently in HKO, the mesoscale NWP model system, to-southeasterly winds blow over the airport in the spring namely, the AIR/NHM [2] using the Nonhydrostatic Model under a stable boundary layer. Also, a sea-breeze circulation is of the Japan Meteorological Agency (JMA-NHM [3]), pro- commonly formed during the late morning due to insolation vides hourly updated model forecasts with horizontal resolu- under synoptic weather patterns for formation of light to tion at 2 km (referred as the 2km NHM hereaer). ft Observa- moderate north-to-northwesterly winds. Wind convergence tions from mesoscale observation networks such as the auto- and disturbance result in low-level windshear and turbulence matic weather stations over Hong Kong and Guangdong and [1]. ground-based remote sensing data such as radar, wind pro- In summer, signicfi ant convection systems or organized filer, and GPS total precipitable water vapour are assimilated thunderstorms affecting HKIA and the Hong Kong Flight in the 3-dimensional variational data assimilation (3DVAR) Information Region (HKFIR) (Figure 1)leadtodisruptionin system. Radar Doppler velocity and retrieved wind data using 2 Advances in Meteorology 25N 20N 200 400 600 800 1000 110E 115E 0 200 400 600 800 Figure 1: Boundary of HKFIR and spatial coverage of PRD-AVM and HKA-AVM. mosaic of radars in Hong Kong, Shenzhen, and Guangzhou (PRD) and the HK Airport (HKA) areas. For brevity, they are are also assimilated in 2 km NHM to improve the short-term referred as PRD-AVM and HKA-AVM. e Th spatial coverage prediction of signicfi ant convection [ 4]. Whileingeneral of PRD-AVM (HKA-AVM) is about 350 km (50 km) in both the2km NHMcan providesomeusefulguidanceonthe east-west and north-south directions (Figure 1). The AVM development of mesoscale weather features, it is inadequate will be operated in hourly update basis: the initial condition to resolve the localized effects due to terrain over the Lantau of each hourly run of PRD-AVM is derived from the forecasts Island and small scale land/sea contrast around the airport. of 2 km NHM with a boundary update frequency of 1 hour. eTh refore, a n fi e-scale modeling system is needed to provide One-waynestingisadoptedinwhichtheinitialandboundary improved guidance on the formation of land/sea-breeze conditions of HKA-AVM are obtained from the forecasts of PRD-AVM. The forecast ranges of PRD-AVM and HKA- effects and other mesoscale phenomena. AVM are 9 hours. In this paper, the development of the Aviation Model (AVM) system based on the Weather Research and Forecast- For a better representation of near-surface weather con- ditions and boundary layer characteristics, about 15 vertical ing(WRF) model[5]isdiscussed. Section 2 introduces the general setup of AVM. Case studies of windshear are pre- levels of data within 1000 m AGL of the model terrain sented in Section 3 including description of impact of model are generated from 2 km NHM forecasts to produce the initial and boundary conditions of PRD-AVM. The orog- tuning. Performance of AVM in forecasting of signicfi ant convection and impact of cloud microphysics schemes are raphy used in both PRD-AVM and HKA-AVM is derived from the Shuttle Radar Topography Mission (SRTM) dataset described in Section 4. Concluding remarks, including the (http://srtm.usgs.gov/) in 3 arc-second of horizontal resolu- current limitations and possible development areas of AVM, are presented in Section 5. tion (approximately at 90 m) in order to resolve spatial varia- tion of the terrain height over Lantau and land-sea contrast in HKIA. 2. Design of the Aviation Model (AVM) System 2.1. General Model Setup. AVM is based on the WRF-ARW 2.2. Model Physical Processes. In WRF-ARW, a number of (Advanced Research WRF)—the Eulerian mass-coordinate optionsare availableineachofthe modelphysicalprocesses dynamical core. The initial setup of AVM, as presented in to perform numerical simulations of atmospheric processes this paper, is based on version 3.2.1 (see Section 4). The at different scales. For instance, one of the settings to run whole AVM system consists of two domains with horizontal WRF-ARW for mesoscale weather simulation (and regional resolutions at 600 m and 200 m covering the Pearl River Delta climateruns, see[6]) is based on the following: (NCAR Advances in Meteorology 3 Community Atmosphere Model) CAM or (Rapid Radiation the runway (Figure 2(b)) during 12-13 HKT (04-05 UTC; Transfer Model) RRTM for longwave and shortwave radia- HKT = UTC + 8 hours). Significant windshear resulting in tion, Mellor-Yamada-Janjic (MYJ) planetary boundary layer a headwind gain of 15–20 knots was encountered by more (PBL), and surface layer process based on similarity theory than 10 aircraft landing from southwest over the north adopted in Eta model. In MYJ scheme, which is a nonlocal runway (07LA corridor). Figure 3(a) shows the forecast from PBL parameterization, Turbulence Kinetic Energy (TKE) is theroutine 2kmNHM.Thebacking of lighttomoderate a prognostic variable, and cloud mixing effect is included to easterly winds to west-to-northwesterly winds over the account for effects of cloud liquid water and cloud ice. The western adjacent waters was captured by the model in the MYJ scheme was applied for a case study of terrain-induced 4-hour model forecast. However, the wind convergence was windshear using a previous version of WRF-ARW (version off from the western end of HKIA. Figure 3(b) shows the 2.2) with horizontal resolution up to 200 m. eTh simulated forecast from HKA-AVM. It couldbeseenthatthe AVMrun wind pattern was comparable to winds derived from LIDAR demonstrated some improvement in the forecast location observations [1]. of the wind convergence. eTh n fi er-resolution model with In the rfi st configuration of AVM using WRF-ARW ver- a more realistic representation of terrain and the land-sea sion 3.2.1, two-dimensional deformation (“km opt” option contrast improved the forecast location of wind convergence. in WRF) is chosen in order to provide consistent treatment It will be useful to provide advance alert to pilots and with the selected planetary boundary layer process, as well aviation users if the model can provide indications of abrupt as for estimate of computation resources required for real- headwind changes upon landing (or take-off) or occurrence time runs in hourly update basis. However, it will be shown of significant windshear situation. Figure 4 shows the sim- in the following section that while the choice of options ulated headwind profile along glide path from HKA-AVM in the previous paragraph can generally produce features (green line) for a selected aircra.ft eTh numerical simulation of mesoscale circulation leading to a sea-breeze-induced was also repeated using NHM with the same grid size windshear, they are yet inadequate to predict ne fi -scale (200 m) to generate the headwind profile for comparison. variation of temperature and wind in forecasting localized The headwind profiles are produced using the direct model wind convergence over HKIA, where the terrain eeff ct of outputs of the 3-dimensional wind components and projected Lantau may contribute to the formation. Thus, more recent along the glide path of aircraft. eTh corresponding aircraft physical parameterization schemes for shortwave and long- headwind profile and the altitudes of the glide path from wave radiation processes, PBL, land surface model, near- 04:43:50 to 04:45:20 UTC are shown by the red line and surface physics, and more advanced numerical procedures to purple line, respectively. It should be noted that the forecast compute diffusion, heat u fl x, and moisture and momentum profiles from HKA-AVM and NHM are taken from the u fl xes in WRF-ARW have been applied to investigate their snapshots of respective model prediction at 05:00 UTC impacts on the model simulation. (T +4hforecast).Theheadwindprofiles amongthe model Given that the AVM is congfi ured at subkilometre reso- forecasts and actual time trace show good agreement with lution, the cumulus parameterization scheme (“cu physics” each other. The two model forecasts indicate an earlier change option) is turned off in both PRD-AVM and HKA-AVM. of headwind to tailwind (04:44:15 UTC to 04:44:40 UTC) Explicit cloud microphysics using 5-class cloud microphysics than the actual at 04:44:50 UTC. eTh differences between (WSM5) was initially chosen in AVM using WRF-ARW the results of HKA-AVM and NHM are small in general, version 3.2.1 as it has been widely adopted for numerical butthe HKA-AVMisabletobettercapture theabrupt simulation of convective systems. The specific humidities of jump in the magnitudes of headwind near the touch-down water vapour, cloud liquid water, rain water, cloud ice, and point (i.e., flight altitude near zero). Both the ne-r fi esolution snow are prognostic variables in the model convective pro- model simulations demonstrate some potential to capture cesses and grid scale precipitation. Initial experiments using windshear and simulate the eeff cts on headwind changes themoreadvanceddouble-moment cloudmicrophysicsin experienced by the aircra.ft a newer version of WRF-ARW (3.4.1) will be discussed in Section 4. 3.2. Impact of Model Physics in Forecast of Localized Wind Convergence (June 25, 2011). The development of local wind 3. Model Case Studies for Windshear due to convergence over HKIA is oeft n complicated by the eeff ct of surrounding terrain that provides localized sensible heat Sea Breeze and Localised Wind Convergence exchange through near-surface processes and blocking of 3.1. Sea-Breeze-Induced Windshear (February 25, 2011). low-level flow to form microscale circulation patterns. An Windshear events occur mostly under nonrainy weather example can be seen from another windshear event that condition in HKIA. Under the insolation, sea-breeze circula- occurred in the early afternoon of June 25, 2011. Throughout tion forms over the Pearl River Estuary and in the vicinity the morning of June 25, a moderate southwesterly wind of the airport during the late morning or early afternoon. prevailed over Lantau (Figures 5(a) and 5(b)). The winds For example, on February 25, 2011, a ridge of high pressure over the eastern adjacent waters of HKIA gradually veered to over eastern China brought light to moderate easterly easterly or southeasterly during 12 to 13 HKT, and windshear winds over HK (Figure 2(a)) and the coastal region of (Figures 5(c) and 5(d)) was encountered by a number of Guangdong. A sea breeze was established where the wind aircrasft descending from northeast and landing at the north convergence was located just over the western end of runway (25RA corridor). 4 Advances in Meteorology 09:00 HKT 12:50 HKT HKIA HKIA (a) (b) Figure 2: AWS wind observations at (a) 09:00 HKT and (b) 12:50 HKT on February 25, 2011. Area of right-side figure is marked in dashed line in (a). HKIA HKIA (a) (b) Figure 3: (a) Four-hour forecast of surface wind and sea-level pressure (contour line) from 2 km NHM run at 0100 UTC February 25, 2011. (b)Four-hour wind forecast from HKA-AVM. Area of (b)ismarkedindashedlinein(a).Locations AWSstationsshown in Figure 2 are marked as dots in (b). T+ 4 h forecast from 01 UTC 2011-02-25 Simulated headwind profiles from AVM (green) and NHM (blue) with horizontal resolution of 200 m Actual headwind profile −5 −10 −15 −20 0 43:50 44:00 44:10 44:20 44:30 44:40 44:50 45:00 45:10 45:20 45:30 Time 2011-02-25 04:mm:ss Actual headwind NHM headwind WRF-ARW headwind Aircraft height (right y axis) Figure 4: Simulated headwind profile from HKA-AVM forecast (green) along the glide path (purple line) and actual flight data (red). For comparison, blue line shows the headwind profile from NHM running at the same horizontal resolution (grid size at 200 m). Headwind speed (knots) Aircraft height (m) Advances in Meteorology 5 09:30 HKT 10:30 HKT (a) (b) 13:30 HKT 12:30 HKT (c) (d) Figure 5: AWS wind observations on 09:30, 10:30, 12:30, and 13:30 HKT on June 25, 2011. The veering of the winds from southwesterly in the radiation schemes based on the RRTMG (a new version morning to westerly over the region to the northeast of HKIA of the Rapid Radiative Transfer Model for more efficient wascapturedinthe 2kmNHM forecasts(notshown). Using and accurate computation of radiation process) generally HKA-AVM (Figure 6(a)), the veering of winds became more improved the forecast temperature over HKIA and the Lantau pronounced due to better model resolution of small scale Island, possibly due to a better treatment of cloud overlap features of wind flow. However, the wind convergence over eeff ctand inclusionofmultiplebands of shortwaveand theeastern endofthe runwayscould notbereproduced. longwave radiation. The use of a different land surface scheme To address the problem on the development of small- by switching from the basic 5-layer thermal diffusion model scale wind features, a series of numerical experiments were to a more sophisticated land surface model (LSM) such as the attempted by inspecting effects on surface wind and tem- NCEP Noah LSM and RUC LSM resulted in a slight positive perature forecasts over HKIA based on all available options impact on the model temperature forecast (not shown). eTh of near-surface physics, boundary layer process, and land sensitivity tests were also performed using different model surface model options. In summary, some selected combi- data on the soil temperature and soil moisture. eTh y included nation of these model physical processes could improve the the NCEP global model (GFS) forecast of horizontal resolu- model forecast of the timing of changes in wind direction or tion at 0.5 degree in latitude/longitude and a higher resolution spatial and temporal variation of temperatures over HKIA. data products from the ECMWF model forecast at 0.125 For instance, the use of recent longwave and shortwave degree in latitude/longitude. However, the resulting changes 6 Advances in Meteorology 18 22 26 30 34 38 18 22 26 30 34 38 ∘ ∘ Temperature ( C) Temperature ( C) (a) (b) Figure 6: Forecast wind and temperature from HKA-AVM from (a) original and (b) new settings of scheme. Refer to Section 3.2 for details. The locations of AWS over HKIA and the Lantau Island are shown in red dots. in forecast temperature and winds over HKIA were relatively 4. Significant Convection Forecasts small. Further study will be conducted to investigate the Using AVM impact of the model data under different weather conditions. 4.1. Widespread Quasi-Stationary Significant Convection over The impact on wind prediction was found to be rather HKFIR (18 September 2011). PRD-AVM was applied to simu- marginal using the existing or newly available PBL schemes in WRF-ARW (e.g., Mellor-Yamada-Nakanishi-Niino third- late a significant convection event that occurred on Septem- ber 18, 2011. During the morning, convection clusters devel- order turbulence closuremodel as adoptedinNHM)with oped over the coastal waters under the presence of a broad corresponding compatible near-surface scheme to diagnose low pressure areas and convergence of easterly airstream thesurface wind,temperature,and moisture.That limitation alongthecoastofGuangdong.Theradarechoesbecamemore is possibly attributed to the insufficient representation of turbulent eddies such that the PBL schemes in WRF (and organizedand formedintobroadbands of signicfi ant convec- tion. eTh y were quasi-stationary and blocked the passage of other mesoscale NWP models) are suitable only for model aircraft from the south in the HKFIR. A series of images horizontal resolution up to one or a few kilometers. In recent versions of WRF, a Large-Eddy Simulation showing radar CAPPI reflectivity at 3 km altitudes is given in Figure 7. (LES) model has been incorporated in the code. LES has been The PRD-AVM forecast was initiated for 2300 UTC widely applied to study turbulence as well as its coherent structures and statistics within PBL. Using this new option September 17, 2011. Its initial and boundary conditions were basedonthe forecastsof2km NHMrun at 2200UTC to (with 3-dimensional diffusion term, 1.5 order TKE closure, ensure a better input to AVM in terms of sufficient spin up of RRTMG shortwave and longwave, and the default MM5 sim- moisture and other model forecast elements. The simulated ilaritynear-surfaceprocess), thesameforecastofwinddistri- maximum reflectivity maps from PRD-AVM in 0000–0500 bution over HKIA is depicted in Figure 6(b),which showsa quite encouraging result on the model simulated windshear UTC are given in Figure 7. eTh hourly accumulated rainfall forecasts from 2 km NHM are also shown for comparison. feature in terms of more realistic location and timing of local- eTh WSM-5schemeemployedinPRD-AVM produced a ized wind convergence and associated microscale anticyclone off the HKIA. Moreover, it should be noted that the simulated reasonable trend of convection development. eTh “simulated radar echoes” were found to intensify and organize into a wind pattern was quite sensitive to the choice of near-surface signicfi ant convection cluster in 3 to 6 hours ahead. However, process (e.g., changing from MM5 similarity to Eta model thelocationofthe wholeforecastconvectionsystemwas type), as well as the method to calculate the momentum foundtolocatemoresouthwest to theactualshown in the u fl x, heat u fl x, and moisture ux fl in the WRF-LES. eTh forecast speed of southwesterly wind could be strengthened radar imagery. Model simulated radar reflectivity of 50 dBZ (or above) together with a higher amount of precipitation using these choices such that the formations of local-scale was obtained when compared to the forecasts from the 2 km anticyclone and wind convergence were suppressed. Advances in Meteorology 7 PRD-AVM 2 km NHM forecast run at forecast run at 0700 HKT 0700 HKT 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 0.5 1 2 5 10 15 20 25 30 40 50 70 100 Max reflectivity (dBZ) (mm) (a) (b) (c) Figure 7: Radar CAPPI reflectivity on 3 km at 07:00, 09:00, 11:00, and 13:00 HKT September 18, 2011. 2 km NHM hourly accumulated rainfall forecast with surface winds in middle column. Simulated radar reflectivity and surface winds from PRD-AVM are shown in the right column; the figure areas are marked in dashed box in 2 km NHM charts. NHM (not shown). In this case, 2 km NHM showed a more 4.2. Initiation of Isolated und Th erstorms (May 20, 2012)— realistic result in terms of a broader coverage of rain bands. It Comparison with Dieff rent Advanced Cloud Microphysics was possibly due to a larger size of domain in 2 km NHM that Parameterizations. Summertime heat thunderstorms, in par- better represented the low-pressure area and the convection ticular associated with the sea-breeze convergence, oeft n system over the coastal waters. develop rapidly into organized rainstorms. Accurate forecasts 8 Advances in Meteorology Figure 8: Radar reflectivity image from 12:00 to 14:30 HKT on May 20, 2012. under Th storm developments over inland Guangdong and the urban areas of HK are labeled in A and B. fortheir initiation anddevelopment arechallenging dueto inclusion of mix-phased hydrometeors and a deeper layer the uncertainties of mesoscale processes. For instance, under of melting and freezing processes. Using the more complex the condition of weak pressure gradient, generally light winds WSM-6, which includes the specific humidity of graupel in and a conditionally unstable environment in the morning the governing equations, a more localized distribution of on May 20, 2012, development of wind convergence over peak rainfall was simulated owing to the fall out of graupel the local territory and coastal areas occurred in the early as well as rain water. However, the scheme showed a bias for aer ft noon leading to rapid development of thunderstorm overpredicting the coverage of light rain. cells. A sequence of radar CAPPI reflectivity images during Recently, more sophisticated double-moment cloud 12 to 14 HKT is shown in Figure 8. With more widespread microphysics schemes have been implemented in WRF- development found over the northern part of the New Terri- ARW that can predict both the mixing ratio and number tories, heavy rain occurred and lasted for a few hours till the concentration of hydrometeor species, such as cloud droplets, evening. cloud ice, rain and snow, to improve the representation of Forecast charts of simulated radar reflectivity with surface cloud processes in forecast of convective weather phenome- winds from the PRD-AVM run initialized at 0800 HKT (0000 na. Three numerical experiments were performed using the UTC) May 20, 2012, are shown in Figure 9.TheWSM-5 cloud WRF-ARW 3.4 using the WDM-6 (WRF 6-class double microphysics scheme was used. Under a rather weak synoptic moment scheme [8]), Morrison double-moment scheme [9] forcing environment, convergence of winds developed due to and Milbrandt-Yau double-moment scheme [10]. Figure 10 theseabreezeoverthecoastalareasandsurfaceheatingunder shows the forecast reflectivity maps in 4–6 hours of forecasts insolation over the inland regions. eTh initiation of convec- where the local sea-breeze-induced convergence and thun- tive cells was forecast by PRD-AVM over these convergence derstorms were initiated. Generally, all the three schemes zones where instabilities were also found. eTh development were capable to predict the development of the localized of intense thunderstorms over inland Guangdong (area A) simulated radar reflectivity with peak intensity at 45 dBZ andHK(area B) wascapturedreasonablyinterms of or above. They have differences in spatial distribution and location and trend. In WRF-ARW, several bulk microphysics locationsthoughthe surfacewindfieldforecastinterms schemes (“mp physics” option)havebeenmadeavailable of its wind convergence was quite similar (not shown). since the release of version 2 for predicting the specific For instance, the forecasts from WDM-6 showed the most humidity of cloud hydrometeors including both water and intense and localized organization of high reflectivity echoes ice components. In particular, Hong et al. [7]comparedthe or storm cells compared to Morrison and Milbrandt-Yau performance of WRF single moment (WSM) 3-class, 5-class, schemes. The development of the thunderstorms was found and 6-class to simulate a mesoscale convective storm. It was to be about 2 hours earlier than the other two schemes. It found that three schemes generally demonstrated similar was discussed in [7] that the WDM-6 was able to provide performance in terms of rainfall distribution and WSM-5 an improved treatment of the variability in cloud and rain tended to produce more realistic rainfall intensity due to number concentrations in order to ameliorate the generation Advances in Meteorology 9 T+ 4:00 12:00 HKT T+ 4:30 12:30 HKT T+ 5:00 13:00 HKT (a) (b) (c) T+ 5:30 13:30 HKT T+ 6:00 14:00 HKT T+ 6:30 14:30 HKT 5 5 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 10 15 20 25 30 35 40 45 50 55 60 65 70 75 10 15 20 25 30 35 40 45 50 55 60 65 70 75 Reflectivity (dBZ) Reflectivity (dBZ) Reflectivity (dBZ) (d) (e) (f) Figure 9: Simulated radar reflectivity from PRD-AVM run at 0000 UTC (0800 HKT) May 20, 2012. eTh development of convective cells in A and B by the model forecasts at 14:00 HKT is indicated for comparison. of widespread light rain as found in its single moment version has been successfully implemented for trial experiments to (WSM-6). simulate windshear events, sea-breeze convergence, and sig- In case for Morrison and Milbrandt-Yau schemes, they nificant convections. Sensitivity experiments using different have similar forecasts in the distribution of simulated reflec- model physical processes have been attempted in order to tivity. In terms of timing and location of development, the obtain a feasible setup and configuration of AVM in forecast- Milbrandt-Yau scheme predicted both the thunderstorms ing the local or microscale weather phenomena near HKIA. (≥35 dBZ) at A and B to occur during 13-14 HKT that were They show encouraging results and thus favorable for routine more coherent with the radar images (Figure 8). Additionally, runs in rapid-update cycle configuration subject to available the Milbrandt-Yau scheme better resembled the convection computing resources. However, there are yet a number of over the coastal waters in T + 4 hour of forecast that were limitations to solve as elaborated in the following paragraphs. also present in WSM-5 (Figure 9). More studies will be eTh 200m HKA-AVMcould reproducemorerealistic performed in future to understand the mechanisms and the wind flow and forecasts for HKIA compared to the current characteristics of the three double moment schemes. operational mesoscale NWP system (2 km NHM). Though a number of choices of physical parameterization process for PBL, near-surface process, land surface models, and cloud 5. Concluding Remarks microphysics from WRF-ARW have been studied or adopted In this paper, the development of AVM is discussed. AVM in other high-resolution NWP studies, they may not be fully is designedtoenhance thecapabilityofNWP to supportthe applicable for the fine-resolution simulation with horizontal aviation forecast in the Hong Kong Observatory. eTh model resolution at 200 m. For instance the PBL process, land 10 Advances in Meteorology T+ 4:00 12:00 HKT T+ 5:00 13:00 HKT T+ 6:00 14:00 HKT WDM-6 (a) Morrison (b) Milbrandt-Yau 5 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 10 15 20 25 30 35 40 45 50 55 60 65 70 75 Reflectivity (dBZ) Reflectivity (dBZ) Reflectivity (dBZ) (c) Figure 10: Simulated radar reflectivity from PRD-AVM run at 0000 UTC (0800 HKT) May 20, 2012 using dieff rent double-moment cloud microphysics scheme: (a) WDM-6, (b) Morrison, and (c) Milbrandt-Yau. Development of convective cells near Guangzhou (A) and over Hong Kong (B) in the model forecasts is shown. Advances in Meteorology 11 surface model, and even the correct specicfi ation of the land Hong Kong observatory,” in Proceedings of the 3rd Interna- tional Workshop on Prevention and Mitigation of Meteorolog- surfacetypeare importanttoaccuratelysimulatethe nfi e- ical Disasters in Southeast Asia,Beppu,Japan,March 2010, scalewindfieldinHKIA. Secondly,the domain size of PRD- http://www.hko.gov.hk/publica/reprint/r882.pdf. AVM may be too limited to forecast the development of the [3] K. Saito, T. Fujita, Y. Yamada et al., “eTh operational JMA convective systems near the boundary, where the boundary nonhydrostatic mesoscale model,” Monthly Weather Review, data areprovidedfrom2km NHM. To mitigate theincon- vol. 134, no. 4, pp. 1266–1298, 2006. sistency, we may consider extending the domain of PRD- [4] W.K.Wong, M. K. Or,P.W.Chan, andC.M.Cheng,“Impact AVM subject to computation resources or including another of radar retrieval winds on data assimilation and forecast of outermodel with coarserresolution(e.g.,gridspacing at 1 a mesoscale convective storm using non-hydrostatic model,” or 2 km) using two-way nesting for a more consistent update in Proceedings of the 14th Conference on Mesoscale Process, of forcing in the model boundary. In fact, two-way nesting American Meteorological Society, Los Angeles, Calif, USA, August 2011. was attempted in the initial setup of WRF but found to be [5] W.C.Skamarock,J.B.Klemp,J.Dudhiaetal.,“Adescription numerically unstable occasionally. With the newer version of the advanced research WRF version 2,” NCAR Tech Notes- of code available (version 3.4), the two-way nesting would 468+STR, 2005. be explored again to check the code robustness and possible [6] X. Z. Liang, M. Xu, X. Yuan et al., “Regional climate-weather gain in computation speed over the existing one-way nesting research and forecasting model,” Bulletin of the American method (a.k.a. nest-down, “ndown” approach). Meteorological Society, vol. 93, pp. 1363–1387, 2012. The whole AVM system is currently initialized by interpo- [7] S.Y.Hong, K. S. Lim, Y. H. Leeetal.,“Evaluation of theWRF lation of the 2 km NHM forecasts. ou Th gh the “3 : 1” nesting double-moment 6-class microphysics scheme for precipitating ratio (2000 m in NHM versus 600 m in PRD-AVM) was convection,” Advances in Meteorology,vol.2010, ArticleID found to be a viable choice from the results of simulation 707253, 10 pages, 2010. experiments, it is no doubt that the model initialization, [8] K.S.S.Lim andS.Y.Hong, “Development of an eeff ctive dou- especially some spin-up problems linked to the cloud micro- ble-moment cloud microphysics scheme with prognostic cloud physics could be alleviated through an implementation of condensation nuclei (CCN) for weather and climate models,” Monthly Weather Review,vol.138,no. 5, pp.1587–1612,2010. data assimilation (DA) in future. eTh implementation of AVM-DA system is not trivial, since the current configura- [9] H. Morrison, G. Thompson, and V. Tatarskii, “Impact of cloud microphysics on the development of trailing stratiform precip- tion of data assimilation using the variational methods itation in a simulated squall line: comparison of one- and two- (3DVAR or 4DVAR) requires a well-tuned background cova- moment schemes,” Monthly Weather Review,vol.137,no. 3, pp. riance model. The correlation of the analysis control variables 991–1007, 2009. is specified based on physical balances in synoptic and [10] J. A. Milbrandt and M. K. Yau, “A multimoment bulk micro- subsynoptic scales of processes to generate an optimal model physics parameterization. Part I: analysis of the role of the spec- analysis as the initial condition. Additionally, the tangent tral shape parameter,” Journal of the Atmospheric Sciences,vol. linear and adjoint models of WRF may not be sufficient to 62,no. 9, pp.3051–3064,2005. represent the mesoscale and cloud microphysical processes. Thiswould be pursuedinthe future research studiesonthe performance of the variational data assimilation of the WRF- ARW, including the technique and related tuning, if available, for constructing a suitable DA system for PRD-AVM and HKA-AVM. Acknowledgments Part of the research work in this paper was conducted by the second author during his attachment to HKO in 2011-2012. aTh nksare also duetoDrs.BetaC.L.Yip andIvy K. Y. Wong for providing technical supports in setting up an experimen- tal version of AVM system using WRF-ARW 3.2.1. References [1] P. W. Chan and T. C. Cheung, “Microscale simulation of terrain- disrupted airflow around the Hong Kong International Airport (HKIA)—comparison of results between numerical models,” in Proceedings of the10thAnnual WRFUsers’Workshop,Boulder, Colo, USA, June 2009. [2] W. K. Wong, “Development of operational rapid update non-hydrostatic NWP and data assimilation systems in the International Journal of Journal of Ecology Mining The Scientific Journal of Geochemistry Scientifica World Journal Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 Journal of Earthquakes Paleontology Journal Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 Journal of Petroleum Engineering Submit your manuscripts at http://www.hindawi.com International Journal of Geophysics Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 Advances in Advances in Journal of Advances in International Journal of Meteorology ClimatologyGeology Oceanography Oceanography Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 Applied & Journal of Environmental International Journal of Journal of International Journal of Computational Soil Science Mineralogy Geological Research Atmospheric Sciences Environmental Sciences Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Volume 2014 Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com http://www.hindawi.com Volume 2014 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Advances in Meteorology Hindawi Publishing Corporation

Aviation Model: A Fine-Scale Numerical Weather Prediction System for Aviation Applications at the Hong Kong International Airport

Download PDF
12 pages

Loading next page...
 
/lp/hindawi-publishing-corporation/aviation-model-a-fine-scale-numerical-weather-prediction-system-for-b64uQOFPi1

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2013 Wai-Kin Wong et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN
1687-9309
eISSN
1687-9317
DOI
10.1155/2013/532475
Publisher site
See Article on Publisher Site

Abstract

Hindawi Publishing Corporation Advances in Meteorology Volume 2013, Article ID 532475, 11 pages http://dx.doi.org/10.1155/2013/532475 Research Article Aviation Model: A Fine-Scale Numerical Weather Prediction System for Aviation Applications at the Hong Kong International Airport 1 2 1 Wai-Kin Wong, Cheong-Shing Lau, and Pak-Wai Chan Hong Kong Observatory, Hong Kong Department of Computing, Polytechnic University of Hong Kong, Hong Kong Correspondence should be addressed to Pak-Wai Chan; pwchan@hko.gov.hk Received 29 December 2012; Revised 16 April 2013; Accepted 22 May 2013 Academic Editor: Richard Leaitch Copyright © 2013 Wai-Kin Wong et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The Hong Kong Observatory (HKO) is planning to implement a fine-resolution Numerical Weather Prediction (NWP) model for supporting the aviation weather applications at the Hong Kong International Airport (HKIA). This new NWP model system, called Aviation Model (AVM), is configured at a horizontal grid spacing of 600 m and 200 m. It is based on the WRF-ARW (Advance Research WRF) model that can have sufficient computation efficiency in order to produce hourly updated forecasts up to 9 hours ahead on a future high performance computer system with theoretical peak performance of around 10 TFLOPS. AVM will be nested inside the operational mesoscale NWP model of HKO with horizontal resolution of 2 km. In this paper, initial numerical experiment results in forecast of windshear events due to seabreeze and terrain eeff ct are discussed. eTh simulation of sea-breeze-related windshear is quite successful, and the headwind change observed from flight data could be reproduced in the model forecast. Some impacts of physical processes on generating the fine-scale wind circu lation and development of significant convection are illustrated. The paper also discusses the limitations in the current model setup and proposes methods for the future development of AVM. 1. Introduction air traffic for in-bound and out-bound flights. Moreover, the airel fi d operation will be greatly affected by widespread or The Hong Kong International Airport [ 1]islocated near prolonged thunderstorms and lightning. Hence, it is partic- the Lantau Island where airflow disturbances are generated ularly essential to provide accurate forecast of the timing and due to the complex mountainous orography. Mountains with severity of the convective weather to the aviation users such as summits of close to 1000 m above ground level (AGL) and the Air Traffic Control to assure ecffi iency and safety of flights valleys of altitudes at around 400 m are found about 4 km and airfield operations, as well as to reduce flight delays and to the southeast of HKIA. Terrain-disrupted airflow could diversions, and to maximize capacity and optimize the flow occuroverand around HKIA when theprevailingwinds of air traffic within the HKFIR. arefromeasttosouthwest,inparticularwhenstrong-east- Currently in HKO, the mesoscale NWP model system, to-southeasterly winds blow over the airport in the spring namely, the AIR/NHM [2] using the Nonhydrostatic Model under a stable boundary layer. Also, a sea-breeze circulation is of the Japan Meteorological Agency (JMA-NHM [3]), pro- commonly formed during the late morning due to insolation vides hourly updated model forecasts with horizontal resolu- under synoptic weather patterns for formation of light to tion at 2 km (referred as the 2km NHM hereaer). ft Observa- moderate north-to-northwesterly winds. Wind convergence tions from mesoscale observation networks such as the auto- and disturbance result in low-level windshear and turbulence matic weather stations over Hong Kong and Guangdong and [1]. ground-based remote sensing data such as radar, wind pro- In summer, signicfi ant convection systems or organized filer, and GPS total precipitable water vapour are assimilated thunderstorms affecting HKIA and the Hong Kong Flight in the 3-dimensional variational data assimilation (3DVAR) Information Region (HKFIR) (Figure 1)leadtodisruptionin system. Radar Doppler velocity and retrieved wind data using 2 Advances in Meteorology 25N 20N 200 400 600 800 1000 110E 115E 0 200 400 600 800 Figure 1: Boundary of HKFIR and spatial coverage of PRD-AVM and HKA-AVM. mosaic of radars in Hong Kong, Shenzhen, and Guangzhou (PRD) and the HK Airport (HKA) areas. For brevity, they are are also assimilated in 2 km NHM to improve the short-term referred as PRD-AVM and HKA-AVM. e Th spatial coverage prediction of signicfi ant convection [ 4]. Whileingeneral of PRD-AVM (HKA-AVM) is about 350 km (50 km) in both the2km NHMcan providesomeusefulguidanceonthe east-west and north-south directions (Figure 1). The AVM development of mesoscale weather features, it is inadequate will be operated in hourly update basis: the initial condition to resolve the localized effects due to terrain over the Lantau of each hourly run of PRD-AVM is derived from the forecasts Island and small scale land/sea contrast around the airport. of 2 km NHM with a boundary update frequency of 1 hour. eTh refore, a n fi e-scale modeling system is needed to provide One-waynestingisadoptedinwhichtheinitialandboundary improved guidance on the formation of land/sea-breeze conditions of HKA-AVM are obtained from the forecasts of PRD-AVM. The forecast ranges of PRD-AVM and HKA- effects and other mesoscale phenomena. AVM are 9 hours. In this paper, the development of the Aviation Model (AVM) system based on the Weather Research and Forecast- For a better representation of near-surface weather con- ditions and boundary layer characteristics, about 15 vertical ing(WRF) model[5]isdiscussed. Section 2 introduces the general setup of AVM. Case studies of windshear are pre- levels of data within 1000 m AGL of the model terrain sented in Section 3 including description of impact of model are generated from 2 km NHM forecasts to produce the initial and boundary conditions of PRD-AVM. The orog- tuning. Performance of AVM in forecasting of signicfi ant convection and impact of cloud microphysics schemes are raphy used in both PRD-AVM and HKA-AVM is derived from the Shuttle Radar Topography Mission (SRTM) dataset described in Section 4. Concluding remarks, including the (http://srtm.usgs.gov/) in 3 arc-second of horizontal resolu- current limitations and possible development areas of AVM, are presented in Section 5. tion (approximately at 90 m) in order to resolve spatial varia- tion of the terrain height over Lantau and land-sea contrast in HKIA. 2. Design of the Aviation Model (AVM) System 2.1. General Model Setup. AVM is based on the WRF-ARW 2.2. Model Physical Processes. In WRF-ARW, a number of (Advanced Research WRF)—the Eulerian mass-coordinate optionsare availableineachofthe modelphysicalprocesses dynamical core. The initial setup of AVM, as presented in to perform numerical simulations of atmospheric processes this paper, is based on version 3.2.1 (see Section 4). The at different scales. For instance, one of the settings to run whole AVM system consists of two domains with horizontal WRF-ARW for mesoscale weather simulation (and regional resolutions at 600 m and 200 m covering the Pearl River Delta climateruns, see[6]) is based on the following: (NCAR Advances in Meteorology 3 Community Atmosphere Model) CAM or (Rapid Radiation the runway (Figure 2(b)) during 12-13 HKT (04-05 UTC; Transfer Model) RRTM for longwave and shortwave radia- HKT = UTC + 8 hours). Significant windshear resulting in tion, Mellor-Yamada-Janjic (MYJ) planetary boundary layer a headwind gain of 15–20 knots was encountered by more (PBL), and surface layer process based on similarity theory than 10 aircraft landing from southwest over the north adopted in Eta model. In MYJ scheme, which is a nonlocal runway (07LA corridor). Figure 3(a) shows the forecast from PBL parameterization, Turbulence Kinetic Energy (TKE) is theroutine 2kmNHM.Thebacking of lighttomoderate a prognostic variable, and cloud mixing effect is included to easterly winds to west-to-northwesterly winds over the account for effects of cloud liquid water and cloud ice. The western adjacent waters was captured by the model in the MYJ scheme was applied for a case study of terrain-induced 4-hour model forecast. However, the wind convergence was windshear using a previous version of WRF-ARW (version off from the western end of HKIA. Figure 3(b) shows the 2.2) with horizontal resolution up to 200 m. eTh simulated forecast from HKA-AVM. It couldbeseenthatthe AVMrun wind pattern was comparable to winds derived from LIDAR demonstrated some improvement in the forecast location observations [1]. of the wind convergence. eTh n fi er-resolution model with In the rfi st configuration of AVM using WRF-ARW ver- a more realistic representation of terrain and the land-sea sion 3.2.1, two-dimensional deformation (“km opt” option contrast improved the forecast location of wind convergence. in WRF) is chosen in order to provide consistent treatment It will be useful to provide advance alert to pilots and with the selected planetary boundary layer process, as well aviation users if the model can provide indications of abrupt as for estimate of computation resources required for real- headwind changes upon landing (or take-off) or occurrence time runs in hourly update basis. However, it will be shown of significant windshear situation. Figure 4 shows the sim- in the following section that while the choice of options ulated headwind profile along glide path from HKA-AVM in the previous paragraph can generally produce features (green line) for a selected aircra.ft eTh numerical simulation of mesoscale circulation leading to a sea-breeze-induced was also repeated using NHM with the same grid size windshear, they are yet inadequate to predict ne fi -scale (200 m) to generate the headwind profile for comparison. variation of temperature and wind in forecasting localized The headwind profiles are produced using the direct model wind convergence over HKIA, where the terrain eeff ct of outputs of the 3-dimensional wind components and projected Lantau may contribute to the formation. Thus, more recent along the glide path of aircraft. eTh corresponding aircraft physical parameterization schemes for shortwave and long- headwind profile and the altitudes of the glide path from wave radiation processes, PBL, land surface model, near- 04:43:50 to 04:45:20 UTC are shown by the red line and surface physics, and more advanced numerical procedures to purple line, respectively. It should be noted that the forecast compute diffusion, heat u fl x, and moisture and momentum profiles from HKA-AVM and NHM are taken from the u fl xes in WRF-ARW have been applied to investigate their snapshots of respective model prediction at 05:00 UTC impacts on the model simulation. (T +4hforecast).Theheadwindprofiles amongthe model Given that the AVM is congfi ured at subkilometre reso- forecasts and actual time trace show good agreement with lution, the cumulus parameterization scheme (“cu physics” each other. The two model forecasts indicate an earlier change option) is turned off in both PRD-AVM and HKA-AVM. of headwind to tailwind (04:44:15 UTC to 04:44:40 UTC) Explicit cloud microphysics using 5-class cloud microphysics than the actual at 04:44:50 UTC. eTh differences between (WSM5) was initially chosen in AVM using WRF-ARW the results of HKA-AVM and NHM are small in general, version 3.2.1 as it has been widely adopted for numerical butthe HKA-AVMisabletobettercapture theabrupt simulation of convective systems. The specific humidities of jump in the magnitudes of headwind near the touch-down water vapour, cloud liquid water, rain water, cloud ice, and point (i.e., flight altitude near zero). Both the ne-r fi esolution snow are prognostic variables in the model convective pro- model simulations demonstrate some potential to capture cesses and grid scale precipitation. Initial experiments using windshear and simulate the eeff cts on headwind changes themoreadvanceddouble-moment cloudmicrophysicsin experienced by the aircra.ft a newer version of WRF-ARW (3.4.1) will be discussed in Section 4. 3.2. Impact of Model Physics in Forecast of Localized Wind Convergence (June 25, 2011). The development of local wind 3. Model Case Studies for Windshear due to convergence over HKIA is oeft n complicated by the eeff ct of surrounding terrain that provides localized sensible heat Sea Breeze and Localised Wind Convergence exchange through near-surface processes and blocking of 3.1. Sea-Breeze-Induced Windshear (February 25, 2011). low-level flow to form microscale circulation patterns. An Windshear events occur mostly under nonrainy weather example can be seen from another windshear event that condition in HKIA. Under the insolation, sea-breeze circula- occurred in the early afternoon of June 25, 2011. Throughout tion forms over the Pearl River Estuary and in the vicinity the morning of June 25, a moderate southwesterly wind of the airport during the late morning or early afternoon. prevailed over Lantau (Figures 5(a) and 5(b)). The winds For example, on February 25, 2011, a ridge of high pressure over the eastern adjacent waters of HKIA gradually veered to over eastern China brought light to moderate easterly easterly or southeasterly during 12 to 13 HKT, and windshear winds over HK (Figure 2(a)) and the coastal region of (Figures 5(c) and 5(d)) was encountered by a number of Guangdong. A sea breeze was established where the wind aircrasft descending from northeast and landing at the north convergence was located just over the western end of runway (25RA corridor). 4 Advances in Meteorology 09:00 HKT 12:50 HKT HKIA HKIA (a) (b) Figure 2: AWS wind observations at (a) 09:00 HKT and (b) 12:50 HKT on February 25, 2011. Area of right-side figure is marked in dashed line in (a). HKIA HKIA (a) (b) Figure 3: (a) Four-hour forecast of surface wind and sea-level pressure (contour line) from 2 km NHM run at 0100 UTC February 25, 2011. (b)Four-hour wind forecast from HKA-AVM. Area of (b)ismarkedindashedlinein(a).Locations AWSstationsshown in Figure 2 are marked as dots in (b). T+ 4 h forecast from 01 UTC 2011-02-25 Simulated headwind profiles from AVM (green) and NHM (blue) with horizontal resolution of 200 m Actual headwind profile −5 −10 −15 −20 0 43:50 44:00 44:10 44:20 44:30 44:40 44:50 45:00 45:10 45:20 45:30 Time 2011-02-25 04:mm:ss Actual headwind NHM headwind WRF-ARW headwind Aircraft height (right y axis) Figure 4: Simulated headwind profile from HKA-AVM forecast (green) along the glide path (purple line) and actual flight data (red). For comparison, blue line shows the headwind profile from NHM running at the same horizontal resolution (grid size at 200 m). Headwind speed (knots) Aircraft height (m) Advances in Meteorology 5 09:30 HKT 10:30 HKT (a) (b) 13:30 HKT 12:30 HKT (c) (d) Figure 5: AWS wind observations on 09:30, 10:30, 12:30, and 13:30 HKT on June 25, 2011. The veering of the winds from southwesterly in the radiation schemes based on the RRTMG (a new version morning to westerly over the region to the northeast of HKIA of the Rapid Radiative Transfer Model for more efficient wascapturedinthe 2kmNHM forecasts(notshown). Using and accurate computation of radiation process) generally HKA-AVM (Figure 6(a)), the veering of winds became more improved the forecast temperature over HKIA and the Lantau pronounced due to better model resolution of small scale Island, possibly due to a better treatment of cloud overlap features of wind flow. However, the wind convergence over eeff ctand inclusionofmultiplebands of shortwaveand theeastern endofthe runwayscould notbereproduced. longwave radiation. The use of a different land surface scheme To address the problem on the development of small- by switching from the basic 5-layer thermal diffusion model scale wind features, a series of numerical experiments were to a more sophisticated land surface model (LSM) such as the attempted by inspecting effects on surface wind and tem- NCEP Noah LSM and RUC LSM resulted in a slight positive perature forecasts over HKIA based on all available options impact on the model temperature forecast (not shown). eTh of near-surface physics, boundary layer process, and land sensitivity tests were also performed using different model surface model options. In summary, some selected combi- data on the soil temperature and soil moisture. eTh y included nation of these model physical processes could improve the the NCEP global model (GFS) forecast of horizontal resolu- model forecast of the timing of changes in wind direction or tion at 0.5 degree in latitude/longitude and a higher resolution spatial and temporal variation of temperatures over HKIA. data products from the ECMWF model forecast at 0.125 For instance, the use of recent longwave and shortwave degree in latitude/longitude. However, the resulting changes 6 Advances in Meteorology 18 22 26 30 34 38 18 22 26 30 34 38 ∘ ∘ Temperature ( C) Temperature ( C) (a) (b) Figure 6: Forecast wind and temperature from HKA-AVM from (a) original and (b) new settings of scheme. Refer to Section 3.2 for details. The locations of AWS over HKIA and the Lantau Island are shown in red dots. in forecast temperature and winds over HKIA were relatively 4. Significant Convection Forecasts small. Further study will be conducted to investigate the Using AVM impact of the model data under different weather conditions. 4.1. Widespread Quasi-Stationary Significant Convection over The impact on wind prediction was found to be rather HKFIR (18 September 2011). PRD-AVM was applied to simu- marginal using the existing or newly available PBL schemes in WRF-ARW (e.g., Mellor-Yamada-Nakanishi-Niino third- late a significant convection event that occurred on Septem- ber 18, 2011. During the morning, convection clusters devel- order turbulence closuremodel as adoptedinNHM)with oped over the coastal waters under the presence of a broad corresponding compatible near-surface scheme to diagnose low pressure areas and convergence of easterly airstream thesurface wind,temperature,and moisture.That limitation alongthecoastofGuangdong.Theradarechoesbecamemore is possibly attributed to the insufficient representation of turbulent eddies such that the PBL schemes in WRF (and organizedand formedintobroadbands of signicfi ant convec- tion. eTh y were quasi-stationary and blocked the passage of other mesoscale NWP models) are suitable only for model aircraft from the south in the HKFIR. A series of images horizontal resolution up to one or a few kilometers. In recent versions of WRF, a Large-Eddy Simulation showing radar CAPPI reflectivity at 3 km altitudes is given in Figure 7. (LES) model has been incorporated in the code. LES has been The PRD-AVM forecast was initiated for 2300 UTC widely applied to study turbulence as well as its coherent structures and statistics within PBL. Using this new option September 17, 2011. Its initial and boundary conditions were basedonthe forecastsof2km NHMrun at 2200UTC to (with 3-dimensional diffusion term, 1.5 order TKE closure, ensure a better input to AVM in terms of sufficient spin up of RRTMG shortwave and longwave, and the default MM5 sim- moisture and other model forecast elements. The simulated ilaritynear-surfaceprocess), thesameforecastofwinddistri- maximum reflectivity maps from PRD-AVM in 0000–0500 bution over HKIA is depicted in Figure 6(b),which showsa quite encouraging result on the model simulated windshear UTC are given in Figure 7. eTh hourly accumulated rainfall forecasts from 2 km NHM are also shown for comparison. feature in terms of more realistic location and timing of local- eTh WSM-5schemeemployedinPRD-AVM produced a ized wind convergence and associated microscale anticyclone off the HKIA. Moreover, it should be noted that the simulated reasonable trend of convection development. eTh “simulated radar echoes” were found to intensify and organize into a wind pattern was quite sensitive to the choice of near-surface signicfi ant convection cluster in 3 to 6 hours ahead. However, process (e.g., changing from MM5 similarity to Eta model thelocationofthe wholeforecastconvectionsystemwas type), as well as the method to calculate the momentum foundtolocatemoresouthwest to theactualshown in the u fl x, heat u fl x, and moisture ux fl in the WRF-LES. eTh forecast speed of southwesterly wind could be strengthened radar imagery. Model simulated radar reflectivity of 50 dBZ (or above) together with a higher amount of precipitation using these choices such that the formations of local-scale was obtained when compared to the forecasts from the 2 km anticyclone and wind convergence were suppressed. Advances in Meteorology 7 PRD-AVM 2 km NHM forecast run at forecast run at 0700 HKT 0700 HKT 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 0.5 1 2 5 10 15 20 25 30 40 50 70 100 Max reflectivity (dBZ) (mm) (a) (b) (c) Figure 7: Radar CAPPI reflectivity on 3 km at 07:00, 09:00, 11:00, and 13:00 HKT September 18, 2011. 2 km NHM hourly accumulated rainfall forecast with surface winds in middle column. Simulated radar reflectivity and surface winds from PRD-AVM are shown in the right column; the figure areas are marked in dashed box in 2 km NHM charts. NHM (not shown). In this case, 2 km NHM showed a more 4.2. Initiation of Isolated und Th erstorms (May 20, 2012)— realistic result in terms of a broader coverage of rain bands. It Comparison with Dieff rent Advanced Cloud Microphysics was possibly due to a larger size of domain in 2 km NHM that Parameterizations. Summertime heat thunderstorms, in par- better represented the low-pressure area and the convection ticular associated with the sea-breeze convergence, oeft n system over the coastal waters. develop rapidly into organized rainstorms. Accurate forecasts 8 Advances in Meteorology Figure 8: Radar reflectivity image from 12:00 to 14:30 HKT on May 20, 2012. under Th storm developments over inland Guangdong and the urban areas of HK are labeled in A and B. fortheir initiation anddevelopment arechallenging dueto inclusion of mix-phased hydrometeors and a deeper layer the uncertainties of mesoscale processes. For instance, under of melting and freezing processes. Using the more complex the condition of weak pressure gradient, generally light winds WSM-6, which includes the specific humidity of graupel in and a conditionally unstable environment in the morning the governing equations, a more localized distribution of on May 20, 2012, development of wind convergence over peak rainfall was simulated owing to the fall out of graupel the local territory and coastal areas occurred in the early as well as rain water. However, the scheme showed a bias for aer ft noon leading to rapid development of thunderstorm overpredicting the coverage of light rain. cells. A sequence of radar CAPPI reflectivity images during Recently, more sophisticated double-moment cloud 12 to 14 HKT is shown in Figure 8. With more widespread microphysics schemes have been implemented in WRF- development found over the northern part of the New Terri- ARW that can predict both the mixing ratio and number tories, heavy rain occurred and lasted for a few hours till the concentration of hydrometeor species, such as cloud droplets, evening. cloud ice, rain and snow, to improve the representation of Forecast charts of simulated radar reflectivity with surface cloud processes in forecast of convective weather phenome- winds from the PRD-AVM run initialized at 0800 HKT (0000 na. Three numerical experiments were performed using the UTC) May 20, 2012, are shown in Figure 9.TheWSM-5 cloud WRF-ARW 3.4 using the WDM-6 (WRF 6-class double microphysics scheme was used. Under a rather weak synoptic moment scheme [8]), Morrison double-moment scheme [9] forcing environment, convergence of winds developed due to and Milbrandt-Yau double-moment scheme [10]. Figure 10 theseabreezeoverthecoastalareasandsurfaceheatingunder shows the forecast reflectivity maps in 4–6 hours of forecasts insolation over the inland regions. eTh initiation of convec- where the local sea-breeze-induced convergence and thun- tive cells was forecast by PRD-AVM over these convergence derstorms were initiated. Generally, all the three schemes zones where instabilities were also found. eTh development were capable to predict the development of the localized of intense thunderstorms over inland Guangdong (area A) simulated radar reflectivity with peak intensity at 45 dBZ andHK(area B) wascapturedreasonablyinterms of or above. They have differences in spatial distribution and location and trend. In WRF-ARW, several bulk microphysics locationsthoughthe surfacewindfieldforecastinterms schemes (“mp physics” option)havebeenmadeavailable of its wind convergence was quite similar (not shown). since the release of version 2 for predicting the specific For instance, the forecasts from WDM-6 showed the most humidity of cloud hydrometeors including both water and intense and localized organization of high reflectivity echoes ice components. In particular, Hong et al. [7]comparedthe or storm cells compared to Morrison and Milbrandt-Yau performance of WRF single moment (WSM) 3-class, 5-class, schemes. The development of the thunderstorms was found and 6-class to simulate a mesoscale convective storm. It was to be about 2 hours earlier than the other two schemes. It found that three schemes generally demonstrated similar was discussed in [7] that the WDM-6 was able to provide performance in terms of rainfall distribution and WSM-5 an improved treatment of the variability in cloud and rain tended to produce more realistic rainfall intensity due to number concentrations in order to ameliorate the generation Advances in Meteorology 9 T+ 4:00 12:00 HKT T+ 4:30 12:30 HKT T+ 5:00 13:00 HKT (a) (b) (c) T+ 5:30 13:30 HKT T+ 6:00 14:00 HKT T+ 6:30 14:30 HKT 5 5 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 10 15 20 25 30 35 40 45 50 55 60 65 70 75 10 15 20 25 30 35 40 45 50 55 60 65 70 75 Reflectivity (dBZ) Reflectivity (dBZ) Reflectivity (dBZ) (d) (e) (f) Figure 9: Simulated radar reflectivity from PRD-AVM run at 0000 UTC (0800 HKT) May 20, 2012. eTh development of convective cells in A and B by the model forecasts at 14:00 HKT is indicated for comparison. of widespread light rain as found in its single moment version has been successfully implemented for trial experiments to (WSM-6). simulate windshear events, sea-breeze convergence, and sig- In case for Morrison and Milbrandt-Yau schemes, they nificant convections. Sensitivity experiments using different have similar forecasts in the distribution of simulated reflec- model physical processes have been attempted in order to tivity. In terms of timing and location of development, the obtain a feasible setup and configuration of AVM in forecast- Milbrandt-Yau scheme predicted both the thunderstorms ing the local or microscale weather phenomena near HKIA. (≥35 dBZ) at A and B to occur during 13-14 HKT that were They show encouraging results and thus favorable for routine more coherent with the radar images (Figure 8). Additionally, runs in rapid-update cycle configuration subject to available the Milbrandt-Yau scheme better resembled the convection computing resources. However, there are yet a number of over the coastal waters in T + 4 hour of forecast that were limitations to solve as elaborated in the following paragraphs. also present in WSM-5 (Figure 9). More studies will be eTh 200m HKA-AVMcould reproducemorerealistic performed in future to understand the mechanisms and the wind flow and forecasts for HKIA compared to the current characteristics of the three double moment schemes. operational mesoscale NWP system (2 km NHM). Though a number of choices of physical parameterization process for PBL, near-surface process, land surface models, and cloud 5. Concluding Remarks microphysics from WRF-ARW have been studied or adopted In this paper, the development of AVM is discussed. AVM in other high-resolution NWP studies, they may not be fully is designedtoenhance thecapabilityofNWP to supportthe applicable for the fine-resolution simulation with horizontal aviation forecast in the Hong Kong Observatory. eTh model resolution at 200 m. For instance the PBL process, land 10 Advances in Meteorology T+ 4:00 12:00 HKT T+ 5:00 13:00 HKT T+ 6:00 14:00 HKT WDM-6 (a) Morrison (b) Milbrandt-Yau 5 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 10 15 20 25 30 35 40 45 50 55 60 65 70 75 Reflectivity (dBZ) Reflectivity (dBZ) Reflectivity (dBZ) (c) Figure 10: Simulated radar reflectivity from PRD-AVM run at 0000 UTC (0800 HKT) May 20, 2012 using dieff rent double-moment cloud microphysics scheme: (a) WDM-6, (b) Morrison, and (c) Milbrandt-Yau. Development of convective cells near Guangzhou (A) and over Hong Kong (B) in the model forecasts is shown. Advances in Meteorology 11 surface model, and even the correct specicfi ation of the land Hong Kong observatory,” in Proceedings of the 3rd Interna- tional Workshop on Prevention and Mitigation of Meteorolog- surfacetypeare importanttoaccuratelysimulatethe nfi e- ical Disasters in Southeast Asia,Beppu,Japan,March 2010, scalewindfieldinHKIA. Secondly,the domain size of PRD- http://www.hko.gov.hk/publica/reprint/r882.pdf. AVM may be too limited to forecast the development of the [3] K. Saito, T. Fujita, Y. Yamada et al., “eTh operational JMA convective systems near the boundary, where the boundary nonhydrostatic mesoscale model,” Monthly Weather Review, data areprovidedfrom2km NHM. To mitigate theincon- vol. 134, no. 4, pp. 1266–1298, 2006. sistency, we may consider extending the domain of PRD- [4] W.K.Wong, M. K. Or,P.W.Chan, andC.M.Cheng,“Impact AVM subject to computation resources or including another of radar retrieval winds on data assimilation and forecast of outermodel with coarserresolution(e.g.,gridspacing at 1 a mesoscale convective storm using non-hydrostatic model,” or 2 km) using two-way nesting for a more consistent update in Proceedings of the 14th Conference on Mesoscale Process, of forcing in the model boundary. In fact, two-way nesting American Meteorological Society, Los Angeles, Calif, USA, August 2011. was attempted in the initial setup of WRF but found to be [5] W.C.Skamarock,J.B.Klemp,J.Dudhiaetal.,“Adescription numerically unstable occasionally. With the newer version of the advanced research WRF version 2,” NCAR Tech Notes- of code available (version 3.4), the two-way nesting would 468+STR, 2005. be explored again to check the code robustness and possible [6] X. Z. Liang, M. Xu, X. Yuan et al., “Regional climate-weather gain in computation speed over the existing one-way nesting research and forecasting model,” Bulletin of the American method (a.k.a. nest-down, “ndown” approach). Meteorological Society, vol. 93, pp. 1363–1387, 2012. The whole AVM system is currently initialized by interpo- [7] S.Y.Hong, K. S. Lim, Y. H. Leeetal.,“Evaluation of theWRF lation of the 2 km NHM forecasts. ou Th gh the “3 : 1” nesting double-moment 6-class microphysics scheme for precipitating ratio (2000 m in NHM versus 600 m in PRD-AVM) was convection,” Advances in Meteorology,vol.2010, ArticleID found to be a viable choice from the results of simulation 707253, 10 pages, 2010. experiments, it is no doubt that the model initialization, [8] K.S.S.Lim andS.Y.Hong, “Development of an eeff ctive dou- especially some spin-up problems linked to the cloud micro- ble-moment cloud microphysics scheme with prognostic cloud physics could be alleviated through an implementation of condensation nuclei (CCN) for weather and climate models,” Monthly Weather Review,vol.138,no. 5, pp.1587–1612,2010. data assimilation (DA) in future. eTh implementation of AVM-DA system is not trivial, since the current configura- [9] H. Morrison, G. Thompson, and V. Tatarskii, “Impact of cloud microphysics on the development of trailing stratiform precip- tion of data assimilation using the variational methods itation in a simulated squall line: comparison of one- and two- (3DVAR or 4DVAR) requires a well-tuned background cova- moment schemes,” Monthly Weather Review,vol.137,no. 3, pp. riance model. The correlation of the analysis control variables 991–1007, 2009. is specified based on physical balances in synoptic and [10] J. A. Milbrandt and M. K. Yau, “A multimoment bulk micro- subsynoptic scales of processes to generate an optimal model physics parameterization. Part I: analysis of the role of the spec- analysis as the initial condition. Additionally, the tangent tral shape parameter,” Journal of the Atmospheric Sciences,vol. linear and adjoint models of WRF may not be sufficient to 62,no. 9, pp.3051–3064,2005. represent the mesoscale and cloud microphysical processes. Thiswould be pursuedinthe future research studiesonthe performance of the variational data assimilation of the WRF- ARW, including the technique and related tuning, if available, for constructing a suitable DA system for PRD-AVM and HKA-AVM. Acknowledgments Part of the research work in this paper was conducted by the second author during his attachment to HKO in 2011-2012. aTh nksare also duetoDrs.BetaC.L.Yip andIvy K. Y. Wong for providing technical supports in setting up an experimen- tal version of AVM system using WRF-ARW 3.2.1. References [1] P. W. Chan and T. C. Cheung, “Microscale simulation of terrain- disrupted airflow around the Hong Kong International Airport (HKIA)—comparison of results between numerical models,” in Proceedings of the10thAnnual WRFUsers’Workshop,Boulder, Colo, USA, June 2009. [2] W. K. Wong, “Development of operational rapid update non-hydrostatic NWP and data assimilation systems in the International Journal of Journal of Ecology Mining The Scientific Journal of Geochemistry Scientifica World Journal Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 Journal of Earthquakes Paleontology Journal Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 Journal of Petroleum Engineering Submit your manuscripts at http://www.hindawi.com International Journal of Geophysics Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 Advances in Advances in Journal of Advances in International Journal of Meteorology ClimatologyGeology Oceanography Oceanography Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 Applied & Journal of Environmental International Journal of Journal of International Journal of Computational Soil Science Mineralogy Geological Research Atmospheric Sciences Environmental Sciences Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Volume 2014 Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com http://www.hindawi.com Volume 2014

Journal

Advances in Meteorology – Hindawi Publishing Corporation

Published: Jun 19, 2013

References

  • Microscale simulation of terrain-disrupted airflow around the Hong Kong International Airport (HKIA)—comparison of results between numerical models,
    P. W. Chan and T. C. Cheung
  • Development of operational rapid update non-hydrostatic NWP and data assimilation systems in the Hong Kong observatory,
    W. K. Wong
  • The operational JMA nonhydrostatic mesoscale model,
    K. Saito, T. Fujita, Y. Yamada et al.
  • Impact of radar retrieval winds on data assimilation and forecast of a mesoscale convective storm using non-hydrostatic model,
    W. K. Wong, M. K. Or, P. W. Chan, and C. M. Cheng
  • A description of the advanced research WRF version 2,
    W. C. Skamarock, J. B. Klemp, J. Dudhia et al.
  • Regional climate-weather research and forecasting model,
    X. Z. Liang, M. Xu, X. Yuan et al.
  • Evaluation of the WRF double-moment 6-class microphysics scheme for precipitating convection,
    S. Y. Hong, K. S. Lim, Y. H. Lee et al.
  • Development of an effective double-moment cloud microphysics scheme with prognostic cloud condensation nuclei (CCN) for weather and climate models,
    K. S. S. Lim and S. Y. Hong
  • Impact of cloud microphysics on the development of trailing stratiform precipitation in a simulated squall line: comparison of one- and two-moment schemes,
    H. Morrison, G. Thompson, and V. Tatarskii
  • A multimoment bulk microphysics parameterization. Part I: analysis of the role of the spectral shape parameter,
    J. A. Milbrandt and M. K. Yau