1. Introduction
Polar mesocyclones (PMCs) are cyclonic vortices that develop poleward of the main polar front and have a horizontal size of meso-α (200–2000 km) to meso-β (20–200 km) scale according to the definition by Orlanski (1975). PMCs include polar lows (PLs), which are intense maritime cyclones with horizontal scales between 200 and 1000 km (Rasmussen and Turner 2003). The Sea of Japan (Fig. 1) is one of the regions where PMCs frequently occur during winter, where they are caused by cold-air outbreaks from the Eurasian continent (Asai 1988, hereafter A88; Ninomiya 1989). A notable characteristic of PMCs over the Sea of Japan is that they are often small (<200 km) but intense. Such PMCs are hazardous because of the associated heavy snowfall (e.g., Miyazawa 1967; A88) and strong winds (e.g., Kuroda 1992; Yamagishi et al. 1992), and can pose a threat to coastal regions and maritime traffic.
Geography and topography around the Sea of Japan. Tracking of PMCs was performed in the blue area.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
Since the 1960s when imagery from meteorological radars and satellites became available, Japanese meteorologists have paid considerable attention to PMCs and have performed a number of observational studies (Miyazawa 1967; Okabayashi and Satomi 1971; Shimizu and Uchida 1974; Asai and Miura 1981). Using satellite images, A88 analyzed the locations of PMC genesis over the Sea of Japan during the winter of 1983/84 and found that PMCs with horizontal scales of less than 300 km are concentrated in two regions: one between the northeastern base of the Korean Peninsula and the central part of Honshu Island, and the other to the west of Hokkaido Island (Fig. 2). A cloud band known as the Japan Sea polar airmass convergence zone (JPCZ; A88) is typically present over these two regions, and several studies indicate a link between PMCs and the JPCZ (Okabayashi and Satomi 1971; A88; Nagata 1993; Tsuboki and Asai 2004; Kato 2005; Watanabe and Niino 2014).
Geographical distribution of the locations of individual PMCs observed in satellite images over the Sea of Japan and its vicinity during the winter of 1983/84 (adapted from A88). The large, medium, and small dots correspond to the PMCs with horizontal scale 150–300, 70–150, and 20–70 km, respectively.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
Several mechanisms have been proposed for the development of PMCs. Asai and Miura (1981) and Nagata (1993) suggested that small PMCs are related to the horizontal shear of the JPCZ, whereas Tsuboki and Wakahama (1992) suggested that PLs are associated with baroclinicity. Diabatic heating caused by condensation also plays an important role in both small PMCs (Kato 2005) and PLs (Yanase et al. 2004). Furthermore, these individual processes can occur at the same time and in a complex manner (Lee et al. 1998; Fu et al. 2004; Shimada et al. 2014; Watanabe and Niino 2014). These differing developmental mechanisms are probably the cause of the various sizes and shapes of PMCs. Yanase and Niino (2007) demonstrate that the cloud patterns and development mechanisms of PLs change significantly when environmental baroclinicity changes. This suggests a strong connection between the features of PMCs and environmental conditions. Climatological studies may be useful if we wish to achieve a more comprehensive understanding of the diverse nature of the PMCs that develop over the Sea of Japan.
Several climatological studies of PMCs over the Sea of Japan have been performed using satellite images and weather charts (A88; Ninomiya 1989; Ookubo 1995). Ninomiya (1989) analyzed several PLs using nephanalysis (analysis chart focusing on the types and amounts of clouds) and weather charts, and described their characteristic environments. Using serial satellite images, Ookubo (1995) demonstrated that there are two types of PMCs: one moves east-northeastward and the other moves southeastward. However, as these studies relied on the subjective detection of PMCs, they were unable to deal with a large amount of data. For example, A88 and Ninomiya (1989) analyzed a period of only one winter, and Ookubo (1995) analyzed only nine PMCs.
Recently, a number of objective tracking algorithms have been developed and applied to objective analysis datasets and dynamic downscale datasets (Zahn and von Storch 2008; Irving et al. 2010; Uotila et al. 2011; Bromwich et al. 2011; Xia et al. 2012; Chen and von Storch 2013; Zappa et al. 2014; Yanase et al. 2016, hereafter Y16). Tracking algorithms are useful in obtaining reliable climatological data for PMCs because they can objectively and automatically detect a large number of PMCs. As for the climatology of PLs over the Sea of Japan, Chen and von Storch (2013) analyzed the entire North Pacific using a regional climate model with a horizontal resolution of 0.4°, and Y16 examined the Sea of Japan in detail using the Japanese 55-year Reanalysis (JRA-55; Kobayashi et al. 2015) with a horizontal resolution of about 60 km. These studies, however, examined only meso-α-scale PLs because of the limited horizontal resolution of the data.
Climatological studies of PMCs using satellite images suggest that many PMCs smaller than 200 km develop over the Sea of Japan (A88) and over the northeastern Atlantic (Harold et al. 1999; Condron et al. 2006; Blechschmidt 2008). Small PMCs have also been observed over the Great Lakes (Hjelmfelt 1990; Laird et al. 2001). Moreover, the ability to objectively detect PMCs is strongly influenced by the size of the PMCs, and a large number of PMCs smaller than 200 km are missed (Condron et al. 2006). Figure 3a presents a visible image of PMCs taken by the Japanese Multifunctional Transport Satellite 2 (MTSAT-2), which shows three meso-β-scale vortices existing simultaneously in the JPCZ. Although JRA-55 marginally resolves the JPCZ, it spreads over a wide band-shaped region and the vortices are not reproduced clearly (Fig. 3b). On the other hand, Fig. 3c shows a corresponding plot based on Mesoscale Analysis (MA) data provided by the Japan Meteorological Agency (2013; see below for details), which indicates that MA has the potential to resolve meso-β-scale PMCs (Fig. 3c).
(a) Visible image from MTSAT-2 and (b),(c) vorticity distributions at 950 hPa at 0600 UTC 29 Jan 2011 (color; 10−4 s−1) obtained from JRA-55 and MA, respectively. Arrows in (a) indicate the locations of PMCs. Note that the color scales in (b) and (c) are different.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
In this study, we present a new algorithm for the detection and tracking of PMCs, including meso-β-scale vortices, using MA. Using this algorithm, we also show some climatology of PMCs over the Sea of Japan. The remainder of this paper is organized as follows. Section 2 describes the datasets, the tracking algorithm, and the validation of the objective tracking. Section 3 shows the tracks of the PMCs and briefly discusses the climatology of the PMCs over the Sea of Japan. Section 4 compares the PMCs detected by the algorithm with those in previous studies and briefly considers the environment of the PMCs. Finally, section 5 provides a summary and conclusions.
2. Data and methodology
a. Data
We used MA data in the tracking analysis of the PMCs. The MA is a product of objective analysis using the Japan Meteorological Agency’s (JMA) mesoscale model, which is based on the JMA nonhydrostatic model (Saito et al. 2006), as a forecasting model. For the data assimilation, MA uses the four-dimensional variational scheme, the JMA-Nonhydrostatic-model-based Variational Analysis Data Assimilation (JNoVA; Honda et al. 2005). The domain of MA is a rectangular area of 3600 km × 2880 km in Lambert conformal conic map projection covering Japan and its surroundings. The original MA dataset has a horizontal resolution of 5 km with 50 vertical levels, and a temporal resolution of 3 h. In the present analysis, we used a dataset prepared by converting the original MA data into geographic coordinates with a horizontal grid spacing of 0.1° in latitude, 0.125° in longitude, and 16 pressure levels by a linear interpolation. This resolution is sufficient to examine the characteristics of meso-β-scale PMCs. As most PMCs occur during the cold season, we analyzed six cold seasons from November 2009 to March 2015. (Note that JNoVA was introduced in April 2009.) Hourly infrared (IR) and visible satellite images from MTSAT-2 were used to validate the detected PMCs. The NOAA 0.25° daily optimum interpolation sea surface temperature (OISST; Reynolds et al. 2007) data were used to examine the environment of the PMCs.
b. Algorithm for tracking PMCs
Relative vorticity at the 950-hPa level is used to track the PMCs. The vorticity field is smoothed using a running mean over 0.4° in latitude and 0.5° in longitude to focus on PMCs with horizontal scales larger than, or equal to, meso-β scale. The algorithm for the tracking of PMCs consists of three parts: the identification of vortices at each time step, the connection of vortices in consecutive time steps, and the exclusion of synoptic-scale disturbances.
1) Ientification of vortices at each time step
The algorithm used to detect vortices is similar to that developed by Shimizu and Uyeda (2012) to detect convective cells in radar reflectivity scans. Their algorithm is based on an adaptive threshold scheme that is able to identify a convective cell that includes a single local maximum of reflectivity from the cell groups and is suitable for identifying PMCs embedded in a larger system such as the JPCZ.
Figure 4 shows a schematic diagram of the algorithm used to identify the vortices. First, the maximum vorticity is detected. Then, a vortex area is defined as a continuous area of vorticity ≥ζmin0 (1.5 × 10−4 s−1) surrounding the maximum (see the red area in Fig. 4a). Similarly, the maximum of vorticity in the rest of the area is again detected and a corresponding vortex area is defined (see the green area in Fig. 4a). This process is repeated until all peaks having vorticity ≥ζmax0 (2.0 × 10−4 s−1) have been detected (Fig. 4a). The vortex area defined here is referred to as an “original vortex area.” There are two kinds of original vortex area: one with a single peak (green area in Fig. 4a) and one with multiple peaks (red area in Fig. 4a). The former is identified as a vortex. A vortex area with a single peak is referred to as an “isolated vortex area” and that with multiple peaks as a “multiple vortex area.”
Schematic diagram of the algorithm used to identify vortices. (right) The horizontal distribution of the vorticity field and (left) the vorticity along line A–B in the right panels. (a) The original vortex areas, which are defined as the green and red areas. (b) An example of a peak (peak 1) identified as a vortex. The gray area in (b) is where vorticity is lower than the temporary threshold ζmin and will be distributed to either the yellow or red vortex area in the last step. (c) An example of a peak (peak 2) identified as a part of the neighboring vortex. (d) The final result of the identification process. The isolated vortex areas are defined as the green, red, and yellow areas in (d). The center of each vortex is shown by cross marks in (d). See text for details.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
To identify vortices in a multiple vortex area, a threshold ζmin used to define a vortex area is increased until a single peak exists within a vortex area (Figs. 4b and 4c). If the difference Δζ between the vorticity of the peak ζmax and the threshold ζmin is more than Δζmin, the peak and corresponding vortex area are identified as a vortex (peak 1 and the yellow area in Fig. 4b). Here Δζmin is defined as γ × ζmax, where γ is a scale factor and is set to 0.25. If Δζ is less than Δζmin, the peak is regarded as part of a neighboring vortex (peak 2 in Fig. 4c). This procedure is continued until all peaks having vorticity ≥ζmax0 in multiple vortex areas are identified. There are two kinds of areas in a multiple vortex area: the first belongs to a vortex area defined by the threshold ζmin that separates two peaks (e.g., red and yellow areas in Fig. 4b), and the other is the rest of the multiple vortex area (gray area in Fig. 4b). For a grid point within the latter area, the distances between the grid point and the edge of the vortex areas are calculated. The grid point is assigned to the vortex area whose edge is closest to it. In this way, each vortex and its vortex area are determined as in Fig. 4d. The center and intensity of each vortex are defined by the location and magnitude of the vorticity peak. The size of each vortex is defined by the area of the isolated vortex area. Note that Wernli and Schwierz (2006) applied similar method to sea level pressure (SLP) field to identify extratropical cyclones. Our method using the vorticity field is more suitable to identify PMCs because they are smaller than extratropical cyclones and sometimes are not accompanied by SLP minimum with closed SLP contours.
2) Connection of vortices at consecutive time steps
To connect vortices at consecutive time steps, a steering wind for each vortex is calculated to estimate the movement of the vortex. This steering wind is defined as a wind averaged over a box that encloses an area of 4° in latitude by 5° in longitude between 1000 and 700 hPa, where the center of the box is chosen to coincide with the vortex center. In some tracking schemes for synoptic-scale systems, the steering wind is taken to be some fraction of average wind partly due to the beta effect (e.g., Simmonds et al. 1999). However, since the sizes of vortices considered here are small, the beta effect might be negligible. Therefore, the averaged wind is directly assumed to be a steering wind.
The position of a vortex at the next time step is estimated by assuming that the vortex is advected by the steering wind. For this estimated position, two requirements are imposed. First, the vortex at the next time step must be located within 1° in longitude and 0.8° in latitude of the estimated position (i.e., the estimated area; Fig. 5a). If there are more than two vortices in the estimated area, the vortex closest to the estimated position is linked to the previous vortex. If there is no vortex that satisfies the first requirement, the second requirement is considered, which is that a part of an isolated vortex area at the next time step overlaps with the estimated area (Fig. 5b). If there are more than two isolated vortex areas for which part of the area overlaps with the estimated area, the vortex at the previous time step is linked to the vortex with an isolated vortex area having the largest amount of overlap with the estimated area.
Schematic diagram of the requirements for connecting vortices at consecutive time steps. The closed circle is the center of a vortex at the previous time step. The double circle is the estimated position of the vortex at the next time step. Open circles are centers of vortices at the next time step. The gray area in (b) is an isolated vortex area of a vortex at the next time step. See text for details.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
If a vortex does not have a corresponding vortex at the previous (next) time step, that time step is defined as the genesis (dissipation) time. If more than two previous vortices are connected to a single vortex at the next time step, they are recognized as being merged. The split of vortices is not explicitly considered in this scheme. When a vortex splits into two vortices, one of them is linked to the vortices at the previous time step and the other is regarded as a vortex that is generated at the time step. In this study, PMCs are defined as vortices with a lifetime exceeding 6 h (three time steps).
3) Exclusion of synoptic-scale disturbances
As we do not use a spatial high-pass filter, the present tracking method also picks up disturbances such as synoptic-scale lows and cold fronts. Therefore, we introduce schemes that discriminate synoptic-scale disturbances from PMCs based on their size and shape.
The size of the vortex area and SLP are used to identify synoptic-scale lows (Fig. 6). First, the sizes of the original vortex areas are examined and SLP minima that are at least 0.5 hPa less than the surroundings are detected. Then, a vortex that is located in an original vortex area larger than 40 000 km2 and that is within 300 km of an SLP minimum is considered a synoptic-scale low (black cross in Fig. 6). Although some large PMCs can be erroneously classified as synoptic-scale lows using these thresholds, almost all synoptic-scale lows are properly classified here. If a vortex is in an original vortex area larger than 40 000 km2 but is located farther than 300 km from an SLP minimum, the vortex is not considered a synoptic-scale low (white cross in Fig. 6). The present scheme is able to distinguish between a synoptic-scale low and a vortex embedded within it. Such multiscale structures are sometimes observed over the Sea of Japan (e.g., Ninomiya et al. 1990; Ninomiya and Hoshino 1990; Ninomiya 1994; Tsuboki and Asai 2004).
Schematic diagram for identifying synoptic-scale lows. The gray area is an original vortex area. Crosses denote centers of vortices and a black dot denotes a minimum of SLP. In this case, the black cross is considered to be a synoptic-scale low but the white cross is not. Note that there is no minimum of SLP within 300 km of the white cross. See text for details.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
Next, the shape of the original vortex area is used to identify cold fronts (Fig. 7). First, the southern and northern edges of each original vortex area are detected. The distance between the two edges (L in Fig. 7) and the clockwise angle between the line connecting the two edges and north (θ in Fig. 7) are calculated. If the distance is greater than 400 km and the angle is between −20° and +60°, the original vortex area is further examined. First, the ridge of the vorticity field in the original vortex area is detected in the following way. A grid point on the ridge is determined by choosing the point where the vorticity is one of the three largest of the nine neighboring grid points. Then, the grid points chosen in this way are fitted by a smooth quadratic curve. If the average curvature of the quadratic curve is less than 0.1 (the curvature is positive when the quadratic curve is convex westward) and the coefficient of determination (R2) is greater than 0.8, the vortices in the original vortex area are considered part of a cold front.
Schematic diagram of the procedure used to identify cold fronts. The gray area is an original vortex area. Triangles indicate the northern and southern edges of the original vortex area, L is the distance between the two edges, and θ is the clockwise angle between the line connecting the two edges and north. Black dots are grid points determined to be ridges of vorticity field. The dashed curve is a fitting curve. See text for details.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
Finally, all time steps from the lifetime of the PMC are examined according to the two schemes. If the number of time steps when the vortex is considered a synoptic-scale low or a cold front is more than one-seventh of the whole lifetime of the PMC, then the PMC is excluded as a synoptic-scale disturbance.
c. Validation of the tracking algorithm
When the present tracking method was applied to the MA data, 642 PMCs were detected over the six cold seasons. Merged PMCs were counted as one PMC within this number and 120 mergers were found. However, most of the mergers occurred at the early stages of a vortex and 23 mergers among them were merger of vortices having lifetime longer than 3 time steps (9 h).
To evaluate the validity of the tracking algorithm, we subjectively examined all of the detected PMCs by using the vorticity and the SLP field from MA and satellite IR images. We first checked whether a detected PMC was subjectively judged to be a PMC using the MA data, and then checked whether the PMC was evident in the IR images. PMCs were then classified into one of three groups: those that were judged to be PMCs using the MA data and were also found in the satellite images at least once during their life time (class A), those that were judged to be PMCs from the MA data but were not found in the satellite images (class B), and those that were not judged to be PMCs (class C). Figure 8a shows the total monthly number of PMCs in each class: 381 PMCs (59%) in class A and 81 PMCs (13%) in class B, with 72% of the PMCs at least judged to be a PMC by subjective detection using the vorticity and SLP field. Some of the class-B PMCs were covered by upper-level cloud in the satellite image, but others were so weak that they were not accompanied by obvious cloud features. Class C contained 180 PMCs (28%) that were either part of a cold front or a weak peak in a synoptic-scale low. The proportion of class-C PMCs was large in November and March (Fig. 8a) when synoptic-scale lows are common over the Sea of Japan. Vorticity peaks associated with these synoptic-scale lows were sometimes erroneously classified to PMCs. To reduce false detections, we introduced criteria for the intensity and lifetime of the PMCs. We selected PMCs that satisfied either of the following criteria (hereafter referred to as strong criteria): 1) the lifetime exceeded 9 h and the vorticity exceeded 3.0 × 10−4 s−1 for at least two time steps, or 2) the lifetime exceeded 6 h and the vorticity exceeded 4.5 × 10−4 s−1 once during the lifetime. A total of 290 PMCs satisfied the strong criteria (Fig. 8b). The numbers of PMCs classified into classes A, B, and C were 226 (78%), 25 (8.6%), and 39 (13%), respectively, and 87% of the PMCs agreed with the subjective detection using the vorticity and SLP field. Thus, the strong criteria reduce the number of false detections, especially in November and March. In the following section we focus on those PMCs that satisfied these criteria to obtain more robust results.
Total monthly number of detected PMCs over the six cold seasons between November 2009 and March 2015. Black, dark-gray, and light-gray bars indicate the numbers of class-A, -B, and -C PMCs, respectively. Statistics are shown for (a) all PMCs and (b) for PMCs that satisfy the thresholds for intensity and lifetime.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
We also subjectively checked the vortices identified as synoptic-scale disturbances by our algorithm using MA and satellite images. A total of 212 of such vortices were found over the six cold seasons. By the subjective analyses, 22 vortices among them were considered to be the vortices that our algorithm should have classified into PMCs.
There were some class-A PMCs that did not satisfy the strong criteria, implying that PMCs found in satellite images sometimes had only a weak surface vorticity signature or none. Thus, our method is likely to miss some PMCs. To evaluate the detection rate, we compared our results with PMCs detected in satellite images. First, we compiled a database of PMCs using MTSAT-2 visible images during daytime (at 0000, 0600, and 0900 UTC) for a cold season from November 2011 to March 2012. When compiling this database, we also referred to a database of PMCs archived by Gurvich (2013) based on Moderate Resolution Imaging Spectroradiometer images. Since we did not track PMCs here, a persistent PMC may be counted several times in successive images. A total of 185 PMCs were detected in this period. Then the PMCs in the database were compared with those detected from the MA data using our detection methodology. If a vortex detected from the MA data was located within 150 km from a PMC in the database, the PMC was considered to be detected by our method. In this operation the persistency of the vortices in the MA data was not considered. Among 185 PMCs, 113 PMCs were detected by our method. Although the database contains some very weak PMCs, our method turns out to detect about 61% of PMCs observed in the visible satellite images.
In the present tracking algorithm, we use several thresholds, which are empirically determined based on subjective analyses of MA and satellite images. The results of the tracking depend on these thresholds. In general, more (less) PMCs are detected with moderate (strict) thresholds, while more (less) false detections are found. The threshold for the intensity of a vortex ζmax0, which is used for identifying vortices at each time step, has large influence on the result. We checked the sensitivity of the results to ζmax0 by changing ζmax0 to 3.0 × 10−4 s−1 and 1.5 × 10−4 s−1 from the original value of 2.0 × 10−4 s−1. With ζmax0 of 3.0 × 10−4 s−1, 171 PMCs that satisfy the strong criteria were detected. The numbers of class-A, -B, and -C PMCs became 141 (82%), 13 (7.6%), and 17 (10%), respectively. Although the larger ζmax0 reduces the proportion of false detection, it also reduces the number of class-A PMCs. With ζmax0 of 1.5 × 10−4 s−1, more PMCs were detected. However, most of them are class-C PMCs. The number of PMCs that satisfy the strong criteria increased to 326 and the numbers of class-A, -B, and -C PMCs became 241 (74%), 35 (9.2%), and 55 (17%), respectively. The number of class-A PMCs was slightly larger than that for ζmax0 of 2.0 × 10−4 s−1, while the number of class-C PMCs was also increased. Since the thresholds depend on the characteristics of the dataset including horizontal resolutions and physical processes incorporated in the parent model, we need to set appropriate thresholds first for each dataset based on subjective analyses. Once appropriate thresholds are set, however, the present method is effective for obtaining the climatology of the PMCs.
3. Climatology of the PMCs
In this section we show the climatology of the PMCs identified using our new tracking method. The greatest numbers of PMCs form during December and January (Fig. 8b). However, as the interannual variation is very large (not shown), analysis is required over a longer time period to provide robust results.
The tracking analysis gives the geographical distribution of the PMCs. Figure 9 shows the tracks and locations at maximum intensity of the PMCs (Fig. 9a) and the average number found within a 1° × 1° grid in the cold season (Fig. 9b). The tracks of the PMCs are concentrated mainly in two regions: the western part of the Sea of Japan from the base of the Korean Peninsula to the center of Honshu Island (the WJ region), and the western part of the northern Sea of Japan to the west of Hokkaido Island (the NJ region). Several PMCs were also found in the central part of the Sea of Japan around 40°N, 138°E (the CJ region). The PMCs were classified into one of these three regional groupings according to the location where they reached their maximum intensity (Fig. 9b). The numbers of PMCs in the WJ, NJ, and CJ regions were 146, 98, and 46, respectively. We examined the typical characteristics of the PMCs in each region, such as their direction of movement (Fig. 10), maximum intensity (Fig. 11), and size (Fig. 12).
(a) Tracks of PMCs over the six cold seasons. Blue crosses are the locations of the genesis of the PMCs. Dots are the locations of the maximum intensity of PMCs and their colors indicate the maximum vorticity (10−4 s−1). (b) Average number of tracks (color) and average number of PMCs reaching the maximum intensity (contours) within a 1° × 1° grid in a cold season. The geographical locations of the WJ, CJ, and NJ regions are indicated by dashed lines.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
Tracks of PMCs for 6 h prior to the time of maximum intensity. Triangles show the location of maximum intensity. Tracks of PMCs that have their maximum intensity at their initial time step are excluded from this plot and those of PMCs whose lifetime is shorter than 6 h prior to the time of maximum intensity are plotted only for their lifetime. Red and blue lines indicate that the meridional movement of PMCs is northward and southward, respectively.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
Boxplot of the maximum vorticity of PMCs in each region. The box shows the interquartile range (IQR) extending from the lower quartile to the upper quartile. The median value is shown by the horizontal line in the box. The vertical lines (“whiskers”) extending upward and downward from the box reach the maximum and minimum within 1.5 × IQR from the upper and lower quartile, respectively. Outlying points are plotted individually as colored dots.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
As in Fig. 11, but for the size of PMCs at their maximum intensity.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
In the WJ region, most of the PMCs form leeward of the mountains to the north of the Korean Peninsula (see also Fig. 1) and move southeastward (Fig. 10). There are also PMCs that move eastward or east-northeastward along the coast of Honshu Island (Fig. 10). Many PMCs attain their maximum intensity at the beginning of their lifespan, whereas other PMCs intensify during their lifetime and reach peak intensity near Honshu Island (Fig. 9). The maximum vorticity of PMCs in this region is small in comparison with other regions (Fig. 11), although a few intense PMCs (e.g., maximum vorticity >8.0 × 10−4 s−1) were also observed. The average size of PMCs in this region was intermediate among the three regions (Fig. 12), although occasional large PMCs (>30 000 km2) were also observed.
PMCs in the NJ region reached their maximum intensity near Hokkaido Island (Fig. 9) and followed two major directions of movement: southward and eastward (Fig. 10). Most PMCs in this group made landfall on Hokkaido Island, although some of the southward-moving PMCs moved into the CJ region along the western coast of northern Honshu Island (Fig. 9a). The maximum vorticity of PMCs in this region was the largest and sometimes exceeded 8.0 × 10−4 s−1 (Fig. 11). The average size of PMCs was smaller than in the other regions, and most were smaller than 20 000 km2 (Fig. 12).
PMCs in the CJ region tended to move eastward or northeastward and reached their maximum intensity off the northwest coast of Honshu Island (Fig. 10). Their average maximum vorticity was intermediate among the three regions, and PMCs with maximum vorticity exceeding 8.0 × 10−4 s−1 were not observed in this region (Fig. 11). These PMCs were relatively large, with more than 25% being larger than 20 000 km2 (Fig. 12).
4. Discussion
a. Comparison with previous works
We now compare the results obtained using our tracking method with previous research into PMCs over the Sea of Japan, which was based mainly on the subjective analysis of satellite images. The geographical distribution of the detected PMCs (Fig. 9) agrees well with the results of A88 (Fig. 2). The two major directions of movement of PMCs around the center of Honshu Island, which are southeastward and east-northeastward (Fig. 10), also agree well with the results of Ookubo (1995). Thus, the tracking algorithm accurately reproduced the characteristics of the PMCs that were derived from the subjective analysis of satellite images.
Previous objective tracking methods for PLs used the difference between SST and temperature at 500 hPa (T500; hereafter SST − T500 is referred to as Tdiff) to discriminate between PLs and synoptic-scale disturbances (Zahn and von Storch 2008; Xia et al. 2012; Chen and von Storch 2013; Zappa et al. 2014; Y16). For example, Y16 uses Tdiff > 43 K as a criterion for PLs. Figure 13 shows histograms of Tdiff averaged in a 1° × 1° box around the center of a vortex at the time of its maximum intensity. In the plot, the PMCs and synoptic-scale disturbances were identified using our algorithm. The synoptic-scale disturbances occur even if Tdiff is small, whereas PMCs tends to develop for large Tdiff. In total 80% of synoptic-scale disturbances developed with a Tdiff of less than 43 K, whereas more than half of the PMCs occurred when Tdiff was greater than 43 K. Although our algorithm for identifying PMCs is based solely on vorticity and SLP (and some of the large PMCs can be categorized as synoptic-scale lows), the result are consistent with those of previous studies that used the Tdiff as the discriminating criterion. The term Tdiff is used because PLs tend to develop in a convectively unstable environment; however, it is not directly linked to vortex development because of the horizontal shear instability, which is one of the significant processes in the generation of meso-β-scale PMCs. Our method is independent of thermal stratification and can deal with PMCs that develop mainly as a result of dynamic instability.
Relative frequency (bars) and cumulative relative frequency (lines) of PMCs and synoptic-scale disturbances for Tdiff. The scales for the relative frequency and cumulative relative frequency are shown on the right and left axes, respectively.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
Y16 also analyzed PLs over the Sea of Japan. Since we analyzed PMCs including meso-β-scale PMCs and weaker PMCs, there are several differences between our results and those of Y16. The number of PMCs identified in this study is 7 times (twice) as large as those for strong (weak) threshold identified in Y16. This is because our result includes meso-β-scale PMCs. Moreover, our algorithm can identify and count individual PMCs appearing simultaneously in a larger system. Note also that our algorithm identifies PMCs in an environment with small Tdiff, such as those developing through barotropic instability. In the present study, a peak in the location of the maximum intensity is found near the base of the Korean Peninsula (Fig. 9b), which is indistinct in Y16 (see Fig. 7 in Y16). The PMCs that develop in this region are so weak that most do not pass the intensity threshold for PLs set in Y16. There is also a difference in the direction of movement of PMCs. In the WJ region, PLs in Y16 tend to move eastward (see Fig. 6 in Y16), whereas we found that PMCs tend to move southeastward (Fig. 10). Most of these PMCs are at the meso-β scale and are associated with the JPCZ. In the NJ region, although both eastward-moving and southward-moving PMCs were detected in the present study, the proportion of eastward-moving PMCs in the present study (Fig. 10) is smaller than that in Y16. As eastward-moving PMCs are relatively large, some are likely to be excluded as a synoptic-scale low by our algorithm.
b. Environment of PMCs
The characteristics of PMCs depend on environmental factors such as the baroclinicity and vertical stability of the atmosphere where the PMCs form and develop (Yanase and Niino 2007). Here, we examine the vertical stability of the atmosphere in which the PMCs developed. Figure 14 shows a scatterplot of the maximum intensity and Tdiff averaged in a 1° × 1° box around the center of the PMC. Although weak PMCs are distributed across a wide range of Tdiff values, the more intense PMCs tend to occur in environments with a large Tdiff. PMCs in the WJ region sometimes occur under less unstable conditions (Tdiff < 35 K), but most are weak. Although these PMCs are likely to be formed by the dynamic instability such as barotropic instability associated with horizontal shear in the JPCZ, they do not intensify further when the stratification is not favorable to their development. In the NJ region, intense PMCs (i.e., maximum vorticity >8.0 × 10−4 s−1) tend to occur under highly unstable conditions (Tdiff > 45 K). Very cold air (T500 < −40°C) frequently intrudes into this region, and this is favorable for the development of PMCs. In the CJ region, there is no obvious relationship between PMC intensity and the vertical stability of the atmosphere, which indicates that the diabatic contribution is likely to be less important for the development of the PMCs in this region and other factors such as baroclinicity might be important.
Scatterplot of maximum vorticity and Tdiff for PMCs in each region.
Citation: Monthly Weather Review 144, 7; 10.1175/MWR-D-15-0349.1
5. Summary and conclusions
To improve our understanding of the climatology of PMCs over the Sea of Japan, we developed a new algorithm for the objective tracking of PMCs. The algorithm is based primarily on the vorticity field and consists of three parts: the identification of vortices, the connection of vortices in consecutive time steps, and the exclusion of synoptic-scale disturbances. First, we identify the location, intensity, and area of a vortex from a vorticity field at each time step, in which isolated vortex areas that include a single local maximum of vorticity are identified from a multiple vortex area. This method is suitable for identifying multiple PMCs embedded in a larger system such as the JPCZ. Second, the movement of a vortex is estimated by the steering wind. Two vortices in consecutive time steps are connected when they satisfy the criteria of the distance between the estimated position and the position of the vortex at the next time step. Finally, we distinguish PMCs from synoptic-scale disturbances using the size and shape of the PMCs.
We have successfully obtained reliable tracks of PMCs using this tracking algorithm. With additional criteria for the lifetime and intensity of PMCs, 86% of the detected PMCs agreed with the PMCs that were subjectively detected using the vorticity and SLP field, and 78% of the detected PMCs were identified using satellite images. The geographical distribution of the PMCs generated using the algorithm agrees well with previous studies based on satellite images. Our method for discriminating between PMCs and synoptic-scale disturbances does not depend on the environmental stratification and have demonstrated that the PMCs tend to occur in less stable environments than extratropical cyclones, which has been assumed in the previous climatological studies. Moreover, our method can deal with PMCs that develop mainly through horizontal shear instability.
Using our new tracking method, we analyzed the climatology of PMCs over the Sea of Japan. PMCs were classified into three groups according to the region where they attained their maximum intensity. PMCs in each region have different characteristics with regard to their direction of movement, size, and intensity. These characteristics are likely to be related to either the environmental conditions at the time of formation or the developmental mechanism. To consider this point in more detail, we are currently performing a composite analysis for which the results will be reported elsewhere.
This is the first research into the objective tracking of PMCs including meso-β-scale vortices. Our method can be applied to a high-resolution dynamical downscaling simulation or regional reanalysis data to examine a long-term climatology of PMCs. We may also apply this method to other regions where PMCs frequently form, in order to study similarity and differences of their characteristics among different regions around the world.
Acknowledgments
We acknowledge helpful comments by staff and students at the Dynamic Marine Meteorology Group, Atmosphere and Ocean Research Institute, The University of Tokyo. We would also like to acknowledge Dr. Irina Gurvich for providing a database of PMCs for us. We also thank two anonymous reviewers for their insightful comments on this paper. Shun-ichi I. Watanabe was supported by JSPS KAKENHI Grant 2611371 and the Program for Leading Graduate Schools, MEXT, Japan. Hiroshi Niino was supported by JSPS KAKENHI Grant 24244074. Wataru Yanase was supported by JSPS KAKENHI Grants 23740349 and 25106703. MA data are collected and distributed by the Research Institute for Sustainable Humanosphere, Kyoto University (http://database.rish.kyoto-u.ac.jp/index-e.html). MTSAT-2 IR images were provided by Kochi University (http://weather.is.kochi-u.ac.jp/). The OISST dataset was provided by NOAA (https://www.ncdc.noaa.gov/oisst/).
REFERENCES
Asai, T., 1988: Meso-scale features of heavy snowfalls in Japan Sea coastal regions of Japan (in Japanese). Tenki, 35, 156–161.
Asai, T., and Y. Miura, 1981: An analytical study of meso-scale vortex-like disturbances observed around Wakasa Bay area. J. Meteor. Soc. Japan, 59, 832–843.
Blechschmidt, A., 2008: A 2-year climatology of polar low events over the Nordic Seas from satellite remote sensing. Geophys. Res. Lett., 35, L09815, doi:10.1029/2008GL033706.
Bromwich, D. H., D. F. Steinhoff, I. Simmonds, K. Keay, and R. L. Fogt, 2011: Climatological aspects of cyclogenesis near Adélie Land Antarctica. Tellus, 63A, 921–938, doi:10.1111/j.1600-0870.2011.00537.x.
Chen, F., and H. von Storch, 2013: Trends and variability of North Pacific polar lows. Adv. Meteor., 2013, 170387, doi:10.1155/2013/170387.
Condron, A., G. Bigg, and I. Renfrew, 2006: Polar mesoscale cyclones in the northeast Atlantic: Comparing climatologies from ERA-40 and satellite imagery. Mon. Wea. Rev., 134, 1518–1533, doi:10.1175/MWR3136.1.
Fu, G., H. Niino, R. Kimura, and T. Kato, 2004: Multiple polar mesocyclones over the Japan Sea on 11 February 1997. Mon. Wea. Rev., 132, 793–814, doi:10.1175/1520-0493(2004)132<0793:MPMOTJ>2.0.CO;2.
Gurvich, I. A., 2013: Intense mesoscale cyclones over the Far Eastern Seas in the cold half of the year according to satellite remote sensing (in Russian). Ph.D. thesis, V.I. Il’ichev Pacific Oceanolocical Institute, Far Eastern Branch Russian Academy of Sciences, 171 pp.
Harold, J., G. Bigg, and J. Turner, 1999: Mesocyclone activity over the North-East Atlantic. Part 1: Vortex distribution and variability. Int. J. Climatol., 19, 1187–1204, doi:10.1002/(SICI)1097-0088(199909)19:11<1187::AID-JOC419>3.0.CO;2-Q.
Hjelmfelt, M. R., 1990: Numerical study of the influence of environmental conditions on lake-effect snowstorms over Lake Michigan. Mon. Wea. Rev., 118, 138–150, doi:10.1175/1520-0493(1990)118<0138:NSOTIO>2.0.CO;2.
Honda, Y., M. Nishijima, K. Koizumi, Y. Ohta, K. Tamiya, T. Kawabata, and T. Tsuyuki, 2005: A pre-operational variational data assimilation system for a non-hydrostatic model at the Japan Meteorological Agency: Formulation and preliminary results. Quart. J. Roy. Meteor. Soc., 131, 3465–3475, doi:10.1256/qj.05.132.
Irving, D., I. Simmonds, and K. Keay, 2010: Mesoscale cyclone activity over the ice-free Southern Ocean: 1999–2008. J. Climate, 23, 5404–5420, doi:10.1175/2010JCLI3628.1.
Japan Meteorological Agency, 2013: Outline of the operational numerical weather prediction at the Japan Meteorological Agency (March 2013). Appendix to WMO technical progress report on the global data-processing and forecasting system and numerical weather prediction, 188 pp. [Available online at http://www.jma.go.jp/jma/jma-eng/jma-center/nwp/outline2013-nwp/index.htm.]
Kato, T., 2005: Momentum energy budget analysis on a vortex disturbance developing along the Japan-Sea Polar-Airmass Convergence Zone (in Japanese). Meteor. Res. Note, 208, 277–284.
Kobayashi, S., and Coauthors, 2015: The JRA-55 Reanalysis: General specifications and basic characteristics. J. Meteor. Soc. Japan, 93, 5–48, doi:10.2151/jmsj.2015-001.
Kuroda, Y., 1992: The convergent cloud band and shipwreck in the Japan Sea (in Japanese). Sea Sky, 67, 261–279.
Laird, N. F., L. J. Miller, and D. A. R. Kristovich, 2001: Synthetic dual-Doppler analysis of a winter mesoscale vortex. Mon. Wea. Rev., 129, 312–331, doi:10.1175/1520-0493(2001)129<0312:SDDAOA>2.0.CO;2.
Lee, T. Y., Y. Y. Park, and Y. L. Lin, 1998: A numerical modeling study of mesoscale cyclogenesis to the east of the Korean Peninsula. Mon. Wea. Rev., 126, 2305–2329, doi:10.1175/1520-0493(1998)126<2305:ANMSOM>2.0.CO;2.
Miyazawa, S., 1967: On vortical mesoscale disturbances observed during the period of heavy snow or rain in the Hokuriku district. J. Meteor. Soc. Japan, 45, 166–176.
Nagata, M., 1993: Meso-β-scale vortices developing along the Japan-Sea Polar-Airmass Convergence Zone (JPCZ) cloud band: Numerical-simulation. J. Meteor. Soc. Japan, 71, 43–57.
Ninomiya, K., 1989: Polar/comma-cloud lows over the Japan Sea and the northwestern Pacific in winter. J. Meteor. Soc. Japan, 67, 83–97.
Ninomiya, K., 1994: A mesoscale low family formed over the northeastern Japan Sea in the northwestern part of a parent polar low. J. Meteor. Soc. Japan, 72, 589–603.
Ninomiya, K., and K. Hoshino, 1990: Evolution process and multi-scale structure of a polar low developed over the Japan Sea on 11–12 December 1985. Part II: Meso-β-scale low in meso-α-scale polar low. J. Meteor. Soc. Japan, 68, 307–318.
Ninomiya, K., K. Hoshino, and K. Kurihara, 1990: Evolution process and multi-scale structure of a polar low developed over the Japan Sea on 11–12 December 1985. Part I: Evolution process and meso-α-scale structure. J. Meteor. Soc. Japan, 68, 293–306.
Okabayashi, T., and M. Satomi, 1971: A study on the snowfall and its original clouds by the meteorological radar and satellite (Part I) (in Japanese). Tenki, 18, 573–581.
Ookubo, A., 1995: Two types of vortical disturbances over the Hokuriku district in winter (in Japanese). Tenki, 42, 705–714.
Orlanski, I., 1975: A rational subdivision of scales of atmospheric processes. Bull. Amer. Meteor. Soc., 56, 527–530.
Rasmussen, E., and J. Turner, 2003: Polar Lows. Cambridge University Press, 612 pp.
Reynolds, R., T. Smith, C. Liu, D. Chelton, K. Casey, and M. Schlax, 2007: Daily high-resolution-blended analyses for sea surface temperature. J. Climate, 20, 5473–5496, doi:10.1175/2007JCLI1824.1.
Saito, K., and Coauthors, 2006: The operational JMA nonhydrostatic mesoscale model. Mon. Wea. Rev., 134, 1266–1298, doi:10.1175/MWR3120.1.
Shimada, U., A. Wada, K. Yamazaki, and N. Kitabatake, 2014: Roles of an upper-level cold vortex and low-level baroclinicity in the development of polar lows over the Sea of Japan. Tellus, 66A, 24694, doi:10.3402/tellusa.v66.24694.
Shimizu, N., and A. Uchida, 1974: An observational study of organized snow echo over the Japan Sea. J. Meteor. Soc. Japan, 52, 289–299.
Shimizu, S., and H. Uyeda, 2012: Algorithm for the identification and tracking of convective cells based on constant and adaptive threshold methods using a new cell-merging and -splitting scheme. J. Meteor. Soc. Japan, 90, 869–889, doi:10.2151/jmsj.2012-602.
Simmonds, I., R. J. Murray, and R. M. Leighton, 1999: A refinement of cyclone tracking methods with data from FROST. Aust. Meteor. Mag. Special Issue, 35–49.
Tsuboki, K., and G. Wakahama, 1992: Mesoscale cyclogenesis in winter monsoon air streams: Quasi-geostrophic baroclinic instability as a mechanism of the cyclogenesis off the west coast of Hokkaido Island, Japan. J. Meteor. Soc. Japan, 70, 77–93.
Tsuboki, K., and T. Asai, 2004: The multi-scale structure and development mechanism of mesoscale cyclones over the Sea of Japan in winter. J. Meteor. Soc. Japan, 82, 597–621, doi:10.2151/jmsj.2004.597.
Uotila, P., T. Vihma, A. B. Pezza, I. Simmonds, K. Keay, and A. H. Lynch, 2011: Relationships between Antarctic cyclones and surface conditions as derived from high‐resolution numerical weather prediction data. J. Geophys. Res., 116, D07109, doi:10.1029/2010JD015358.
Watanabe, S. I., and H. Niino, 2014: Genesis and development mechanisms of a polar mesocyclone over the Japan Sea. Mon. Wea. Rev., 142, 2248–2270, doi:10.1175/MWR-D-13-00226.1.
Wernli, H., and C. Schwierz, 2006: Surface cyclones in the ERA-40 dataset (1958–2001). Part I: Novel identification method and global climatology. J. Atmos. Sci., 63, 2486–2507, doi:10.1175/JAS3766.1.
Xia, L., M. Zahn, K. I. Hodges, F. Feser, and H. V. Storch, 2012: A comparison of two identification and tracking methods for polar lows. Tellus, 64A, 17196, doi:10.3402/tellusa.v64i0.1719.
Yamagishi, Y., M. Doi, N. Kitabatake, and H. Kamiguchi, 1992: A polar low which accompanied strong gust (in Japanese). Tenki, 39, 27–36.
Yanase, W., and H. Niino, 2007: Dependence of polar low development on baroclinicity and physical processes: An idealized high-resolution numerical experiment. J. Atmos. Sci., 64, 3044–3067, doi:10.1175/JAS4001.1.
Yanase, W., G. Fu, H. Niino, and T. Kato, 2004: A polar low over the Japan Sea on 21 January 1997. Part II: A numerical study. Mon. Wea. Rev., 132, 1552–1574, doi:10.1175/1520-0493(2004)132<1552:APLOTJ>2.0.CO;2.
Yanase, W., H. Niino, S. I. Watanabe, K. Hodges, M. Zahn, T. Spengler, and I. A. Gurvich, 2016: Climatology of polar lows over the Sea of Japan using the JRA-55 reanalysis. J. Climate, 29, 419–437, doi:10.1175/JCLI-D-15-0291.1.
Zahn, M., and H. von Storch, 2008: Tracking polar lows in CLM. Meteor. Z., 17, 445–453, doi:10.1127/0941-2948/2008/0317.
Zappa, G., L. Shaffrey, and K. Hodges, 2014: Can polar lows be objectively identified and tracked in the ECMWF operational analysis and the ERA-Interim reanalysis? Mon. Wea. Rev., 142, 2596–2608, doi:10.1175/MWR-D-14-00064.1.
