210 likes | 227 Vues
This study presents a new TC-initialization scheme improving typhoon simulations via WRF model. It outlines the methodology, results, and statistical evaluations for 2006 typhoons over the South China Sea, emphasizing the need for precise vortex specifications and model adjustments. The proposed technique optimizes vortex structures and environmental adaptability, enhancing forecast accuracy. Statistical verification and comparisons with various initialization methods are discussed, validating the effectiveness of the new approach.
E N D
Typhoon Track and Intensity Simulations by WRF with a New TC-Initialization Scheme HIEP VAN NGUYEN and YI-LENG CHEN Department of Meteorology, University of Hawaii, Honolulu, HI 96822, USA
Outline + Review on TC-initialization + A new TC-initialization technique + Results + Statistical evaluation for four typhoons over the South China Sea in 2006 + Summary
Review on TC-initialization + TCs form and develop over open ocean where observations are few. + The initial conditions for meso-scale models are interpolated from coarse resolution global analysis + The analyzed initial vortices are too large and too weak (Kurihara et al., 1993). + Vortex specification: removing the analyzed vortex, construct a bogus vortex, and insert the bogus vortex to the observed TC location (Iwasaki et al., 1987; Mathur, 1991; Kurihara et al., 1993)
TC initialization can be divided into three groups: • a bogus vortex constructed by model integration; • (Kurihara et al., 1993; Lui et al., 1997) • (2) a bogus vortex constructed by analytic empirical functions; (Fujita, 1952; Holland, 1980; Chan and Williams, 1987; Iwasaki et al., 1987; Mathur, 1991; Nam and Davis, 2001; Kwon and Cheong, 2009). • (3) a bogus vortex constructed by three- (3DVAR) or four-dimensional variational data assimilation (4DVAR) with bogus data as one of observation sources. ( Zou and Xiao, 2000; Pu and Braun, 2001; Zhao at al., 2007; Zhang et al., 2007; Wang et al., 2008)
Objectives: To propose a new TC-initialization technique for WRF model with an initial bogus vortex which is well adjusted to the environment and compatible with the WRF model employed. Assumptions: + In a short period of time (<1 hour), the TC moves, however, its structure does not change significantly. + TC structure at the initial time is a function of environment conditions including SST, land surface properties, environment winds and other environment meteorological variables. Methodology: Integrate model for a short period of time (dt=1hour), the vortex structure at t=t0+dt is used to construct vortex structure at the initial time (t=t0) for the next cycle run. The initial TC in the model is well adjusted to the environment after a number of cycle runs.
Separate the vortex part The cycle process only applies for the vortex part and in a radius of R from the current TC center. A variable, F, is first separated into vortex part and environment Where: FV, FE are the vortex part and environment part of F, respectively
Separate the vortex part (cont) FE is computed using modified methods of Kurihara et. al. (1993) Where • [ ] round an argument to the nearest whole number • Delta is horizontal grid resolution in km for each domain.
Cycle runs After each cycle, the vortex part is updated with Where dx, dy are the differences x and y directions between observed TC center and simulated TC center at t+dt • The weighting function, W, is similar to that in Kwon and Cheong (2009)
Typhoon Morakot (2009) Time: 00 UTC 6 August, Observed: Pmin=960 mb Vmax=38 m s-1 The difference in vortex part between the analysis and after vortex initialization for the NT run for (a) SLP (hPa) and (b) 10 m wind vector (m s-1
Time: 00 UTC 6 August, Observed: Pmin=960 mb Vmax=38 m s-1 SLP U10 CTRL East-west cross section along TC center (23.0 N) Thick: SLP Thin: 10 m wind speed SLP U10 WB SLP U10 NT
MORAKOT (2009) Radar reflectivity at 1200 UTC 8 August, 2009 for (a) observed simulated reflectivity of more than 25 dBz for (b) the NT, (c) CTRL and (d) WB runs after 60 h of integrations CTRL obs NT WB
Compare with QuikScat at 10 UTC Aug 6 2009 Black cross is satellite estimated TC center CTRL QuikScat Core is too weak, eye is too large, not right location, wind is too weak Morakot 2009 QuikScat 12 CTRL
Compare with QuikScat at 10 UTC Aug 6 2009 Black cross is satellite estimated TC center QuikScat WRF Bogus Wind near the core is too weak, so does TC intensity, wind is too strong outside Morakot 2009 13
Compare with QuikScat at 10 UTC Aug 6 2009 Black cross is satellite estimated TC center New scheme QuikScat Morakot 2009 14
Typhoon Chanchu (2006) Vertical-eastwest cross section through the center at 0000 UTC 15 May + horizontal wind speed (contour, m s-1), + vertical wind vector (vector, m s-1) + total condensate mixing ratio (g kg-1) for cycle number (a) 4, (b) 20, and (c) 80. 15
Statistical verification for the four typhoons over the South China Sea in 2006 The name, number of run cases, and initial time in 2006 for the four TCs used for the experiments
Super Typhoon Chanchu (2006) • Mean absolute errors for four typhoons (2006) over the South China Sea WB CTRL NT CTRL WB NT CTRL WB NT
Summary on model initialization + A model self-bogus vortex was constructed for WRF to provide high-resolution initial conditions for tropical cyclone simulations. + Three separate simulations including CTRL, WB, NT were performed for seven typhoons over NWP +The NT runs show advantages in generating realistic vortex features including SLP, winds, warm core, and TC size. +The NT scheme shows significant improvements in TC simulations including asymmetric structure, track + The NT scheme shows advantages in intensity simulation at forecast period up to 36h
Acknowledgements This work was funded by the Pacific Disaster Center, Kihei, Hawaii. We also would like to thank the USDA Forest Service; the University of Hawaii/Maui High Performance Computing Center (UH/MHPCC) for helping to fund this research; the Joint Institute of Marine and Atmospheric Research (JMAR)/NOAA for funding the publication costs; Profs. F. F. Jin, P.-S. Chu, D. E. Stevens, and K. F. Cheung for their comments; Mei-Yu Chang and Dr. S.-C. Lin of Central Weather Bureau, Taiwan, for the land surface data and rainfall data used in this research. 20
Main reference Nguyen, H. V., and Y.-L. Chen, 2011: High Resolution Initialization and Simulations of Typhoon Morakot (2009). Mon. Wea. Rev.139,1463-1491 Nguyen, H. V., and Y.-L. Chen, 2011: WRF Initialization and Simulations of Four Typhoons in 2006 over the South China Sea. (In Preparation).