Observations of fogs with a millimeter-wave scanning Doppler radar were conducted at Kushiro in Hokkaido, Japan, in the summer seasons of 1999 and 2000. Three typical types of plan position indicator (PPI) displays were observed: cellular echoes with high radar reflectivity factors (∼−10 dBZ), uniformly distributed echoes with high reflectivities (∼−10 dBZ), and uniformly distributed echoes with low reflectivities (∼−30 dBZ). The authors focused on advection fog with cellular echoes observed on 5 August 1999 and 31 July 2000. Echoes showed structures of cells with a reflectivity of −10 dBZ and with intervals of about 1 km. This echo pattern moved northward (i.e., from the sea to the land). There was a vertical shear of the horizontal wind at a height around 200 m in both cases, and structures of each cell were upright above the shear line and were leaning below it. The direction and the speed of the echo pattern in both PPI and range–height indicator (RHI) displays agreed well with that of the horizontal wind at heights above the shear (200 m). In the echo cells, existence of drizzle drops is implied.
On the Pacific Ocean east coast of Hokkaido, Japan, fogs (especially sea fogs) frequently appear, especially in the boreal summer season. Observational studies in this region have been performed for a long time (e.g., Research Mobilization Council 1945; Hori 1953; Karatsu et al. 1963; Uyeda and Yagi 1984; Sea Fog Research Team 1985; Yanagisawa et al. 1986). Sea fogs in other regions have also been studied (e.g., Pilié et al. 1979; Leipper 1994). Since most of them were based on observations at one (or more) point(s), spatial distribution or structure of fogs has not been sufficiently clarified. It is necessary to obtain distribution, structure, and movement of fogs to understand their evolution and advection.
It is effective to apply remote sensing techniques to the study of fog, and actually several fog observations have been performed using them. Welch and Wielicki (1986) studied fogs using Landsat satellite images and reported that the horizontal structure of fog showed cellular structure. However, satellites cannot provide any information on vertical distribution or inner structure of fogs, and it is also difficult to obtain movement of fogs from satellite imaging because of its scarce time resolution. Lidar observations were also conducted (e.g., Tomine et al. 1991; Jyumonji and Uchiyama 1996). Tomine et al. (1991) obtained vertical cross section of the attenuation coefficient using a lidar. Jyumonji and Uchiyama (1996) obtained horizontal and vertical density profiles of sea fog and land fog. Millimeter-wave (mm wave) radars have sufficient sensitivity to observe inner structure of fogs because of its short operational wavelength. Furthermore, mm-wave radars have higher temporal resolution than radiosondes and satellite images. Cloud and precipitation studies using 35-GHz radars (e.g., Hobbs et al. 1985, Takeda and Horiguchi 1986; Moran et al. 1998), including the scanning Doppler radar of the National Oceanic and Atmospheric Administration (NOAA) Environmental Technology Laboratory (ETL; Pasqualucci et al. 1983; Kropfli et al. 1995; Kropfli and Kelly 1996), 95-GHz radars (e.g., Lhermitte 1987; Pazmany et al. 1994; Horie et al. 2000), and scanning radars using dual frequency of both (e.g., Mead et al. 1994; Iwanami et al. 2001) were reported. However, there have been only a few studies of fog observations using mm-wave radars. Yanagisawa et al. (1986) observed sea fogs with a 35-GHz radar, and Mead et al. (1989) conducted a preliminary observation of a radiation fog using a 215-GHz radar. Special observations of sea fogs using a mm-wave radar were conducted at Kushiro in 1981 and 1982 (Sea Fog Research Team 1985; Yanagisawa et al. 1986; Sawai 1988). Yanagisawa et al. (1986) showed that there were two types of echo distribution: “coastal fog” and “advection fog.” The former echoes are distributed from the coastline to 4 km out to the sea, and their intensities do not change in time. The latter echoes move toward land from an area of sea more than 20 km from the coast. They found cellular structure as found in cloud by using a vertical-pointing mm-wave radar. The mm-wave radar used by Yanagisawa et al. (1986) did not have a steerable antenna so they could not obtain two- or three-dimensional distribution and structure of fogs. Moreover, this radar could not provide Doppler velocities, corresponding to motion of the fog droplets.
The Research Institute for Sustainable Humanosphere (RISH) of Kyoto University and Mitsubishi Electric Corporation have developed a mm-wave Doppler radar (34.75 GHz) in 1997 (Hamazu et al. 2000, 2003). This radar has a scanning antenna, and provides spatial distribution of fog echo and Doppler velocity. In this study, we conducted radar observations in the Kushiro district in July and August 1999 and 2000. Hamazu et al. (2003) showed the initial result of three-dimensional structure of sea fog using the data in 1999. In this paper, we will show the relationship among three-dimensional structure, its movement, and horizontal wind velocity.
The site location is shown in Fig. 1. The first observations were conducted on 22–27 July and 5–7 August 1999. We placed the car-mounted mm-wave Doppler radar at Kushiro Airport (43.02°N, 144.12°E) in Kushiro, Hokkaido, Japan. Kushiro Airport is located 20 km west of downtown Kushiro, 4.8 km north of the Pacific Ocean coastline, and 90 m above sea level. The radar beam covered the airport runway and the ocean, though the radar vehicle and trees behind the radar prevented observation to the northeast area of the observation range.
The second set of observations was conducted on 30 July–10 August 2000. The radar was installed on a plateau in the town of Shiranuka (43.00°N, 144.11°E, 46 m above sea level), where we had an unobstructed view of the Pacific Ocean. The site was located 0.9 km north of the Pacific Ocean coastline and 4.4 km southwest of Kushiro Airport. The radar vehicle behind the radar prevented observations to the west of the observation range, and the radar beam covered both Kushiro Airport and the ocean.
Visibility was routinely observed with Runway Visible Range (RVR) meters installed in three places at Kushiro Airport. Surface meteorological parameters, including wind direction, wind speed, temperature, relative humidity, and air pressure were also observed by the Kushiro Airport Meteorological Observatory.
Millimeter-wave Doppler radar
The mm-wave Doppler radar (Hamazu et al. 2000, 2003) operates at a frequency of 34.75 GHz. The radar has a 2-m parabolic antenna and a 100-kW peak output power, with a range resolution of 125 m, and sufficient performance to detect scattered echoes from fog droplets. However, we could not obtain data within 750-m range from the radar because of limitations of the modulator. Specifications of the radar are shown in Table 1. Minimum detectable radar reflectivity factor when signal-to-noise (S/N) ratio is 0 dB is −40, −30, and −20 dBZ when the range from the radar is 2, 5, and 10 km, respectively (Hamazu et al. 2003). Reliability of Doppler measurement was confirmed by Hamazu et al. (2003). To obtain an accurate radar reflectivity factor, we have carefully checked system gain and loss factors in prior to observations.
In the present observations, we regarded large radar reflectivities (>0 dBZ) as ground clutter echoes and removed them. For the 1999 observations, we operated the radar in pulse-pair mode, in which the Nyquist velocity was ±9.7 m s−1 (Hamazu et al. 2003). For the 2000 observations, we operated the radar in single-pulse mode to obtain more accurate Doppler velocities. Since the Nyquist velocity was ±1.94 m s−1 (Hamazu et al. 2003), we corrected velocity ambiguities using surface wind velocity measurements at the radar site and Kushiro Airport Meteorological Observatory. We used a combination of plan position indicator (PPI), constant-altitude PPI (CAPPI), and range–height indicator (RHI) antenna scanning modes.
General characteristics of radar reflectivity
Figure 2 shows some examples of radar reflectivity on a PPI display. The first type shows an echo area (>−10 dBZ) with cellular, or line-shaped, structures and extension beyond 10 km (Fig. 2a). In most cases, this echo retained its structure as the fog moved, and the direction was generally from sea (south) to land (north). The second type shows a uniformly distributed echo area with relatively strong (∼−10 dBZ) radar reflectivity (Fig. 2b). The area of high radar reflectivity repeatedly stretched and shrank with time. The last type is an echo area with no cellular structure and small reflectivity factor (∼−30 dBZ) (Fig. 2c). The range of observed echoes is shorter than 4 km in this case. The sensitivity of the mm-wave radar is limited to about −40 and −30 dBZ at the range of 2–5 km, and the mm-wave radar might fail to detect fogs in the far region.
According to classification by Yanagisawa et al. (1986), we categorize the first and second type as “advection fog” and the last type as “coastal fog,” respectively. In Yanagisawa et al. (1986) there are two echo types in advection fog. In one echo type, echo power temporarily increases and range of the observed echo expands, and in the other echo type, observed echo moves from sea to land. If the latter type echo is observed, echo has an inner structure, such as band-like structures, and moves at a certain speed. However, even if the former type is observed, as long as beam direction is fixed in one direction, it is difficult to distinguish between the following two causes: 1) observed echo range expands because the fog region spatially expands or detectable range expands owing to increase of echo power (i.e., growth of fog droplets), and 2) band-like echo moves in the direction perpendicular to the radar beam. Yanagisawa et al. (1986) reported that in one case the range of observed echo temporarily expanded because of an increase of echo power at first, however, after the radar beam direction was changed from westward to southward, a bandlike echo that moved from sea to land was observed. Using a scanning radar, we can determine whether echoes have inner structures such as cellular or bandlike echoes (see Fig. 2a) or do not (see Figs. 2b,c), and can capture the three-dimensional motion of the transition and movement of echo structures. In this study, we will focus on the cellular structure of echoes and its movement, and show two typical cases in the following section.
Structure and motion of cellular fog
We will show two cases of echoes having cellular structure, which were observed on the mornings of 5 August 1999 and 31 July 2000, respectively. Radar observations were conducted during 0200–1000 LT in both cases.
Synoptic and surface weather conditions
Roach (1995) explains that most sea fogs arise in moist airstreams carried poleward over colder sea in the warm sectors of cyclones or around the western flanks of large anticyclones. Especially in the northwestern Pacific in summers, typical sea fogs form when southerly flow at the western flanks of Pacific high pressure is cooled by the colder sea due to the Oyashio (Sawai 1982). Surface weather charts at 0300 LT 5 August 1999 and 31 July 2000 are shown in Fig. 3. On 5 August 1999, high pressure was located in the Pacific Ocean, and Kushiro was located at the western edge of the high pressure. An isobar stretched from south to north around Kushiro and the geostrophic wind tended to be southerly, by which sea fog was advected from south to north. On 31 July 2000, high pressure was located in the Pacific Ocean far to the east of Hokkaido, low pressure was located in Siberia, and a stationary front existed over the Japan Sea. Kushiro was again located at the western edge of the high pressure, as in the case of 1999, with a southerly geostrophic wind. Sea temperature was measured during all observation periods in 2000 by a research vessel, the Koufumaru, of Hakodate Marine Meteorological Observatory, and Usui (2001) showed that the sea temperature was 1°–5°C lower than the air temperatures. It is considered that in both cases moist southerly flow due to the Pacific high pressure was cooled by the sea and therefore sea fog formed. These conditions agree well with typical conditions for sea fogs (Roach 1995; Sawai 1982).
Figure 4 shows temporal variations of visibility, temperature, dewpoint temperature, relative humidity, wind direction, and wind velocity obtained at the surface at Kushiro Airport Meteorological Observatory on 5 August 1999 and 31 July 2000. Visibility was sometimes less than 1000 m in both cases. Both temperature and dewpoint were gradually increasing until 1000 LT 5 August 1999 and during the entire observation period on 31 July 2000. Surface wind was almost northerly or easterly during the whole observation period on 5 August 1999 and in the first half of the observation period on 31 July 2000, which was different from geostrophic wind. When we consider the formation of sea fog, it is necessary to consider not only surface wind but also the wind profile at the whole height where fog exists.
Structure of cellular echoes
Figure 5 shows PPI displays of radar reflectivity factor having cellular structure at 0742 LT 5 August 1999 (Fig. 5a) and at 0457 LT 31 July 2000 (Fig. 5b). In each case, cellular structure was observed, and the echo pattern moved from south to north while retaining its pattern. Similar horizontal cellular structure of fog was observed in central California and eastern South Dakota using the Landsat satellite (Welch and Wielicki 1986).
Figure 6 shows RHI displays of radar reflectivity and Doppler velocity for the two cases. Though the Doppler velocity includes both horizontal and vertical movements of fog droplets, we neglected the vertical-movement components of Doppler velocity since the elevation angle is small—for example, 5.7° at the height of 500 m at 5-km distance from the radar. Because the radar beam was pointed southward, the Doppler velocity in this case shows only meridional wind components. A positive (negative) value indicates southerly (northerly) wind.
The echo top is almost flat at a height of 600 and 500 m in Fig. 6a (upper) and Fig. 6b (upper), respectively. In both cases, radar reflectivity changes from less than −30 to ∼−10 dBZ at intervals of about 1 km. The size of each cell is 200–1000 m in the horizontal and 500–600 m in the vertical dimension; therefore, the shape of each cell is boxlike. Each cell vertically extends above 200 m and leans northward with height below 200 m. In Kushiro, existence of these cellular structures had been suggested by Yanagisawa et al. (1986). They observed echoes containing a spatial structure with a vertically fixed mm-wave radar, and the horizontal scales of cellular echoes were estimated as 1.0 km from the wind velocity. By using the mm-wave scanning Doppler radar, we directly observed the horizontal–vertical distribution of the echo structure and confirmed the estimation by Yanagisawa et al. (1986) is correct.
Vertical shear of meridional wind exists at 200 m height in both cases. At 0746:30 LT 5 August 1999, the wind is southerly at 3–7 m s−1 above 200 m and northerly at 1–2 m s−1 below 200 m (see Fig. 6a, lower). At 0503:02 LT 31 July 2000, southerly of 2.0–3.5 m s−1 prevails above 200 m and weak southerly or northerly < 0.5 m s−1 exists below 200 m (see Fig. 6b, lower). Figure 7 shows the height profiles of horizontal wind estimated with the velocity–azimuth display (VAD) method at 0459–0501 LT 31 July 2000. Vertical shear of the horizontal wind exists at a height of approximately 200 m. A southerly wind of approximately 2.5–3.0 m s−1 was blowing above the shear, while an almost easterly wind was present at heights below the shear. Surface wind at the Kushiro Airport (see Fig. 4b) and at the radar site at 0500 LT (not shown here) were approximately easterly with a speed of 0.3–1.0 m s−1. This is consistent with Doppler velocity in RHI display (see Fig. 6b, lower). Wind directions above and below the shear are almost consistent with the geostrophic wind at 0300 LT (see Fig. 3b) and the surface wind at 0500 LT (see Fig. 4b), respectively. In comparison with the structure of cells and wind velocity, the break in the vertical structure of each cell exists at the height of 200 m, which is the same height as the shear line.
Some Doppler velocity fluctuations of 2.0–3.5 m s−1 exist above 200 m (see Fig. 6b, lower), and as well as radar reflectivity several cells exist at intervals of 1–2 km. Such structure was not found at 0746:30 LT 5 August 1999 (see Fig. 6a, lower). We cannot determine the source of spatial structure in Doppler velocity from only these observational data, but we interpret it qualitatively as due to convection or falling of fog droplets or drizzle drops.
Relation between moving velocity of echo pattern and horizontal wind
Figure 8 shows the time–range plot of reflectivity in the meridional plane. The reflectivities in Fig. 8a were obtained every 1 min at elevation angles of 0.3°–1.1° during 0205–1001 LT 5 August 1999, while the reflectivities in Fig. 8b were obtained every 4 min at an elevation angle of 0.7° during 0153–1023 LT 31 July 2000. Only data to the south of the radar site are displayed in Fig. 8a, and data to the north and south of the radar site are displayed in Fig. 8b. Some data are missing because of the radar operation in different scanning modes. Missing pockets of data in the range means that ground clutter echoes overwhelmed the atmospheric return. Elevation angles of 0.7° and 1.1° correspond to heights of 122 and 192 m at a range of 10 km, respectively. These figures represent the meridional movement of the echo pattern at heights of 0–200 m.
In both figures the bandlike areas of high radar reflectivity were moving from south to north. Slopes of the echo represent translational speeds of the echo patterns shown in Figs. 2a and 5. The translated speeds of the echo pattern were nearly constant and were 4.2 and 2.6 m s−1 on 5 August 1999 (see Fig. 8a) and 31 July 2000 (see Fig. 8b), respectively. These speeds agree well with the wind velocity at heights above 200 m on both 5 August 1999 (see Fig. 6a, lower) and 31 July 2000 (see Fig. 6b, lower), respectively, although elevation angles in PPI displays (see Fig. 8) are 0.3°–1.1° and therefore the height of the scanning area is below 200 m.
Figure 9 shows temporal changes of the radar reflectivities in RHI displays. The reflectivities in Fig. 9a were obtained at 18-s intervals during 0745:37–0747:03 LT 5 August 1999, and the reflectivities in Fig. 9b were at 36-s intervals during 0503:02–0505:40 LT 31 July 2000. In both figures, each echo cell moved from south to north while preserving its shape. On 5 August 1999, during the 71 s (0745:37–0746:48) each echo cell moved 300 m at all heights (see Fig. 9a). It shows that the echo moves meridionally with a speed of 4.2 m s−1. On 31 July 2000, each echo cell moved 350–500 m for the 143-s period (see Fig. 9b), with a speed of 2.4–3.5 m s−1. These movements agree with the wind velocity above 200 m shown in Figs. 6a and 6b, and also with the horizontal translational speed of the echo pattern shown in Figs. 8a and 8b, respectively. This indicates that the echo cells at all heights moved with the same speed as the wind above the shear.
Figure 10 shows temporal changes in the Doppler velocity at RHI displays for the same period as Fig. 9b, where inner structure was also observed. The shear line at the height of 200 m remained during this period, and the fluctuating patterns of Doppler velocity above 200 m moved from south to north.
Echo cells, whose slope changed at the shear height, moved with their structure retained, and with almost the same velocity as wind velocity at the height higher than the shear. This implies that the fog droplets or the drizzle drops with certain falling speeds are being seeded from above. If there are drops with larger size than ordinary fog droplets, we can insist that movement of the echo cells indicates falling drizzle drops while blown by the background wind.
On the Pacific east coast of Hokkaido, existence of drizzle in fog layers has been reported in several articles. Takahashi (1949) observed drop-size distributions of fog and drizzle at Kushiro with drop-size diameters greater than 200 μm. Yanagisawa et al. (1986) did not observe drops of diameter larger than 100 μm, but they calculated liquid water contents (LWC) from the radar reflectivity factors that reached 4.0 g m−3, and they concluded that drizzle existed in the fog layer. Homma et al. (1962) observed drizzle at Nemuro, about 100 km east of Kushiro, with maximum drop diameters of more than 200 μm, mostly when the thickness of the fog layer was more than 400 m. At Nemuro, another observation (Kuroiwa and Okita 1959) showed that diameters of 180–270 μm were dominant in drop-size distributions of drizzle drops at a height of 200 m. Therefore, as far as previous research shows, it is possible for larger drops to exist in fog layers.
In our data, it is difficult to determine drop size because we did not observe either the LWC or the drop-size distribution. Suppose that we assume that the drop-size distribution is uniform and the LWC is constant. The radar reflectivity factor of the echo cells at a height of 300 m is approximately −7 and −12 dBZ at 0746:30 LT 5 August 1999 and at 0503:02 LT 31 July 2000, respectively. If drop-size distribution is assumed to be monodispersed, the relationship between LWC W (g m−3) and drop diameter D (m) under the observed radar reflectivity factor Z (mm6 m−3) is (Hamazu et al. 2003)
Figure 11 shows the relationship between W and D under the observed Z (dBZ) = 10 × log10Z (mm6 m−3). Yanagisawa et al. (1986) reported that when the radar reflectivity factor was more than −20 dBZ, measured LWCs of fogs at Kushiro varied from 0.04 to 0.8 g m−3. For LWCs of 0.04 and 0.8 g m−3, a reflectivity of −7 dBZ implies drop diameters of 138 and 51 μm, and for −12 dBZ, these values are reduced to 94 and 35 μm, respectively. If we follow the definition of drizzle by French et al. (2000), the diameter of drizzle drops is larger than 50 μm. Therefore, existence of drizzle is strongly implied especially in the case of −7 dBZ.
Considering that drop-size distribution is not uniform, there may be small numbers of large diameter drops in the cells. Radar reflectivity is sixth-power proportional to drizzle diameter, and French et al. (2000) pointed out that if drop-size distribution is bimodal or even if size distribution of drizzle is almost flat, drizzle is dominant in reflectivity-weighted size distribution relative to cloud droplets. Previous studies show that drop-size distribution of sea fog at the east Pacific coast of Hokkaido can be bimodal (Sea Fog Research Team 1985; Kuroiwa and Okita 1959). To confirm the existence of drizzle drops in echo cells, measurement of drop-size distributions is required in future studies.
In our study, cellular structures were accompanied by horizontal wind shear. These cellular structures imply that there is some cell-type circulation pattern. Many investigators have studied the cellular circulation patterns in the atmospheric boundary layer. Numerous references can be found in reviews by Young et al. (2002), Etling and Brown (1993), and Brown (1980). Possibilities of inflection-point instability, parallel instability, and convective instability have been pointed out when cellular circulation patterns are observed. Studies of dynamical and thermal problems are required in future studies.
We have studied three-dimensional structure and movement of fogs at Kushiro in Hokkaido, Japan, using observations from the mm-wave scanning Doppler radar. We discussed general characteristics of the radar reflectivities of these fogs. There were three broad types of fog echoes: cellular types with a high radar reflectivity factor of approximately −10 dBZ, uniformly distributed types with high radar reflectivity factors, and a uniformly distributed type with weak radar reflectivities. We focused on the two cases having cellular structures of 5 August 1999 and 31 July 2000, and the following results were obtained.
Cellular echo structure at intervals of about 1 km was observed.
There was a vertical shear of the horizontal wind at a height around 200 m.
Vertical structure of each cell was upright above the shear region, and was tilted below the shear region.
The echo patterns from the cellular structures moved from sea (south) to land (north), with retaining their structure.
The direction and the speed of the echo-moving pattern is almost equal to the horizontal wind above the shear.
To advance our understanding of fog and drizzle, simultaneous measurements of LWCs, using a lidar, for example, and drop-size distributions together with reflectivity factors from a mm-wave radar, are required.
The authors thank Mr. Kenji Akaeda of Japan Meteorological Agency and Mr. Akira Yamamoto of Meteorological Research Institute for their support to the present observations and providing surface meteorological data at Kushiro Airport. The authors also thank the staff of Kushiro Airport Meteorological Observatory for their support for fog observations. We thank Dr. Gernot Hassenpflug for his careful reading of the original manuscript. The present study was financially supported by Grants-in-Aids (12740270) of the Japan Society for the Promotion of Science (JSPS).
Corresponding author address: Akihisa Uematsu, Research Institute for Sustainable Humanosphere, Kyoto University, Gokasho, Uji, Kyoto 611-0011, Japan. firstname.lastname@example.org