A CASE STUDY OF POLAR LOW OVER THE JAPAN SEA ------ 2

A Case Study of Polar Low over the Japan Sea -- Part Two: The Numerical Simulation FU Gang^1,2, LIU Qinyu¹ H. NIINO², R. KIMURA² and WU Zengmao¹ (received 1997/9/18, revised 1998/4/25, accepted 1998/5/15) ABSTRACT

Following the results of the analysis of a typical polar low event that occurred on February 26, 1996 over the Japan Sea, the JSM (Japan Spectral Model) was used to simulate the development of this polar low event. Based on the detailed JSM outputs, the evolution and structure of this polar low is discussed in two stages, corresponding to the initial and mature stages. At the initial stage, the polar low generates in a baroclinic zone. In addition, the PV (Potential Vorticity) analysis indicates that an approaching upper-level PV anomaly later intensified the low-level cyclonic circulation. At the mature stage, strong sea surface fluxes from the relatively warm waters of the Japan Sea give rise to the instability of the low-level. Consequently, the vertical convection occurs due to the low-level convergence and the release of latent heat takes place through the convection. The CISK-like process supplies a mechanism for further development of the polar low. The analysis of vorticity indicates that the Changbai Mountain provides the positive vorticity contribution to the polar low.

(Key words: polar lows, Japan Spectral Model, numerical simulation, CISK-like process)

¹College of Physical and Environmental Oceanography, Ocean University of Qingdao, Qingdao 266003

² Division of Marine Meteorology, Ocean Research Institute, University of Tokyo, 164 Tokyo, Japan

INTRODUCTION Thus far, only several numerical simulations of polar lows based on observations have appeared in literature. As pointed out by Businger and Reed (1989), the modeling of the polar low poses a challenging problem to the numerical models. This is mainly due to the relatively coarse resolution of the most models (as compared to the small scale of the polar low), to a lack of detailed initial conditions, and to the incomplete parameterization schemes used for boundary layer and convective processes. In recent years, several studies with numerical models have shown that high horizontal resolution, accurate SST, and appropriate convective parameterization schemes are important for the simulation of polar lows.

Although numerical research on polar lows has made some progress in areas such as the Norwegian Sea, the Gulf of Alaska, and the Labrador Sea (Grf nås, et al., 1987; Nordeng, 1987; Nordeng and Rasmussen, 1992; Mailhot, et al., 1996), there have been comparatively less numerical studies of polar lows over the Japan Sea . In the second part of this paper, following our preliminary results of the analysis of a typical polar low, which occurred over the Japan Sea on February 26, 1996, we will use JSM to investigate in more detail various aspects of the evolution and structure of this polar low.

THE MODEL DESCRIPTION JSM is a limited-area spectral model which was developed by the JMA (Japan Meteoro-logical Agency) for operational forecasts (Segami et al., 1989). The main features of JSM are summarized in Table 1. The model is formulated in terms of the primitive equations in ? coordinates. The spectral method with a time-dependent lateral boundary condition (Tatsumi, 1986) is adopted in the model. The vertical levels are 19-? levels in the present study. The model expresses horizontal fields of model variables by the double Fourier series. The transform grids are 97´ 97 points in the zonal and meridional directions, respectively. The horizontal spacing of the model grids is chosen to be 40 km at 60^oN for the north polar stereographic projection. The distribution of the SST is the daily-analyzed value.

Table. 1

The horizontal momentum diffusion in the model is employed as the linear fourth-order Laplacian form, i.e., F_x = -K_dÑ⁴u, F_y = - K_dÑ⁴v. K_d is the diffusion coefficient. The precipitation processes in the model are the moist convective adjustments for the sub-grid-scale convection, which was proposed by Gadd and Keers (1970), large-scale condensation and evaporation of raindrops. Surface fluxes are calculated by a bulk method. The turbulent closure model is used for vertical diffusions. The ground temperature is calculated with a four-layer model. Radiation is taken into account only for the calculation of the ground temperature.
THE FULL-PHYSICS SIMULATION The calculation domain, as well as the topography and the sea surface temperature, of the present study are shown in Figure 1. Some features, such as the Changbai Mountain in the Eurasian Continent and Fuji Mountain in the Japan Islands, for example, are relatively represented in the topography field. Although most areas of topography are well depicted, the model mountains are not quite high enough in comparison to the real topography. As emphasized in numerous studies, polar lows are sensitive to the variations of SST, so detailed SST is analyzed in order to obtain a good representation of the polar low development. The initial value of the simulation was set at 00 UTC on February 26, 1996. The boundary value of the model is given by GANAL (JMA global objective analysis) data.
Fig. 1 The fields of topography (solid line, every 200 m) model and the sea surface temperature (dashed line, every 0.5^oC) on February 26, 1996, used by the JSM.
The description of the initial situation

At the initial time of the simulation, a surface low center with 1013 hPa was found at approximately 124.8^oE, 43.5^oN (Figure 2a). On the western side of this low center, northwesterly surface winds advected colder air southeastward. A core of cold air with the temperature of --10 ^oC or below was found. As stated in the first part of this paper, a strong baroclinic zone in the lower troposphere (1000 hPa ~ 700 hPa) was maintained over the area 36^oN ~ 42^oN, 115^oE ~ 125 ^oE.

Fig. 2 The initial conditions at 00 UTC on February 26, 1996. (a) The sea level pressure (solid line, in hPa) and surface temperature (dashed line, in ^oC), (b) The 500 hPa geopotential height (solid line, in m) and relative vorticity (dashed line, in 10^-5s^-1). The position of the surface low center is indicated by a black dot. The upper level situation was that a 500 hPa short wave trough was located around 121.5^oE, 43.8^oN, about 500 km upstream of the surface low (Figure 2b). The maximum 500 hPa vorticity associated with the trough was 2.7´ 10^-4s^-1. The large-scale situation described here provided a favorable condition for the development of a baroclinic cyclone.

The evolution and structure of the polar low

Figure 3a ~ 3d show the 500 hPa geopotential height and temperature fields at 6-h intervals since 12 UTC on February 26, 1996, for this simulation. The location of the surface low center is indicated by a black dot. The 500 hPa short wave amplifies and propagates rapidly southeastward during the first 12-h integration. By 12 UTC on February 26, 1996 (12-h forecast), the surface low has deepened 7 hPa (drops to 1008 hPa), and the surface low center is located around 300 km southeast underneath the 500 hPa geopotential height field center (Figure 3a). The lower-level atmosphere has developed a more idealized baroclinic structure. The propagating 500 hPa short wave supports the development of the surface low. This pattern persists over the next time period, while both the surface low and the 500 hPa short wave continue to move eastward. During their eastward movement (Figure 3b), the 500 hPa short wave propagates faster than the surface low and overtakes the surface low. By 24-h (00 UTC on February 27, 1996, Figure 3c) the surface low is almost directly underneath the 500 hPa geopotential height field center. The upper and lower level circulations are coupled with no vertical tilt. By 30-h (06 UTC on February 27, 1996, Figure 3d), the 500 hPa geopotential height field center has moved to the east of the surface low center. The condition unfavorable to further development has formed and suggests the end of the mature stage of the polar low and the beginning of its decay.

The 12-h forecast valid at 12 UTC on February 26, 1996, shows that the surface polar low center with 1008 hPa has reached 129.9^oE, 41.5^oN (Fig. 4a). It is a very interesting phenomenon that at this time the rear-half of the surface polar low system is still situated over the land. However, its preceding half part has reached to the Japan Sea. As indicated in the first part of this paper, the SST in the Japan Sea is much warmer. The underlying surface in which the temperatures are quite different will affect the rear and preceding half parts of the polar low quite differently. The analysis of heat fluxes indicates that there are larger surface heat fluxes from the Japan Sea into the preceding half part of the polar low in association with the strong winds. However, the heat fluxes from the land into the rear part of the polar low are rather small; in addition the winds are also much weaker.

The instability of the air over the sea can be clearly seen from the vertical cross section along the line EF indicated in Figure 4a because of ¶ q e /¶ p > 0 (Figure 4b). But the rear part of the air over the land is stable, for ¶ q e/¶ p < 0. This contrast between different underlying surfaces clearly shows that the stable air from land will become unstable after moving over to the warm sea surface. As time passes, the whole of the polar low will gradually move from the land to the sea. By 24-h (00 UTC on 27 February 1996), the surface low center with 1002 hPa has reached 133.4^oE, 42.3^oN (Figure 5a). The maximum surface wind speed reaches 20 s^-1. The 500 hPa geopotential field and temperature field coincide in phase with each other, and the relative vorticity reaches 2.8 f (Figure 5b). Combined with the analysis of GMS-5 satellite images, it is suggested that the polar low was in the mature stage at that time.

Fig. 3 The 500 hP a geopotential height (solid line, in m) and temperature (dashed line, in ^oC) for the full-physics simulation. (a) 12-h forecast at 12 UTC on February 26, 1996, (b) 18-h forecast at 18 UTC on February 26, 1996, (c) 24-h forecast at 00 UTC on February 27, 1996, (d) 30-h forecast at 06 UTC on February 27, 1996. The position of the surface low center is indicated by a black dot. Figure 6a shows the PV (Potential Vorticity) anomaly in the layer of 300 hPa to 400 hPa after 24-h integration (00 UTC on February 27, 1996). This figure clearly indicates that there exists a positive PV anomaly in the upper-level. The maximum PV value in the layer of 300 hPa to 400 hPa is around 6.5 PVU ( 1 PVU = 10^-6 KPa^-1m s^-3).

Figure 6b, the vertical cross section of PC along the line GH, shows that air of stratospheric origin marked by PV with a value >1 PVU, extended from upper level to about 700 hPa. The decent of high-PV air from the upper-level will intensify the low-level cyclonic circulation (Hoskins et al., 1985).

Fig. 4 The 12-h forecast at 12 UTC on February 26, 1996. (a) Sea level pressure (solid line, in hPa). The position of the cross-section is indicated by line EF, (b) The vertical distribution of the equivalent potential temperature along the line EF(in K).
Figure 7a shows the horizontal distribution of surface fluxes of sensible and latent heat at 00 UTC on February 27, 1996. A higher anomaly of heat fluxes on the order of 550 Wm^-2 is found in the vicinity of the polar low center. The equivalent potential temperature q e, along the line GH, shows the presence of instability (¶ q e/¶ p > 0) in the lower atmosphere (Figure 7b). The creation of this low-level instability is attributed to destabilization by larger surface fluxes of sensible and latent heat in this region. At the same time, the surface divergence field indicates that there is a strong convergence center that corresponds to the heat fluxes anomaly center. The low-level atmosphere is moisture because the relative humidity at the surface is high in the area of the polar low center. In the upper level, a high divergence center in concurrence with the convergence center below is found. The low-level moisture convergence and divergence aloft in the polar low central area drives a strong upward motion over -20 hPa hour^-1 over the surface polar low center (Figure 7b). This indicates that shallow convection occurred. Fig. 5 The 24-h forecast at 00 UTC on February 27, 1996. (a) Sea level pressure (solid line, in hPa) and wind speed (in m s^-1). Line GH is the cross-section line for later use, (b) The 500 hPa geopotential height (solid line, in m) and temperature (dashed line, in ^oC). The shaded area indicates the relative vorticity greater than 1.5´ f ( f: Coriolis parameter). Fig. 6 The 24-h forecast at 00 UTC on February 27, 1996. (a) The PV anomaly in the 300 hPa to 400 hPa layer (in PVU, 1 PVU = 10^-6KPa^-1m s^-3), (b): The cross-section of PV (solid line, in PVU) and potential temperature (dashed line, in K) along the line GH. The thick arrow indicates the surface polar center.
The vertical convection occurs due to the environmental low-level moisture convergence, and the release of latent heat takes place through the convection. The diabatic heating drives the low-level convergence to maintain a favorable condition for further convection. This CISK-like process supplies a mechanism for further development of the polar low. Fig. 7 After 24-h forecast at 00 UTC on February 27, 1996. (a) The surface fluxes of sensible heat and latent heat (in Wm^-2). The shaded area indicates the heat fluxes greater than 520 Wm^-2, (b): The vertical p-velocity (solid line, in hPa/hour) and equivalent potential temperature (dashed line, in K) along the line GH. The position of the surface polar low center is indicated with a thick arrow. The dissipated stage of the polar low is not only marked by the upper-level vorticity center having moved to the east of the surface low center, but also by the ceasing convection (¶ q e/¶ p » 0). As there is the lack of energy supplies, the polar low tends to decay after 30-h integration.

The effect of Changbai Mountain

Although the strengthening effects upon the polar low due to the topography of Greenland, Iceland and the Spitsbergen mountains have been noted by Grf nås et al. (1987), the mechanism has not been analyzed in detail.

As indicated in the first part of this paper, the polar low passed over Changbai Mountain around 12 UTC on February 26, 1996. Here, the question of what the influence of Changbai Mountain is upon this polar low will be examined.

According to the vorticity equation in isobaric coordinates:

¶ z /¶ t = - V· Ñ (z +f) - w ¶ z /¶ p - (z +f) Ñ · V + k· (¶ V/¶ p ´ Ñ w) + k· Ñ ´ F

where the term on the left-hand-side of the equation is the local time tendency term of the vorticity. On the right-hand-side, the first term is the vorticity advection term, and the second term is called the vertical convection term. The third one is the divergence term, and the fourth term is the tilting or twisting term. The last term is due to the friction effect.

After integration of 12-h (at 12 UTC on February 26, 1996), at the lee of Changbai Mountain there exists a positive vorticity zone at 850 hPa. At level 1000 hPa, a positive vorticity zone is also found. These positive vorticity regions correspond to a short wave trough of 500 hPa aloft. As indicated in textbooks of fluid mechanics, each term in the vorticity equation can produce the vorticity in different way. Here our concern is mainly on the friction term. We calculate the contribution of the friction term, which is mainly due to the influence of Changbai Mountain (Figure 8a). The calculation shows that Changbai Mountain provides the positive vorticity contribution to the polar low with the maximum of 3.1´ 10^-5s^-2 .

The question of how Changbai Mountain provides the positive vorticity to the polar low will be investigated. Figure 8b shows that the northerly winds crossing Changbai Mountain become southwesterly on its eastern side. The mountain blocks the lower-level flows, and the flows are forced to steer around to the eastern side of the mountain. This large mountain topography causes the deformation of the horizontal flow and forms the horizontal wind shear (both in direction and speed). Obviously, the positive vorticity field is induced by this horizontal wind shear.

DISCUSSION We now present the results of the full-physics simulation for comparison with the objective analysis and satellite image. The 24-h prediction is relatively accurate in terms of the position of the low and its intensity. However, the prediction of the low's position after 24-h is less successful. The model simulates the polar low slightly far to the north, and the travel speed is slightly faster than the objective analysis.
Fig. 8 At 850 hPa after 12-h forecast at 12 UTC on February 26, 1996. The shaded area indicates the topography higher than 1000 m. (a) The vorticity contribution due to the effect of Changbai Mountain (in 10^-5s^-2), (b) The horizontal wind shear due to the topography effect.
However, the warm core of this polar low event is not reproduced successfully in this model. The warm core is a very interesting phenomenon. In the study of Mailhot et al. (1996), the warm core structure of a polar low event in the Labrador Sea was successfully reproduced. The authors thought that this warm core was probably due to the combined effects of warm air seclusion by cold air wrapping around the surface low, and the diabatic heating from intense convection. It should be emphasized that that polar low event was in many aspects stronger than in the present case. At 500 hPa the core of cold air reached a temperature of -- 47^oC (about 8^oC lower than in the present case), and the surface pressure of 989.5 hPa (about 12 hPa lower than in the present case) of that polar low had been reported. Particularly, after the 12-h integration, the maximum value reached 1400 Wm^-2 for the sensible heat flux, while the smaller values of 500~600 Wm^-2 were found for the latent heat flux. However, the maximum value of sensible heat and latent heat of the present simulation is only about 550 Wm^-2 in the most mature stage of the polar low. The diabatic heating of sensible heat and latent heat is not so strong. And the warm core is also a weak and shallow system, as indicated in the first part of this paper.

The failure of reproducing the warm core in the model is probably caused by a number of factors. It is perhaps related to the incomplete parameterization of this model. The moist adjustment scheme used in the model is the simplest convective scheme that was proposed by Gadd and Keers (1970). There is no detailed meso-scale physical processes described in this scheme. There is only a final state of conditional neutrality being supposed, just like a "black box", to calculate the effect of convection. Furthermore, the coarse resolution of the present model (40 km) is probably another reason for failure.

SUMMARY AND CONCLUSIONS In the second part of this paper, we present the JSM numerical simulation results of a typical polar low that occurred over the Japan Sea on February 26, 1996. The model was initialized at 00 UTC on February 26, 1996, and the simulation showed the polar low at approximately the right position and time. Based on the detailed outputs, the evolution and structure of the polar low over the Japan Sea is discussed in two stages corresponding to the initial and mature stages. At the initial stage, the polar low generated in a baroclinic zone. In addition, the PV analysis indicates that an approaching upper-level PV anomaly later intensified the low-level cyclonic circulation. At the mature stage, strong sea surface fluxes from the relatively warm waters of the Japan Sea give rise to the instability of the lower atmosphere. Consequently, the vertical convection occurs due to the environmental low-level moisture convergence, and the release of latent heat takes place through the convection. The CISK-like process supplies a mechanism for further development of the polar low. The analysis of vorticity suggests that Changbai Mountain provides the positive vorticity contribution to the polar low by blocking the low-level flows and forming the horizontal wind shear both in direction and speed.

In summary, the structure of the polar low simulated by JSM is found to be consistent with the available observations. However, it is also worth noting that the model develops the polar low slightly far to the north, and the travel speed is a bit faster than in the observation. The warm core phenomenon was not successfully reproduced by this model.

Finally, it should be pointed out that numerical experiments are necessary in order to examine the effects of various physical factors upon the development of the polar low. Another paper will continue to deal with this problem.

ACKNOWLEDGMENTS The first author would like to express his hearty thanks to Dr. K. Tsuboki for his help in running the JSM. He is very grateful to Prof. Zhou Faxiu for reading the manuscript and giving very helpful suggestions. Thanks are also extended to the staffs & students in the Division of Marine Meteorology, Ocean Research Institute, University of Tokyo, for their kind assistance in this research during his study in Japan. GrADS (Grid Analysis and Display System) is used to draw some figures.
References

Businger, S. and R. J. Reed (1989) Cyclogenesis in polar air masses. Wea. Forecasting, 4, 133-156.

Gadd, A. J. and J. F. Keers (1970) Surface exchanges of sensible and latent heat in a 10-level model atmosphere. Quart. J. Roy. Meteor. Soc., 96, 297-308.

Grf nås, S., A. Foss and M. Lystad (1987) Numerical simulations of polar lows in the Norwegian Sea. Tellus, 39A, 334-353.

Hoskins, B. J., M. E. McIntyre and A. W. Robertson (1985) On the use and significance of potential vorticity maps. Quart. J. Roy. Meteor. Soc., 111, 877-946.

Mailhot, J., D. Hanley, B. Bilodeau and O. Hertzman (1996) A numerical case study of a polar low in the Labrador Sea. Tellus, 48A, 383-402.

Nordeng, T. E. (1987) The effect of vertical and slantwise convection on the simulation of polar lows. Tellus, 39A, 354-375.

Nordeng, T. E. and E. A. Rasmussen (1992) A most beautiful polar low. A case study of a polar low in the Bear Island region. Tellus, 44A, 81-99.

Segami, A, K. Kurihara, H. Nakamura, M. Ueno, I. Takano and Y. Tatsumi (1989) Operational Mesoscale Weather Prediction with Japan Spectral Model. J. Metor. Soc. Japan, 67, 907-924.

Tatsumi, Y. (1986) A spectral limited-area model with time-dependent lateral boundary conditions and its application to a multi-level primitive equation model. J. Metor. Soc. Japan, 64, 637-663.