The band-shaped heavy rainfall observed in northern Kyushu on 29 June 1999 was brought by the passage of a cold front that had been rapidly enhanced during a few hours. A southwesterly low-level jet existing between 2 and 3 km high was observed to the south of the cold front. In the present study, the enhancement process of the cold front and the intensification of a low level jet associated with this front are examined by using the special observation data of X-BAIU-99 and the MRI-nonhydrostatic mesoscale model (Saito and Kato, 1996, MRI-NHM). Special observations were conducted by using Doppler radar, GPS, wind profiler, 4 vessels of JMA, and so on (Fig. 1). Used are the special upper sounding data at Nagashima in the present study.
Fig.1 X-BAIU-99 special observation
* Corresponding author address: Teruyuki Kato, Meteorological Research Institute, 1-1 Nagamine, Tsukuba, Ibaraki 305-0035, Japan
e-mail: tkato@mri-jma.go.jp
Fig.2 Surface weather mapat 09 JST 29 June 1999
When a heavy rainfall was observed at Fukuoka, located northern part of Kyushu, at 08 JST on 29 June 1999, a cold front associated with a synoptic low pressure having moved eastward on the East China Sea was passing over northern Kyushu (Fig. 2).
The maximum precipitation of 99.5 mm hr-1 was observed in the present event. Figure 3 shows the observed hourly-accumulated rainfall at 08 JST on 29 June when a heavy rainfall was observed at Fukuoka. A line-shaped precipitation system was organized associated with the cold front. At 30 km north of Fukuoka, the cold front passed at 06 JST but the temperature hardly dropped. On the other hand, the drop of temperature associated with the passage of cold front at Fukuoka at 08 JST was about 4 degrees. Therefore, the cold front strongly enhanced between these times. This enhancement is also ascertained by infrared image of Satellite data.
Fig. 3
Observed hourly-accumulated precipitation at 08 JSTRadar observation showed that a line-shaped rainfall area passed over Fukuoka at 08 JST, and convective cells moved eastward with the speed of 15 m/s and northward with 20 m/s (Fig. 4). The direction of their movements is north-northeast. This direction has more northward component than that of the line-shaped rainfall area.
Fig.4 (a) South-north and (b) east-west components of convective cell's movement analized by JMA's operational radar.
Two models are used in the present study. One is the JMA's operational regional spectral model (RSM), horizontal resolution is 20 km, grids 129 x 129 in horizontal and 36 layers in vertical. For the precipitation schemes, moist convective adjustment, prognostic Arakawa-Schubert and large-scale condensation are used. The other is the MRI-NHM, whose horizontal resolution is 5 km and 2 km, grids 200 x 200 in horizontal for 5 km model, 340 x 265 in horizontal for 2 km model, and 38 layers in vertical. For the precipitation scheme, a cold rain scheme explicitly predicting the mixing ratios of cloud water and ice, rainwater, snow and graupel is used.
The initial data of RSM are obtained from the regional objective analysis and boundary data are from GMS forecast. For the initial data of 5km-NHM, the 3 hour forecast of RSM is interpolated, and for those of 2km-NHM, the 3 hour forecast of 5km-NHM is interpolated.
4.1. Intensification of the cold front
The intensity of RSM-predicted rainfall (not shown) is weak and the RSM could not predict a line-shaped form. Figure 5 shows the rainfall areas simulated by the 2km-NHM. The line-shaped rainfall area was reproduced successfully, and the simulated precipitation intensity corresponds well with the observation in addition to the form of precipitation system (Fig.3).
Fig.5 Same as Fig.3 but 2km-NHM-predicted rainfall.
Nest, discussed is the enhancement process of cold front in the present case. Simulated equivalent potential temperature fields show that a cold air advecting from the northwest in the middle layer flowed into the west of northern Kyushu earlier than that in the lower layer. This environment with intense convective instability is suitable for moist convection to develop when a low-level humid air flows from the south into this region.
4.2. Low-level jet intensification process
Usually, a heavy rainfall observed over Kyushu during the Baiu season is brought from the inflow of an air with the specific humidity larger than 20 g kg-1. In the present case, however, the specific humidity is under 18 g kg-1 to the south of the cold front. Further, as the characteristic features of the Baiu frontal zone, a near-surface air is nearly saturated. This means that an intense ageostrophic flow is produced in the lower layer due to the convection-induced low pressure produced by latent heat release.
Well then, how strong will the low-level ageostrophic flow be intensified by the convection-induced low pressure? Figure 6 show the height-time cross section of horizontal winds simulated by the 5km-NHM. The wind speed in the lower layer became from under 20 m s-1 to over 30 m s-1, while that in the middle layer decreased at first, and then increase around the rainband. This indicates that the enhanced horizontal momentum in the lower layer was transported to the middle layer by moist convection, and supplied a great deal of water vapor to moist convection to induce a heavy rainfall.
This transportation is ascertained by the simulation result. Figure 6 also shows the upward transportation of large horizontal momentum around 07 JST. This time corresponds with that when the cold front was enhanced. This upward transportation is also found in the special upper sounding observations at Nagashima (not shown).
The intensification of a low-level jet during the Baiu season is explained by several hypotheses. One is the downward transportation of upper jet due to the convective mixing (Matsumoto and Ninomiya, 1971). Another is the Coriolis acceleration of the ageostrophic wind produced by adiabatic heating (Chou et al., 1990). These studies are the result using coarse mesh models with convection parameterization schemes, so that vertical transportation of horizontal momentum is not clear. On the other hand, the result using high mesh model with an explicit precipitation scheme in the present study seems to be considerably exact. The present results correspond with those of Kato (1998).
Fig.6 Height-time cross section of horizontal winds simulated by the 5km-NHM. The speed is averaged among
32N,129.5E within 1 degree.
Fig.7 Same as Fig. 4 but analyzed by the 2km-NHM simulated mixing ratio of rainwater.
4.3. Structure of the cold front
By using the maximum mixing ratios of rainwater in the vertical columns simulated by 2km-NHM, the movements of convective cells are analyzed (Fig. 7). This analysis reveals that the rainband associated with the cold front consists of several meso-
?-scale convective systems (M?CSs) lying in a more northward direction than the rainband. The horizontal interval of each M?CS is almost 100 km. Figure 7a also shows that the rainband moves southward at the speed less than 5 m s-1, while the M?CSs move northward at the speed of about 5 m s-1. New convective cells successively generate south of each M?CS, then move northward at the speed larger than 10 m s-1 with developing, and merge into the M?CS. This formation process of convective system indicates that the present M?CSs have a multicell structure. Figure 7b shows that the eastward speed of M?CSs is 10 m s-1, while convective cells forming repeatedly with an interval of about 30 min move eastward at the speed of about 20 m s-1. Since the polar-stereo map projection with the standard longitude of 140oE is used in the NHM, the direction in the Kyushu area rotates clockwise by about 10 degree. Thus, simulated convective cells move northwestward, and its direction almost agrees with that of the radar observation (Fig.4).Figure 8 shows the 3-dimensional structure of clouds predicted by the 2 km-NHM. This is visualized by predicted cloud water larger than 0.5 g kg-1. It can be seen that convection forms at the edge of rainband, and moves northeastward with developing higher. Another convection, which is still shallow, is also found at the middle area of rainband.
Fig. 8 3-dimensional structure of clouds predicted by the 2 km-NHM.
4.4. Height of convective cells
The height of convective cells is estimated by using the 2km-NHM simulated upward velocity. It is determined that a convective cell exists in a vertical column where the maximum upward velocity exceeds 2 m s-1, The height of convective cell is determined to be the position at which the upward motion changes to downward.
Figure 9 shows the distribution of convection height, averaged in a south-north direction between 06 JST and 09 JST. Most of convective cells develop to the height of 5-7 km in the west part of rainband, while many convective cells develop to the tropopause (about 14 km) in the east part. Further, the second peak is found in the height of 5-7 km in the east part. This result hardly changes even in the sensitive experiment removing the ice-phase process from the precipitation scheme. Therefore, two peaks of convection height could be not induced by the release of latent heat of solidification.
Fig.9 Vertical distribution of convection height averaged in a south-north direction over the area shown in the figure between 06 JST and 09 JST.
Horizontal scale of upward motion core is calculated to examine the formation of deep convections (Fig. 10). This scale can estimate the scale of convective cells. The scale for the upward motion larger than 1 m s-1 at the western edge of rainband is about 4 km, while that in the east part becomes about 10 km. In the west part of rainband where the scale of convective cells is small, convective cells can not develop above the middle layer with low equivalent potential temperature (
qe), because a low qe air intrudes into them to suppress their development. On the other hand, large-scale convective cells in the east part of rainband can develop high, because the intrusion of a low qe air into them could be little.