Article Contents

Influence of Boundary Layer Mixing Mechanisms on the Simulation of Typhoon Wutip: A Direct Hit on the Xisha Islands in 2013

Funding:

National Natural Science Foundation of China 42075064

Guangxi Key Technologies R&D Program AB22080101

Science and Technology Planning Project of Guangdong Province 2023B1212060019


doi: 10.3724/j.1006-8775.2024.019

  • In this study, we simulated the tropical cyclone (TC) Wutip, which originated in the South China Sea in 2013, using three planetary boundary layer (PBL) parameterized schemes within the Weather Research and Forecasting model, i.e., Medium-Range Forecast (MRF), Yonsei University scheme (YSU), and Asymmetric Convective Model Version 2 (ACM2), with different vertical mixing mechanisms. We investigated the effects of different PBL mixing mechanisms on the simulation of TC track, intensity, structure, and precipitation. The results reveal that the surface flux and vertical mixing of PBL jointly influenced the TC throughout its lifecycle, and the simulated TC intensity was closely correlated with the eyewall structure. These three schemes were all first-order and nonlocal closure schemes. However, the MRF scheme was over-mixed, which led to a relatively dryer and warmer near-surface layer, a wetter and colder upper PBL, and thus a simulated eyewall with the smallest wet static energy and weaker convection. Moreover, the MRF scheme produced the smallest 10-m wind speed, which was closest to the observation, and the weakest TC warm-core structure and intensity. The YSU scheme was similar to the MRF scheme, yet it distinguished itself by incorporating an explicit treatment of the entrainment process at the top of the PBL and developing thermal-free convection above the PBL of the eyewall, which significantly increased the wet static energy over the TC eyewall. Thus, the simulated eyewall was more contracted and steeper with stronger upward motion while the eye area became even warmer, finally leading to the strongest TC. The precipitation distribution simulated by the YSU scheme was the most consistent with the observation. The ACM2 scheme used the nonlocal upward and local downward mixed asymmetric convection modes, which reduced the excessive development of thermal-free convection at the eyewall, and avoided restricting the dynamically forced turbulent motion outside the eyewall, leading to a larger radius of the maximum wind speed, and thus more reasonable structural characteristics of PBL and TC intensity. In summary, compared with the YSU scheme and the MRF scheme, the ACM2 scheme demonstrated superior performance in capturing the structure, track, and intensity of Typhoon Wutip. It is important to note that this analysis was based on a specific case study, which might have inherent limitations due to its modest focus.
  • 加载中
  • Figure 1.  Comparison of the simulated TC track with observations (OBS) based on CMA's TC database.

    Figure 2.  Comparison of the simulated MSLP with observations (OBS).

    Figure 3.  Comparison of the simulated MSW with observations (OBS).

    Figure 4.  Comparison of the simulated 24-h cumulative precipitation (units: mm) from 06:00 on September 29 to 06:00 on September 30 with observations (GPM).

    Figure 5.  Comparison of the simulated 10-m wind (units: m s–1) at 06:00 on September 29 with ERA5 data.

    Figure 6.  The radial distribution of surface sensible heat flux over time (units: W m–2).

    Figure 7.  The radial distribution of surface latent heat flux over time (units: W m–2).

    Figure 8.  The radial distribution of friction velocity over time (units: m s–1).

    Figure 9.  The average radial velocity (units: m s–1) at radius height at 06:00 on September 29.

    Figure 10.  The average equivalent potential temperature (units: K) at radius height at 06:00 on the 29th.

    Figure 11.  The average specific humidity (units: g kg–1) at radius-height at 06:00 on September 29.

    Figure 12.  The average virtual potential temperature (units: K) at radius-height at 06:00 on September 29.

    Figure 13.  The average virtual potential temperature (left, units: K) and specific humidity (right, units: g kg–1) within a radius of 50 km in the TC eye area at 06:00 on September 29.

    Figure 14.  Radial profiles of Km (units: m2 s–1) within the lowest 1 km layer at 06:00 on September 29.

    Figure 15.  The average tangential wind and radial wind (left column, contour line represents tangential wind, colored shadow represents radial wind), and vertical velocity (right column) at radius height at 06:00 on September 29 (units: m s–1).

    Table 1.  Main differences between these three PBL schemes.

    Scheme Description Turbulence equation Key feature
    MRF First-order non-local closure scheme. Inverse gradient flux is used to deal with heat and moisture in unstable conditions. Ct=z[Kc(Czγc)] It makes up for the limitations of K-theory and solves the problem of inverse gradient transport caused by large eddies. For unstable or well-mixed boundary layer conditions, the simulation results are better. When the wind speed is relatively high, there is a problem of excessive mixing, which will lead to a decrease in convective precipitation.
    YSU First-order non-local closure scheme. Compared with the MRF scheme, the treatment of entrainment processes at the top of the boundary layer is added in the YUS scheme. Ct=Z[Kc(Czγc)(wc)h¯(zh)3] The shortcoming of excessive mixing in the MRF scheme is effectively solved with the thermal turbulent motion increased and the dynamic-forced one reduced. The limitation is that the convective available potential energy of mixed layer associated with deep convection environment is underestimated.
    ACM2 Asymmetric convection model of non-local upward mixing and local downward mixing. Different from MRF and YSU schemes, the parameter weighting factor fconv is introduced to the turbulence equation in the ACM2 scheme to control the proportion of non-local and local action. Cit=fconvMuC1fconvMdiCi+fconvMdi+1Ci+1zi+1zi+z[KC(1fconv)Ciz] Combining the advantages of ACM1 boundary layer scheme and eddy diffusion model, the ACM2 scheme allows the transport process generated by large-scale turbulent vortices to be simulated under convective conditions, and the small-scale turbulent mixing process of subgrid can be reflected.
    Note: C is the prediction variable, z refers to height, Kc represents the eddy diffusion coefficient, γc is the local gradient correction term, h is the height of the boundary layer, Mu is the non-local convective mixing rate upward from the bottom layer of the model, and Md denotes the downward mixing rate.
    DownLoad: CSV

    Table 2.  Physical parameterization schemes used in this study.

    Parameter YSU MRF ACM2
    Microphysics WSM7 WSM7 WSM7
    Cumulus physics Kain-Fritsch Kain-Fritsch Kain-Fritsch
    Longwave radiation RRTMG RRTMG RRTMG
    Shortwave radiation RRTMG RRTMG RRTMG
    PBL physics YSU MRF ACM2
    Surface layer MM5 Monin-Obukhov MM5 Monin-Obukhov MM5 Monin-Obukhov
    Land surface Noah Noah Noah
    DownLoad: CSV
  • [1] EMANUEL K. Increasing destructiveness of tropical cyclones over the past 30 years [J]. Nature, 2005, 436(1): 686–688, https://doi.org/10.1038/nature03906
    [2] MENDELSOHN R, EMANUEL K, CHONABAYASHI S, et al. The impact of climate change on global tropical cyclone damage [J]. Nature Climate Change, 2012, 2(3): 205–209, https://doi.org/10.1038/NCLIMATE1357
    [3] HUANG M F, WANG Q, LIU M F, et al. Increasing typhoon impact and economic losses due to anthropogenic warming in Southeast China [J]. Scientific Reports, 2022, 12(1): 14048, https://doi.org/10.1038/s41598-022-17323-8
    [4] PANDEY R S, LIOU Y. Typhoon strength rising in the past four decades [J]. Weather and Climate Extremes, 2022, 36(1): 100446, https://doi.org/10.1016/j.wace.2022.100446
    [5] UTSUMI N, KIM H. Observed influence of anthropogenic climate change on tropical cyclone heavy rainfall [J]. Nature Climate Change, 2022, 12(5): 436–440, https://doi.org/10.1038/s41558-022-01344-2
    [6] HU L T, WEN T, SHAO Y C, et al. Economic impacts of tropical cyclone-induced multiple hazards in China [J]. Earth's Future, 2023, 11(9): e2023EF003622, https://doi.org/10.1029/2023EF003622
    [7] GAO S Z, ZHANG S J, LU X Y, et al. Circulation characteristics and thermal and dynamic conditions 48 hours before typhoon formation in the South China Sea [J]. Journal of Applied Meteorological Science, 2021, 32(3): 272–288, https://doi.org/10.11898/1001-7313.20210302
    [8] KRISHNAMURTI T N, PATTNAIK S, STEFANOVA L, et al. The hurricane intensity issue [J]. Monthly Weather Review, 2005, 13(7): 1886–1912, https://doi.org/10.1175/MWR2954.1
    [9] ROGERS R, ABERSON S, BLACK M, et al. The intensity forecasting experiment: A NOAA multiyear field program for improving tropical cyclone intensity forecasts [J]. Bulletin of the American Meteorological Society, 2006, 87(11): 1523–1537, https://doi.org/10.1175/BAMS-87-11-1523
    [10] TAKAHASHI T, NOLAN D S, MCNOLDY B D. The vortex structure and near-surface winds of Typhoon Faxai (2019) during landfall, Part Ⅱ: Evaluation of WRF simulations [J]. Quarterly Journal of the Royal Meteorological Society, 2024, 150(760): 1643–1667, https://doi.org/10.1002/qj.4663
    [11] LI X L, PU Z X. Sensitivity of numerical simulation of early rapid intensification of hurricane Emily (2005) to cloud microphysical and planetary boundary layer parameterizations [J]. Monthly Weather Review, 2008, 136(12): 4819–4838, https://doi.org/10.1175/2008MWR2366.1
    [12] RICCHI A, MIGLIETTA M M, BARBARIOL F, et al. Sensitivity of a Mediterranean tropical-like cyclone to different model configurations and coupling strategies [J]. Atmosphere, 2017, 8(5): 92, https://doi.org/10.3390/atmos8050092
    [13] HONG S Y, LIM J. The WRF single-moment 6-class microphysics scheme (WSM6) [J]. Asia-pacific Journal of Atmospheric Sciences, 2006, 42(2): 129–151.
    [14] EFSTATHIOU G A, ZOUMAKIS N M, MELAS D, et al. Sensitivity of WRF to boundary layer parameterizations in simulating a heavy rainfall event using different microphysical schemes: Effect on large-scale processes [J]. Atmospheric Research, 2013, 132(1): 125–143, https://doi.org/10.1016/j.atmosres.2013.05.004
    [15] SUN W Q, LI C Y. A review of atmospheric boundary layer parameterization schemes in numerical models [J]. Journal of Marine Meteorology, 2018, 38(3): 11–19, https://doi.org/10.19513/j.cnki.issn2096-3599.2018.03.002, in Chinese with English abstract
    [16] TYMVIOS F, DEMETRIS C, SILAS M, et al. Intercomparison of boundary layer parameterizations for summer conditions in the eastern Mediterranean island of Cyprus using the WRF-ARW model [J]. Atmospheric Research, 2018, 208(1): 45–59, https://doi.org/10.1016/j.atmosres.2017.09.011
    [17] LI X, DAVIDSON N E, DUAN Y H, et al. Analysis of an ensemble of high-resolution WRF simulations for the rapid intensification of super typhoon Rammasun (2014) [J]. Advances in Atmospheric Sciences, 2020, 37(2): 187–210, https://doi.org/10.1007/s00376-019-8274-z
    [18] GOPALAKRISHNAN S, HAZELTON A, ZHANG J A. Improving hurricane boundary layer parameterization scheme based on observations [J]. Earth and Space Science, 2021, 8(3): 1–13, https://doi.org/10.1029/2020EA001422
    [19] STULL R B. An Introduction to Boundary Layer Meteorology [M]. Amsterdam: Kluwer Academic Publishers, 1988.
    [20] STENSRUD D J. Parameterization Schemes: Keys to Understanding Numerical Weather Prediction Models [M]. Cambridge: Cambridge University Press, 2007.
    [21] STULL R B. Static stability-an update [J]. Bulletin of the American Meteorological Society, 1991, 72(10): 1521–1529, https://doi.org/10.1175/1520-0477(1991)072<1521:ssu>2.0.co;2 doi:
    [22] RUAN Z X, LI J N, LI F Z, et al. Effects of local and non-local closure PBL schemes on the simulation of super typhoon Mangkhut (2018) [J]. Frontiers of Earth Science, 2022, 16(2): 277–290, https://doi.org/10.1007/s11707-020-0854-9
    [23] SHEN W Q, LU Z B, YE G L, et al. Exploring the impact of planetary boundary layer schemes on rainfall forecasts for typhoon Mujigae, 2015[J]. Atmosphere, 2022, 13(2): 220–237, https://doi.org/10.16032/10.3390/atmos13020220
    [24] HUANG W Y, SHEN X Y, WANG W G, et al. Comparison of the thermal and dynamic structural characteristics in boundary layer with different boundary layer parameterizations [J]. Chinese Journal of Geophysics, 2014, 57(5): 1399–1414, https://doi.org/10.1002/cjg2.20123
    [25] NOLAN D S, ZHANG J A, STERN D P. Evaluation of planetary boundary layer parameterizations in tropical cyclones by comparison of in situ observations and high-resolution simulations of Hurricane Isabel (2003), Part Ⅰ: Initialization, maximum winds, and the outercore boundary layer [J]. Monthly Weather Review, 2009, 137(11): 3651–3674, https://doi.org/10.1175/2009MWR2785.1
    [26] NOLAN D S, ZHANG J A, STERN D P. Evaluation of planetary boundary layer parameterizations in tropical cyclones by comparison of in situ observations and high-resolution simulations of Hurricane Isabel (2003), Part Ⅱ: Inner-core boundary layer and eyewall structure [J]. Monthly Weather Review, 2009, 137(11): 3675–3698, https://doi.org/10.1175/2009MWR2786.1
    [27] DING C H, LI J N, ZHAO Y J, et al. The influence of boundary layer parameterization schemes on autumn typhoon Sarika (2016) in South China Sea [J]. Journal of Tropical Meteorology, 2018, 34(5): 657–673, https://doi.org/10.16032/j.issn.1004-4965.2018.05.008, in Chinese with English abstract
    [28] LIU J J, ZHANG F M, PU Z X. Numerical simulation of the rapid intensification of Hurricane Katrina (2005): Sensitivity to boundary layer parameterization schemes [J]. Advances in Atmospheric Sciences, 2017, 34(4): 482–496, https://doi.org/10.1007/s00376-016-6209-5
    [29] ROTUNNO R, BRYAN G H. Effects of parameterized diffusion on simulated hurricanes [J]. Journal of the Atmospheric Sciences, 2012, 69(7): 2284–2299, https://doi.org/10.1175/JAS-D-11-0204.1
    [30] WEN X P, LONG X, ZHANG S W, et al. Numerical studies of planetary boundary layer parameterization sciences on super typhoon Sanba (2012) during its initial stage [J]. Journal of Tropical Meteorology, 2018, 24(3): 288–299, https://doi.org/10.16555/j.1006-8775.20l8.03.003
    [31] HONG S Y, NOH Y, DUDHIA J. A new vertical diffusion package with an explicit treatment of entrainment processes [J]. Monthly Weather Review, 2006, 134(9): 2318–2341, https://doi.org/10.1175/MWR3199.1
    [32] HONG S Y, PAN H L. Nonlocal boundary layer vertical diffusion in a medium-range forecast model [J]. Monthly Weather Review, 1996, 124(10): 2322–2339, https://doi.org/10.1175/1520-0493(1996)124<2322:NBLVDI>2.0.CO;2 doi:
    [33] WANG C X. Experiments of influence of planetary boundary layer parameterization on Muifa typhoon prediction [J]. Advance in Earth Sciences, 2013, 28(2): 197–208, https://doi.org/10.11867/j.issn.1001-8166.2013.02.0197
    [34] LOH W T, JUNENG L, TANGANG F T. Sensitivity of typhoon Vamei (2001) simulation to planetary boundary layer parameterization using PSU/NCAR MM5[J]. Pure and Applied Geophysics, 2011, 168(10): 1799–1811, https://doi.org/10.1007/s00024-010-0176-z
    [35] JIANG L F, YIN Y, LIU C X. Comparison of the thermal and dynamic boundary layer structure with different boundary layer parameterizations during typhoon "Molave" [J]. Marine Forecasts, 2017, 34(4): 20–31, https://doi.org/10.11737/j.issn.1003-0239.2017.04.003
    [36] WANG Y X, ZHONG Z, SUN Y, et al. The mechanism analysis of the track deviation of tropical cyclone Megi (2010) simulated with two planetary boundary layer schemes [J]. Chinese Journal of Geophysics, 2017, 60(7): 2545–2555, https://doi.org/10.6038/cjg20170704
    [37] ZHANG F M, PU Z X, WANG C H. Effects of boundary layer vertical mixing on the evolution of hurricanes over land [J]. Monthly Weather Review, 2017, 145(6): 2343–2361, https://doi.org/10.1175/MWR-D-16-0421.1
    [38] ZHU P, MENELAOU K, ZHU Z D. Impact of subgrid-scale vertical turbulent mixing on eyewall asymmetric structures and mesovortices of hurricanes [J]. Quarterly Journal of the Royal Meteorological Society, 2014, 140(679): 416–438, https://doi.org/10.1002/qj.2147
    [39] SMITH R K, THOMSEN G L. Dependence of tropical-cyclone intensification on the boundary-layer representation in a numerical model [J]. Quarterly Journal of the Royal Meteorological Society, 2010, 136(652): 1671–1685, https://doi.org/10.1002/qj.687
    [40] KANADA S, WADA A, NAKANO M, et al. Effect of planetary boundary layer schemes on the development of intense tropical cyclones using a cloud-resolving model [J]. Journal of Geophysical Research: Atmospheres, 2012, 117(3): 107–120, https://doi.org/10.1029/2011JD016582
    [41] ZHANG J A, ROGERS R F, TALLAPRAGADA V. Impact of parameterized boundary layer structure on tropical cyclone rapid intensification forecasts in HWRF [J]. Monthly Weather Review, 2016, 145(4): 1413–1426, https://doi.org/10.1175/MWR-D-16-0129.1
    [42] ZHANG J A, NOLAN D S, ROFERS R F, et al. Evaluating the impact of improvements in the boundary layer parameterization on hurricane intensity and structure forecasts in HWRF [J]. Monthly Weather Review, 2015, 143(8): 3136–3155, https://doi.org/10.1175/MWR-D-14-00339.1
    [43] CHEN X M, BRYAN G H, ZHANG J A, et al. A framework for simulating the tropical cyclone boundary layer using large-eddy simulation and its use in evaluating PBL parameterizations [J]. Journal of the Atmospheric Sciences, 2021, 78(11): 3559–3574, https://doi.org/10.1175/JAS-D-20-0227.1
    [44] CHEN X M, BRYAN G H, HAZELTON A, et al. Evaluation and improvement of a TKE-Based Eddy-Diffusivity Mass-Flux (EDMF) planetary boundary layer scheme in hurricane conditions [J]. Weather and Forecasting, 2022, 37(6): 935–951, https://doi.org/10.1175/WAF-D-21-0168.1
    [45] ZHONH Q J, ZHANG L F, LI J P, et al. Estimating the predictability limit of tropical cyclone tracks over the western North Pacific using observational data [J]. Advances in Atmospheric Sciences, 2018, 35(12), 1–34, https://doi.org/10.1007/s00376-018-8008-7
    [46] ZHONG Q J, LI J P, ZHANG L F, et al. Predictability of tropical cyclone intensity over the Western North Pacific using the IBTrACS dataset [J]. Monthly Weather Review, 2018, 146(9): 2741–2755, https://doi.org/10.1175/MWR-D-17-0301.1
    [47] PLEIM J E. A combined local and nonlocal closure model for the atmospheric boundary layer, Part Ⅰ: Model description and testing [J]. Journal of Applied Meteorology and Climatology, 2007, 46(9): 1383–1395, https://doi.org/10.1175/JAM2539.1

Get Citation+

GUO Tian-yun, LI Jiang-nan, PANG Si-min, et al. Influence of Boundary Layer Mixing Mechanisms on the Simulation of Typhoon Wutip: A Direct Hit on the Xisha Islands in 2013 [J]. Journal of Tropical Meteorology, 2024, 30(3): 211-222, https://doi.org/10.3724/j.1006-8775.2024.019
GUO Tian-yun, LI Jiang-nan, PANG Si-min, et al. Influence of Boundary Layer Mixing Mechanisms on the Simulation of Typhoon Wutip: A Direct Hit on the Xisha Islands in 2013 [J]. Journal of Tropical Meteorology, 2024, 30(3): 211-222, https://doi.org/10.3724/j.1006-8775.2024.019
Export:  

Share Article

Manuscript History

Manuscript received: 20 November 2023
Manuscript revised: 15 May 2023
Manuscript accepted: 15 August 2024
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Influence of Boundary Layer Mixing Mechanisms on the Simulation of Typhoon Wutip: A Direct Hit on the Xisha Islands in 2013

doi: 10.3724/j.1006-8775.2024.019
Funding:

National Natural Science Foundation of China 42075064

Guangxi Key Technologies R&D Program AB22080101

Science and Technology Planning Project of Guangdong Province 2023B1212060019

Abstract: In this study, we simulated the tropical cyclone (TC) Wutip, which originated in the South China Sea in 2013, using three planetary boundary layer (PBL) parameterized schemes within the Weather Research and Forecasting model, i.e., Medium-Range Forecast (MRF), Yonsei University scheme (YSU), and Asymmetric Convective Model Version 2 (ACM2), with different vertical mixing mechanisms. We investigated the effects of different PBL mixing mechanisms on the simulation of TC track, intensity, structure, and precipitation. The results reveal that the surface flux and vertical mixing of PBL jointly influenced the TC throughout its lifecycle, and the simulated TC intensity was closely correlated with the eyewall structure. These three schemes were all first-order and nonlocal closure schemes. However, the MRF scheme was over-mixed, which led to a relatively dryer and warmer near-surface layer, a wetter and colder upper PBL, and thus a simulated eyewall with the smallest wet static energy and weaker convection. Moreover, the MRF scheme produced the smallest 10-m wind speed, which was closest to the observation, and the weakest TC warm-core structure and intensity. The YSU scheme was similar to the MRF scheme, yet it distinguished itself by incorporating an explicit treatment of the entrainment process at the top of the PBL and developing thermal-free convection above the PBL of the eyewall, which significantly increased the wet static energy over the TC eyewall. Thus, the simulated eyewall was more contracted and steeper with stronger upward motion while the eye area became even warmer, finally leading to the strongest TC. The precipitation distribution simulated by the YSU scheme was the most consistent with the observation. The ACM2 scheme used the nonlocal upward and local downward mixed asymmetric convection modes, which reduced the excessive development of thermal-free convection at the eyewall, and avoided restricting the dynamically forced turbulent motion outside the eyewall, leading to a larger radius of the maximum wind speed, and thus more reasonable structural characteristics of PBL and TC intensity. In summary, compared with the YSU scheme and the MRF scheme, the ACM2 scheme demonstrated superior performance in capturing the structure, track, and intensity of Typhoon Wutip. It is important to note that this analysis was based on a specific case study, which might have inherent limitations due to its modest focus.

GUO Tian-yun, LI Jiang-nan, PANG Si-min, et al. Influence of Boundary Layer Mixing Mechanisms on the Simulation of Typhoon Wutip: A Direct Hit on the Xisha Islands in 2013 [J]. Journal of Tropical Meteorology, 2024, 30(3): 211-222, https://doi.org/10.3724/j.1006-8775.2024.019
Citation: GUO Tian-yun, LI Jiang-nan, PANG Si-min, et al. Influence of Boundary Layer Mixing Mechanisms on the Simulation of Typhoon Wutip: A Direct Hit on the Xisha Islands in 2013 [J]. Journal of Tropical Meteorology, 2024, 30(3): 211-222, https://doi.org/10.3724/j.1006-8775.2024.019
  • Tropical cyclones (TCs) are among the strongest weather disasters. With global warming, TCs occur and develop more frequently and violently, seriously threatening human life and properties (Emanuel [1]; Mendelsohn et al. [2]; Huang et al. [3]; Pandey et al. [4]; Utsumi et al. [5]; Hu et al. [6]). Local TCs generated in the South China Sea greatly influence southern China, with an average of over 6.2 TCs generated per year (Gao et al. [7]). Numerical models, both global and mesoscale, have been extensively utilized in TC simulation studies. The continuous advancement of these models has notably improved the prediction of TC tracks. However, accurately predicting TC intensity remains a challenge (Krishnamurti et al. [8]; Rogers et al. [9]; Takahashi et al. [10]). Recognizing the crucial role of planetary boundary layer (PBL) processes in the formation and development of TCs (Li et al. [11]; Ricchi et al. [12]), researchers have extensively investigated the influence of PBL physics schemes on TCs. Despite these efforts, uncertainties still exist (Hong et al. [13]; Efstathiou et al. [14]; Sun et al. [15]; Tymvios et al. [16]; Li et al. [17]; Gopalakrishnan et al. [18]).

    The boundary layer of a TC plays a vital role in the exchange of water vapor, heat, and momentum. These exchanges occur through friction mixing and radiation, affecting the thermal and dynamic structure of the troposphere by influencing turbulence and entrainment effects within cumulus clouds. In the PBL scheme, the turbulence effect mainly consists of two parts: the closure order of turbulence and the application of local or nonlocal mixing methods (Stull [19]; Stensrud [20]). For example, Stull [21] found that the nonlocal closure scheme could simulate the circulation of the deep boundary layer more accurately compared with the local closure scheme; however, the simulation results of the local closure scheme could be improved by increasing the closure order of turbulence. Meanwhile, the TC simulated using the local quasi-normal-scale elimination scheme produced stronger thermal flux and vertical mixing, leading to a more intense TC (Ruan et al. [22]). Shen et al. [23] found that the nonlocal PBL scheme delivered the most accurate precipitation results in both the inner core and outer regions of the TC. However, some studies revealed that the nonlocal closure PBL scheme was notably superior to the local closure PBL scheme (Huang et al. [24]). Different PBL schemes reflect the diverse characteristics of the PBL and also have different impacts on TCs (Nolan et al. [25,26]).

    The intensity of simulated TC is influenced by thermal and dynamic forces, and variations in TC intensity are closely related to changes in TC structure (Ding et al. [27]). Liu et al. [28] found that both the surface flux and the vertical mixing of the PBL had a significant impact on the forecast of TC intensity. The horizontal diffusion in the PBL also plays a critical role in simulating hurricane intensity (Rotunno et al. [29]). The choice of PBL schemes have different effects on TCs of different intensities (Wen et al. [30]). For the simulation of weak TCs, the YSU scheme (Hong et al. [31]) appears to be superior to the MRF scheme (Hong et al. [32]), which has better track simulation for strong TCs (Wang et al. [33]). The near-equator TC intensity and track are not sensitive to different PBL schemes (Loh et al. [34]), but differences in simulation results are most significant during the TC weakening stage. Jiang et al. [35] investigated the effects of different PBL schemes on the dynamical and thermal structure of TCs. They found that the simulation of TC structure varied greatly among the PBL schemes. The local K-theory provides a more accurate calculation of turbulent diffusion under stable layer conditions, whereas the non-local scheme offers closer approximations to actual wind speed and barometric pressure simulations. Wang et al. [36] observed that the YSU scheme exhibited stronger vertical mixing and water vapor transport in the PBL compared to the MYJ scheme, which influenced the TC track. Zhang et al. [37] found that vertical turbulent mixing had a strong effect on modelled TC intensity, with weaker vertical mixing leading to stronger TC intensity over land. Zhu et al. [38] conducted a series of numerical experiments on the parameterization of vertical turbulent mixing at different subgrid scales using the WRF model, and proved that the vertical turbulent mixing scheme played an important role in the asymmetric structure of TCs and the formation of eyewall mesovortices.

    In summary, many studies have highlighted the significant impact of different vertical turbulent mixing schemes on the dynamics and thermal structures of TCs. While much of the research focuses on the discussion of different PBL schemes, it is worth noting that PBL schemes with the same mixing method and closure order also can produce different prediction results. For example, Smith et al. [39] found that different diffusion coefficients of turbulence in different PBL schemes could lead to different TC intensities. Similarly, Kanada et al. [40] found that different vertical eddy diffusivity coefficient resulted in different TC intensities, inner core structures, and the relationship between maximum surface wind (MSW) and central pressures. Particularly, the large diffusion coefficient of the vertical vortex in the low layer (height < 300 m) led to large heat and water vapor transfer, resulting in extremely intense TCs accompanied by an upright, contracted eyewall structure. Zhang et al. [41] emphasized the important effect of parameterized PBL structure on the change in TC intensity. Furthermore, PBL turbulent processes also play a crucial role in the development of TC structure and intensity (Zhang et al. [42]; Chen et al. [43]; Chen et al. [44]).

    To explore the impact of PBL mixing mechanisms on TC simulation, this study selected the same land surface process scheme and near-surface layer scheme, along with three first-order and nonlocal closure PBL schemes with different mixing mechanisms to simulate TC Wutip in 2013. TC Wutip was generated in the South China Sea (SCS), where the predictability of TC tracks and intensity is relatively low (Zhong et al. [45,46]). Moreover, the Xisha Islands are susceptible to TC attacks. Due to the unique geographical location of the Xisha Islands, TCs directly striking them have been rarely studied. This paper focused on the influence of different PBL mixing mechanisms on the simulation of TC Wutip's impact on the Xisha Islands.

    The remainder of the paper is structured as follows. Section 2 describes data and provides a brief review of TC Wutip in 2013. Section 3 outlines the experimental design. Section 4 provides the study results, and conclusions and discussion are summarized in Section 5.

  • The data used in this study included the following: (1) The final reanalysis (FNL) data provided by the National Centers for Environmental Protection (NCEP) and the National Centers for Atmospheric Research (NCAR), featuring a horizontal resolution of 1° × 1° and updated every 6 hours. (2) GPM half-hour precipitation data provided by NASA, with a horizontal resolution of 0.1° × 0.1°. (3) The best track dataset from the Joint Typhoon Warning Center, which included the central latitude and longitude of TCs, minimum sea level pressure (MSLP), and MSW. (4) 10-m wind field data from the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis v5 (ERA5), with a horizontal resolution of 0.25° × 0.25°.

    At 06:00 on September 26, 2013 (UTC, hereinafter the same), a tropical disturbance around Huangyan Island developed into a tropical depression (TD). At 18:00, it was coded as Wutip, and started moving westward. At 06:00 on September 28, it developed into a TC. At 04:00 on September 29, the MSW suddenly increased to 45 m s–1 and remained at that level until 08:00. Four hours later at 12:00, TC Wutip began to weaken. At 09:00 on September 30, it landed in Vietnam, with an MSW of 35 m s–1 and an MSLP of 970 hPa. During the lifespan of the typhoon, the MSLP was 948 hPa and the MSW was 51.4 m s–1. The structural adjustment of TC Wutip during the stabilization stage was unusually active, with repeated cycles of eyewall dissipation, formation, and dissipation.

    Precipitation was mainly concentrated on the left side of the TC's track. Under the impact of TC Wutip, Hainan Island and the Xisha Islands experienced saliently severe strong storms. The rainfall on North Reef and Duncan Island reached more than 500 mm. During the 24 hours from 12:00 on 29 September to 12:00 on 30 September, the maximum rainfall on Hainan Island reached 141.3 mm, with 36 towns and villages receiving more than 50 mm of rainfall. The Xisha Coral Sea Islands faced a severe situation, with a large number of fishing boats at risk and fishermen reported missing. This event marked the most prolonged and severe TC on record to impact the Xisha Islands, as per meteorological data.

  • In this study, the mesoscale numerical model WRF 4.1 was employed to simulate atmospheric conditions. The vertical direction was divided into 39 non-equidistant layers, featuring a horizontal spacing of 3 km, a grid configuration of 550 × 734, and a model top pressure set at 50 hPa. The initial field and boundary conditions were obtained from the FNL data of the grids provided by NCEP/NCAR. The simulation started from 12:00 on September 25 to 18:00 on September 30, 2013 (a total of 126 h), and the results were output every 3 h. To investigate the impact of the boundary layer mixing mechanism on simulation results, we selected three PBL schemes for the experiment: the first-order closure and nonlocal YSU scheme, the first-order closure and nonlocal MRF scheme, and the nonlocal and local mixing ACM2 scheme proposed by Pleim [47]. The main differences between the three PBL schemes are shown in Table 1. The physical parameterization schemes are shown in Table 2.

    Scheme Description Turbulence equation Key feature
    MRF First-order non-local closure scheme. Inverse gradient flux is used to deal with heat and moisture in unstable conditions. Ct=z[Kc(Czγc)] It makes up for the limitations of K-theory and solves the problem of inverse gradient transport caused by large eddies. For unstable or well-mixed boundary layer conditions, the simulation results are better. When the wind speed is relatively high, there is a problem of excessive mixing, which will lead to a decrease in convective precipitation.
    YSU First-order non-local closure scheme. Compared with the MRF scheme, the treatment of entrainment processes at the top of the boundary layer is added in the YUS scheme. Ct=Z[Kc(Czγc)(wc)h¯(zh)3] The shortcoming of excessive mixing in the MRF scheme is effectively solved with the thermal turbulent motion increased and the dynamic-forced one reduced. The limitation is that the convective available potential energy of mixed layer associated with deep convection environment is underestimated.
    ACM2 Asymmetric convection model of non-local upward mixing and local downward mixing. Different from MRF and YSU schemes, the parameter weighting factor fconv is introduced to the turbulence equation in the ACM2 scheme to control the proportion of non-local and local action. Cit=fconvMuC1fconvMdiCi+fconvMdi+1Ci+1zi+1zi+z[KC(1fconv)Ciz] Combining the advantages of ACM1 boundary layer scheme and eddy diffusion model, the ACM2 scheme allows the transport process generated by large-scale turbulent vortices to be simulated under convective conditions, and the small-scale turbulent mixing process of subgrid can be reflected.
    Note: C is the prediction variable, z refers to height, Kc represents the eddy diffusion coefficient, γc is the local gradient correction term, h is the height of the boundary layer, Mu is the non-local convective mixing rate upward from the bottom layer of the model, and Md denotes the downward mixing rate.

    Table 1.  Main differences between these three PBL schemes.

    Parameter YSU MRF ACM2
    Microphysics WSM7 WSM7 WSM7
    Cumulus physics Kain-Fritsch Kain-Fritsch Kain-Fritsch
    Longwave radiation RRTMG RRTMG RRTMG
    Shortwave radiation RRTMG RRTMG RRTMG
    PBL physics YSU MRF ACM2
    Surface layer MM5 Monin-Obukhov MM5 Monin-Obukhov MM5 Monin-Obukhov
    Land surface Noah Noah Noah

    Table 2.  Physical parameterization schemes used in this study.

  • During the period from the 24th to the 48th hour of simulation (Fig. 1), the track of the TC shifted toward the north-northwest direction. All three schemes accurately captured this characteristic. At the 48th hour, the TC track changed to the west-northwest direction, and all three schemes managed to capture this change. It is worth noting that the simulated TC center positions were slightly more eastward compared to the actual situation. As the simulation progressed beyond 72 hours, the TC track exhibited a more southerly trend compared to the real scenario. Over the entire 126-hour simulation period, the YSU scheme had an average track deviation of 50.79 km, the MRF scheme had a larger deviation of 81.83 km, and the ACM2 scheme showed the smallest average deviation of 49.42 km. Specifically, the MRF scheme had the highest average track error among the three schemes. In terms of TC speed, the YSU scheme produced the fastest simulated TC speeds, followed by the MRF scheme, and the ACM2 scheme had the slowest simulated TC speeds. The simulated average TC speeds were 14.3 km h–1, 14.1 km h–1, and 13.8 km h–1, respectively.

    Figure 1.  Comparison of the simulated TC track with observations (OBS) based on CMA's TC database.

  • All the three schemes successfully captured the general trend of the tropical cyclone's (TC) intensity, with an initial increase and subsequent decrease (Figs. 2 and 3). However, there were considerable differences in the intensities simulated by the three schemes. The YSU scheme produced the strongest simulated TC intensity, while the MRF scheme resulted in the weakest intensity. At the peak moment of the TC, the YSU scheme simulated an MSLP that was 40 hPa lower than the MRF scheme. During the TC mature stage (90–114 h), the MSLP simulated by the ACM2 scheme was closest to the actual conditions. The MSW simulated by the YSU scheme was the strongest and stronger than the actual value in the TC development stage (12–90 h) and the mature stage. The MSW simulated by the MRF was the weakest and always weaker than the actual value. In terms of the TC intensity, the ACM2 scheme was the most skillful among the three schemes.

    Figure 2.  Comparison of the simulated MSLP with observations (OBS).

    Figure 3.  Comparison of the simulated MSW with observations (OBS).

  • The precipitation was mainly located on the left side in the direction of TC's track (Fig. 4), indicating a northwest–southeast trend. The MRF scheme simulated weaker precipitation compared to the other two schemes, and the precipitation distribution was skewed towards the southern region, possibly influenced by the track of the TC. The precipitation simulated by the YSU and ACM2 were approximately equal to the observation. Furthermore, in terms of the overall precipitation distribution, the YSU scheme demonstrated the closest resemblance to the observed patterns.

    Figure 4.  Comparison of the simulated 24-h cumulative precipitation (units: mm) from 06:00 on September 29 to 06:00 on September 30 with observations (GPM).

  • At the peak intensity of the TC, the 10-m wind field showed a cyclonic distribution with a small central wind speed (Fig. 5). During this period, the TC center simulated by the ACM2 scheme closely matched the observation, while the TC center simulated by the MRF scheme was located further south. The YSU scheme, on the other hand, simulated a TC center that was shifted more to the west, corresponding to a faster TC movement. Moreover, the TC eye simulated by the YSU was the smallest with a tighter TC structure, and the windy region simulated by the scheme was the largest. Consequently, the YSU scheme simulated the strongest TC, with 10-m wind speeds exceeding 50 m s–1, surpassing those recorded in the ERA5 dataset. Conversely, the MRF scheme simulated the smallest wind speeds, which were closest to the observed values.

    Figure 5.  Comparison of the simulated 10-m wind (units: m s–1) at 06:00 on September 29 with ERA5 data.

    In terms of sensible heat flux and latent heat flux (Figs. 6c, 7c, 6b, and 7b), the ACM2 scheme indicated the maximum values, while the MRF scheme indicated the minimum values. The maximum value ranges for both schemes varied within the radius of 40–50 km. The YSU scheme simulated the maximum values of sensible heat flux and latent heat flux near the 30 km radius, closer to the TC center, which corresponded to a compact TC eye (Figs. 6a and 7a). Furthermore, the maximum value ranges simulated by the YSU scheme appeared earlier, persisted for a longer duration, and contributed to a faster development and increased intensity of the TC. In contrast, the MRF scheme exhibited slow and limited increases in sensible heat flux and latent heat flux, resulting in the weakest TC. Moreover, the MRF scheme simulated the weakest friction velocity, while the YSU scheme simulated the strongest friction velocity (as shown in Fig. 8). The friction velocity simulated by the YSU scheme increased earlier, and the maximum value range varied at a radius of 30 km. This indicated that the strongest turbulence of vertical transport simulated by the YSU was closer to the TC center and that TC could be enhanced more easily.

    Figure 6.  The radial distribution of surface sensible heat flux over time (units: W m–2).

    Figure 7.  The radial distribution of surface latent heat flux over time (units: W m–2).

    Figure 8.  The radial distribution of friction velocity over time (units: m s–1).

  • The PBL convergence simulated by the YSU scheme was remarkably stronger than those of the other two schemes (Fig. 9). This strong convergence was observed below 0.5 km and was most prominent in the radius of 50 km, closer to the TC center. On the other hand, the ACM2 scheme simulated weaker low-layer convergence than the YSU scheme did, and the strongest inflow airflow was mainly distributed in the radius of 50–100 km.

    Figure 9.  The average radial velocity (units: m s–1) at radius height at 06:00 on September 29.

    During the mature stage of the TC, the PBL within the inner core zone, as simulated by the three schemes, remained in an unstable stratification state (as depicted in Fig. 10), with the YSU scheme exhibiting the most noticeable effect. The YSU scheme simulated a significantly higher equivalent potential temperature in the lower layer compared to the other two schemes, resulting in lower pressure and a stronger TC. In the YSU-simulated eye wall (approximately within a radius of 30 km), the variation of the equivalent potential temperature with height tended to be constant, indicating violent turbulence in this region. Meanwhile, water vapor, heat, and momentum were well mixed within the PBL. Similar phenomena were observed in the eyewalls simulated by the ACM2 and MRF schemes (within a radius of approximately 40–50 km).

    Figure 10.  The average equivalent potential temperature (units: K) at radius height at 06:00 on the 29th.

    In the TC eye zone, the YSU-simulated specific humidity had a prominent peak (Fig. 11). Compared to Fig. 10, the YSU simulated a deep layer of high temperature and high humidity, which was more conducive to convective development and the strengthening of the TC. Below 0.5 km, the MRF scheme simulated significantly lower water vapor content, indicating a relatively dry atmospheric environment. As a result, the turbulence in the PBL as simulated by the YSU scheme was stronger and exhibited more effective mixing in the inner core of the eyewall.

    Figure 11.  The average specific humidity (units: g kg–1) at radius-height at 06:00 on September 29.

    The virtual potential temperatures simulated by the three schemes were distributed differently in the TC eye zone (Fig. 12). A warm-core structure was evident at approximately 1 km above the eye zone center, with the YSU scheme showing the most pronounced effect. When comparing the 314-K virtual potential temperature contours simulated by the three schemes, we found that the contour simulated by the YSU scheme dipped to its lowest point at 0.6 km above the TC eye center. The 314-K contour simulated by the MRF scheme and the ACM2 scheme was located above 1 km in the center of the TC eye. Above 2 km, the YSU scheme simulated virtual potential temperature reaching 330 K, significantly exceeding those of the other two schemes. Based on the comparison of the three schemes in terms of the average virtual potential temperature and specific humidity in the TC radius of 50 km (Fig. 13), the MRF scheme simulated a relatively dry near-surface layer, with the upper level of the boundary layer being relatively wet and dry. This resulted in poor TC development and the weakest TC intensity.

    Figure 12.  The average virtual potential temperature (units: K) at radius-height at 06:00 on September 29.

    Figure 13.  The average virtual potential temperature (left, units: K) and specific humidity (right, units: g kg–1) within a radius of 50 km in the TC eye area at 06:00 on September 29.

    The distribution of turbulent diffusivity (Km) in the TC eye area varied among the schemes, with differences in the location of the maximum Km values (Fig 14). The region with maximum Km simulated by the YSU scheme was closer to the TC center, indicating a more compact TC eye and the strongest TC intensity. Conversely, the MRF scheme produced the furthest location from the TC center, resulting in a more dispersed TC structure and the weakest TC intensity.

    Figure 14.  Radial profiles of Km (units: m2 s–1) within the lowest 1 km layer at 06:00 on September 29.

  • The YSU-simulated tangential wind was saliently stronger than those of other schemes (Fig. 15). The simulated main circulation reached a higher height, and the TC system was deeper. The ACM2 scheme ranked second in terms of tangential wind intensity and the MRF-simulated tangential wind was the minimum. The strongest tangential wind simulated by the three schemes occurred in a radius of 50 km and at a height of 0.5 km; however, the YSU-simulated tangential wind was closer to the TC center. The comparison of the circulation gradients of tangential wind shows that the YSU-simulated tangential wind had a larger gradient than those of the other two schemes and that the TC structure was tighter.

    Figure 15.  The average tangential wind and radial wind (left column, contour line represents tangential wind, colored shadow represents radial wind), and vertical velocity (right column) at radius height at 06:00 on September 29 (units: m s–1).

    The inflow mainly occurred at a low level below 2 km (Fig. 15), and the outflow mainly occurred at a high level above 14 km. Meanwhile, the strongest outflow airflow occurred at a height of 16 km. In terms of inflow and outflow airflow, the YSU scheme simulation demonstrated the highest intensity, followed by the ACM2 scheme, while the MRF scheme had the weakest airflow.

    A strong upward airflow occurred around the TC eyewall (Fig. 15). The upward and downward airflow simulated by the YSU were the strongest, with the ACM2 being weaker than the YSU. The MRF simulated the minimum vertical velocity, and the maximum vertical velocity occurred at a height of 14 km.

    Thus, the YSU-simulated tangential wind, radial wind, and vertical velocity were larger than other schemes in the TC development stage, followed by ACM2, and those of the MRF were the smallest. Emanuel's Wind-Induced Surface Heat Exchange (WISHE) mechanism explained the influence of wind-induced surface heat exchange in the TC development process. It emphasized the important role of surface energy in the TC reinforcement process. However, the vertical mixing in the PBL was also very important for the strengthening of TC. The YSU increased the entrainment effect of the PBL, and the TC intensity was the strongest. The MRF scheme lacked proper handling of PBL entrainment, and this indicated the limitation of the excessive mixing of thermal and dynamic effects. Consequently, the simulated friction velocity, tangential and radial wind velocity, and vertical mixing were the weakest. The TC structure appeared scattered and vertical transport was insufficient. The MRF simulation exhibited the weakest TC intensity, with minimal convective precipitation. The ACM2 scheme used the nonlocal upward and local downward mixed asymmetric convection modes. It reduced the excessive development of thermal-free convection at the eyewall. Consequently, it produced a more reasonable TC structure and the TC intensity was closest to the observations.

  • In this study, three first-order and nonlocal closure PBL parameterization schemes were selected (i.e., YSU, MRF, and ACM2) for simulation and comparison experiments on the local TC Wutip, which originated in the South China Sea in 2013, using the WRF mesoscale model. The three experiments shared the same near-surface layer scheme (MM5 Monin-Obukhov) and land surface process scheme (Noah). However, they employed different mixing mechanisms in the PBL scheme, which led to the differences in the simulated TC track, intensity, dynamics, and thermal structure.

    In terms of TC tracks, the ACM2 scheme exhibited the smallest deviation compared to observations, while the MRF scheme showed the largest deviation. The YSU scheme simulated the strongest TC intensity, while the MRF scheme portrayed the weakest intensity. The ACM2 scheme yielded the most consistent results with observations in terms of TC intensity. When it comes to TC speeds, the YSU scheme simulated the fastest speeds, while the ACM2 scheme simulated the slowest. The strong precipitation zone simulated by the three schemes distributed in the northwest-southeast direction. The YSU-simulated precipitation distributed relatively consistent with the observation. The MRF-simulated precipitation was generally weaker than the observation. In terms of 10-m wind speed, all three schemes simulated stronger winds than observed. The YSU scheme simulated the highest wind speed, while the MRF scheme's simulation was the closest to the observed wind speed.

    Although the TC development process can be explained by the conditional instability of the second kind, the ultimate TC intensity is closely correlated with the eyewall structure via physical processes. In the simulation using the MRF scheme, the TC exhibited the weakest warm core and a relatively dispersed structure, with an eye area radius of approximately 40 km. Consequently, the TC intensity was the weakest. The YSU scheme simulated a TC with a warmer eye area and a more compact structure, characterized by an eye area radius of about 30 km. As a result, it had the strongest TC intensity. The ACM2 scheme simulated the eye area radius similar to that of the MRF, but with a stronger warm core structure. Therefore, the TC intensity simulated by ACM2 was stronger than that of MRF and was closest to the observation.

    Moreover, the three schemes varied in simulating the dynamic and thermodynamic structure of the TC. The MRF scheme resulted in a relatively dryer and warmer near-surface layer and a relatively wetter and colder upper PBL. The eyewall had minimal wet static energy and weaker convection. The surface sensible heat flux, latent heat flux, friction velocity, wind speed, and warm-core structure simulated by the MRF scheme were the weakest. The surface sensible heat flux, latent heat flux, and friction velocity simulated by the YSU scheme appeared earlier in the time of sustained enhancement around the radius of 30 km. The turbulence simulated by the YSU scheme was stronger, and the mixing in the PBL was more adequate. The corresponding upward motion and secondary circulation were intensified. The low-layer radial wind showed a stronger convergence, while the tangential wind formed a stronger circulation, which contributed to the strongest TC. The ACM2 scheme showed a stronger surface sensible heat flux and latent heat flux. However, it also had a more dispersed eye area compared to the YSU scheme, with a radius of approximately 40 km. As a result, the strongest heating was far away from the TC center. Outside the eye wall, the radius of the maximum wind velocity was considerably larger than the YSU scheme, which avoided restricting the dynamically forced turbulent motion outside the eyewall. Moreover, The YSU scheme simulated more reasonable structural characteristics of the PBL and TC.

    In summary, among the three PBL schemes, the MRF scheme was over-mixed. The YSU scheme added an explicit processing for the top entrainment process of the PBL. The ACM2 scheme applied the nonlocal upward and local downward mixed asymmetric convection modes. These variations led to significant differences in the structural characteristics of the temperature, humidity, and wind fields simulated by the model with three PBL schemes. These characteristics also influenced the distribution of surface fluxes and led to the differences in TC intensities. This study provides insights into the impact of different mixing mechanisms employed by various schemes on the simulation of TCs based on individual cases. However, it is important to note that this study focused on a single case of TC Wutip, and further experiments with multiple TC cases of varying tracks and intensities, using multiple PBL schemes, are necessary to explore the interactions of different physical processes in the simulation.

Reference (47)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return