2021 Volume 99 Issue 5 Pages 1351-1369
This study presents the impacts of evaporative cooling from raindrops on precipitation over western Japan associated with the Baiu front during a heavy rainfall event from 5 to 8 July 2018. We conducted analyses of the dynamic and thermodynamic features of the stationary Baiu front using the Japanese 55-year reanalysis. During this period, large amounts of water vapor were continuously transported to the stationary Baiu front, supporting the record-breaking rainfall. The 299K isentropic surface was identified as the frontal surface. Along the isentropic surface, warm moist air adiabatically ascended, became saturated at an altitude of approximately 500 m, and initiated active precipitation systems. We found that the diabatic cooling near the tip of the frontal surface played an important role in retaining the position of the frontal surface and stalling its northward retreat. Additionally, numerical sensitivity experiments were conducted to examine the impacts of evaporative cooling and topography on heavy rainfall formation using a cloud-resolving non-hydrostatic numerical model (the Japan Meteorological Agency Non-Hydrostatic Model: JMA-NHM) with a horizontal resolution of 3 km. A heavy precipitation area extending from the Chugoku region to central Kinki was simulated regardless of whether the terrain was flattened or not. Precipitation formed primarily as a result of updrafts above a frontal surface at a potential temperature of 300 K. This precipitation area shifted northward by more than 100 km when raindrop evaporation was excluded from the experiment. Raindrop evaporation suppressed the northward retreat of the frontal surface by maintaining cold airmass amounts below the frontal surface.
From 28 June to 8 July 2018, a record-breaking heavy rainfall event occurred primarily in western Japan and in the Tokai region (see Fig. 1). The Japan Meteorological Agency (JMA) designated it as the “Heavy Rain Event of July 2018”. During this period, the total amount of rainfall exceeded 1,000 mm in some regions, reaching as much as 2–4 times the typical monthly precipitation amounts for July (Japan Meteorological Agency 2018). The Baiu front, which typically extends from west to east across Japan in early summer, experienced unusual stagnation in western Japan from 5 to 8 July. Consequently, accumulated precipitation amounts in this area increased significantly.
Shimpo et al. (2019) pointed out the main factors of the heavy rainfall from 5 to 8 July; two extremely lowlevel wet streams from the South China Sea and the tropical western North Pacific merged over western Japan. The continuous updraft over the Baiu front caused heavy rainfall. The stationary Baiu front was considerably enhanced between the North Pacific subtropical and Okhotsk highs. Line-shaped heavy precipitation systems caused heavy mesoscale rainfall in some areas (Tsuguti et al. 2019). Takemura et al. (2019) demonstrated that the moisture convergence during 5–7 July was the largest in western Japan since 1958. A strong southerly wind enhanced evaporation from the surrounding ocean, particularly along the Kuroshio Current, and transported water vapor to the frontal region (Sekizawa et al. 2019).
Moteki (2019) reported that typhoon T1807 (Prapiroon) strongly transferred cold air from the Sea of Okhotsk to the Sea of Japan. At the same time, the Okhotsk High rapidly expanded southward. Both the southward pressure gradient from the Okhotsk High and the northward pressure gradient from the North Pacific Subtropical High stagnated the Baiu front in western Japan. Moteki further stated that strong ascending of warm air continued above thick cold air over the Sea of Japan and led to heavy rainfall along the Baiu front.
Few analyses have been conducted to separate the upward flow into adiabatic and diabatic components in this case. Accordingly, it is unclear on which isentropic surface warm moist air ascended dominantly and caused strong condensation during this period. The stagnation factors of the Baiu front have been studied from a dynamic perspective (Moteki 2019) but not intensively from the perspective of diabatic processes. Markowski and Richardson (2010) reported that the motion of surface fronts is not only dependent on front-normal isallobaric gradients but is also influenced by diabatic effects. Accordingly, the following were identified as possible factors involved in the stagnation of fronts, i.e., the strength of warm and cold air advection in the surrounding area, the enhancement of convergence corresponding to rainfall phenomena, and diabatic heating and cooling caused by physical processes, such as cloud microphysical processes, radiative processes, and planetary boundary layer processes. The present study particularly focused on the impacts of evaporative cooling from raindrops.
Using a semigeostrophic frontogenesis model, Huang and Emanuel (1991) reported that evaporative cooling forms a strong and concentrated downdraft at the cold side of a frontal zone. The evaporative cooling of raindrops has in some cases been reported to have had an impact on Baiu front systems. Ishihara et al. (1995) reported for the Baiu front environment with a small horizontal temperature gradient that cold air generated in the downdraft by evaporative cooling diverged within the lower layer and formed a gust front between the cold pool and the warm moist southwesterly flow, thereby maintaining convective activities. Davis and Lee (2012) described a case in which the evaporative cooling from raindrops contributed to the initiation and intensification of convection through maintenance of a coastal front by forming a cold pool. Nagata and Ogura (1991) showed in a sensitivity experiment that the evaporative cooling from stratiform precipitation in the Baiu front strengthened frontogenesis by forming a cold pool below the front; they also indicated that a convective precipitation area, produced in an experiment without the evaporation from stratiform raindrops, moved eastward faster. Moteki et al. (2004) reported from a case study that a cold pool, formed by evaporative cooling, pushed the Baiu front southward approximately 50 km within 2 hours. Jeong et al. (2016) presented another example in which the evaporative cooling of raindrops prevented mesoscale convective systems from moving northward.
Evaporative cooling does not always contribute to Baiu frontal precipitation. Sensitivity experiments conducted by Kato and Goda (2001) demonstrated that the evaporative cooling of raindrops did not affect the formation of a quasi-stationary rain band observed in the Niigata area, central Japan, on 4 August 1998. Liu and Moncrieff (2017) showed that a weak cold pool, generated by evaporative cooling, negligibly affected convection initiation in shear-parallel mesoscale convective systems in an idealized low-inhibition and unidirectional shear environment of the Baiu moisture front. Therefore, it is necessary to investigate how and in what specific case evaporative cooling contributes to Baiu frontal precipitation system.
This study was conducted to identify the isentropic surface on which warm air ascending was the most active in the latter period of the heavy rainfall event that occurred in western Japan in July 2018. The aim of doing so was to clarify the mechanisms of precipitation formation and to investigate the influences of diabatic processes on the position of the isentropic surface. Furthermore, we also examined the impacts of evaporative cooling from raindrops in sensitivity experiments using a non-hydrostatic model (JMA-NHM; Saito et al. 2006).
The remainder of this paper is organized as follows. Section 2 describes the data, the methods for isentropic and cold airmass analyses, and the numerical models used in this study. Section 3 presents the analyses using data from the Japanese 55-year reanalysis (JRA-55; Kobayashi et al. 2015), the global atomspheric reanalysis conducted by the JMA. Section 4 presents the results of the sensitivity experiments that were conducted using the JMA-NHM. In this section, we show and describe the impacts of evaporative cooling on the Baiu frontal precipitation system. Finally, the study conclusions are presented in Section 5.
The regions described in this article (the East China Sea, the Pacific Ocean, the Sea of Japan, and the Kyushu, Chugoku, Shikoku, Kinki, and Tokai regions) are presented in Fig. 1. The location of the Sea of Okhotsk outside the range of Fig. 1 is depicted in Fig. 3. Kochi Prefecture, described in Section 4.1, is located on the Pacific side of the Shikoku region.
Large-scale features of the stationary Baiu front were examined using JRA-55. In the JRA-55 reanalysis system, the JMA's Global Spectral Model (JMAGSM; Japan Meteorological Agency 2013) is used as a forecasting model for data assimilation. We used six-hourly snapshots of analysis values with a horizontal resolution of 1.25° × 1.25° at fixed isobaric levels. The JMA radar observation data with a horizontal resolution of 1 km were used to validate the precipitation that was reproduced in numerical experiments using the JMA-NHM.
2.2 Isentropic analysis and cold airmass analysisGenerally, an isentropic surface located at the warmer end of a potential temperature gradient zone is called a frontal surface. An intersection between a frontal surface and the ground, or an isobaric surface, is called a front. In this study, based on this definition, the structure and temporal evolution of the Baiu front were studied using the cold airmass analysis proposed by Iwasaki et al. (2014). Cold airmass amount DP and horizontal cold airmass flux MF are given as follows:
 |
 |
 |
. This equation shows that temporal change of the cold airmass amount is determined from two terms, i.e., the convergence or divergence of the horizontal cold airmass flux (the first term on the right-hand side in Eq. 3) and the decrease or increase of the cold airmass amount by diabatic heating or cooling (the second term). The diabatic change (the second term on the right-hand side of Eq. 3) can be indirectly estimated using the conservation equation of cold airmass amount.
In this study, we analyzed the vertical flow as follows. The vertical p velocity ω in the isentropic coordinate system is given as follows:
 |
In this study, we also conducted sensitivity experiments to assess the impacts of evaporative cooling from raindrops using the JMA-NHM. The initial and lateral boundary conditions were provided by JMA mesoscale analyses with a horizontal resolution of 5 km every 3 hours (Japan Meteorological Agency 2013). The sea surface temperature (SST) was fixedly given by the merged satellite and in-situ data Global Daily Sea Surface Temperature (MGDSST; Kurihara et al. 2006) on 5 July 2018.
In addition to the control experiment CNTL, a sensitivity experiment NOEVAP was conducted without raindrop evaporation, i.e., the transformation of raindrops into water vapor was turned off. It is noted that the evaporation from other hydrometers, including cloud water, was not turned off, even in the NOEVAP experiment. The combined use of a cloud microphysics scheme and a cumulus parameterization scheme makes it difficult to evaluate the impacts of evaporative cooling from raindrops. Accordingly, a high-resolution cloud-resolving model with 3 km grid-spacing was used to avoid the use of a cumulus parameterization scheme. To assess the influences of the terrain, we also conducted ideal FLAT and NOEVAP_FLAT experiments, in which the terrain over western Japan was flattened for the CNTL and NOEVAP experiments, respectively. The computational domain is colored in Fig. 1. The flattened area in the FLAT and NOEVAP_FLAT experiments is enclosed by a square. At the boundary of this flattened area, the terrain was smoothed so that the slope of the terrain between adjacent grids was not too steep.
The computational domain of the numerical experiments. The topographic height is shown alongside a color scale. The topography within the domain surrounded by the black square was flattened for conducting the FLAT and NOEVAP_FLAT experiments.
The detailed model settings were as follows. There were 501 × 501 horizontal grid points with a 3 km grid spacing and 50 vertical levels (Δz = 40 m at the bottom level, Δz = 540 m at the top level, and 13.2 km at the top). The time integration period was 96 hours from 1500 UTC 3 July to 1500 UTC 7 July. The NOEVAP and NOEVAP_FLAT experiments were conducted without the raindrop evaporation from the start to the end of the calculation. For parameterizing cloud microphysics processes, the model used a two-moment bulk scheme that predicts the mixing ratios of cloud water, rain, cloud ice, snow, and graupel, as well as the numeric concentrations for cloud ice, snow, and graupel (Ikawa and Saito 1991). The Mellor-Yamada Nakanishi-Niino level 2.5 scheme (Nakanishi and Niino 2004) was used as a planetary boundary layer parameterization scheme. The model terrain was prepared using the GTOPO30 (Global 30 arc-second elevation) with a horizontal resolution of approximately 1 km.
The target event in this study was characterized by the long stationarity of the Baiu front with a wide range of active precipitation. For that reason, the heavy precipitation for a long time caused severe damages in western Japan.
Figure 2 shows the geopotential heights and the horizontal wind vectors at 975 hPa and 700 hPa, respectively, during 4–6 July 2018 and illustrates the cold airmass amount below the isentropic surface of 296 K. The 296 K surface was identified as the top of cold air blowing from the northeast (described in the next subsection). The extratropical cyclone transformed from Typhoon T1807 over the Sea of Japan on 4 July and traveled east-northeastward, reaching the northern Pacific on 6 July. Following the passage of this extratropical cyclone over the Sea of Japan, two distinct high-pressure systems prevailed over the Sea of Okhotsk and the subtropical northern Pacific Ocean. A low-pressure region over the East China Sea extended eastward between the two high-pressure systems.
The geopotential height (m, contour) and wind velocity (m s−1, vector) at 975 hPa at (a), (b), and (c) 0000 UTC during 4–6 July 2018, (d) at 700 hPa on 0000 UTC 6 July, and (e) cold airmass amount (hPa, colored) and cold airmass flux (hPa m s−1, vector) below the 296 K isentropic surface at 0000 UTC 6 July, derived from JRA-55. Contour intervals are 20 m in (a)–(d).
Behind the extratropical cyclone that had transformed from Typhoon T1807, cold air blew into the Sea of Japan (see Figs. 2b, c, and the horizontal distribution of potential temperature in Fig. 4a). As pointed out by Moteki (2019), the southwestward cold airflow on the back of Typhoon T1807 was enhanced by the southward migration of the Okhotsk High. The southwestward cold airflow was not found at 700 hPa (Fig. 2d), which meant that the cold airflow was located to a limited degree in the lower layer. This cold airflow was able to effectively push the Baiu front southward. On 6 July, the cold air prevailed over the entire Sea of Japan.
The daily mean of vertically integrated horizontal water vapor flux (kg m−1 s−1, vector) and its divergence (kg m−2 s−1, colored), derived from JRA-55 during 4–7 July 2018.
Another important airflow was the subtropical southerlies, which transported warm moist air toward the Baiu front. This airflow was caused by the large pressure gradient force between the low-pressure region over the East China Sea and the North Pacific Subtropical High. Unlike the cold airflow over the Sea of Japan, the warm airflow around the North Pacific Subtropical High at 700 hPa closely resembled that at 975 hPa. Figure 3 shows that the subtropical southwesterly wind transported large amounts of water vapor to the Baiu front. The water vapor flux convergence increased in western Japan and reached a maximum on 6 July, and which was maintained until 7 July. Takemura et al. (2019) reported that significant amounts of water vapor were brought toward the convergence area by the confluence of the deeply moist southwesterly wind and the low-level southerly wind. They also reported that the successive formation of active convection over the southern part of the East China Sea generated low-level cyclonic vorticity, and, consequently, large northeastward water vapor transportation was maintained.
As a result, the stationary Baiu front was strongly enhanced over western Japan through close contact between the strong southwestward cold airflow and the northeastward warm airflow. The warm moist northeastward airflow rose over the cold southwestward airflow. These synoptic situations persisted during 5–8 July and caused the record-breaking accumulated precipitation amounts.
(a): The potential temperature (K, contour) and horizontal wind (m s−1, vector) at 975 hPa, averaged over two days, 5 and 6 July 2018, and vertical cross-sections of potential temperature (K, contour) and (v, ω) vector (v, meridional wind; ω, vertical p velocity), averaged for the same period along (b) A–B and (c) C–D, shown as blue straight lines in (a). These values were derived from JRA-55. The red and light blue solid lines in each figure (b), (c), respectively, represent 299 K and 296 K.
Figure 4 presents the geographical pattern of potential temperature and horizontal winds, averaged over two days during 5–6 July at 975 hPa, and meridional cross-sections (along lines A–B and C–D) of the time-averaged potential temperature and (v, ω) wind vector (v, meridional wind). At 975 hPa, cold air flowed out from the Okhotsk High toward western Japan via the Sea of Japan as shown in Fig. 2. An area with a large horizontal gradient of potential temperature extended over western Japan from the Okhotsk High. The 299 K isentropic surface was located at the southern end of this area. In vertical cross-sections, shown in Figs. 4b and 4c, the southernmost isentropic surface that intersected the ground can be observed at 299 K. Thus, we regarded the 299 K isentropic surface as a frontal surface. Unsaturated warm air ascended northward over the 299 K surface as long as the warm air traveled adiabatically. A frontal surface often corresponds with a zone of sharp change in the wind direction. However, a near-surface boundary between the southerly and northerly wind areas was in the vicinity of the tip of the 296 K isentropic surface, approximately 200–300 kilometers north of the 299 K isentropic surface. Therefore, the 296 K isentropic surface represented the top of the cold air blowing from the Okhotsk High.
Next, the vertical velocity of the warm air was examined based on Eq. (4). The temporal change in pressure on the isentropic surface (the first term on the right-hand side in Eq. 4) was relatively small compared to adiabatic and diabatic velocities (as confirmed in Fig. 6). Figures 5a and 5b show adiabatic and diabatic vertical velocities averaged longitudinally between lines A–B and C–D as shown in Fig. 4a. The maximum adiabatic ascending was found around the 299 K isentropic surface (Fig. 5a). An adiabatic ascending area extended vertically above the 299 K isentropic surface. The warm moist air ascending on the 299 K surface was anticipated to release large amounts of condensation heating. South of western Japan, the lifting condensation level (LCL), estimated from the lower layer of 975 hPa, was approximately 500 m (not shown). After the lower air reached this level, strong diabatic ascending was added to the adiabatic ascending. A stronger diabatic ascending area also extended vertically but was located slightly north of the adiabatic ascending area (Fig. 5b). Additionally, the air being lifted to the level of free convection (LFC) initiated active convection.
The meridional cross-sections of (a) adiabatic and (b) diabatic vertical velocities (hPa h−1, colored), averaged over the same period as in Fig. 4 and averaged from 131.25–135.00°E longitude (between A–B and C–D in Fig. 4a). The adiabatic and diabatic vertical velocities were calculated using JRA-55. The contours in the respective figures show geopotential height (m) averaged similarly. The gray shaded area is the undefined area below the ground level.
The three daily averaged components of the vertical p velocity [(a) the temporal change in pressure, (b) adiabatic vertical velocity, (c) diabatic vertical velocity], (d) the total vertical velocity (hPa h−1) at the 299 K isentropic surface, and (e) the observed daily accumulated precipitation (mm) on 6 July 2018. The total vertical velocity is the sum of all the three components on the right-hand side in Eq. (4), i.e., it is equivalent to the vertical p velocity. The three components and the total vertical velocity were calculated using JRA-55. The gray shaded area is the undefined area below the ground level. Similar conditions continued during 5–7 July 2018.
Figure 6 shows the daily averaged three components on the right-hand side in Eq. (4), and the total vertical velocity on the 299 K isentropic surface, as well as the observed daily accumulated precipitation amounts on 6 July. Note that the total vertical velocity is the sum of all three components, i.e., it is equivalent to the vertical p velocity. There was weak ascending due to the temporal change in pressure on the isentropic surface from the northern East China Sea to the Kyushu region (Fig. 6a). However, its magnitude was relatively smaller than those of the other two components (Figs, 6b, c) within a broad area of western Japan. An adiabatic ascending area predominately extended over the southern coast of western Japan (Fig. 6b). A diabatic ascending area was found on the north side of the adiabatic ascending area (Fig. 6c). This diabatic ascending was weaker than that in the upper layer because the 299 K surface was located at the lower end of the diabatic ascending area as shown in Fig. 5b. A weak diabatic descending area existed over the southern coast of western Japan. In this area, the adiabatic ascending was greater than the diabatic descending. As a result, the total vertical velocity showed a broad ascending area over western Japan (Fig. 6d). The diabatic ascending area corresponded to the precipitation area, but the adiabatic ascending area was slightly off to the south. This result indicated that warm air, at first, adiabatically ascended as it traveled northward and then further ascended as it released condensation heat, thereby bringing about precipitation. The strong adiabatic ascending continued along the southern coast of western Japan until 7 July. However, the daily accumulated precipitation amount on that day was clearly less than those on the previous two days (not shown). Additionally, we confirmed the weakening of diabatic ascending on 7 July (not shown). This ascertained that the precipitation weakened as a result of the reduction in the water vapor transport to western Japan (presented in Fig. 3). In other words, a large amount of moisture transport caused the heavy rainfall on 5 and 6 July; however, its reduction weakened the precipitation on 7 July.
3.3 The dynamic and thermodynamic balance of the stationary frontThis section discusses the stationarity of the Baiu front during 5–6 July, which brought about longlasting and heavy precipitation over western Japan. Figure 7 shows the 299 K isentropic line in the vertical section along 132.5°E longitude every 6 hours from 0000 UTC 5 to 0000 UTC 7 July. The tip of the 299 K surface (< 925 hPa), where the adiabatic ascending was active (Figs. 4b, c), fluctuated within 100–200 km in the meridional direction during this period. Warm moist air continuously uplifted over the 299 K surface and triggered the initiation of active convection.
A meridional cross-section of the 299 K isentropic surface (contour) along 132.5°E longitude every 6 h from 0000 UTC 5 to 0000 UTC 7 July 2018, derived from JRA-55.
Here, vertically integrated cold airmass amounts below the 299 K isentropic surface were simply quantified by the pressure difference between the 299 K isentropic surface and the surface pressure (Eq. 1), and they are referred to as “cold airmass amounts” in this section. Dynamic and thermodynamic effects on the temporal change of cold airmass amounts are examined based on Eq. (3).
Figure 8 shows the daily averages of horizontal cold airmass fluxes and their convergence (upper panels), as well as the diabatic change rates of cold airmass amounts (lower panels) on 5 and 6 July. The reduction of cold airmass amounts due to their flux divergence was compensated for by their increase due to diabatic cooling near the southern coast of western Japan. Here, we introduce the vertical mean cold airmass transport velocity to identify the detailed dynamic processes (Yamaguchi et al. 2019), as per the following equation:
 |
 |
(a), (b) Daily averages of cold airmass flux MF below 299 K (hPa m s−1, vector) and its convergence (hPa h−1, colored), and (c), (d) the change rate of the cold airmass amounts 
(hPa h−1, colored) because of diabatic processes on 5 and 6 July 2018. These values were calculated using JRA-55. The gray shaded area is the undefined area below the ground level.
The previous section implied that some diabatic cooling was necessary to maintain cold airmass amounts near the frontal surface. In this section, we focus on the impacts of evaporative cooling from raindrops using the numerical experiments conducted by using the JMA-NHM with 3 km grid spacing as described in Section 2.3.
4.1 Geographical distributions of cold airmass amounts and precipitationThe 24 h accumulated precipitation distributions simulated in the CNTL experiment on 5 and 6 July were validated using the radar observations as shown in Fig. 9. The control experiment well captured the observed precipitation patterns including heavy rainfall in northern Kyushu (around 33.5°N, 131°E), Kochi Prefecture (around 33.5°N, 134°E), the Chugoku region (around 34.5°N, 133.5°E), and central Kinki (around 35°N, 135.5°E). Looking closely, the northward spread of the precipitation area was slightly smaller in the CNTL experiment than that observed on both days. Observed weak precipitation areas over the southern sea of the Shikoku region were not well reproduced on 5 July. Except for these differences, the model's performance might be sufficient for examining the impacts of evaporative cooling on the geographical distribution of precipitation amounts. In addition, we distinguished between front-induced and orographic precipitation types. These precipitation types refer to precipitation initiated and maintained by ascending of warm moist air along the frontal surface and mountain surface, respectively. Note that “Baiu frontal precipitation” simply refers to precipitation associated with the Baiu front. The orographic precipitation might not have been sensitive to the change in cold airmass amounts because the updraft along mountain slopes initiates active convection. On the other hand, the front-induced precipitation might have been sensitive to cold airmass amounts below the frontal surface. Therefore, we examined the impacts of evaporative cooling from raindrops just only on front-induced precipitation using the FLAT and NOEVAP_FLAT experiments to exclude the effects of orographic precipitation.
The distributions of 24 h accumulated precipitation (mm) obtained from the synthetic radar echo intensity on (a) 5 and (b) 6 July, and from the CNTL experiment on (c) 5 and (d) 6 July 2018.
In the analysis using JRA-55, the frontal surface was identified as the 299 K isentropic surface. In these mesoscale numerical experiments, the 300 K isentropic surface was more appropriate to the threshold potential temperature, based on the same definition as in Section 3.2 (Fig. S2).
Figure 10 presents cold airmass amount DP below the 300 K isentropic surface, averaged over the period from 1200 UTC on 5 July to 0000 UTC on 6 July, and the differences between those in the experiments with and without raindrop evaporation. In this period, the differences in the cold airmass amounts were significant. Regardless of whether the terrain had been flattened or not, the cold airmass amounts around the Chugoku region and central Kinki were significantly smaller in the numerical experiments without evaporation (NOEVAP and NOEVAP_FLAT). A difference exceeding 60 hPa was found in the vicinity of the Chugoku region. In addition, an area with a slightly lower difference in cold airmass amounts (< 30 hPa) extended from the Chugoku region to the southwest (see the dotted ellipses in Figs. 10c, f). This may have resulted from the disappearance of evaporative cooling under sporadic convective clouds traveling northeastward. The sporadic convective clouds can be confirmed as precipitation areas extending from the southwest to the northeast (see the dotted ellipses in Fig. 11). Note that the difference in the cold airmass amounts in the inland areas of the Kyushu and Shikoku regions, found in the experiments with flattened terrain, was not found in those with realistic terrain (Figs. 10c, f). This was because an increase in the cold airmass amounts due to evaporative cooling in the lower layer, which occurred in the FLAT experiment, hardly occurred in the CNTL experiment; this was due to the presence of mountains equal to or higher than the 300 K isentropic surface in those areas.
The distributions of 12 h averages of the cold airmass amount DP (hPa) below the 300 K isentropic surface [(a) CNTL, (b) NOEVAP, (d) FLAT, (e) NOEVAP_FLAT] and the difference (hPa) between experiments with evaporation and those without evaporation [(c) CNTL – NOEVAP, (f) FLAT – NOEVAP_FLAT] from 1200 UTC on 5 to 0000 UTC on 6 July 2018.
The distributions of 12 h accumulated precipitation (mm) until 0000 UTC on 6 July 2018: (a) CNTL, (b) NOEVAP, (c) FLAT, and (d) NOEVAP_FLAT.
Figure 11 shows 12 h accumulated precipitation amounts until 0000 UTC on 6 July in the numerical experiments. In the CNTL experiment, a heavy rainfall area extended in the zonal direction from the Chugoku region to central Kinki (Fig. 11a). This heavy rainfall area was also simulated in the FLAT experiment, although it was located slightly south of that in CNTL. This result indicated that the front-induced precipitation dominated the heavy rainfall in the Chugoku region and central Kinki. Meanwhile, the rainfall area was significantly shifted northward in the numerical simulations without the raindrop evaporation (Figs. 11b, d). In contrast, a rainfall area with the local maximum in the Shikoku region (around 33.5°N, 134.0°E) was simulated in almost the same area in both the CNTL and NOEVAP experiments. Although the rainfall maximum was found in the experiments with the flattened terrain, the precipitation amounts in the Shikoku region were less than half of those found in the experiments with the realistic terrain. Therefore, orographic effects were more critical than frontal effects for this rainfall maximum.
4.2 Evaporation-induced shift of the front-induced precipitationIn this subsection, we examine how the evaporative cooling from raindrops changed the distribution of heavy rainfall. The most distinct change was the northward shift of the front-induced precipitation in the Chugoku region when raindrop evaporation was turned off. This northward shift was similarly simulated, regardless of whether the terrain had been flattened or not. In the experiments with the flattened terrain, front-induced precipitation dominated the heavy rainfall in the Chugoku region and central Kinki as mentioned before. In fact, the rainfall peak was located near a steep, sloped region of the isentropic surface (as shown in Fig. 12). Therefore, we were able to examine the impacts of raindrop evaporation on the front-induced precipitation in detail by comparing the results of the FLAT experiment with those of the NOEVAP_FLAT experiment.
(a) A meridional cross-section of 12-h accumulated precipitation until 0000 UTC on 6 July 2018 averaged for 132–135°E longitude (black, FLAT; blue, NOEVAP_FLAT) and (b) a meridional vertical cross-section of 12 h average for isentropic surfaces averaged for 132–135°E longitude (black solid lines, FLAT; blue broken lines, NOEVAP_FLAT). The black solid lines and the blue broken lines in (b), respectively, show 298 K through 302 K at 1 K intervals.
Figure 12 presents the meridional cross-sections of 12 h accumulated precipitation amounts and isentropic surfaces averaged for 132–135°E longitude. In the NOEVAP_FLAT experiment, the exclusion of raindrop evaporation significantly reduced the cold airmass amounts below the 300 K surface between 32–37°N latitude and shifted the location with a steep slope of the 300 K surface northward. Correspondingly, the precipitation peak was displaced more than 100 km northward when raindrop evaporation was turned off. Figure 13 shows the meridional vertical crosssections of diabatic heating in cloud microphysics processes, (v, w) vector (w, vertical wind), and equivalent potential temperature, averaged for 132–135°E longitude from 1200 UTC on 5 to 0000 UTC on 6 July. We focused on two evaporative cooling areas that existed in the FLAT experiment (Fig. 13a). One was found near the ground surface, immediately below a strong heating area by condensation around 34°N. Another area vertically extended close to the north of the strong heating area around 35°N. However, these two cooling areas did not exist in the NOEVAP_FLAT experiment (Fig. 13b). The strong heating area associated with strong updrafts was located in higher latitudes in the NOEVAP_FLAT experiment (∼ 35°N) compared to the FLAT experiment (∼ 34°N). This also corresponded to the northward retreat of the location with a steep slope of the frontal surface in the NOEVAP_FLAT experiment (Fig. 12b). Therefore, the evaporative cooling from raindrops contributed to restraining the northward shift of the frontal surface. Specifically, the evaporative cooling below the strong heating area compensated for the reduction in the cold airmass amounts due to the northward advection of warm air. Furthermore, the evaporative cooling area close to the north of the strong heating area may have increased the cold airmass below the frontal surface. Clarifying how much each type of cooling contributed to the positioning of the frontal surface is an issue that should be addressed in future research.
Meridional vertical cross-sections of diabatic heating (K day−1, colored) because of cloud microphysics process, (v, w) vector (v, meridional wind; w, vertical wind) and the equivalent potential temperature (K, black contour), averaged in the same manner as in Fig. 12b: (a) the FLAT and (b) the NOEVAP_FLAT experiments. The blue solid line in each figure represents the 300 K isentropic line, averaged similarly to other values in this figure.
Next, we focused on the strong heating area. The convection indicated by the strong heating area was initiated by an ascent of warm air with an equivalent potential temperature of approximately 348 K in the lower layer (Fig. 13). However, it was considered that the equivalent potential temperature of the ascending warm air had not been preserved (but slightly lowered) because of mixing with the surrounding air. The bottom height of the heating area was approximately 400 m in the FLAT and 500 m in the NOEVAP_FLAT experiments, respectively (Fig. 13). This was consistent with the evaluation that the LCL of low-level warm air on the windward side was from 300 m to 500 m in the FLAT experiment, and from 400 m to 600 m in the NOEVAP_FLAT experiment, respectively (Figs. S3a, b). After reaching the LCL by ascending along the frontal surface, the warm air released condensation heat. After arriving at the LFC of 1,000 to 1,500 m (Figs. S3c, d), it formed the front-induced convective precipitation. Considering this initiation process of the convection, the raindrop evaporation contributed to maintaining the position of the front-induced precipitation by suppressing the northward retreat of the area, where low-level warm air reached the LCL and LFC. Conducting forward trajectory analysis confirmed that the ascending area of low-level warm air from the south was located at higher latitudes in the NOEVAP_FLAT compared to the FLAT experiment (Fig. S4).
Although not discussed above, a cooling area existed at approximately 500 m high, from 38–40°N, in both experiments. This cooling area was likely the result of cloud-water evaporation. This cooling is beyond the scope of the present study and is not fully discussed in this paper. The phenomenon corresponding to this cooling area is briefly described in the supplementary file (Fig. S5).
4.3 The environment for evaporative cooling from raindropsNext, the environmental relative humidity (RH) should be examined because it affects the evaporation efficiency from raindrops. Figure 14 presents the meridional cross-sections of the vertical velocity, the temporal change in the mixing ratio of water vapor, and RH in the FLAT experiment, averaged between 132–135°E longitude, from 1200 UTC on 5 to 0000 UTC on 6 July. A relatively dry area (RH < 90 %) was found on the north side of the large condensation (strong updraft) area and above 1 km, where an evaporation area extended vertically, and weak downdrafts were dominant. Jeong et al. (2016) also pointed out a similar coincidence between evaporative cooling and downdraft in another case. The downdrafts contributed not only to the downward advection of the relatively dry air from the mid-layer on the north of the strong heating area but also to the lowering of the RH by adiabatically raising the temperature. Therefore, while the raindrop evaporation increased the mixing ratio of the descending air, the downdrafts increased the saturated mixing ratio due to the temperature rise. Another raindrop evaporation area below the condensation area (altitude < 0.5 km) corresponded to no clear downdraft area (Fig. 14a). Relatively dry air (RH < 90 %) below 0.5 km flowed into this area from the south (Figs. 13a, 14c), which was regarded as having contributed to the raindrop evaporation in the lower layer.
Meridional vertical cross-sections of (a) vertical velocity (m s−1; colored), (b) temporal change of mixing ratio of water vapor (kg kg−1 s−1; colored), and (c) relative humidity (%; colored) in FLAT, averaged for 132–135°E longitude from 1200 UTC on 5 to 0000 UTC on 6 July 2018.
This study investigated the heavy precipitation mechanisms associated with the stationary Baiu front over western Japan in July 2018, with a particular focus on the impacts of evaporative cooling from raindrops on the Baiu frontal precipitation. In the analysis using JRA-55, the cold airflow from the northeast had a southernmost surface potential temperature of approximately 296 K. The warm airflow from the south and/or southwest had the bottom surface with a potential temperature of 299 K, which was identified as the frontal surface. Near the tip of the 299 K surface, the diabatic cooling compensated for the decrease in the cold airmass amounts below the frontal surface due to the northward advection of warm air. The frontal surface was almost stationary for two days from 5 July. Warm moist air from the south reached the LCL at an altitude of approximately 500 m by adiabatically ascending above the frontal surface, and then further ascended with the release of diabatic heating. When lifted to the LFC, the air initiated active convection. As a result, heavy rainfall was broadly recorded over western Japan.
Sensitivity experiments using JMA-NHM were conducted to examine the impacts of raindrop evaporation and the topography. Regardless of whether the terrain was flattened or not, the exclusion of raindrop evaporation considerably reduced the cold airmass amounts over western Japan, resulting in a northward shift of the heavy precipitation area in the Chugoku region. In experiments without raindrop evaporation, the northward shift in the precipitation area followed that of the frontal surface. The heavy precipitation area, extending from the Chugoku region to central Kinki, was more sensitive to raindrop evaporation than to terrain effects. Raindrop evaporation maintained the position of the heavy front-induced precipitation by suppressing the northward retreat of the frontal surface. On the other hand, the experiments with the flattened terrain suggested that the precipitation in the Shikoku region had been enhanced by the terrain.
Based on the above considerations, we illustrated the front-induced precipitation as shown in Fig. 15. Low-level warm moist air first ascended adiabatically along the frontal surface. After being saturated at the LCL, it released condensation heat. Then, reaching the LFC, it initiated moist convection. The evaporative cooling from raindrops maintained the cold airmass amounts below the frontal surface and prevented the northward retreat of the frontal surface. As a result, the front-induced precipitation area stagnated in the same region and did not retreat northward. Two evaporative cooling areas were found below a condensation heating area and in an area in a northerly direction, close to the heating area. However, it was not clear how much the latter area contributed to the stagnation of the frontal surface. This problem requires the detailed study of complicated precipitation systems. Clarification of this point should be considered as a future task.
A schematic illustration of the mechanism of the front-induced precipitation and the distribution of the diabatic heating and cooling. The solid light-blue lines schematically illustrate the isolines of the potential temperature.
Although the impacts of evaporative cooling from raindrops in the Baiu front environment have already been reported in some cases, the impacts have depended on the cases. What controls the cases and the magnitude of the impacts of raindrop evaporation is a significant problem that remains unsolved. In the current case, two relatively dry areas contributed to the occurrence of raindrop evaporation. One flowed from the south into the lower layer below the strong condensation heating area, while another area existed in a close northerly direction to the strong heating area and accompanied weak downdrafts, in which the raindrop evaporation was enhanced by an increase in the saturated mixing ratio. The Baiu front remained almost stationary for a long time. Accordingly, the cold airmass amounts may have been effectively maintained by the evaporative cooling from raindrops and accumulated in the lower layer. In this case, the evaporative cooling from raindrops significantly controlled the precipitation distribution through the maintenance of the cold airmass amounts below the frontal surface because the front-induced precipitation was more predominant than the orographic precipitation. The evaporation rate from falling raindrops must be under the influence of thermodynamic parameters, such as temperature and humidity. A systematic survey of this issue is beyond the scope of this work. It remains a subject to be examined in future studies.
To confirm which of the two terms on the righthand side in Eq. (6) mainly contributed to the decrease in the cold airmass amounts, the contributions of these two terms were calculated. Figure S1 illustrates the daily averages of contributions to the temporal change in the cold airmass amounts below the 299 K surface due to the (a), (b) first and (c), (d) second terms on the right-hand side in Eq. (6) on 5 and 6 July 2018, respectively. Focusing on the area near the tip of the 299 K surface (near the southern coast of western Japan), the contribution of the first term was almost zero or slightly negative. On the other hand, that of the second term was clearly negative. This means that the northward advection by southerly wind mainly contributed to the decrease in the cold airmass amounts.
S.2 Description of Fig. S2Based on the same definition as that in Section 3.2, we identified a frontal surface in the CNTL experiment to apply the same cold airmass analysis as the analysis using JRA-55. Figure S2 shows the meridional vertical cross-sections of the potential temperature averaged from 132°E to 135°E longitude every 12 h from 0000 UTC on 4 to 1200 UTC on 5 July 2018 in the CNTL experiment. The 300 K isentropic surface was identified as the frontal surface. At 0000 UTC on 4 July, as shown in Fig. 2 (Section 3.1), because Typhoon T1807 was passing over the Sea of Japan north of the Chugoku region, no cold air flowed out to western Japan. Therefore, the horizontal gradient of the potential temperature had not been strengthened in western Japan. After the passage of Typhoon T1807, a gradient zone of the potential temperature was formed at around 34°N latitude by the outflow of cold air from the Okhotsk High. The southernmost isentropic surface, which was located at the tip of the gradient area of the potential temperature and intersected the ground, was the 300 K isentropic surface. Note that the isentropic surface of 300 K extended southward of western Japan before the arrival of the cold air from the Okhotsk High. The cold air outflow increased the vertical inclination of the 300 K surface.
S.3 Description of Fig. S3Figure S3 illustrates the LCL and LFC, when the air parcel at an altitude of 100 m was lifted above, averaged from 1200 UTC on 5 to 0000 UTC on 6 July in the FLAT and NOEVAP_FLAT experiments. The LCL of the warm air on the windward side of the frontal surface was about 400 m in the FLAT experiment and about 500 m in the NOEVAP_FLAT experiment, respectively. This was consistent with the bottom height of the strong heating area extended vertically due to condensation (Figs. 13a, b). The LFC of the warm air on the windward side of the frontal surface was more than from 1,000 m to 1,500 m in both the FLAT and NOEVAP_FLAT experiments.
S.4 Description of Fig. S4To show the shift of the ascending area corresponding to the northward shift of the inclined frontal surface, we conducted a forward trajectory analysis of the low-level warm moist air. In the FLAT and NOEVAP_FLAT experiments, trajectories were calculated for 12 h; 189 parcels were initially located at (x, y, z) (x represents one of 21 values from 130.5°E to 132.5°E longitude with an interval of 0.1°E; y represents one of three values from 31.5°N to 32.5°N latitude with an interval of 0.5°N; z represents one of three values from 0.1 km to 0.5 km with an interval of 0.2 km) at 1200 UTC on 5 July. Data used in this calculation had a time interval of 3 min. We calculated the trajectories using the Runge–Kutta fourth-order method with the integration time step of 1 min. Figure S4 presents three-dimensional forward trajectories and their projections to the surface for two experiments: FLAT and NOEVAP_FLAT. Air parcels were transferred northeastward at a low level, lifted up rapidly near the heavy precipitation area, and transferred east-northeastward in the middle troposphere. The active ascending area was located at higher latitudes in NOEVAP_FLAT than in FLAT. The two-dimensional surface projection illustrates clearly that the parcels reached higher latitudes in NOEVAP_FLAT than in FLAT.
S.5 Description of Fig. S5Figure S5 shows diabatic heating because of radiation processes in the FLAT experiment and the difference between air temperature at 50 m and SST. From a comparison of Fig. S5a to Fig. 13a, one can confirm that a radiative cooling area exists near the top of the condensation heating paired with a cooling area over the Sea of Japan (38–40°N). This radiative cooling was regarded as cooling by long-wave radiation from near the top of the lower cloud extending over the Sea of Japan. In addition, SST was about 1–2 K higher than the air temperature over the Sea of Japan (Fig. S5b). Therefore, the lower clouds were regarded as having been formed and maintained in the lower atmosphere destabilized by the radiative cooling near the top of the clouds and heating from the sea surface.
This work was supported by the activity of the Core Research Cluster of Disaster Science at Tohoku University. The numerical sensitivity experiments were conducted with the non-hydrostatic numerical model (JMA-NHM) developed at the Japan Meteorological Agency using supercomputing resources at Cyberscience Center, Tohoku University. Synthetic radar echo intensity data are distributed by Research Institute for Sustainable Humanosphere, Kyoto University (http://database.rish.kyoto-u.ac.jp/index-e.html) and the Japanese 55-year Reanalysis (JRA-55) is provided by the Japan Meteorological Agency and achieved at the Data Integration & Analysis System (DIAS), which is developed and operated by the Ministry of Education, Culture, Sports, Science, and Technology. We are grateful to Dr. Junshi Ito and two anonymous reviewers for their constructive comments and suggestions.
