WMO/CAS/WWW

FIFTH INTERNATIONAL WORKSHOP on TROPICAL CYCLONES


Topic 3.1: Tropical Cyclone Motion: Numerical and Statistical Model Guidance and Improvements

Rapporteur: Noel E. Davidson
Bureau of Meteorology Research Centre
PO Box 1289K
Melbourne, Australia. 3001

E-mail: n.davidson@bom.gov.au
Fax: +613-9669-4660


Working Group: Mike Fiorino, Jim Goerss, Akhilesh Gupta, Julian Heming, Joe Kwon, Masashi Nagata, David Richardson, Keon-tae Sohn, Naomi Surgi, Xu Yiming.

Abstract:

Due to advanced data assimilation methods, continuous model development, higher resolutions, and vortex specification, very significant progress is being made on numerical prediction of storm tracks. We document the current state-of-the-art for a single model in each of the TC basins. Such progress is expected to continue.

To improve short-term prediction of the details in tracks, major issues are mesoscale analysis, vortex initialization and the performance of physical parameterizations at high resolution.

Diagnosis of systematic errors and major forecast failures would identify system weaknesses for forecasters and provide feedback to system developers. Careful diagnosis of the impact of vortex specification on short to medium range prediction should be undertaken. Establishing a simple procedure to facilitate model intercomparison by exchange of initial conditions is suggested. Further diagnostic studies and tools are also required for interpretation of model forecasts.


So that improvements in numerical prediction eventually result in improved global-wide, operational forecasts, it is recommended that operational centers which use vortex specification extend the bogussing to all basins, and all centers transmit forecast positions out to 120 hours. Real-time verification of forecast tracks would also be valuable. Further support is also recommended for the WGNE intercomparison project, as implemented by JMA, to provide web sites for display and intercomparison of forecast tracks and NWP output, both in real-time and at the end of each season.

3.1.1: Introduction


Since 1995, forecast track errors from numerical prediction systems have decreased by approximately 25% at 24 hours and 50% at 72 hours (eg, McAdie and Lawrence, 2000). This has been due to (i) more and better use of observational data to define the large-scale environment and outer structure of storms, (ii) continuous model development including improvements in physical parameterizations, (iii) higher resolution of forecast systems, and (iv) for short lead times, the use of some form of vortex specification (bogussing). Track predictions have improved to such an extent that forecasters require very strong evidence to disregard model guidance.

There is no doubt that ongoing development on (i) - (iv) above will continue to result in improved track predictions. Leslie et al. (1998) suggest improvements of 30% are still possible. But are there specific tropical cyclone issues that should be addressed? Can the range of useful forecasts be extended to at least 5 days, and can improvements be made in the prediction of the details in tracks out to say 2 days?



3.1.2: Current Status of Numerical Track Prediction


Table 1 shows track forecast errors for the NW Pacific for the 2002 season up to September. The first row shows errors for JTWC’s consensus aid (see report from working group 3.2) which utilizes interpolated versions of 8 models : NOGAPS (Hogan and Rosmond, 1991, Barker, 1992, Goerss and Jeffries, 1994), UKMO (Cullen, 1993, Heming et al, 1995), JMA (JMA, 2002, Nagata et al. 2001a,b), NCEP (Surgi et al.,1998 and references contained therein), MM5 (run by US Air Force), GFDN (GFDL model run by FNMOC using NOGAPS [Kurihara et al., 1998, Rennick, 1999]), and COAMPS (Hodur, 1997). The second row shows the range of forecast errors for the individual models. JTWC’s official track forecast errors are very similar to the consensus errors.


The quality of the consensus and even the individual system forecasts truly indicates the skill that can be achieved using numerical forecast guidance. An interesting feature is the spread in the quality of model forecasts. It appears that even some relatively poor model forecasts contain useful information. We note however that these statistics are for a single year and that substantial interannual variability exists in the quality of forecasts.



Forecast Duration (hrs)

24

48

72

96

120
Consensus Error (km)

128

216

289

387

503
Range of individual model errors(km)

140 – 181

240 – 303

335 – 459

435 – 527

592 – 666

Table 1 : Mean forecast track errors (km) for the NW Pacific up to September 2002 for consensus of 8 models and for individual models.


Table 2 shows mean track forecast errors from the UKMO global system for the three latest seasons up to northern summer 2002 for each of the six TC basins. The basins are ordered top to bottom in terms of absolute skill in the 48-hour forecast. The number of forecasts is indicated in brackets. These statistics document current state-of-the-art for each basin in terms of guidance from an individual numerical forecast system. Because relatively few forecasts are made beyond 72 hours, the statistics at these longer lead times are less robust. Forecasts in the northern hemisphere are more skilful than those in the southern hemisphere, reflecting the data availability issue and also possibly the greater variability in motion, and hence difficulty, in the southern hemisphere. Performance is best in the NE Pacific and worst in the SW Indian Ocean. Since persistence in the motion affects the degree of difficulty in the forecast, verification against CLIPER (eg, Bessafi et al, 2002) is needed to help define the source of this difference. The generally skilful performance over all basins likely reflects the improved use of synthetic TC and remotely-sensed observational data. There is some evidence that error growth is largest in the first 24 hours, suggesting improvements in initialization are still possible.

3.1.3: Systematic Errors

An important issue in numerical track forecasting is that of systematic errors. The CAS/JSC Working Group on Numerical Experimentation (WGNE) has an ongoing project on intercomparison of typhoon track forecasts from operational models (Tsuyuki et al., 2002). The following results are based upon that project and input from working group members.

(i) Larger track errors occur for weak storms. Difficulties arise with the vertical structure of storms and appropriate steering levels. Environments also tend to be less well-defined, making tracks more difficult to forecast.

(ii) Most models possess a slow bias, possibly associated with deficiencies in the large-scale wind forecast.

(iii) The BMRC model has larger along-track (cross-track) errors for storms moving zonally (meridionally).

(iv) In the past, most systematic errors have been related to resolution and the representation of moist processes. There are preliminary indications that these factors still contribute to systematic performance

Further work on verification and statistical diagnosis are required to quantify systematic errors in track prediction under different synoptic conditions. This would not only be of great assistance to forecasters but also identify specific NWP weaknesses that need to be addressed. Elsberry et al. (1999), Fraedrich et al. (2000) and Kwon and Sohn (2002, personal communication) report on success with statistical post-processing of numerical model outputs. Also, is it possible to use the “Systematic Approach” of Carr and Elsberry (2000a), (2000b), Carr et al. (2001) to identify the cause of systematic errors?




Basin

Forecast Duration (hours)

        

24

48

72

96

120

NEP

102 (258)

188 (181)

290 (123)

434 (88)

606 (61)

NI

130 (54)

220 (36)

313 (20)

305 (11)

278 (6)

NA

132 (330)

241 (257)

332 (194)

384 (145)

476 (111)

NWP

134 (517)

251 (397)

371 (279)

508 (195)

679 (118)

AUST

160 (207)

269 (142)

382 (91)

476 (53)

632 (30)

SWI

157 (270)

276 (224)

370 (181)

475 (137)

580 (99)


Table 2 : Mean forecast track errors (km) from UKMO for 3 seasons prior to northern summer 2002, for six TC basins : North East Pacific, Northern Indian Ocean, North Atlantic, North West Pacific, Australian, and South West Indian Ocean. The number of forecasts in each category is shown in brackets.



3.1.4: Major Forecast Failures

Examples of recent major forecast failures for individual numerical systems over the NW Pacific have been Rex (1998), Saomai (2000), and Nari (2001). Aside from the potential for a major disaster, these forecast failures also affect verification statistics, and the skill of intensity forecasts. The vast majority of forecasts provide very useful guidance. Unfortunately, a few are extremely poor and have the potential to result in catastrophes. Table 3 shows for 48-hour forecasts, the number of track errors that fell within 100 km bins from the BMRC TC system (Davidson and Weber, 2001) for the NW Pacific in 2001. Relatively few (5 out of 89 forecasts) have errors larger than 500 km. The mean track error for the same sample is a quite respectable 240 km. The forecast failures were single base times for Kong-rey, Danas, Nari, Krosa and Haiyan.




Bin(km)

0-100

100-200

200-300

300-400

400-500

500-600

600-700

>700

Number

14

25

27

10

8

4

1

0


Table 3. For 48-hour forecasts from BMRC TC system, number of errors in 100km bins.

Are major forecast failures a natural consequence of numerical prediction, and will always occur? Can they be eliminated? Should they undergo detailed investigation (eg, Henderson et al., 1999)? Are they purely related to deficiencies in initial conditions or model configurations? How do ensemble methods perform on such events? Should they be the subject of intensive data impact studies (eg, Velden et al., 1998, Goerss et al., 1998)? It would be valuable to develop an archive of major forecast failures so that some of these questions can be addressed. It would also be useful to establish a procedure to facilitate model intercomparisons by exchange of initial conditions via a web or ftp site, similar to what as been suggested by Nagata et al., (2001).

3.1.5: Research Issues

Developments in data assimilation and numerical models will continue, independent of the TC forecast problem. Ongoing improvements to 3D and 4DVAR assimilation methods (eg, Rabier et al., 2000, Zou et al., 2001), with the inclusion of new data sources, and the use of physical initialization procedures (eg, Krishnamurti et al., 1997) will continue to produce improvements in the definition of the large scale environment and outer structure of storms (and hence in track forecasts). For those basins fortunate enough to have aircraft reconnaissance, availability of dropsonde data and eventually advanced NWP targeting strategies should result in further significant improvements in track forecasting (eg, Aberson and Franklin, 1999).


The question of vortex specification remains open. Its use is possibly dependent on the application. For short-term track and intensity forecasts it would seem to be necessary in the near future. For longer-term forecasts it may not be required and may even have a negative impact. Careful diagnosis of its impact on short to medium range prediction should be undertaken. With increases in resolution, the question also arises of how much real TC structure can or needs to be represented in initial conditions. The issue of vortex initialization (eg, Zou and Xiao, 2000), particularly for intense storms, and even without a synthetic vortex (eg, Liu et al., 2002), will become increasingly important and should be investigated.

With the continuous improvement in numerical guidance on track, there appears to have been a slight reduction in basic research on TC motion. For example, it has been suggested that at least two upper tropospheric cold lows might have affected the track of Typhoon Rex. How do upper tropospheric cold lows affect the motion, or how do they affect midlevel steering flows? The problem of these environmental interactions, erratic tracks, stalling or rapidly-accelerating storms, and recurvature generally, are still worthy of investigation. Some recent studies relating to these issues are : Shapiro and Franklin (1999), Wu and Kurihara (1996), Wu and Wang (2000). Other issues affecting motion include horizontal and vertical shear, as discussed for example by Wang et al. (1997) and Wang and Holland (1996).

For short-term track forecasts, is it possible to improve the details in forecasts via improved mesoscale assimilation techniques (3DVAR, 4DVAR) that take advantage of the latest observational data sets with high spatial and temporal resolution, eg, scatterometer, radar, rainrate and remotely-sensed data (eg, Peng and Chang, 1996, LeMarshall and Leslie, 1998, Marecal et al., 2001)? At these high resolutions, initialization and the performance of physical parameterizations become significant research issues in themselves.

The effect of surrounding MCSs and convective asymmetries on TC motion is mostly unexplained (Harr and Elsberry, 1996, Ritchie and Elsberry, 2000). Now that models can generally handle the “smooth” motion of storms, is it possible to predict details in the tracks, that might be related to such mesoscale features in the environment?

For short-term, high-resolution prediction, should attempts be made to simulate eyewall cycles and their potential effect on small-scale track deviations and trochoidal motion?

Questions relating to storm track, structure and rainfall near topography and at landfall are still mostly unanswered. The (subjective) BMRC experience has been that numerical forecasts tend to be less consistent from one base time to the next for storms moving along or across the coast. Further work is needed to determine how to initialize circulations in such circumstances and to understand the interactions between the storm’s circulation and the underlying surface. In this regard, it is believed that advanced atmosphere-ocean-land assimilation-prediction systems may be required.

3.1.6: Operational Issues

To take advantage of the progress made in numerical prediction of storms in operational forecasting centers, the following issues are recommended for serious consideration :

  1. that all major NWP centers, even those who do not use vortex specification, transmit forecast TC positions out to 120 hours from every run of their model,
  2. that those major NWP centers (apart from UKMO, FNMOC, and NCEP) that routinely use TC vortex specification extend their bogussing to all basins.
  3. request that major NWP centers involved in short-term, high resolution TC prediction transmit forecast tracks and intensities, and provide access to web sites for evaluation of detailed structure change (eg, size, asymmetries, rainfall).
  4. that further support be given to the WGNE intercomparison project, as implemented by JMA, to provide central web sites for display and intercomparison of forecast tracks and NWP output, both in real-time and at the end of each season.
  5. that development of diagnostic tools for model interpretation be undertaken.
  6. that significant research results be identified, and resources made available to transition them into operations.

3.1.7: Future Improvements

For extended range forecasts (beyond 3 days), improvements in track prediction will mostly result from points (i)–(iv) in the introduction. More and better use of remotely-sensed data will undoubtedly be a key to ongoing improvements.

For further improvements in short-term forecasts, a major issue will be mesoscale analysis and initialization, eventually within a coupled atmosphere-ocean-land system. To define details in storm structure and mesoscale aspects of the storm’s environment is a challenging problem. It will no doubt require very high resolution, use of some of the data sets previously mentioned, and in the short-term, the use of a synthetic vortex. At high resolution and for intense circulations, other significant issues are initialization, and the performance of physical parameterizations. These are all important areas of research.

3.1.8: Summary

Since 1995, there have been substantial improvements in numerical prediction of storm tracks. This has been due to (i) more and better use of observational data, (ii) continuous model development including improvements in physical parameterizations, (iii) higher resolution of forecast systems, and (iv) for short lead times, the use of synthetic (bogus) vortices in initial conditions. These have led to improved initial state specification of the large-scale environment and structure of storms, and to a reduction in systematic errors in models. Such improvements are expected to be ongoing.

Opportunities exist to improve short-term prediction, which will likely involve mesoscale assimilation, vortex initialization methods and the use of high temporal and spatial resolution observational data such as, scatterometer, radar, rain-rate, and remotely-sensed data generally. These should better define the storm’s environment and mesoscale structure. However, attention is drawn to the significant issues of initialization and the performance of physical parameterizations at very high resolution.

Verification and statistical analysis of forecast tracks are needed to understand systematic errors in all models. This will be valuable for operational forecasters, possibly form the basis for statistical correction of forecast tracks, and provide objective guidance on numerical system deficiencies. As part of the verification issue, major forecast failures should be documented and diagnosed at a later time.

Careful diagnosis of the impact of vortex specification on short to medium range timescales should be undertaken. Establishing a simple procedure to facilitate model intercomparison by exchange of initial conditions would also be valuable. Further diagnostic studies and tools are still required for interpretation of model forecasts.

So that improvements in numerical prediction eventually result in improved global-wide, operational forecasts, it is recommended that operational centers which use vortex specification extend the bogussing to all basins, and all centers transmit forecast positions out to 120 hours. Real-time verification of forecast tracks would also be valuable. We recommend that further support be given to the WGNE intercomparison project, as implemented by JMA, to provide web sites for display and intercomparison of forecast tracks and NWP output, both in real-time and at the end of each season.

Finally, the group would like to emphasize the need for ongoing international collaboration on both modeling capabilities and warning strategies. Perhaps as a first step it would be valuable to establish a simple procedure to exchange initial conditions to facilitate model intercomparisons on events of particular interest.

Bibliography


Aberson, S.D. and J.L. Franklin, 1999: Impact on hurricane track and intensity forecasts of GPS dropwindsonde observations from the first-season flights of the NOAA Gulfstream-IV jet aircraft. Bull. Amer. Met. Soc., 80, 421-427.

Baker, N., 1992. Quality control for the Navy operational atmospheric database. Wea. Forecasting, 7, 250-261.

Barker, E., 1992. Design of the Navy's multivariate optimum interpolation analysis system. Wea. Forecasting, 7, 220-231.

Bessafi, M., A. Lassere-Bigorry, C.J. Neumann, F. Pignolet-Tardan, D. Payet and M. Lee-Ching-Ken,2002 : Statistical prediction of tropical cyclone motion: An analog-CLIPER approach. Wea. Forecasting. 17, 821-831.

Carr, L.E. III and R.L. Elsberry, 2001: Beta test of the systematic approach expert system prototype as a tropical cyclone track forecasting aid. Wea. Forecasting, 16, 355-368.

Carr, L.E. III and R.L. Elsberry, 2000a: Dynamical tropical cyclone track forecast errors. Part I: Tropical region error sources. Wea. Forecasting, 15, 641-661.

Carr, L.E. III and R.L. Elsberry, 2000b: Dynamical tropical cyclone track forecast errors. Part I: Midlatitude circulation influences. Wea. Forecasting, 15, 662-681.

Carr, L.E. III and R.L. Elsberry, 1998: Objective diagnosis of binary tropical cyclone interactions for the western north Pacific basin. Mon. Wea. Rev., 126, 1734-1740.

Chandrasekar, A. and B.N. Goswami, 1996: Feedback between convective heating and dynamics and movements of tropical cyclones. Meteor. Atm. Phys., 60, 165-178.

Cullen, M.J.P., 1993: The unified forecast/climate model. Meteorol. Mag., 122, 81-122.

http://www.metoffice.com/research/nwp/numerical/index.html
http://www.metoffice.com/research/nwp/numerical/unified_model &
/new_dynamics.html


Davidson, N.E. and H.C. Weber, 2001. The BMRC high-resolution tropical cyclone prediction system: TC-LAPS. Mon. Wea. Rev., 128, 1245-1265.

Elsberry, R.L., M.A. Boothe, G.A. Ulses and P.A. Harr, 1999: Statistical postprocessing of NOGAPS tropical cyclone track forecasts. Mon. Wea. Rev., 127, 1912-1919.

Fulton, S.R., 2001: An adaptive multigrid barotropic tropical cyclone track model. Mon. Wea. Rev., 129, 138-151.

Fraedrich, K., R. Morison and L.M. Leslie, 2000: Improved tropical cyclone track predictions using error recycling. Met. Atm. Phys., 74, 51-56.

Goerss, J.S., C.S. Velden and J.D. Hawkins, 1998: The impact of multispectral GOES-8 wind information on Atlantic tropical cyclone track forecast in 1995. Part II: NOGAPS forecast. Mon. Wea. Rev., 126, 1219-1227.

Goerss, J., and R. Jeffries, 1994. Assimilation of synthetic tropical cyclone observations into the Navy Operational Global Atmospheric Prediction System. Wea. Forecasting, 9, 557-576.

Goerss, J., and P. Phoebus, 1992. The Navy's operational atmospheric analysis. Wea. Forecasting, 7, 232-249.

Gu, Jian-feng, X.-A. Jiang, B. Bao, and Y.-M. Xu, 2000: The operational prediction results of the East China Sea tropical cyclone model and improvement of the model. J. Trop. Met. 16, 54-61.

Harr, P.A. and R.L. Elsberry, 1996: Structure of a mesoscale convective system embedded in Typhoon Robyn during TCM-93. Mon. Wea. Rev., 124, 634-652.

Heming, J.A., A.M. Radford and J. C.-L. Chan, 1995: A new scheme for the initialisation of tropical cyclones in the UK Meteorological Office global model. Meteor. App., 2, 171-184.

http://www.metoffice.com/sec2/sec2cyclone/sec2cyclone.html

Henderson, J.M., G.M. Lackmann and J.R. Gyakum, 1999: An analysis of Hurricane Opal’s forecast track errors using quasi-geostrophic potential vorticity inversion. Mon. Wea. Rev., 127, 292-307.

Hodur, R., 1997. The Naval Research Laboratory's Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS). Mon. Wea. Rev.,125, 1414-1430.

Hogan, T., and T. Rosmond, 1991. The description of the Navy Operational Global Atmospheric Prediction System's spectral forecast model. Mon. Wea. Rev., 119, 1786-1815.

Japan Meteorological Agency, 2002: Outline of the Operational Numerical Weather Prediction at the Japan Meteorological Agency, Appendix to WMO Numerical Weather Prediction Progress Report. JMA 157pp.
http://www.jma.go.jp/JMA_HP/jma/jma-eng/jma-center/nwp/outline-nwp/index.htm

Krishnamurti, T. N., R. Correa-Torres, G. Rohaly, D. Oosterhof, and N. Surgi, 1997: Physical Initialization and Hurricane Ensemble Forecasts. Wea. Forecasting, 3, 503–514.

Kurihara, Y., M. Bender, and R. Ross, 1993. An initialization scheme of hurricane models by vortex specification. Mon. Wea. Rev., 121, 2030-2045.

Kurihara, Y., M. Bender, R. Tuleya, and R. Ross, 1995. Improvements in the GFDL hurricane prediction system. Mon. Wea. Rev., 123, 2791-2801.

Kurihara, Y., R. Tuleya, and M. Bender, 1998. The GFDL hurricane prediction system and its performance in the 1995 hurricane season. Mon. Wea. Rev., 126, 1306-1322.

LeMarshall, J.L. and L.M. Leslie, 1998: Tropical cyclone track prediction using high resolution satellitre data with a new methodology. Aust. Meteor. Mag., 47, 261-266.

Leslie, L.M., R.F. Abbey and G.J. Holland, 1998: Tropical cyclone track predictability. Meteor. Atm. Phys., 65, 223-231.

Li, Y.S. and J.C.L. Chan, 1999: Momentum transports associated with tropical cyclone recurvature. Mon. Wea. Rev., 127, 1021-1037.

Liu, Q., S. Lord, N. Surgi, Pan, H-L, and F. Marks, 2002: Hurricane initialization using reconnaissance data in GFDL hurricane forecast model. Preprints, 25th Conference on Hurricanes and Tropical Meteorology. 29 April – 3 May, 2002, San Diego, California.

Marecal, V., E. Gerard, J.-F. Mahfouf and P. Bauer, 2001: The comparative impact of the assimilation of SSM/I and TMI brightness temperatures in the ECMWF 4D-Var system. Quart. J. Roy. Met. Soc., 127, 1123-1142.

McAdie, C.J. and M.B. Lawrence, 2000: Improvements in tropical cyclone track forecasting in the Atlantic basin. Bull. Amer. Met. Soc., 81, 989-997.

Nagata, M. and co-authors, 2001a: Third COMPARE workshop: A model intercomparison experiment of tropical cyclone intensity and track prediction. Bull. Amer. Met. Soc.,82, 2007-2020.

Nagata, M. and co-authors, 2001b: A mesoscale model intercomparison: A case of explosive development of a tropical cyclone (COMPARE III). J. Meteor. Soc. Japan, 79, 999-1033.

Peng, M.S. and S.W. Chang, 1996: Impacts of SSM/I retrieved rainfall rates on numerical prediction of a tropical cyclone. Mon. Wea. Rev., 124, 1181-1198.

Rennick, M., 1999. Performance of the Navy's tropical cyclone prediction model in the western North Pacific basin during 1996. Wea. Forecasting, 14, 3-14.

Ritchie, E.A. and R.L. Elsberry, 2000: Simulated impacts of a mesoscale convective system on the track of typhoon Robyn during TCM-93. Mon. Wea. Rev., 128, 2232-2251.

Shapiro, L.J. and J.L. Franklin, 1999: Potential vorticity asymmetries and tropical cyclone motion. Mon. Wea. Rev., 127, 103-123.

Soden, B.J., C.S. Velden and R.E Tuleya, 2001: The impact of satellite winds on experimental GFDL hurricane model forecasts. Mon. Wea. Rev., 129, 835-852.

Surgi, N., H.-L. Pan, and S.J. Lord, 1998: Improvement of the NCEP Global Model over the Tropics: An Evaluation of Model Performance during the 1995 Hurricane Season. Mon. Wea. Rev., 126, 1287–1305.

Tsuyuki, T., R. Sakai and H. Mino, 2002: The WGNE intercomparison typhoon track forecasts from operational global models for 1991-2000. WMO Bulletin, July 2002, Vol 51, No 3, 253-257.

Velden, C.S., T.L. Olander and S. Wanzong, 1998: The impact of multispectral GOES-8 wind information on Atlantic tropical cyclone track forecast in 1995. Part I: Dataset methodology, description and case analysis. Mon. Wea. Rev., 126, 1202-1218.

Wang, B. X. Li and L. Wu, 1997: Direction of hurricane beta drift in horizontally sheared flows. J. Atmos. Sci., 54, 1462-1471.

Wang, Y. and G.J. Holland, 1996: Tropical cyclone motion and evolution in vertical shear. J. Atmos. Sci., 53, 3313-3332.

Weber, H.C., 2001: Hurricane track prediction with a new barotropic model. Mon. Wea. Rev., 129, 1834-1858.

Wu, C.-C. and Y. Kurihara, 1996: A numerical study of the feedback mechanisms of hurricane-environment interaction on hurricane movement from a potential vorticity perspective. J. Atmos. Sci., 53, 2264-2282.

Wu, L. and B. Wang, 2001: Movement and vertical coupling of adiabatic baroclinic vortices. J. Atmos. Sci., 58, 1801-1814.

Wu, L. and B. Wang, 2001: Effects of convective heating on movement and vertical coupling of tropical cyclones: A numerical study. J. Atmos. Sci., 58, 3639-3649.

Wu, L. and B. Wang, 2000: A potential vorticity tendency diagnostic approach for tropical cyclone motion. Mon. Wea. Rev., 128, 1899-1911.

Zou, X., J. Derber, J.G. Sela, R. Treadon, I.M. Navon and B. Wang, 2001: Four-dimensional variational data assimilation with a diabatic version of the NCEP global spectral model: System development and preliminary results. Quart. J. Roy. Met. Soc., 127, 1095-1122.

Zou, X. and Q. Xiao, 2000: Studies on the initialisation and simulation of a mature hurricane using a variational bogus data assimilation scheme. J. Atmos. Sci., 57, 836-860.