**3. Experimental design**

This study presents the results of two numerical experiments. One is a series of idealized numerical experiments applying MRINHM. MRINHM covers a 600 km x 600 km square computational domain with a horizontal grid spacing of 2 km. The other is a series of numerical simulations of Typhoon Choi-Wan in 2009 performed using JMANHM. JMANHM covers a 2700 km x 3600 km rectangular domain with a horizontal grid spacing of either 6 km or 12 km. Both MRINHM and JMANHM have 40 vertical levels with variable intervals from 40 m for the lowermost (near-surface) layer to 1180 m for the uppermost

Effect of the Ocean on TC 47

Typhoon Choi-Wan in 2009 was simulated during its mature and decaying phases. Choi-Wan was initiated when a tropical depression evolved into a TC around 15.4ºN, 150.9ºE at 18:00 UTC on 12 September 2009. Choi-Wan moved initially west-northwestward but changed to a northwestward track as it rapidly intensified. From 12:00 UTC on 15 September to 18:00 UTC on 16 September, the minimum central pressure (MCP) was 915 hPa and MWS was 105 knots (~54 m s-1). At a location around 23.2N, 138.9E, Choi-Wan

gradually slowed and changed to a north-northeastward track as it began to decay.

Table 2. Key data used for numerical simulations of Typhoon Choi-Wan in 2009

calculated using the MOVE system were used as initial oceanic conditions.

consideration in cloud physics.

The initial time for all simulations was 00:00 UTC on 17 September 2009, when Choi-Wan entered its mature phase. Two sets of atmospheric initial and boundary conditions were used. One was derived from six-hourly global objective analysis data (GA data hereafter) from JMA with a grid spacing of 20 km. The other was derived from six-hourly data from the JMA Climate Data Assimilation System (JCDAS hereafter) with latitude and longitude grid spacings of 1.25º. Daily oceanic reanalysis data with two grid spacings (0.1º and 0.5º )

Table 2 provides the key specifications of the JMANHM numerical simulations of Choi-Wan. Both NHM and coupled NHM-wave-ocean models were used to investigate the effect of the ocean on TC simulations from viewpoints of existence or non-existence of Choi-Waninduced SSC. Warm-rain experiments did not take snow and ice-cloud phases into

**3.2 Typhoon Choi-Wan in 2009** 

layer. Both MRINHM and JMANHM have maximum height approaching nearly 23 km. The time step of MRINHM is 6 s and that of JMANHM is 15 s (6 km grid spacing) or 30 s (12 km grid spacing). The length of the time step of the ocean model is six times those of both MRINHM and JMANHM. The Coriolis parameter is uniformly set to 5.0 x 10-5 (nearly 20ºN) in MRINHM and varies in JMANHM depending on the grid latitude.

Water depth in the multilayer ocean model coupled with MRINHM is uniformly set to 1000 m. Initial SST is set to 30ºC, the initial temperature at the base of the mixed layer to 29ºC, the initial temperature at the base of the thermocline to 18ºC and the initial temperature at the bottom to 5ºC. Initial salinity is set to 35 at all levels. The initial mixed-layer depth is set to be 30 m, the initial thermocline thickness to 170 m and the initial third-layer thickness to 800 m. The third layer thickness is assumed to be unaffected by entrainment. In JMANHM, the initial depth of the mixed layer is determined from oceanic reanalysis data by assuming a difference in the value of density from the surface of no more than 0.25 kg m-3 and the depth of the mixed layer is limited to 200 m. The base of the thermocline is limited to 600 m and water depth is limited to 2000 m. The oceanic reanalysis data are calculated using the MRI ocean variational estimation (MOVE) system (Usui et al., 2006).

#### **3.1 Idealized experiment**

Table 1 summarizes the idealized numerical experiments performed using MRINHM with and without coupling with the ocean model. The initial TC-like vortex and thermal conditions were as given by (Wada, 2009). The integration time was 81 h with results output every 30 min. The sensitivity of vertical turbulent mixing in the ocean model was evaluated using two tuning parameters: *md* = 17.5 and *md* = 175. Parameter *md* is associated with turbulent kinetic energy flux produced by breaking surface waves.


Table 1. Summary of key parameters of idealized MRINHM numerical experiments with and without coupling with the ocean model.

Experiment CTL was a control run. 'D2' in the experiment index indicates that the representative mixing-length scale was determined as a two-dimensional geometric mean, and 'K35' indicates that drag coefficients leveled off when 10-m wind speed exceeded 35 m s-1, This saturation of drag coefficients has been reported by (Powell et al., 2003) and (Donelan et al., 2004). Drag coefficients and mixing-length scales were as given by (Deardorff, 1980).

layer. Both MRINHM and JMANHM have maximum height approaching nearly 23 km. The time step of MRINHM is 6 s and that of JMANHM is 15 s (6 km grid spacing) or 30 s (12 km grid spacing). The length of the time step of the ocean model is six times those of both MRINHM and JMANHM. The Coriolis parameter is uniformly set to 5.0 x 10-5 (nearly 20ºN)

Water depth in the multilayer ocean model coupled with MRINHM is uniformly set to 1000 m. Initial SST is set to 30ºC, the initial temperature at the base of the mixed layer to 29ºC, the initial temperature at the base of the thermocline to 18ºC and the initial temperature at the bottom to 5ºC. Initial salinity is set to 35 at all levels. The initial mixed-layer depth is set to be 30 m, the initial thermocline thickness to 170 m and the initial third-layer thickness to 800 m. The third layer thickness is assumed to be unaffected by entrainment. In JMANHM, the initial depth of the mixed layer is determined from oceanic reanalysis data by assuming a difference in the value of density from the surface of no more than 0.25 kg m-3 and the depth of the mixed layer is limited to 200 m. The base of the thermocline is limited to 600 m and water depth is limited to 2000 m. The oceanic reanalysis data are calculated using the MRI

Table 1 summarizes the idealized numerical experiments performed using MRINHM with and without coupling with the ocean model. The initial TC-like vortex and thermal conditions were as given by (Wada, 2009). The integration time was 81 h with results output every 30 min. The sensitivity of vertical turbulent mixing in the ocean model was evaluated using two tuning parameters: *md* = 17.5 and *md* = 175. Parameter *md* is associated with

Table 1. Summary of key parameters of idealized MRINHM numerical experiments with

2004). Drag coefficients and mixing-length scales were as given by (Deardorff, 1980).

Experiment CTL was a control run. 'D2' in the experiment index indicates that the representative mixing-length scale was determined as a two-dimensional geometric mean, and 'K35' indicates that drag coefficients leveled off when 10-m wind speed exceeded 35 m s-1, This saturation of drag coefficients has been reported by (Powell et al., 2003) and (Donelan et al.,

in MRINHM and varies in JMANHM depending on the grid latitude.

ocean variational estimation (MOVE) system (Usui et al., 2006).

turbulent kinetic energy flux produced by breaking surface waves.

and without coupling with the ocean model.

**3.1 Idealized experiment** 

#### **3.2 Typhoon Choi-Wan in 2009**

Typhoon Choi-Wan in 2009 was simulated during its mature and decaying phases. Choi-Wan was initiated when a tropical depression evolved into a TC around 15.4ºN, 150.9ºE at 18:00 UTC on 12 September 2009. Choi-Wan moved initially west-northwestward but changed to a northwestward track as it rapidly intensified. From 12:00 UTC on 15 September to 18:00 UTC on 16 September, the minimum central pressure (MCP) was 915 hPa and MWS was 105 knots (~54 m s-1). At a location around 23.2N, 138.9E, Choi-Wan gradually slowed and changed to a north-northeastward track as it began to decay.


Table 2. Key data used for numerical simulations of Typhoon Choi-Wan in 2009

The initial time for all simulations was 00:00 UTC on 17 September 2009, when Choi-Wan entered its mature phase. Two sets of atmospheric initial and boundary conditions were used. One was derived from six-hourly global objective analysis data (GA data hereafter) from JMA with a grid spacing of 20 km. The other was derived from six-hourly data from the JMA Climate Data Assimilation System (JCDAS hereafter) with latitude and longitude grid spacings of 1.25º. Daily oceanic reanalysis data with two grid spacings (0.1º and 0.5º ) calculated using the MOVE system were used as initial oceanic conditions.

Table 2 provides the key specifications of the JMANHM numerical simulations of Choi-Wan. Both NHM and coupled NHM-wave-ocean models were used to investigate the effect of the ocean on TC simulations from viewpoints of existence or non-existence of Choi-Waninduced SSC. Warm-rain experiments did not take snow and ice-cloud phases into consideration in cloud physics.

Effect of the Ocean on TC 49

D2DIM, K35, and K35D2 but a rapid decrease in SST in experiment OC (Fig. 2b). The decrease in SST becomes small when the vortex reaches its mature phase. It should be noted that SST in experiment OC at 81 h (i.e., at the mature phase) is the smallest among all experiments (Table 1), but CP at that time is almost the same in all experiments except AT. This suggests that neither the evolution of the idealized vortex nor the final value of CP is

Intensification of the vortex is more suppressed with *md* = 175 (Fig. 1b) than with *md* = 17.5. A difference in CPs for each experiment begins to appear after around 15 h, when the vortex undergoes slow intensification in experiment CTL, whereas rapid intensification of the vortex is apparent at this time in experiment OC. The vortices in experiments CTL, D2DIM, K35 and K35D2 begin to intensify rapidly at 36 h, later than in the experiments AT and OC. It is interesting that CP in experiment OC eventually reaches a value similar to those reached in experiments CTL, D2DIM, K35 and K35D2 at 81 h, suggesting that the evolution

When the location of CP coincides with the circulation center, it is presumed that deep convection occurs easily there owing to enhancement of updraft. Deep convection leads to rapid intensification of the vortex. The distance between CP and the circulation center varies in an oscillatory manner from the initial time to the end of the period of rapid intensification (Fig. 3). On the other hand, variation of these distances is rarely evident from the end of the intensification phase through the mature phase (Fig. 3), indicating that rapid intensification occurs when the distance is small. Differences in the evolution of the distance between *md* = 17.5 (Fig. 3a) and *md* = 175 (Fig. 3b) imply that SSC has a negative effect on reducing the

Fig. 3. Time series (as in Fig. 1) of distance between the CP and the circulation center of the

The results presented in Figs 1-3 indicate that SSC plays a crucial role in TC intensification. In contrast, the effects of the atmospheric boundary layer (experiment D2DIM) and drag coefficient (experiment K35) on TC intensification are smaller than the effect of vortexinduced SSC. To investigate the effect of the atmospheric boundary layer and drag coefficient on the vortex, the relationships between MWS and MCP were examined (Fig.4); the results show that the MWS-MCP relationship is remarkably different when the vortex is strong. For the same value of MCP, MWS tends to be high in experiments K35 and K35D2, implying that low surface friction causes not only a super-gradient inflow due to small drag coefficients for winds exceeding 35 m s-1 but also weak wind stresses, resulting in low

dependent on SST directly below the circulation center.

distance.

vortex.

of SSC and the final value of SST may not be related to final TC intensity.

#### **4. Results**

#### **4.1 Intensity and structural change**

#### **4.1.1 Vortex intensity**

Figure 1a indicates that central pressure (CP) is high during the integration when the MRINHM is coupled with the ocean model. Figure 1b indicates that a higher value of *md* (175) greatly reduces intensification, resulting in high CP. A difference in CP between the OC and CTL experiments is evident after 24 h and is unrelated to the values of *md*. The vortex intensifies slowly in experiment CTL, whereas intensification is rapid in experiment OC. CP maintains their values within ranges of 950-960 hPa (*md* = 17.5) and 980-990 hPa (*md* = 175) during the mature phase (except in experiment AT). These results indicate minimal intensification of the mature vortex despite continued vortex-induced SSC.

Fig. 1. Time series of CP of the idealized TC-like vortex at 30 min intervals for (a) *md* = 17.5 and (b) *md* = 175 for each experiment shown in Table 1.

SST is defined in this study as the temperature directly below the circulation center of the vortex. The circulation center is defined as the position where surface wind speed is at a minimum and the difference between CP and sea-level pressure is less than 4 hPa. The evolution of SST at the circulation center shows that the variation of SST is small during slow intensification. SST decreases rapidly during rapid intensification with *md* = 17.5 (Fig. 2a). The high value of *md* (175) produces a gradual decrease in SST in experiments CTL,

Fig. 2. Time series (as in Fig. 1) of SST immediately below the idealized TC-like vortex.

Figure 1a indicates that central pressure (CP) is high during the integration when the MRINHM is coupled with the ocean model. Figure 1b indicates that a higher value of *md* (175) greatly reduces intensification, resulting in high CP. A difference in CP between the OC and CTL experiments is evident after 24 h and is unrelated to the values of *md*. The vortex intensifies slowly in experiment CTL, whereas intensification is rapid in experiment OC. CP maintains their values within ranges of 950-960 hPa (*md* = 17.5) and 980-990 hPa (*md* = 175) during the mature phase (except in experiment AT). These results indicate minimal intensification of the mature vortex despite continued vortex-induced

Fig. 1. Time series of CP of the idealized TC-like vortex at 30 min intervals for (a) *md* = 17.5

SST is defined in this study as the temperature directly below the circulation center of the vortex. The circulation center is defined as the position where surface wind speed is at a minimum and the difference between CP and sea-level pressure is less than 4 hPa. The evolution of SST at the circulation center shows that the variation of SST is small during slow intensification. SST decreases rapidly during rapid intensification with *md* = 17.5 (Fig. 2a). The high value of *md* (175) produces a gradual decrease in SST in experiments CTL,

Fig. 2. Time series (as in Fig. 1) of SST immediately below the idealized TC-like vortex.

and (b) *md* = 175 for each experiment shown in Table 1.

**4. Results** 

SSC.

**4.1.1 Vortex intensity** 

**4.1 Intensity and structural change** 

D2DIM, K35, and K35D2 but a rapid decrease in SST in experiment OC (Fig. 2b). The decrease in SST becomes small when the vortex reaches its mature phase. It should be noted that SST in experiment OC at 81 h (i.e., at the mature phase) is the smallest among all experiments (Table 1), but CP at that time is almost the same in all experiments except AT. This suggests that neither the evolution of the idealized vortex nor the final value of CP is dependent on SST directly below the circulation center.

Intensification of the vortex is more suppressed with *md* = 175 (Fig. 1b) than with *md* = 17.5. A difference in CPs for each experiment begins to appear after around 15 h, when the vortex undergoes slow intensification in experiment CTL, whereas rapid intensification of the vortex is apparent at this time in experiment OC. The vortices in experiments CTL, D2DIM, K35 and K35D2 begin to intensify rapidly at 36 h, later than in the experiments AT and OC. It is interesting that CP in experiment OC eventually reaches a value similar to those reached in experiments CTL, D2DIM, K35 and K35D2 at 81 h, suggesting that the evolution of SSC and the final value of SST may not be related to final TC intensity.

When the location of CP coincides with the circulation center, it is presumed that deep convection occurs easily there owing to enhancement of updraft. Deep convection leads to rapid intensification of the vortex. The distance between CP and the circulation center varies in an oscillatory manner from the initial time to the end of the period of rapid intensification (Fig. 3). On the other hand, variation of these distances is rarely evident from the end of the intensification phase through the mature phase (Fig. 3), indicating that rapid intensification occurs when the distance is small. Differences in the evolution of the distance between *md* = 17.5 (Fig. 3a) and *md* = 175 (Fig. 3b) imply that SSC has a negative effect on reducing the distance.

Fig. 3. Time series (as in Fig. 1) of distance between the CP and the circulation center of the vortex.

The results presented in Figs 1-3 indicate that SSC plays a crucial role in TC intensification. In contrast, the effects of the atmospheric boundary layer (experiment D2DIM) and drag coefficient (experiment K35) on TC intensification are smaller than the effect of vortexinduced SSC. To investigate the effect of the atmospheric boundary layer and drag coefficient on the vortex, the relationships between MWS and MCP were examined (Fig.4); the results show that the MWS-MCP relationship is remarkably different when the vortex is strong. For the same value of MCP, MWS tends to be high in experiments K35 and K35D2, implying that low surface friction causes not only a super-gradient inflow due to small drag coefficients for winds exceeding 35 m s-1 but also weak wind stresses, resulting in low

Effect of the Ocean on TC 51

The time series of the vertical section of horizontally averaged Vor (Fig. 5b) shows that Vor below 1 km altitude increases after 15 h. Vor increases markedly at around 24 h when the vortex starts to rapidly intensify. The rapid increase of Vor continues until around 43 h, after which the profile changes little through the mature phase. These results indicate that variations of PV are controlled mainly by variations of Vor. In addition, the vertical gradient of horizontally averaged potential temperature becomes low from around 2 to 9 km altitude, corresponding temporally with the change of Vor (not shown). Thus, the evolution of the Vor profile depends on the phase of the vortex and affects both the thermal and PV profiles.

(d)

(e)

(f)

Fig. 5. Time series for experiment AT of the vertically averaged profiles of (a) PV, (b) Vor, (c)

Upward vertical velocity increases suddenly between 10 to 12 km altitudes at 24, 43 and 51 h (Fig. 5c), corresponding temporally to sudden increases in total water (Fig. 5d) due to the production of cloud ice and snow. Total water in this study is defined as the sum of cloud water, rain, cloud ice, snow and graupel. The sudden increase in total water causes a sudden increase in relative humidity (Fig. 5e) at around 9 km altitude. The sudden increase in

vertical velocity, (d) total water, (e) relative humidity and (f) horizontal wind speed. Averages are calculated over a 120 km x 120 km domain (corresponding to the scale of the

initial vortex) centered in the computational domain. Black solid contours indicate

equivalent potential temperature.

(a)

(b)

(c)

Fig. 4. Relationship between MWS and MCP in each of the numerical experiments shown in Table 1.

vortex-induced SSC. High MWS and small SSC, however, have little effect on the lowering of MCP in experiments K35 and K35D2.

In other words, differences in the MWS-MCP relationships have little effect on the evolution of CP, SST and the phases of the vortex (Figs. 1-2). In particular, the sensitivity of the tuning parameter *md* on the evolutions is noteworthy. The question then arises: How does vortexinduced SSC affect the evolutions of CP, the phase of the vortex and subsequent SSC?

#### **4.1.2 Structural changes of the vortex**

PV was used to investigate the structural changes of the vortex and the role of SSC in its evolution. PV is formulated as

$$PV = -g\left(\zeta\_{\theta} + f\right)\frac{\partial\theta}{\partial p},\tag{1}$$

where *g* is the acceleration of gravity, the relative vorticity (Vor hereafter) on the isentropic surface (surface), and *p* the vertical pressure coordinate.

The time series of the vertical section of horizontally averaged PV for experiment AT (Fig. 5a) shows that there are four phases of the vortex: a spin-up phase from the initial time to 15 h, a slow intensification phase from 15 to 24 h, a rapid intensification phase from 24 to 43 h and a mature phase after 43 h (Fig. 1).

PV increases between altitudes of around 4 to 8 km during the slow intensification phase. Then at the start of the rapid intensification phase, PV above 1 km altitude increases and after 37 h, there are further increases above 10 km. After 37 h, the PV profile changes little, except above 10 km, where marked temporal oscillations are evident.

Fig. 4. Relationship between MWS and MCP in each of the numerical experiments shown in

vortex-induced SSC. High MWS and small SSC, however, have little effect on the lowering

In other words, differences in the MWS-MCP relationships have little effect on the evolution of CP, SST and the phases of the vortex (Figs. 1-2). In particular, the sensitivity of the tuning parameter *md* on the evolutions is noteworthy. The question then arises: How does vortexinduced SSC affect the evolutions of CP, the phase of the vortex and subsequent SSC?

PV was used to investigate the structural changes of the vortex and the role of SSC in its

*PV g f* ,

surface), and *p* the vertical pressure coordinate.

The time series of the vertical section of horizontally averaged PV for experiment AT (Fig. 5a) shows that there are four phases of the vortex: a spin-up phase from the initial time to 15 h, a slow intensification phase from 15 to 24 h, a rapid intensification phase from 24 to 43 h

PV increases between altitudes of around 4 to 8 km during the slow intensification phase. Then at the start of the rapid intensification phase, PV above 1 km altitude increases and after 37 h, there are further increases above 10 km. After 37 h, the PV profile changes little,

except above 10 km, where marked temporal oscillations are evident.

*p*

(1)

the relative vorticity (Vor hereafter) on the

Table 1.

of MCP in experiments K35 and K35D2.

**4.1.2 Structural changes of the vortex** 

where *g* is the acceleration of gravity,

and a mature phase after 43 h (Fig. 1).

evolution. PV is formulated as

isentropic surface (

The time series of the vertical section of horizontally averaged Vor (Fig. 5b) shows that Vor below 1 km altitude increases after 15 h. Vor increases markedly at around 24 h when the vortex starts to rapidly intensify. The rapid increase of Vor continues until around 43 h, after which the profile changes little through the mature phase. These results indicate that variations of PV are controlled mainly by variations of Vor. In addition, the vertical gradient of horizontally averaged potential temperature becomes low from around 2 to 9 km altitude, corresponding temporally with the change of Vor (not shown). Thus, the evolution of the Vor profile depends on the phase of the vortex and affects both the thermal and PV profiles.

Fig. 5. Time series for experiment AT of the vertically averaged profiles of (a) PV, (b) Vor, (c) vertical velocity, (d) total water, (e) relative humidity and (f) horizontal wind speed. Averages are calculated over a 120 km x 120 km domain (corresponding to the scale of the initial vortex) centered in the computational domain. Black solid contours indicate equivalent potential temperature.

Upward vertical velocity increases suddenly between 10 to 12 km altitudes at 24, 43 and 51 h (Fig. 5c), corresponding temporally to sudden increases in total water (Fig. 5d) due to the production of cloud ice and snow. Total water in this study is defined as the sum of cloud water, rain, cloud ice, snow and graupel. The sudden increase in total water causes a sudden increase in relative humidity (Fig. 5e) at around 9 km altitude. The sudden increase in

Effect of the Ocean on TC 53

numerical experiments suggest that the activity of the mesovortices that accumulated around the circulation center during the intensification phase contributes to formation of the complete annular ring, which is closely related to the warm-core structure of the vortex. This relationship is similar to that between TC intensity and accumulated tropical cyclone heat potential (Wada & Usui, 2007) in that the vortex merge effect coincides with accumulation of upper-ocean heat content directly below the vortex. The annular ring of Vor is accompanied by a robust warm-core structure even at 4 km altitude, which is lower than

In experiment CTL, the differences of horizontally averaged profiles of PV, Vor, vertical velocity, total water, relative humidity and horizontal wind speed are smaller during the spin-up phase than those in experiment AT (Fig. 7), except that the total water and wind speed in experiment CTL are even lower at around 10 h. A dry-air feature at around 6 to 8 km altitude at the initial time is maintained during the spin-up phase in both experiments AT and CTL (Figs. 5e and 7e). In contrast, distant differences between the AT and CTL experiments begin to appear after the slow intensification phase. Slow intensification continues until 36 h in experiment CTL (Figs 1 and 5), which is longer than that in

(d)

(e)

(f)

the general warm-core altitude (nearly 12 km).

Fig. 7. As in Fig. 5, but for experiment CTL with *md* = 175.

experiment AT.

(a)

(b)

(c)

upward vertical velocity and relative humidity corresponds temporally to strengthening of horizontal wind speed (Fig. 5f). Because the increase in wind speeds is linked to increases in Vor, the intensification of the vortex can be explained by vertical transfers of heat and moisture due to the sudden increase of Vor and the production of cloud ice and snow around 10 to 12 km altitude, which in turn causes a sudden increase in Vor.

The horizontal distributions of Vor at altitudes of 1 and 4 km show clear differences for the slow intensification, rapid intensification and mature phases of experiment AT (Fig. 6). At the start of the slow intensification phase (Figs. 6a and 6d), Vor at the center of the vortex is low with mesovortices with horizontal diameters of up to 10 km scattered around the circulation center. At the start of the rapid intensification phase, the surviving mesovortices are concentrated closer to the circulation center at both 1 and 4 km altitudes (Figs. 6b and 6e) and those at 4 km altitude begin to show a spiral distribution as the vortex merger events far from the circulation center become enhanced by the eyewall-shrinking process (Fig. 3). After the rapid intensification phase, the vortex merger events cause the formation of an annular ring in the Vor distribution (Figs. 6c and 6f), with high Vor values also along the spirally bands at 4 km altitude. After formation of the annular ring, vortex intensification ceases during the mature phase (Fig. 1).

Fig. 6. Horizontal distributions of Vor at 4 km altitude for experiment AT at (a) 15 h, (b) 24 h, and (c) 43 h, and at 1 km altitude at (d) 15 h, (e) 24 h, and (f) 43 h. Black solid contours indicate positive Vor and black dashed lines indicate negative Vor. Pink contours indicate equivalent potential temperature at the same altitudes and times.

In other words, when the vortex merger events are extensive before formation of the annular ring of Vor, the mesovortices (high relative vorticity at 10 km scale accompanied by updraft) play a crucial role in the vertical transfer of heat and moisture. The locally-scattered horizontal distributions of equivalent potential temperature at 15 and 24 h are also strongly affected by the formation and enhancement of mesovortices (Fig. 6). Thus, the results of the

upward vertical velocity and relative humidity corresponds temporally to strengthening of horizontal wind speed (Fig. 5f). Because the increase in wind speeds is linked to increases in Vor, the intensification of the vortex can be explained by vertical transfers of heat and moisture due to the sudden increase of Vor and the production of cloud ice and snow

The horizontal distributions of Vor at altitudes of 1 and 4 km show clear differences for the slow intensification, rapid intensification and mature phases of experiment AT (Fig. 6). At the start of the slow intensification phase (Figs. 6a and 6d), Vor at the center of the vortex is low with mesovortices with horizontal diameters of up to 10 km scattered around the circulation center. At the start of the rapid intensification phase, the surviving mesovortices are concentrated closer to the circulation center at both 1 and 4 km altitudes (Figs. 6b and 6e) and those at 4 km altitude begin to show a spiral distribution as the vortex merger events far from the circulation center become enhanced by the eyewall-shrinking process (Fig. 3). After the rapid intensification phase, the vortex merger events cause the formation of an annular ring in the Vor distribution (Figs. 6c and 6f), with high Vor values also along the spirally bands at 4 km altitude. After formation of the annular ring, vortex intensification ceases

Fig. 6. Horizontal distributions of Vor at 4 km altitude for experiment AT at (a) 15 h, (b) 24 h, and (c) 43 h, and at 1 km altitude at (d) 15 h, (e) 24 h, and (f) 43 h. Black solid contours indicate positive Vor and black dashed lines indicate negative Vor. Pink contours indicate

In other words, when the vortex merger events are extensive before formation of the annular ring of Vor, the mesovortices (high relative vorticity at 10 km scale accompanied by updraft) play a crucial role in the vertical transfer of heat and moisture. The locally-scattered horizontal distributions of equivalent potential temperature at 15 and 24 h are also strongly affected by the formation and enhancement of mesovortices (Fig. 6). Thus, the results of the

equivalent potential temperature at the same altitudes and times.

(d) (e) (f)

(a) (b) (c)

around 10 to 12 km altitude, which in turn causes a sudden increase in Vor.

during the mature phase (Fig. 1).

numerical experiments suggest that the activity of the mesovortices that accumulated around the circulation center during the intensification phase contributes to formation of the complete annular ring, which is closely related to the warm-core structure of the vortex. This relationship is similar to that between TC intensity and accumulated tropical cyclone heat potential (Wada & Usui, 2007) in that the vortex merge effect coincides with accumulation of upper-ocean heat content directly below the vortex. The annular ring of Vor is accompanied by a robust warm-core structure even at 4 km altitude, which is lower than the general warm-core altitude (nearly 12 km).

In experiment CTL, the differences of horizontally averaged profiles of PV, Vor, vertical velocity, total water, relative humidity and horizontal wind speed are smaller during the spin-up phase than those in experiment AT (Fig. 7), except that the total water and wind speed in experiment CTL are even lower at around 10 h. A dry-air feature at around 6 to 8 km altitude at the initial time is maintained during the spin-up phase in both experiments AT and CTL (Figs. 5e and 7e). In contrast, distant differences between the AT and CTL experiments begin to appear after the slow intensification phase. Slow intensification continues until 36 h in experiment CTL (Figs 1 and 5), which is longer than that in experiment AT.

Fig. 7. As in Fig. 5, but for experiment CTL with *md* = 175.

Effect of the Ocean on TC 55

The value of *md* represents the amount of turbulent kinetic energy flux produced by breaking surface waves. When *md* is low, vortex-induced SSC becomes low. Therefore, the calculated SST may be relatively high directly beneath the vortex. Strengthening of the vortex due to low SSC is greater than that caused by the high SSC induced by vertical turbulent mixing enhanced by high *md* (Figs. 1 and 2). However, the evolution of the calculated SST is not determined simply by the value of *md* because the intensity of the vortex also affects the turbulent kinetic energy flux produced by breaking surface waves (Wada et al., 2010), which then affects subsequent vortex-induced SSC. The high value of *md* causes a rapid decrease in SST soon after the MRINHM-ocean coupled model starts to run. In contrast, the low value of *md* causes a moderate decrease in SST, resulting in a higher SST around the vortex (Fig. 2). The high SST then leads to rapid vortex intensification followed by a rapid decrease in SST directly below the vortex. It should be noted that SSTs calculated by the coupled models are close to 22ºC at 81 h and differs little among all numerical experiments except for experiments AT and OC, both with *md* = 175, although CPs differ between the numerical experiments with *md* = 17.5 and *md* = 175. The intensification processes shown in Fig. 5 correspond to the processes shown in Fig. 9. The numerical result in experiment OC with *md* = 17.5 suggests that CP is not easily determined by the SST

(d)

(e)

(f)

Fig. 9. As in Fig. 5 but for experiment OC with *md* = 17.5.

(a)

(b)

(c)

After 37 h, the vortex intensifies gradually with an increase in PV at around 4 to 8 km altitude (Fig. 7a). The intensification produces a strong SSC effect (Fig. 2) and the subsequent decrease of SST suppresses the intensification of the vortex (Fig. 1). When the relative humidity becomes low at 5 to 10 km altitude (Fig. 7e), SST directly below the vortex continues to decrease (Fig. 2) and wind speed is maintained after 72 h (Fig. 7f). Therefore, SSC plays an essential role in determining the dynamic and thermal frameworks of the vortex and its intensification. However, it remains unclear which phases of vortex-induced SSC affect vortex intensification and control the maximum intensity reached. This may be because the vortexinduced SSC interacts with the vortex by different mechanisms according to the vortex phases.

Considering the effect of vortex-induced SSC on vortex intensification after 27 h in experiment OC with *md* = 175, an increase in PV above 10 km altitude is suppressed, which is associated with an increase in Vor is suppressed in the lower troposphere at altitudes lower than 2 km (Figs. 8a and 8b). The sudden increase in upward vertical velocity at altitudes of around 10 to 12 km in experiment AT (Fig. 5c) is not as marked as in experiment OC with *md* = 175 (Fig. 8c). This difference is related to the reduction of total water, particularly cloud ice and snow at the same altitude (Fig. 8d) and also to the reduction of relative humidity at altitudes of around 4 to 10 km (Fig. 8e). These reductions affect wind speed and the maximum intensity reached (Fig. 8f).

Fig. 8. As in Fig. 5, but for experiment OC with *md* = 175.

After 37 h, the vortex intensifies gradually with an increase in PV at around 4 to 8 km altitude (Fig. 7a). The intensification produces a strong SSC effect (Fig. 2) and the subsequent decrease of SST suppresses the intensification of the vortex (Fig. 1). When the relative humidity becomes low at 5 to 10 km altitude (Fig. 7e), SST directly below the vortex continues to decrease (Fig. 2) and wind speed is maintained after 72 h (Fig. 7f). Therefore, SSC plays an essential role in determining the dynamic and thermal frameworks of the vortex and its intensification. However, it remains unclear which phases of vortex-induced SSC affect vortex intensification and control the maximum intensity reached. This may be because the vortexinduced SSC interacts with the vortex by different mechanisms according to the vortex phases. Considering the effect of vortex-induced SSC on vortex intensification after 27 h in experiment OC with *md* = 175, an increase in PV above 10 km altitude is suppressed, which is associated with an increase in Vor is suppressed in the lower troposphere at altitudes lower than 2 km (Figs. 8a and 8b). The sudden increase in upward vertical velocity at altitudes of around 10 to 12 km in experiment AT (Fig. 5c) is not as marked as in experiment OC with *md* = 175 (Fig. 8c). This difference is related to the reduction of total water, particularly cloud ice and snow at the same altitude (Fig. 8d) and also to the reduction of relative humidity at altitudes of around 4 to 10 km (Fig. 8e). These reductions affect wind

(d)

(e)

(f)

speed and the maximum intensity reached (Fig. 8f).

(a)

(b)

(c)

Fig. 8. As in Fig. 5, but for experiment OC with *md* = 175.

The value of *md* represents the amount of turbulent kinetic energy flux produced by breaking surface waves. When *md* is low, vortex-induced SSC becomes low. Therefore, the calculated SST may be relatively high directly beneath the vortex. Strengthening of the vortex due to low SSC is greater than that caused by the high SSC induced by vertical turbulent mixing enhanced by high *md* (Figs. 1 and 2). However, the evolution of the calculated SST is not determined simply by the value of *md* because the intensity of the vortex also affects the turbulent kinetic energy flux produced by breaking surface waves (Wada et al., 2010), which then affects subsequent vortex-induced SSC. The high value of *md* causes a rapid decrease in SST soon after the MRINHM-ocean coupled model starts to run.

In contrast, the low value of *md* causes a moderate decrease in SST, resulting in a higher SST around the vortex (Fig. 2). The high SST then leads to rapid vortex intensification followed by a rapid decrease in SST directly below the vortex. It should be noted that SSTs calculated by the coupled models are close to 22ºC at 81 h and differs little among all numerical experiments except for experiments AT and OC, both with *md* = 175, although CPs differ between the numerical experiments with *md* = 17.5 and *md* = 175. The intensification processes shown in Fig. 5 correspond to the processes shown in Fig. 9. The numerical result in experiment OC with *md* = 17.5 suggests that CP is not easily determined by the SST

Fig. 9. As in Fig. 5 but for experiment OC with *md* = 17.5.

Effect of the Ocean on TC 57

(a) (b)

(c) (d)

Fig. 11. Horizontal distribution of Vor at 4 km altitude at 81 h in experiments (a) CTL with

experiments. Black solid contours indicate positive Vor and black dashed contours indicate negative Vor. Pink contours indicate equivalent potential temperature at the same altitudes

The location of the circulation center differs clearly from that of CP during the slow intensification phase, but these locations are almost the same after the rapid intensification phase (Fig. 3). These spatial differences may be related to a change of the budget of absolute angular momentum (AAM) for the vortex with a diameter of almost 100 km, which is given as the initial TC-like vortex (Wada, 2009). This subsection addresses the relation of the

> *v u <sup>f</sup> x y*

, (2)

*u v wv wu*

**v** , (3)

where *x* and *y* are zonal and meridional directions, and *u* and *v* are wind velocities in the *x*

*t z x y xz yz*

where *z* is the vertical direction and *w* the vertical wind velocity. The upward direction for *z* and *w* are indicated by positive values. The four terms on the right-hand side of Eq. (3) indicate (from left to right) horizontal advection, vertical advection, divergence and titling.

 

*md* = 175 and (b) OC with *md* = 175 and (c)-(d) at 1 km altitude at 81 h in the same

deepening of CP and intensification of vortex circulation to the budget of AAM.

*a*

and *y* directions, respectively. The variation of AAM with time is expressed as

 

*<sup>a</sup>*) is the sum of Vor and the Coriolis parameter (*f*):

*a a h ha <sup>a</sup>*

*w*

and times.

AAM (

**4.1.3 Absolute angular momentum analysis** 

directly below the vortex, but is influenced by the background effects of CP and SST evolution.

The horizontal distributions of Vor at altitudes of 1 and 4 km show clearly that there are differences in TC phases between the CTL and OC experiments with *md* = 175 and the OC experiment with *md* = 17.5 (Fig. 10). An annular Vor-ring does not form in the CTL experiment with *md* = 175 (Figs. 10a and 10d), but it does form completely in the OC experiment with *md* = 17.5 (Figs. 10c and 10f). Comparison of the CTL experiment with *md* = 175 (Figs. 10a and 10d) with the OC experiment with *md* = 175 (Figs. 10c and 10f) clearly shows a difference of TC phases between the experiments of CTL and OC, implying that the slow formation of the annular ring is affected by SSC at both spin-up and slowintensification phases (Figs. 7b and 8b). The effect of vortex-induced SSC on the annular Vor-ring at the mature phase differs from that during the other phases. The annular Vorring in experiment OC with *md* = 175 becomes small at 81 h (Fig. 11) as CP gradually increases during the mature phase (Fig. 1). The amplitude of Vor at 81 h at 4 km altitude in experiment CTL with *md* = 175 (Fig. 11a) is almost the same as that of experiment OC with *md* = 175 (Fig. 11b), even though the amplitude at 1 km altitude in experiment CTL with *md* = 175 (Fig. 11c) is higher than that in experiment OC with *md* = 175 (Fig. 11d). In addition, the SST in experiment OC with *md* = 175 is lower than that in experiment CTL with *md* = 175 (Fig. 2). These results are consistent with the small amplitude of Vor at 81 h in experiment OC with *md* = 175. Therefore, low vortex-induced SST has little effect on the structure of the vortex at 4 km altitude and that influence is limited to 1 km altitude during the mature phase.

Fig. 10. Horizontal distribution of Vor at 4 km altitude at 43 h in experiments (a) CTL with *md* = 175, (b) OC with *md* = 175, and (c) OC with *md* = 17.5, and (d)-(f) at 1 km altitude at 43 h in the same experiments. Black solid contours indicate positive Vor and black dashed contours indicate negative Vor. Pink contours indicate equivalent potential temperature at the same altitudes and times.

directly below the vortex, but is influenced by the background effects of CP and SST

The horizontal distributions of Vor at altitudes of 1 and 4 km show clearly that there are differences in TC phases between the CTL and OC experiments with *md* = 175 and the OC experiment with *md* = 17.5 (Fig. 10). An annular Vor-ring does not form in the CTL experiment with *md* = 175 (Figs. 10a and 10d), but it does form completely in the OC experiment with *md* = 17.5 (Figs. 10c and 10f). Comparison of the CTL experiment with *md* = 175 (Figs. 10a and 10d) with the OC experiment with *md* = 175 (Figs. 10c and 10f) clearly shows a difference of TC phases between the experiments of CTL and OC, implying that the slow formation of the annular ring is affected by SSC at both spin-up and slowintensification phases (Figs. 7b and 8b). The effect of vortex-induced SSC on the annular Vor-ring at the mature phase differs from that during the other phases. The annular Vorring in experiment OC with *md* = 175 becomes small at 81 h (Fig. 11) as CP gradually increases during the mature phase (Fig. 1). The amplitude of Vor at 81 h at 4 km altitude in experiment CTL with *md* = 175 (Fig. 11a) is almost the same as that of experiment OC with *md* = 175 (Fig. 11b), even though the amplitude at 1 km altitude in experiment CTL with *md* = 175 (Fig. 11c) is higher than that in experiment OC with *md* = 175 (Fig. 11d). In addition, the SST in experiment OC with *md* = 175 is lower than that in experiment CTL with *md* = 175 (Fig. 2). These results are consistent with the small amplitude of Vor at 81 h in experiment OC with *md* = 175. Therefore, low vortex-induced SST has little effect on the structure of the vortex at 4 km altitude and that influence is limited to 1 km altitude during the mature

Fig. 10. Horizontal distribution of Vor at 4 km altitude at 43 h in experiments (a) CTL with *md* = 175, (b) OC with *md* = 175, and (c) OC with *md* = 17.5, and (d)-(f) at 1 km altitude at 43 h in the same experiments. Black solid contours indicate positive Vor and black dashed contours indicate negative Vor. Pink contours indicate equivalent potential temperature at

(a) (b) (c)

(d) (e) (f)

evolution.

phase.

the same altitudes and times.

Fig. 11. Horizontal distribution of Vor at 4 km altitude at 81 h in experiments (a) CTL with *md* = 175 and (b) OC with *md* = 175 and (c)-(d) at 1 km altitude at 81 h in the same experiments. Black solid contours indicate positive Vor and black dashed contours indicate negative Vor. Pink contours indicate equivalent potential temperature at the same altitudes and times.

#### **4.1.3 Absolute angular momentum analysis**

The location of the circulation center differs clearly from that of CP during the slow intensification phase, but these locations are almost the same after the rapid intensification phase (Fig. 3). These spatial differences may be related to a change of the budget of absolute angular momentum (AAM) for the vortex with a diameter of almost 100 km, which is given as the initial TC-like vortex (Wada, 2009). This subsection addresses the relation of the deepening of CP and intensification of vortex circulation to the budget of AAM.

AAM (*<sup>a</sup>*) is the sum of Vor and the Coriolis parameter (*f*):

*a v u <sup>f</sup> x y* , (2)

where *x* and *y* are zonal and meridional directions, and *u* and *v* are wind velocities in the *x* and *y* directions, respectively. The variation of AAM with time is expressed as

$$\frac{\partial \boldsymbol{\xi}\_{a}}{\partial t} = -\mathbf{v}\_{h} \nabla\_{h} \boldsymbol{\xi}\_{a} - w \frac{\partial \boldsymbol{\xi}\_{a}}{\partial z} - \boldsymbol{\xi}\_{a} \left(\frac{\partial u}{\partial x} + \frac{\partial v}{\partial y}\right) - \left(\frac{\partial w}{\partial x} \frac{\partial v}{\partial z} - \frac{\partial w}{\partial y} \frac{\partial u}{\partial z}\right)\_{\text{V}} \tag{3}$$

where *z* is the vertical direction and *w* the vertical wind velocity. The upward direction for *z* and *w* are indicated by positive values. The four terms on the right-hand side of Eq. (3) indicate (from left to right) horizontal advection, vertical advection, divergence and titling.

Effect of the Ocean on TC 59

compression in the upper troposphere and tilting in the lower troposphere play significant

In experiment CTL with *md* = 17.5, SSC had a marked effect on the evolution of horizontally averaged AAM from the slow intensification phase onward, particularly the vertical advection and divergence terms of Eq. (3) (Fig. 13). These results indicate that vortexinduced SSC contributes to a decrease of AAM because of a reduction of vertical advection near the surface and a reduction of stretching at around 6 km altitude. These decreases decay the start of compression in the upper troposphere, which is compensated for by the reduction of vertical advection. In addition, vortex-induced SSC results in a decrease of AAM at around 3 km altitude in response to the effect of tilting during the spin-up phase. Therefore, vortex-induced SSC leads to suppression of the intensification of the vortex in the

(c)

(d)

roles in suppressing the vortex at the mature phase.

Fig. 13. As in Fig. 12, but for experiment CTL with *md* = 17.5.

In experiment CTL with *md* = 17.5, SSC has a marked effect on the evolution of horizontally averaged AAM from the slow intensification phase onward, particularly the vertical In experiment OC with *md* = 17.5, the effect of SSC on horizontally averaged AAM commences during the slow intensification phase with the evolution of horizontally averaged vertical advection in the lower troposphere and stretching at around 6 km altitude (Fig. 14). Decreases in the vertical advection and divergence terms of AAM at the mature phase lead to decay of the acceleration of vortex intensification, weakening of vertical advection, and suppression of AAM. Therefore, budget analysis of AAM suggests that the budget depends on the phase of the vortex. The effects of vertical advection and stretching are essential for intensification of the vortex, whereas tilting at around 3 km altitude and compression in the upper troposphere suppress vortex intensification at the mature phase. SSC plays a crucial role in the decrease of AAM in response to vertical advection and stretching. The decrease of AAM subsequently leads to decay of the acceleration of vortex intensification and

lower troposphere (Fig. 13).

(a)

(b)

The divergence term expresses vertical intensification (suppression) of a vertical vortex in response to stretching (compression); the tilting term expresses transformation from a horizontal vortex to a vertical vortex. The averages of these terms were calculated over a 120 km square at the center of the computational domain.

The contributions of each of the term on the right-hand side of Eq. (3) to the budget of AAM to an altitude approaching 15 km in experiment AT is shown in Fig. 12. Horizontally averaged horizontal advection is positive near the surface throughout the integration (Fig. 12a) and repetitive filament-like positive and negative features are evident during the spinup and slow-intensification phases. However, horizontal advection is low over the entire integration, particularly after the rapid intensification phase.

Fig. 12. Time series of the vertical profile of experiment AT for AAM terms averaged over a 120 km x 120 km domain (corresponding to the scale of the initial vortex) at the center of the computational domain: (a) horizontal advection, (b) vertical advection, (c) divergence, and (d) tilting.

Horizontally averaged vertical advection is positive at all levels during the spin-up phase, particularly below 2 km altitude (Fig. 12b). The time series of both vertical advection and stretching show that these processes enhance vortex intensification until the rapid intensification phase. After the rapid intensification phase, the increase in AAM due to vertical advection is almost balanced by the decrease of AAM above 10 km altitude due to compression (Fig. 12c).

The tilting effect is highly negative around 3 km altitude after the slow intensification phase (Fig. 12d). In addition, from the slow intensification phase to the mature phase, it is locally highly negative at around 8 km altitude during the period when total water is also locally high (Fig. 5d). The AAM decreases in response to tilting at around 4 km altitude, but this is partly offset by vertical advection. The AAM budget analysis suggests that both

The divergence term expresses vertical intensification (suppression) of a vertical vortex in response to stretching (compression); the tilting term expresses transformation from a horizontal vortex to a vertical vortex. The averages of these terms were calculated over a 120

The contributions of each of the term on the right-hand side of Eq. (3) to the budget of AAM to an altitude approaching 15 km in experiment AT is shown in Fig. 12. Horizontally averaged horizontal advection is positive near the surface throughout the integration (Fig. 12a) and repetitive filament-like positive and negative features are evident during the spinup and slow-intensification phases. However, horizontal advection is low over the entire

(c)

(d)

Fig. 12. Time series of the vertical profile of experiment AT for AAM terms averaged over a 120 km x 120 km domain (corresponding to the scale of the initial vortex) at the center of the computational domain: (a) horizontal advection, (b) vertical advection, (c) divergence, and

Horizontally averaged vertical advection is positive at all levels during the spin-up phase, particularly below 2 km altitude (Fig. 12b). The time series of both vertical advection and stretching show that these processes enhance vortex intensification until the rapid intensification phase. After the rapid intensification phase, the increase in AAM due to vertical advection is almost balanced by the decrease of AAM above 10 km altitude due to

The tilting effect is highly negative around 3 km altitude after the slow intensification phase (Fig. 12d). In addition, from the slow intensification phase to the mature phase, it is locally highly negative at around 8 km altitude during the period when total water is also locally high (Fig. 5d). The AAM decreases in response to tilting at around 4 km altitude, but this is partly offset by vertical advection. The AAM budget analysis suggests that both

km square at the center of the computational domain.

integration, particularly after the rapid intensification phase.

(d) tilting.

(a)

(b)

compression (Fig. 12c).

compression in the upper troposphere and tilting in the lower troposphere play significant roles in suppressing the vortex at the mature phase.

In experiment CTL with *md* = 17.5, SSC had a marked effect on the evolution of horizontally averaged AAM from the slow intensification phase onward, particularly the vertical advection and divergence terms of Eq. (3) (Fig. 13). These results indicate that vortexinduced SSC contributes to a decrease of AAM because of a reduction of vertical advection near the surface and a reduction of stretching at around 6 km altitude. These decreases decay the start of compression in the upper troposphere, which is compensated for by the reduction of vertical advection. In addition, vortex-induced SSC results in a decrease of AAM at around 3 km altitude in response to the effect of tilting during the spin-up phase. Therefore, vortex-induced SSC leads to suppression of the intensification of the vortex in the lower troposphere (Fig. 13).

Fig. 13. As in Fig. 12, but for experiment CTL with *md* = 17.5.

In experiment CTL with *md* = 17.5, SSC has a marked effect on the evolution of horizontally averaged AAM from the slow intensification phase onward, particularly the vertical In experiment OC with *md* = 17.5, the effect of SSC on horizontally averaged AAM commences during the slow intensification phase with the evolution of horizontally averaged vertical advection in the lower troposphere and stretching at around 6 km altitude (Fig. 14). Decreases in the vertical advection and divergence terms of AAM at the mature phase lead to decay of the acceleration of vortex intensification, weakening of vertical advection, and suppression of AAM. Therefore, budget analysis of AAM suggests that the budget depends on the phase of the vortex. The effects of vertical advection and stretching are essential for intensification of the vortex, whereas tilting at around 3 km altitude and compression in the upper troposphere suppress vortex intensification at the mature phase. SSC plays a crucial role in the decrease of AAM in response to vertical advection and stretching. The decrease of AAM subsequently leads to decay of the acceleration of vortex intensification and

Effect of the Ocean on TC 61

Fig. 15. Evolution of CPs for various simulations of Typhoon Choi-Wan in 2009 from 00:00 UTC 17 September and JMA best-track CP. Model specifications are (a) 6 km horizontal grid spacing and GA, (b) 6 km horizontal grid spacing and JCDAS, (c) 12 km horizontal grid spacing and GA and (d) 12 km horizontal grid spacing and JCDAS. Results shown include those obtained by both the NHM and NHM-wave-ocean coupled model in combination with the oceanic reanalysis dataset with a horizontal grid spacing of 0.1º or 0.5º (see Table 2).

(a) (c)

(a) (c)

(b) (d)

(b) (d)

Fig. 16. Various simulated tracks for Typhoon Choi-Wan in 2009 from 00:00 UTC 17

GA and (d) 12 km horizontal grid spacing and JCDAS. Results shown include those obtained by both the NHM and NHM-wave-ocean coupled model in combination with the oceanic reanalysis dataset with a horizontal grid spacing of 0.1º or 0.5º (see Table 2).

September and the JMA best track. Model specifications are (a) 6 km horizontal grid spacing and GA, (b) 6 km horizontal grid spacing and JCDAS, (c) 12 km horizontal grid spacing and

Fig. 14. As in Fig. 12, but for experiment OC with *md* = 17.5.

suppression of the intensification process. The negative processes thus lead to weakening of adiabatic heating in the upper troposphere and contribute to vortex Rossby wave activity.
