prediction of track and intensity of the tropical cyclonic ... 35.pdf · prediction of track and...

Prediction of Track and Intensity of the Tropical Cyclonic using High

Resolution WRF-ARW Model

Md. A. E. Akhter*1,2

and M. M. Alam1

1Department of Physics, Khulna University of Engineering & Technology, Bangladesh

2SAARC Meteorological Research Centre, Bangladesh

*Corresponding author : [email protected]

Abstract:

High resolution Advanced Research WRF (ARW) meso-scale model version 3.5.1 adopted from

National Center for Atmospheric Research (NCAR) is used to simulate the track and intensity of the

tropical cyclonic of the Bay of Bengal. WSM3 microphysics and Kain Fritisch’s cumulus

parameterization scheme has been used with Yonsei University planetary boundary layer. Horizontal

resolution of 9 km is used. The objective of this study is to simulate the track, and intensity of the

tropical cyclone Nargis, Thane and Mahasen to observe the performance of WRF (ARW) model. The

model simulated track, sea level pressure and wind are compared with those taken from Indian

Meteorological Department (IMD). Model simulates landfall position and time with fair accuracy.

Keyword: WRF Model, Tropical Cyclone, Intensity, Track, Landfall Position.

1. INTRODUCTION

Tropical cyclones (TCs) are one of the most dangerous natural calamities throughout the globe. Every year, they

cause considerable loss of life and do immense damage to property due to strong MSWs, torrential rainfall and

storm surges. These occur during the pre-monsoon (April–May), early monsoon (June), late and post-monsoon

(September– November) periods. Originating in both the Bay of Bengal and Arabian Sea, tropical cyclones

often attain velocities of more than 160 km/h and are notorious for causing intense rain and storm tides (surges)

as they cross the coast of India, Bangladesh and other coasts. Strong MSWs, heavy and torrential rains and the

cumulative effect of storm surges and astronomical tides are the three major elements of tropical cyclone

disaster. This is also the basin where TCs do the most damage due to the combined effects of terrain, geography,

and lack of communications. The need for sea surface temperature (SST) greater than 26.5◦C and a deep lower

level moist layer and absence of strong vertical shear for development from a tropical low to a cyclonic storm

and further intensification has been referred to in many studies [1]. More than 200,000 people perished in the

Bhola cyclone of 1970 in Bangladesh.

Predictions of tropical cyclones have improved steadily over the last three decades, mostly due to a

combination of better observations, especially the satellite and radar, improvement in dynamical models and

improved understanding of physical processes and mechanisms that govern the motion of tropical cyclones. The

current emphasis of international tropical cyclone research is to achieve greater accuracy of tropical cyclone

track and intensity prediction, especially in the short range (48–72 h in advance), by maximizing the use of non-

conventional data, meso-scale analysis, and the physical parameterization in nonhydrostatic environment at

higher model resolution. In India, development of objective techniques for forecasting track of TCs began in

1972 [3] by using a computer oriented half persistence and half climatology technique. There is a study [3]

related to forecasting movement of TCs by adopting a primitive equation barotropic model. Some authors have

used multi-level primitive equation models with parameterization of physical processes [4, 5]. Quasi-Lagrangian

Model (QLM) has also been implemented in IMD for operational track forecast up to 72 h in the year 2000.

QLM [6] have been evaluated in 2003 and its performance is found to be even better. The effect of different

initial condition on the track and intensity of tropical cyclone are done using MM5 and WRF models [7]. The

simulation of track and landfall of tropical cyclones over the Bay of Bengal is done [8,9], Prediction of severe

tropical cyclones over the Bay of Bengal during 2007–2010 using high-resolution mesoscale model is also done

[10].

In this study, the high resolution Advanced Research WRF (ARW) meso-scale model adopted from NCAR

is used to simulate the track and intensity of the tropical cyclone Nargis, Thane and Mahasen. The objective of

the present study is to evaluate the performance of the model towards simulation of track and intensity of

tropical cyclone.

Page 73

P-ID 35 International Conference on Physics Sustainable Development & Technology (ICPSDT-2015)(August 19-20, 2015)

Department of Physics, CUET.

2. MODEL DESCRIPTION AND INITIAL CONDITIONS

The cumulus convection used in this case is modified Kain-Fritisch scheme [912]. The cloud microphysics

scheme is WRF Single-Moment (WSM) 3-class simple ice scheme, which is a simple efficient scheme with ice

and snow processes suitable for mesoscale grid sizes [10, 11 13,14]. It replaces NCEP 3 scheme. The long-wave

radiation parameterization is the Rapid Radiative Transfer Model (RRTM) scheme, which is an accurate scheme

using look-up tables for efficiency accounts for multiple bands, trace gases, and microphysics species [12,15].

The short-wave radiation scheme is as per the Dudhia scheme, which allows simple downward integration for

efficient cloud and clear-sky absorption and scattering [13,16]. The Planetary Boundary Layer (PBL)

parameterization is the Yonsei University Scheme (YSU) [14,17], which is the next generation MRF–PBL. An

overview of the model used in this study is provided in Table 1.

3. DESCRIPTION OF THE SYSTEM

Very Severe Cyclonic Strom Nargis

The system ‘Nargis’ initially appeared as a low pressure area over southeast Bay of Bengal on 26 April and

concentrated into a depression at 0300UTC of 27 April. The system slightly moved northwestwards and

intensified into a depression and then into a cyclone at 00UTC of 28 April over southwest and adjoining

southeast and west central Bay of Bengal about 550 km east of Chennai. The system remained practically

stationary for some time and further intensified into an SCS at 09UTC of 28 April and into a VSCS at 0300UTC

of 29 April over west central and adjoining southwest and southeast Bay of Bengal. The system moved slightly

to the north and then eastwards for a period of about 30 h from 0600UTC of 1 May till 1500UTC of May

towards the Myanmar coast and crossed the coast of Myanmar between 1730 and 1930 h IST on 2 May 2008 as

a ‘VSCS’ (figure 1). Even after the landfall the system was maintaining its intensity as a VSCS for a period of

about 12 h. The VSCS ‘Nargis’ devastated large parts of the low-lying delta region of the Irrawaddy River.

MSWs exceeded 190 kmph as the storm ripped through the Myanmar’s biggest city Yangon (estimated

population 6 million) for over more than ten hours. Homes were flattened, more sturdy structures damaged, trees

uprooted and power lines downed. In rural parts of the country up to 95% of all homes were destroyed. In the

history of Myanmar it had never faced such a disastrous weather system in past 40 years.

Cyclone Thane

and into a depression in the same evening over the north Interior Tamil Nadu. It weakened further and lay as a

A depression formed over southeast Bay of Bengal at 0900 UTC of 10th

May 2013 near latitude 5.0°N and

longitude 92.0°E. It moved northwestwards and intensified into a deep depression in the evening of the same

day. Continuing its northwestward movement, it further intensified into a cyclonic storm, Mahasen in the

morning of 11th

May 2013. Under the influence of the anticyclonic circulation lying to the east, the cyclonic

storm changed its direction of movement initially from northwesterly to northerly and then to north-

Page 74



First the NCAR version 3.1 of Advanced Research WRF (ARW) model has been used. Detail description of the

model is documented in [11]. For the diagnostic study the GFS data at 1◦×1◦ grid is used from 00 UTC of

December 26 to 00 UTC of December 31, 2011.

in the early morning of 26th

December, 2011 and into a cyclonic storm ‘THANE’ in the same midnight. It then

moved west-northwestwards and intensified into a severe cyclonic storm in the afternoon and into a very severe

A depression formed over southeast Bay of Bengal in the evening of 25th

December, 2011 and lay centred about

1000 km southeast of Chennai. It gradually moved north-northwestwards and intensified into a deep depression

December, 2011 with a MSW speed of 120-140 kmph. After landfall, the system rapidly weakened into a severe

cyclonic storm over north coastal Tamil Nadu at 0830 hrs IST of 30th

and into a deep depression around noon

cyclonic storm in the evening of 28th

December, 2011. It then moved west-southwestwards and crossed north

Tamil Nadu & Puducherry coast between Cuddalore and Puducherry within 0630 and 0730 hrs IST of 30th

well marked low pressure area over north Kerala and neighbourhood in early morning of 31st December, 2011.

Cyclone Mahasen

northeasterly on 13th

and 14th May respectively. On 15

th May, it further came under the influence of the mid-

latitude westerly trough running roughly along 77°E, which further helped in enhancing the north-northeastward

speed of the cyclonic storm. The cyclonic storm crossed Bangladesh coast near lat.22.8°N and long. 91.4°E,

about 30 km south of Feni around 0800 UTC of 16th

May 2013 with a sustained maximum surface wind speed

of about 85-95 kmph. After the landfall, it continued to move north-northeastwards and weakened gradually due

to interaction with land surface. It weakened into a deep depression over Mizoram in the evening and into a

depression over Manipur around mid-night of 16th

. It further weakened into a well marked low pressure area

over Nagaland in the early morning and moved away towards Myanmar as a low pressure area in the morning of

17th

May.

Table 1: Domain design of the model

Horizontal grid distance 9 km

Integration time step 30 s

Map projection Mercator

Horizontal grid system Arakawa-C grid

Vertical co-ordinate Sigma co-ordinates 28 σ levels

Time integration scheme Third-order Runge–Kutta scheme

Spatial Differencing scheme Second to Sixth order schemes

Radiation parameterizations RRTM, Dudhia scheme

Surface layer parameterizations Monin–Obukhov scheme, thermal diffusion scheme

Cumulus parameterization Kain- Fritisch schemes

PBL parameterization YSU scheme

Microphysics WSM 3 simple ice scheme

Land surface Unified Noah Land surface Model

4. RESULTS AND DISCUSSION

The results obtained from several numerical experiments using WRF model are used to predict the intensity and

track of the tropical cyclones Nargis, Thane and Mahasen those form in the Bay of Bengal. The model was run

for 96 hours for intensity prediction and 96 and 72 hours for track prediction. Model output of each cyclone has

been produced at three hours interval and processed using Grid Analysis and Display System (GrADS). Model

simulated Minimum Sea Level Pressure (MSLP), Maximum Sustained Wind (MSW) and tracks are compared

with those reported by India Meteorological Department (IMD) and discussed in the following sections.

4.1. MSLP and MSW of the Tropical Cyclone Events

MSLP and MSW of a TC are of great importance as they help to measure the intensity of a TC. Since TCs

develop over vast oceanic areas, where observations are sparse or not available, it is of great difficulty to make

any validation of model simulated MSLP and MSW with real observable data from sea before the landfall. But

now meteorologists are able to estimate MSLP and MSW using interpretations of satellite products. To obtain

model simulated intensity (MSLP and MSW); model is run for 96 hours for the cyclone Nargis, Thane and

Mahasen. These run are based on the initial fields of 00 UTC on 29 April 2008, 00 UTC of 27 December 2011,

00 UTC of 13 May 2013 for the Cyclone Nargis, Thane and Mahasen respectively. Time variation of model

simulated and observed MSLP and MSW of TC Nargis, Thane and Mahasen are presented as in Figure 1. The

lowest MSLP and highest MSW with their obtaining time are tabulated in the Table 2.

4.1.1. MSLP and MSW of the Tropical Cyclone Nargis

It appears from the Figure1 that model simulated MSLP gradually drops (having little bit oscillation) with time

and attains peak intensity with minimum pressure 956 hPa at 00 UTC on 02 May 2008 and thereafter MSLP

increases gradually. Finally just before the landfall the MSLP is 960 hPa at 03 UTC of 02 May 2008. On the

other hand, the observed MSLP 962 hPa is obtained at 06 UTC of 02 May and remain same up to 12 UTC of 02

May 2008 and thereafter MSLP increases gradually. Landfall of the system occurs at 12 UTC of 02 May with

observed value of MSLP 962 hPa. The variation of model simulated MSLP compare to that of observed with

time shows that the model simulates realistic temporal variation of MSLP but simulated value is higher than

observed values up to 18 UTC of 30 April and then becomes lower up to 03 UTC of May 02 and then becomes

Page 75



Dynamics Nonhydrostatic

Model domain 4.590–26.56

0N, 76.35

0–97.64

0E

higher up to 00 UTC of May 03. Again, the simulated highest MSW is 44 m/s and obtained at 00 UTC of 02

May 2008 where as that for observed is 46 m/s and obtained at 06 UTC of 02 May 2008 and retains this value

up to 15 UTC on 02 May 2008. These lowest values MSLP and highest MSW with their obtaining time are

tabulated in Table 2. From the Figure 1 and Table 2, it is clear that the simulated intensity of cyclone Nargis

intensify earlier than the observation.

4.1.2. MSLP and MSW of the Tropical Cyclone Thane

Figure 1 shows the comparative evolution of observed MSLP and simulated MSLP of TC Thane. It appears

from the Figures 1 that model simulated and observed MSLP gradually drops with time and attains peak

intensity just before the landfall and thereafter MSLP increases. The Model simulated MSLP of 973 hPa is

obtained at 12 UTC of 29 December whereas the observed MSLP 969 hPa is obtained at 21 UTC of 29

December 2011. The model simulated MSLP at the centre of the cyclone is obtained before 09 hours from the

observed MSLP. The variation of model simulated MSLP compared to that of observed with time shows that

both the models simulated realistic temporal variation of MSLP. The simulated highest MSW is 36 m/s and

obtained at 15 UTC of 29 December 2011 where as that for observed is 40 m/s and obtained at 03 UTC of 29

December 2011 retains this value up to 00 UTC on 30 December 2011. These lowest values MSLP and highest

MSW with their obtaining time are tabulated in Table 2. The simulated MSLP of cyclone Thane intensifies

earlier than the observation but the simulated MSW of cyclone Thane intensifies later than the observation.

4.1.3. MSLP and MSW of the Tropical Cyclone Mahasen

Figure 1 shows the comparative evolution of observed MSLP and simulated MSLP of TC Mahasen. It appears

from the Figures 1 that model simulated and observed MSLP gradually drops with time and attains peak

intensity just before the landfall and thereafter MSLP increases. The Model simulated MSLP of 952 hPa is

obtained at 15 UTC of 14 May 2013 and attain this value up to 21 UTC of 14 May 2013 whereas the observed

MSLP 990 hPa is obtained at 06 UTC of 15 May 2013 and attain this value up to 06 UTC of 16 May 2013. The

model simulated MSLP at the centre of the cyclone is obtained before 15 hours from the observed MSLP. The

variation of model simulated MSLP compared to that of observed with time shows a realistic temporal variation

of MSLP. Simulated MSLP are lower than those of observed. The simulated highest MSW is 49 m/s and

obtained at 15 UTC of 14 May 2013 where as that for observed is 23 m/s and obtained at 06 UTC of 15 May

2013 and retains this value up to 06 UTC on 16 May 2013. These lowest values MSLP and highest MSW with

their obtaining time are tabulated in Table 2. The simulated intensity of cyclone Mahasen intensifies earlier than

the observation.

The Absolute Error (AE) and Root Mean Square Error (RMSE) for the MSLP and MSW are tabulated in Table

3. Minimum errors are found for cyclone Thane.

4.2. Track Pattern of Tropical cyclones

Model simulated track of TC Nargis, Thane and Mahasen along with observed track are plotted in the Figures 2.

4.2.1. Track Pattern of Tropical cyclone Nargis

The track forecasts of TC Nargis for 96 and 72 hrs are based on the initial fields of 00 UTC on 29 April and 00

UTC on 30 April 2008 respectively. It is seen from Figure 2 that the model simulated track for 96 and 72 hours

are parallel to observed track but it is deviated south and north side of the observed track. Figure shows that

model is able to generate northwest, north and northeast movement of the system very well. It reveals that track

obtained from 72 hrs simulation are more close to the India Meteorological Department (IMD) best track

compared to the track obtained from 96 hrs simulation. However, there are some errors in the landfall with

respect to position and time landfall tabulated in Table 4. Simulated landfall time are 0450 UTC of 02 May and

0750 UTC of 02 May 2008 using 96 and 72 hrs forecast where as observed landfall time is 1200 UTC of 02

May 2008. So, 96 and 72 hours forecast shows landfall time 7 and 2 hours earlier than the observed landfall

time. Again, Simulated landfall positions are (18.450 N, 94.30

0 E) and (16.35

0 N, 94.31

0 E) using 96 and 72 hrs

forecast where as observed landfall position is (16.000 N, 94.30

0 E). So, 96 and 72 hours forecast shows landfall

position 274 and 43 km north of the observed landfall position. Again, RMSE and AE for 96 and 72 hours

forecast are 278 and 133 km respectively (Table 5).

Page 76



4.2.2. Track Pattern of Tropical cyclone Thane

Model simulated track of TC Thane along with observed track are plotted in the Figures 2. The track forecasts

of TC Thane for 96 and 72 hrs are based on the initial fields of 00 UTC of 27 December and 00 UTC of 28

December 2011 respectively. It is seen from Figure 2 that the simulated track obtained by running the model for

96 and 72 hours is parallel to observed track. 96 hours simulated track is mainly north side of observed track but

72 hours forecast track is overlapping north and south side of observed track. Finally 72 hours forecast run is

parallel and close to observed one.

However, there are some errors in the landfall with respect to position and time landfall tabulated in Table 4.

Simulated landfall time are 0915 UTC of 30 December and 2300 UTC of 29 December 2011 using 96 and 72

hrs forecast where as observed landfall time is 0100 UTC of 30 December 2011. So, 96 and 72 hours forecast

shows landfall time 8.75 and 2 hours earlier than the observed landfall time. Again, Simulated landfall positions

are (14.400 N, 80.20

0 E) and (11.89

0 N, 79.99

0 E) using 96 and 72 hrs forecast where as observed landfall

position is (11.600 N, 79.77

0 E). So, 96 and 72 hours forecast shows landfall position 314 and 41 km north of

the observed landfall position. Again, RMSE and AE for 96 and 72 hours forecast are 241 and 47 km

respectively (Table 5).

4.2.3. Track Pattern of Tropical cyclone Mahasen

Models simulated track of TC Mahasen along with observed track are plotted in the Figures 2. The track

forecasts of TC Mahasen for 96 and 72 hrs are based on the initial fields of 00 UTC of 13 May and 14 May

2013 respectively. Both of tracks are parallel to observed one and deviated east side of the observed track.

However, there are some errors in the landfall with respect to position and time landfall tabulated in Table 4.

Simulated landfall time are 1900 UTC of 16 May and 1930 UTC of 16 May 2013 using 96 and 72 hrs forecast

where as observed landfall time is 0715 UTC of 16 May 2013. So, 96 and 72 hours forecast shows landfall time

11.75 and 12.25 hours later than the observed landfall time. Again, Simulated landfall positions are (22.520 N,

91.680 E) and (22.40

0 N, 91.78

0 E) using 96 and 72 hrs forecast where as observed landfall position is (22.80

0 N,

91.350 E). So, 96 and 72 hours forecast shows landfall position 48 and 64 km east of the observed landfall

position. Again, RMSE and AE for 96 and 72 hours forecast are 168 and 229 km respectively (Table 5).

Again average landfall position are 212 and 49 km for 96 and 72 hours forecast respectively and average

landfall time are 8.83 and 6.08 hours for 96 and 72 hours forecast respectively (Table 4). Again, mean RMSE

and AE for 96 and 72 hours forecast are 229 and 136 km respectively (Table 5). Track obtained from 72 hours

forecast shows better performance Thane 96 hours forecast track.

Table 2: Lowest MSLP and highest MSW

Cyclone Lowest MSLP Highest MSW

Simulated observed simulated observed

Value

(hPa)

Date/time

(DD/UTC)

Value

(hPa)

Date/time

(DD/UTC)

Value

(m/s)

Date/time

(DD/UTC)

Value

(m/s)

Date/time

(DD/UTC)

Nargis 956 02/00 962 02/06 44 02/00 46 02/06

Thane 973 29/12 969 29/21 36 29/15 40 29/03

Mahasen 952 14/15 990 15/06 49 14/15 23 15/06

Table 3: RMSE and AE of intensity of selected TCs

Cyclone Error for MSLP Error for MSW

Absolute Error

(hPa)

RMSE

(hPa)

Absolute Error

(m/s)

RMSE

m/s)

Nargis 13.58 15.13 10.30 12.67

Thane 4.17 5.39 5.43 6.79

Mahasen 30.19 32.12 17.53 18.29

Average 15.98 17.55 11.09 12.58

Page 77



Table 4: Landfall point and time error of TCs

Cyclone obs/

models

Forecast

Hours

Initial

condition

Date/Tim

e (UTC)

Landfall

time

Date/Time

(UTC)

Landfall position Error

Lat)N LonE Distance

(km)

Time

(hours)

Nargis Obs 02/1200 16.00 94.30

Model 96 29/0000 02/0450 18.45 94.30 274 n 7E

72 30/0000 02/0750 16.35 94.31 43 n 4E

Thane Obs 30/0100 11.60 79.77

Model 96 27/0000 30/0915 14.40 80.20 314 n 8.75 D

72 28/0000 29/2300 11.89 79.99 41 n 2 E

Mahasen Obs 16/0715 22.80 91.35

Model 96 05/1300 16/1900 22.52 91,68 48 e 11.75D

72 05/1400 16/1930 22.40 91.76 64 e 12.25D

Mean 96 212 8.83

72 49 6.08

e indicates deviation toward eastern direction, w indicates deviation west of the actual landfall

position, n indicates deviation toward northern direction, D indicates forecast landfall time is

delayed compared to actual time, and E indicates forecast landfall time is earlier to actual

landfall time.

Table 5: RMSE and AE of track of TC

Cyclone Forecast Hours RMSE of track

(km)

Average Error of

track (km)

Nargis 96 278 257

72 133 127

Thane 96 241 202

72 47 42

Mahasen 96 168 132

72 229 171

Mean 96 229 197

72 136 113

5. CONCLUSION

In this study, WRF ARW model was used to predict the intensity and track of the tropical cyclones. From the

simulations the following results come forward:

1. Simulated SLPs of TC are comparable with those obtained from IMD except for cyclone Mahasen but

time variation patterns are similar in all cases.

2. Simulated 10 m MSWs of TC are comparable with those obtained from IMD except for cyclone

Mahasen but time variation patterns are similar in all cases.

3. The track from 72 hours simulation track is better than that of 96 hours simulation except for cyclone

Mahasen.

4. The landfall position error and time error for 72 hours simulation is very little than that of 96 hours

simulation except for cyclone Mahasen.

Hence it may be concluded that the WRF model can be used to forecast the tropical cyclone of the Bay of

Bengal.

Page 78



Acknowledgements: The authors gratefully acknowledge the NCEP/NCAR for their reanalysis data sets used

for the present study. One of the authors Md. Abdullah Elias Akhter is thankful to the Ministry of Science and

Information & Communication Technology, Bangladesh for the sanction of fellowship.

REFERENCES

1. W. M. Gray, Tropical Cyclone Disasters (eds) James Lighthill, Greg Holland, Zheng Zhemin and Keny

Ommanuel, Peking University Press (1992)

2. D. R. Sikka and R. Suryanarayana, Indian J. Met. Geophys. 23, pp.35–40 (1972).

3. S. S. Singh and K. R. Saha, Indian J. Met. Geophys. 29, pp.367–374 (1978).

4. U. C. Mohanty and Gupta Akhilesh, Mausam 48, pp.257–272 (1997).

5. A. Gupta and R. K. Bansal, NCMRWF, New Delhi Tech. report (1997).

6. K. Prasad and Y. V. Rama Rao, Meteorol. Atmos. Phys. 83 , pp.173–185 (2003).

7. S. Pattanayak and U. C. Mohanty, Ncurrent Science, 95, No. 7, pp923-936 (2008)

8. Pattanaik D R and Rama Rao Y V 2009; J. Earth Syst. Sci. 4 309–329.

9. Kumar A, Done J, Dudhia J and Niyogi D, 2011; Meteorol Atmos Phys, 114, 123-137, DOI

10.1007/s00703-011-0161-9.

10. Raju P V S, Potty J, and Mohanty U C, 2012; Nat Hazards, 63, 1361–1374, DOI 10.1007/s11069-011-

9918-1.

11. C. William, J. B. Skamarock and J. Dudhia : D. O. Gill, D. M. Barker, Michael G. Duda, X. Y. Huang, W.

wang and J. G. Power, NCAR Tech Note. NCAR/TN- 475+STR: 117 (2009).

12. J. S. Kain, J. Meteor., 43, pp.170-181 (2004).

13. S. Y. Hong, H. M. Juang and Q. Zhao, Mon. Wea. Rev. 126, pp.2621–2639 (1998).

14. S. Y. Hong, J. Dudhia and S. H. Chen, Mon. Wea. Rev. 132, pp.103–120 (2004).

15. E. J. Mlawer S. J. Taubman, P. D. Brown, M. J. Lacono and S. A. Clough, J. Geophys. Res. 102(D14), pp.

16,663–16,682 (1997).

16. J. Dudhia (1989), J. Atmos. Sci. 46, pp.3077–3107 (1989).

17. S. Y. Hong, Y. Noh, and J. Dudhia, Mon. Wea., 134, (2006), pp.2318-2341 (2006).

9



Figure 1: Model simulated and observed MSLP and MSW of TC Nargis, Thane and Mahasen

TC Nargis TC Thane TC Mahasen

Figure 2: Model simulated and observed tracks of TC Nargis, Thane and Mahasen

0



prediction of track and intensity of the tropical cyclonic ... 35.pdf · prediction of track and...

Documents