Elevated Ducts and Low Clouds over the Central Western Pacific Ocean in Winter Based on GPS Soundings and Satellite Observation

LI Xiaodong, SHENG Lifang, , *, and WANG Wencai

Elevated Ducts and Low Clouds over the Central Western Pacific Ocean in Winter Based on GPS Soundings and Satellite Observation

LI Xiaodong1), SHENG Lifang1), 2), *, and WANG Wencai2)

1) College of Oceanic and Atmospheric Sciences, Ocean University of China, Qingdao 266100, China 2) Key Laboratory of Physical Oceanography, Ocean University of China, Qingdao 266100, China

Both low clouds and elevated ducts are common phenomena in the oceanic atmosphere. Low clouds affect elevated ducts by changing the structure of atmospheric temperature and humidity. However, due to the limitation of met-ocean measurements, research on them is still scattered. This paper presents the distribution of elevated ducts and clouds over the central Western Pacific Ocean (WPO) based on Global Position System (GPS) sounding data and Himawari-8 satellite products from November 2015 to January 2016. Results show that the frequency of elevated ducts detected by ship-based GPS soundings was as high as 77% over the central WPO. The height and frequency of elevated ducts are closely related to the low clouds. If there are no clouds, the occurrence probability and mean base height of the elevated ducts are 14% and 730m, respectively. By comparison, the occurrence probability and mean base height increase up to 24% and 1471m, respectively, in the presence of cumulus (Cu) clouds, and 22% and 1511m, respectively, in the presence of stratocumulus (Sc) clouds. Elevated ducts occur near the cloud top. The analysis of geopotential height and wind fields from the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis dataset (ERA-interim) shows that the study area is covered by a strong and stable subtropical high, and slowly sinking dry air masses inside the subtropical high are above the moist boundary-layer air mass. The appearance and evolution of low clouds will adjust the temperature and humidity structure of the lower troposphere. If there are no clouds, the marine boundary layer (MBL) is the classic mixed boundary layer. Humidity gradient and subsidence inversion are formed atop the mixed layer. When low clouds are present, long wave radiation and entrainment atop clouds form a strong temperature inversion and humidity gradient, which strengthen elevated ducts. However, when Sc clouds are decoupled, a weaker temperature inversion and humidity gradient may occur between the surface mixed layer and subcloud layer, leading to a weak elevated duct atop the mixed layer.

elevated duct; cumulus cloud; stratocumulus cloud; marine boundary layer; subtropical high

1 Introduction

When electromagnetic waves propagate in the atmosphere, they are refracted, absorbed, and scattered. If the part of electromagnetic waves is trapped in the thin layer of a certain thickness, it is called the atmospheric wave- guide propagation of electromagnetic waves, and the atmosphere of a certain thickness formed by the atmosphericwaveguide propagation is called the duct (Freehafer and Kerr, 1988). Elevated ducts generally occur in the troposphere below 3000m over the ocean (Lopez, 2009) and are caused by anomalous atmospheric refractive structures associated with the vertical distributions of temperature and moisture.They can change the energy distribution and propagation direction of electromagnetic waves. Since electromagnetic waves are trapped by elevated ducts, shadow zones of radars could be formed in the tropo-spheric atmosphere and they affect naval low altitude operations (Ziemba, 2013; Dinc and Akan, 2014).

The central region of the Western Pacific Ocean (WPO) is located in the area of Hadley cell subsidence and is often covered by a subtropical high in winter. On the one hand, warm surface water in the WPO in winter continuously provides water vapor for the marine boundary layer (MBL) through evaporation and turbulence (Song and Yu, 2013). On the other hand, slowly sinking dry air mass within the subtropical high inhibits the development of vertical convection and the condensation of water vapor. Consequently, there is sufficient moisture in the MBL and the weather is fine over the central WPO in winter (Stevens, 2005). Sporadic low clouds, which are mainly cumulus (Cu) and stratocumulus (Sc), appear at the top of MBL over the central WPO (Schubert, 1995). The radiation and entrainment mechanisms of low clouds can feed back to the structure of the MBL (Horn, 2015). The vertical distributions of temperature and moisture inside the MBL are changed by the actions of long wave radiation (Wood, 2012) and entrainment (Stevens, 1999; Katzwinkel, 2012) atop low clouds. Once the vertical distributions of temperature and moisture inside the MBL are not uniform, it may lead to the occurrence of elevated ducts (Hacck, 2010; Wang, 2016).

Many scholars have studied the MBL, marine low clouds, and marine elevated ducts over multiple sea areas using Global Position System (GPS) sounding data (Ding, 2013; MacKellar, 2013; Shupe, 2013; Manjula, 2016). Carrillo(2015) indicated the strong stratification observed in the lower troposphere over the east side of the subtropical North Atlantic, with the strengthening of stability centered near 900 and 800hPa. This double structure is associated with the top of the MBL (at 900hPa) and trade-wind inversion (at 800hPa). Johansson(2005) found that the double-layer structure in the MBL over the Baltic Sea mainly occurs during autumn. The structure is caused by the advection of land boundary-layer air over the convective MBL or by the development of Sc clouds in weak frontal zones connected to low pressure systems. Viher(2013) pointed out that the elevated ducts are more likely to occur over the Northern Adriatic, and Prtenjak(2015) pointed out that significant areas of elevated ducts above the Adriatic surface can occur in accordance with local bora jets and wakes. Although much related research is present, the correspondence among elevated ducts, the MBL, and marine low clouds is still uncertain.

The structure of the MBL affects the evaporation of water, the distribution of turbulence, and the marine low clouds. Meanwhile, low clouds feed back to the MBL and change the vertical distributions of temperature and moisture through long wave radiation and entrainment at the top of low clouds. Therefore, low clouds are closely related to the formation, dispersion, and location of elevated ducts (Sinclair, 2010; Laing and Evans, 2011; Alappattu, 2016; Guo, 2016). The central region of the WPO is located in the warm pool with high latent heat release (Yu and Weller, 2007) and is always controlled by the subtropical high with downwelling in winter. Turton(1988) showed that the process of subsidence inversion is more favorable to the formation of elevated ducts. Overall, it is necessary to research the relationship among marine low clouds, the MBL, and marine elevated ducts over the central WPO. The statistical characteristics of elevated ducts over the central WPO in winter are studied using ship-based and island-based GPS sounding data in the present work. Correspondences between elevated ducts and the MBL with low cloud cover are also described.

This paper is organized as follows: Data and methods used are briefly described in Section 2. The characteristics of the elevated ducts over the south-central WPO from November 2015 to January 2016 are given in Section 3. Section 4 describes the effect of the subtropical high on the elevated ducts over the central WPO in winter. Section 5 presents the statistical analysis of low clouds based on Himawari-8 satellite data. Section 6 describes the influences of low clouds on the elevated ducts. Conclusions are presented in Section 7.

2 Data and Methods

2.1 Data

2.1.1 Ship-based GPS sounding data

Ship-based GPS sounding data were collected at the three sections of the WPO from November 2015 to January 2016 (Fig.1). They are the 146˚E meridional section from 22˚N to 36˚N, 143˚E meridional section from 1˚N to 22˚N, and equatorial zonal section from 137˚E to 161˚E.

A total of 66 valid observation data sets are present. Every set includes temperature, pressure, dew point temperature, specific humidity, wind speed, and wind direction data at different heights.

Radiosondes used for the soundings were independently developed by the China Aerospace Science and Industry Corporation (CASIC) (Table 1). Temperature, pressure, dew point temperature, specific humidity, height, wind speed, and wind direction data were recorded at 1-s intervals using automated computer systems.

Fig.1 The locations of the ship-based (red marks) and island-based (pink marks) GPS sounding stations in the WPO. The green shaded area is land, and the blue shaded area is ocean.

Table 1 The specifications of the GPS radiosondes independently developed by CASIC

2.1.2 Island-based GPS sounding data

Twice daily (0000 UTC and 1200 UTC) sounding data obtained at four islands (Table 2) were provided by the Department of Atmospheric Science of the University of Wyoming (http://weather.uwyo.edu/upperair/sounding.html).

Table 2 The details of four island stations with GPS soundings

2.1.3 ERA-interim dataset

ERA-interim data were downloaded from the European Centre for Medium-Range Weather Forecasts (ECMWF) Data Server (http://apps.ecmwf.int/datasets/). ERA-interim data including meteorological parameters such as geopotential height, temperature, specific humidity, meridional wind, zonal wind, and sea surface temperature (SST) have the horizontal resolution of 0.5 degree with 37 pressure levels and provide reanalysis data at four times (0000, 0600, 1200, and 1800 UTC) every day (Dee, 2011).

2.1.4 Himawari-8 satellite data

Himawari-8, which is a new generation geostationary meteorological satellite with high time resolution and short acquisition time, was launched on October 7, 2014 and became operational on July 7, 2015. Himawari-8 is equip- ped with more advanced multispectral imagers (Advanced Himawari Imager) than other planned geostationary meteorological satellites (Bessho., 2016). The imager has 16 observation bands from visible to infrared. Level-2 products with the horizontal resolution of 0.05 degrees including cloud parameters such as cloud type and cloud top height and one-hourly temporal resolutions are used from November 2015 to January 2016 (http://www.data. jma.go.jp/mscweb/en/himawari89/space_segment/sample_hisd.html).

2.2 The Screening and Processing of Island-Based GPS Sounding Data

The structure of the atmospheric boundary layer can be classified into three types: convective boundary layer(CBL), stable boundary layer (SBL), and neutral boundary layer (NBL). Liu and Liang (2010) have proposed a method for identifying boundary structures in tropical and subtropical regions using GPS sounding data:

whereis potential temperature in Kelvin, and its subscript number denotes the data level index assuming a surface level at=1.δis theincrement for the minimum strength of the stable layer above the CBL top or below the SBL top. Generally,δis assumed to be equal to 0.2K.

Considering the lower vertical resolution of the island-based sounding data, the data for the island-based sounding whose height is the closest to the second and fifth level data of the ship-based soundings are used to identify the boundary structure through Eq. (1).

MBL stability can be calculated using Eq. (1). The potential temperature data is from ship-based and island-based GPS soundings. Because the structure of the atmospheric boundary layer is related to the characteristics of the underlying surface, the difference between the underlying surfaces of the island and ocean cannot be ignored.Guam,Yap,Chichijima,and Minami-Tori-Shimaare544km2, 100.20km2, 24.00km2, and 1.52km2, respectively. Differences between them and the ship-based sounding data should be known. Accordingly, the stability of the boundary layer over marine underlying surface (from ship-based GPS sounding data) and island underlying surface (from island-based GPS sounding data) is compared (Table 3).

Table 3 The statistics of boundary layer stability from ship-based and island-based GPS soundings from November 2015 to January 2016

Table 3 shows that a significant difference is present between the boundary layer structure over the marine and island underlying surfaces during nighttime. The SST (see Fig.2) in the south-central region of the WPO is always higher than the air temperature (not shown). Radiational cooling during nighttime has little effect on the surface temperature owing to the large specific heat capacity of the ocean, but it can reduce the island surface temperature. Therefore, the thermal effect of the island underlying surface on the atmosphere is mainly reflected at night. During nighttime, the atmosphere above the ocean surface can remain NBL or occasionally becomes SBL whereas the atmosphere above the islands often becomes SBL. The larger the island is, the higher the occurrence probability of SBL. Even at Minami-Tori-Shima, the smallest island, there is a 74.74% probability of SBL occurrence during night.

Conversely, the boundary layer structure over the marine underlying surface is highly similar to that over the island underlying surface at daytime. The probability is highest when the boundary layer is NBL, which is followed by CBL and SBL. Kim and Yum (2011) obtained similar results. The types of MBL over the west coast of Korea are mainly NBL and CBL through the statistics of the sounding data. During the daytime, due to the sufficient solar radiation reaching the underlying surface, the formation of the SBL is difficult (Chen and Houze, 1997). Although a positive heat flux is present over marine and island underlying surfaces, the clouds and subtropical high over the study region would inhibit the exchange of heat and mass between the boundary layer and free atmosphere (Medeiros, 2005; Zhang, 2017). Turbulence and entrainment can make it easier to mix in the boundary layer and form NBL.

Considering boundary layer changes caused by differences in underlying surfaces, only the daytime island-based GPS sounding data were used to analyze the elevated duct in the present study.

Fig.2 The spatial distribution of average sea surface temperature from November 1, 2015 to January 31, 2016. The solid points and five-star marks are the ship-based and island-based GPS sounding stations, respectively.

2.3 The Determination of Lifting Condensation Level

Lifting condensation level (LCL) is formally defined as the height at which the relative humidity of an air parcel will reach 100% with respect to liquid water when it is cooled by dry adiabatic lifting. Assuming that the lapse rate for dry adiabatic lifting is 9.8Kkm−1, and the lapse rate of dew point is 1.8Kkm−1, the LCL can be calculated by surface meteorological parameters:

wheresurfaceis surface level temperature in Kelvin,surfaceis surface level dew point temperature in Kelvin, andsurfaceis surface level altitude in meters.

2.4 The Determination of the Ceiling Height of the Marine Mixed Layer

Liu(2010) proposed a method for discriminating the ceiling height of the marine mixed layer (mix):

whereδis theincrement for the minimum strength of the unstable layer,is thevertical gradient per heightandis its minimum strength for the overlying inversion layer. Here,can be considered as the overshooting threshold of the rising parcel. The value ofδandare 0.1K and 0.5Kkm−1, respectively. The lowest level=that meets Eqs. (3), (4), and (5) can be found, and the altitude of k-level data ismix.

2.5 The Determination of the Elevated Duct

The atmospheric duct in the troposphere is determined by the vertical gradients of modified refractivity (Bean and Dutton, 1968):

whereis the modified refractivity in-unit,is the temperature in Kelvin,is the atmospheric pressure in hectopascals,is the water vapor pressure in hectopascals, andis the height above sea level in meters.

In general,increases asincreases. When an atmospheric duct occurs,decreases asincreases. Therefore, the atmospheric duct is judged by

According to the verticalprofile and the position of the atmospheric duct, the atmospheric duct can be classified as evaporation duct, surface duct, and elevated duct. A diagram of the structure of an elevated duct is shown in Fig.3. Theat the top of elevated duct’s trapping layer is greater than that at the sea surface. If there are multilayer elevated ducts occurring at the same time over one site, it is called a complex duct.

To eliminate the negative effects of atmospheric turbulence and instrumental noise, the verticalprofile is smoothed using irregularly spaced five-point cubic polynomials (Cheng, 2016).

Numerous weak and thin elevated ducts, which cannot effectively represent stable atmospheric structures, may be obtained in verticalprofiles. Therefore, the maximum cut-off wavelength (maxin meters) is quoted (Zhu and Atkinson, 2005):

where the coefficient=5.66×10−3is for an elevated duct,is the thickness of the duct in meters, and δis the difference of thethrough the trapping layer in- units (Fig.3).

Fig.3 A structural diagram of an elevated duct and its characteristics.

maxis the maximum wavelength of electromagnetic waves trapped by the elevated duct. The larger themaxis, the stronger the elevated duct is. In the present research, an elevated duct is effective whenmax≥1.

3 The Characteristics of Elevated Ducts

The structure of elevated ducts is mainly characterized by base height,, and δ(Zhao., 2013). All the ship-based GPS sounding data from three sections in the WPO were used to calculate elevated ducts. The result showed that the occurrence probability was 47%. The mean base height,, δ, andmaxof the elevated ducts were 1188m, 205m, 11.7-units, and 2.7m, respectively. Complex ducts accounted for 26% of the elevated ducts.

The research area was divided into an equatorial region (0˚–5˚N) and a non-equatorial region (5˚–30˚N) to compare the difference in elevated ducts. Over the equatorial region, the elevated ducts’ occurrence probability was 22%.The mean base height,, δ, andmaxof the elevated ducts were 1104m, 174m, 9.4-units, and 2.0m, respectively. Only one complex duct was present, which accounted for 13% of the elevated ducts.

Outside the equatorial region, the elevated ducts’ probability was 77%. The mean base height,, δ, andmaxof the elevated ducts were 1212m, 214m, 12.3-units, and 2.9m, respectively. There were seven complex ducts, which accounted for 30% of the elevated ducts (Table 4).

Comparatively, the occurrence probability and strengths of the elevated ducts over the central region of the WPO were far larger than those over the equatorial region of the WPO. Hence, the rest of this analysis focused on this region.

Island GPS sounding data (see in Table 2) at 0000 UTC for each day from November 2015 to January 2016 were also used to calculate elevated ducts. Statistic results (Table 5) indicate that elevated ducts occur frequently. Occurrence probability was 78% in Guam, and mean base height,, δ, andmaxof the elevated ducts were 1675m, 233m, 14.3-units, and 3.5m, respectively, and the complex ducts accounted for 48% of the elevated ducts. Guam is the nearest island to the ship-based GPS sounding stations (see in Fig.1). Statistical results at Guam were virtually the same as those from the ship-based GPS soundings.

Table 4 The statistical characteristics of elevated ducts based on ship-based GPS soundings for the period from November 2015 to January 2016

Table 5 The statistical characteristics of elevated ducts based on island-based GPS soundings for the period from November 2015 to January 2016

4 Weather Conditions over the Elevated Duct Region

Both the ship-based and island-based GPS sounding data show that temperature increased with increasing height and humidity decreased rapidly with increasing height in the trapping layer when elevated ducts occurred. This phenomenon is consistent with previous results (Hermann., 2002).

To explore the relationship between elevated ducts and the synoptic situation over the central WPO, the ERA-interim dataset from November 1, 2015 to January 31, 2016 was used to analyze circulation patterns in upper layers.

From the 500-hPa level weather map (Fig.4a), the WPO was covered by the subtropical high (see the 588 geopo-tential decameter isopleth in Fig.4a) with an extremely low specific humidity air mass. Along the 16˚N meridional profile (Fig.4b), a weak downdraft was present inside the subtropical high, and the downward motion caused the lower-level divergence. Specific humidity was low in the upper atmosphere and high in the MBL, which means the upper dry-lower wet structure of the marine atmosphere. Specific humidity isopleths were dense and the humidity gradient was strong near the interface (about 800hPa) between dry and wet air masses.

A reasonable deduction is that the slowly sinking dry air mass interacts with the wet marine boundary air mass, and the strong relative humidity gradient and subsidence inversion (see in Fig.5) are formed within the region of the subtropical high over the WPO in winter, leading to the occurrence of elevated ducts.

Fig.4 (a) The spatial distribution of average specific humidity, geopotential height, and wind field at 500-hPa level from November 1, 2015 to January 31, 2016. (b) The meridional profile of average vertical velocity and specific humidity along 16˚N from November 1, 2015 to January 31, 2016. The area in the black bold solid bordered rectangle is the focus area for cloud analysis. The shaded areas are (a) average specific humidity in grams per kilogram and (b) average vertical velocity (negative values indicate descending motion) in meters per second. The contour lines are (a) average geopotential height in geopotential decameters and (b) average specific humidity in grams per kilogram. The vectors are average wind field in meters per second and the solid points are the ship-based GPS sounding stations. The five-star marks are island-based GPS sounding stations. The dotted line is the schematic of the section position.

Fig.5 The vertical profiles of (a) specific humidity, (b) temperature, and (c) M observed by the ship-based GPS sounding at 1919 UTC, November 20, 2015. The station is located at 23˚N and 146˚E. The shaded area is the region where the elevated duct occurs.

5 The Characteristics of Clouds over the Elevated Duct Region

Cloud is an important indicator of boundary layer structure. The cloud type and top height in daytime (2300, 0000, 0100, 0200, 0300, 0400, and 0500 UTC) from November 1, 2015 to January 31, 2016 from Himawari-8 satellite were used to explore the relationship between clouds and elevated ducts. The scope of the data area is the subtropical high-controlled area (5˚–30˚N, 130˚–160˚E, see the black bold solid bordered rectangle in Fig.4a). This scope covers ship-based and island stations.

The central WPO has the highest probability of cloudlessness, followed by Cu, Sc, and Stratus (St) with the top heights of approximately 2000m (Fig.6). The occurrence probabilities of cloudless and low clouds are 30.74% and 25.75%, respectively.

Fig.7 shows the spatial distribution of cloud types over the WPO central region. The mixture of Ci, Ac, and Sc was banded with the cloudless sky.

The time series of low cloud probability and its average top height in daytime from November 1, 2015 to January 31, 2016 are shown in Fig.8. It was observed that the occurrence probability of cloudless, Cu, Sc, St, and their average top heights did not exhibit strong volatility and trend changes during the statistical period. Meanwhile the occurrence probability of cloudless, Cu, Sc, St, and their average top heights at different times on the same day also kept consistent.

6 The Influences of Low Clouds on Elevated Ducts

The number of daytime ship-based data in the region selected for cloud analysis is too small to research the correspondence among elevated ducts, the MBL, and low clouds. Hence,mix, LCL, the based height of elevated duct, and other physical parameters calculated from the four is- lands’ sounding data were analyzed from Himawari-8 satellite data.

Fig.6 The occurrence probability histogram of (a) cloud type and (b) its cloud top height in daytime from November 1, 2015 to January 31, 2016 over 5˚–30˚N, 130˚–160˚E regions.

Fig.7 The spatial distribution of clouds over 5˚–30˚N, 130˚–160˚E regions at 0000 UTC, December 26, 2015. The shaded area is cloud type. The contour lines are geopotential height in geopotential decameters. The vectors are wind field in meters per second.

Fig.8 The time series of the occurrence probability of (a) cloudless, (b) Cu, (d) Sc, and (f) St over 5˚–30˚N, 130˚–160˚E regions in daytime from November 1, 2015 to January 31, 2016. The time series of the average cloud top height of (c) Cu, (e) Sc, and (g) St over 5˚–30˚N, 130˚–160˚E regions in daytime from November 1, 2015 to January 31, 2016.

6.1 Under Cloudless Conditions

The time series ofmixand the base heights of elevated ducts at Guam, Yap, Chichijima, and Minami-Tori-Shima under cloudless skies are shown in Fig.9. The figure showsthat the elevated ducts could occur near themix. It was as- sumed that the averageof the elevated ducts over each island (see Table 5) is the diagnosis threshold. The elevated duct with a base height betweenmix–andmix+was regarded as the elevated duct near themix(see the red dots inside the yellow shaded area in Fig.9).

Fig.9 The time series of elevated duct base height and hmix over (a) Guam, (b) Yap, (c) Chichijima, and (d) Minami-Tori-Shima islands under cloudless skies from November 1, 2015 to January 31, 2016. The black solid lines are hmix. The red dot marks are the base heights of elevated ducts. The shaded area is the height range between hmix–d and hmix+d.

The occurrence probability of elevated ducts near themixwas 14%. The mean base height,, δ, andmaxof the elevated ducts were 730m, 219m, 9.8-units, and 2.6m, respectively.

The averagemixand LCL were 805m and 792m, respectively. In winter, SST is higher than the air temperature in the central region of the WPO. The latent heat flux drives the mixing of the boundary layer. Moreover, this region is mainly controlled by the subtropical high. Lo- wer-level divergence makes themixapproximate the LCL, which is not conducive to water vapor condensation.

Overall, the structural feature of the MBL under cloudless skies is the typical mixed boundary layer. We used the observation at Yap island as an example to illustrate this structure (Fig.10). Themixis low. The distributions of both temperature and humidity are uniform in the mixed boundary layer. Wind speed basically increases linearly with height. A temperature inversion and humidity gradient are present at themix, leading to the occurrence of elevated ducts.

6.2 Under Cu Cloud Conditions

The time series ofmix, the base height of elevated duct, and cloud top height (Cu) at Guam, Yap, Chichijima, and Minami-Tori-Shima under Cu cloud conditions are shown in Fig.11.

Equally, it was assumed that the averageover each island is the diagnosis threshold. The elevated duct with the base height betweenCu–andCu+was regarded as the elevated duct near the top height of the Cu cloud layer (see the red dots inside the yellow shaded area in Fig.11).

The elevated ducts occurred near the cloud top height with a 24% probability. The mean base height,, δ, andmaxof the elevated ducts were 1471m, 213m, 10.9-units, and 2.7m, respectively.

Fig.10 The vertical profiles of (a) M, (b) potential temperature, (c) specific humidity, (d) wind speed, and (e) wind direction over Yap at 0000 UTC, November 11, 2015. The black solid lines are hmix. The shaded area is the region where the elevated duct occurs.

Fig.11 The time series of elevated duct base height, hmix, and hCu over (a) Guam, (b) Yap, (c) Chichijima, and (d) Minami-Tori-Shima when Cu clouds were present from November 1, 2015 to January 31, 2016. The black solid lines are hmix. The brown dashed lines are hCu. The red dot marks are the base heights of elevated ducts. The shaded area is the height range between hCu–d and hCu+d.

The averagemixand LCL were 874m and 764m, respectively. The latent heat release from the WPO central region in winter provides enough water vapor in the marine mixed layer. When themixis much higher than the LCL, the air reaches saturation and the water vapor starts to condense above the LCL. Finally, it becomes the cumulus convection.

The structural feature of the MBL under Cu cloud conditions is similar to that under cloudless conditions (not shown), but the cloudless skies above Cu clouds controlled by the subtropical high is very advantageous to long wave radiation at the low cloud top. The inhibition of subsidence in the subtropical high and the long wave radiation at the cloud top make the temperature inversion stronger and the structure of the MBL more stable. Compared with elevated ducts under cloudless skies, elevated ducts near Cu cloud tops occur more frequently.

6.3 Under Sc Cloud Conditions

The time series of the elevated ducts base height,mix, and cloud top height (Sc) over Guam, Yap, Chichijima, and Minami-Tori-Shima under Sc clouds are shown in Fig.12.

Similarly, it was assumed that the averageof the elevated duct over each island is the diagnosis threshold. The elevated duct with the base height betweenSc–andSc+was regarded as the elevated duct near the cloud top height of Sc clouds (see the red dots inside the yellow shaded area in Fig.12).

The elevated ducts occurred near the cloud top height with a 22% probability. The mean base height,, δ, andmaxof the elevated ducts were 1511m, 213m, 12.5-units, and 3.1m, respectively.

There are two situations of Sc clouds. Some Sc cloud tops are close to themix, and some are higher than themix. These two situations are discussed separately below.

6.3.1 Coupled Sc clouds (St clouds)

The MBL structure is shown in Fig.13 when Sc cloud tops are close to themix. The top layer of Sc clouds gains negative buoyancy due to long wave radiation and entrainment. The interaction between this negative buoyancy and the positive buoyancy produced by the latent heat of the sea surface cause the heat and water vapor in the MBL to mix well from the sea surface to the Sc cloud top height. The Sc clouds are coupled. The entrainment interfacial layer atop the Sc clouds forms a strong temperature inversion and humidity gradient, resulting in the strong elevated ducts near the cloud top height.

The MBL structure under St clouds is similar to that under Sc clouds (not shown). Strong elevated ducts also occur near the cloud top height. However, the occurrence probability of St clouds over the central region of the WPO in winter is quite low (see in Fig.6a), so there are no more descriptions of St clouds.

6.3.2 Decoupled Sc clouds (St clouds)

Wyant(1997) indicated that the increase of SST makes Sc clouds deepening and warming. The deepening and warming continuously lead to stronger entrainment at the top of Sc clouds. In this way, it is difficult for long wave radiation and entrainment at the top of Sc clouds to mix the whole Sc cloud boundary layer uniformly, and theSc clouds become decoupled. Nicholls and Leighton (1986) suggested that besides the mixed layer that is inside the Sc clouds, the decoupled Sc clouds also have a typical surfacemixed layer caused by the sea surface latent heat. There is a subcloud layer between these two mixed layers.

Fig.12 The time series of elevated duct base height, hmix, and hSc over (a) Guam, (b) Yap, (c) Chichijima, and (d) Minami-Tori-Shima under Sc clouds from November 1 to January 31, 2016. The black solid lines are hmix. The brown dashed lines are hSc. The red dot marks are the base height of the elevated duct. The shaded area is the height range between hSc–d and hSc+d.

Fig.13 The vertical profiles of (a) M, (b) potential temperature, (c) specific humidity, (d) wind speed, and (e) wind direction over Guam at 0000 UTC, January 12, 2016. The brown dashed lines are hSc, and the shaded area is the elevated duct region.

St clouds are always deepening, warming, and decoupling over the warm ocean. When the Sc cloud top is much higher than the marine mixed layer (decoupled Sc clouds), the structural feature of the MBL is as shown in Fig.14. The temperature inversion and humidity gradient caused by the entrainment interfacial layer atop Sc clouds still exist, and the elevated duct occurs near the Sc cloud top height (See in Figs.14a–e). However, there may be weak temperature inversion and humidity gradient between the surface mixed layer and subcloud layer, leading to a weak elevated duct near themixbesides the top height of Sc clouds.

Fig.14 The vertical profiles of (a) M, (b) potential temperature, (c) specific humidity, (d) wind speed, and (e) wind direction over Minami-Tori-Shima at 0000 UTC, December 26, 2015. The vertical profiles of (f) M, (g) potential temperature, (h) specific humidity, (i) wind speed, and (f) wind direction over Guam at 0000 UTC, January 31, 2016. The brown dashed lines are hSc. The black solid lines are hmix. The shaded area is the region where the elevated duct occurs.

7 Conclusions

In this paper, the distribution characteristics of elevated ducts, low clouds, and the MBL structure over the central region of the WPO in winter are studied.

Elevated ducts frequently occur in the central region of the WPO. The occurrence probability, mean base height,, and δof elevated ducts detected by ship-based GPS soundings are 77%, 1212m, 214m, and 12.3-units, respectively. The occurrence probability, mean base height,, and δof elevated ducts obtained by four island-based GPS soundings are 62%, 1736m, 224m, and 13.2-units, respectively.

Under the control of the subtropical high, a slowly sinking dry air mass resides above the wet MBL’s air mass, and the humidity gradient and subsidence inversion are formed at the top of the MBL, thereby leading to the occurrence of elevated ducts.

The occurrence probabilities of cloudless conditions and low clouds are 30.74% and 25.75%, respectively. The feedback mechanisms between the MBL and lows clouds affect the generation and disappearance of elevated ducts. Under cloudless skies, elevated ducts are able to occur nearmix(17% probability). The mean base height,, and δof elevated ducts are 703m, 219m, and 9.8-units, respectively. When there are low clouds, elevated ducts occur near the cloud top with the average height of 2116m. The strength of elevated ducts near the top of Sc and St clouds is greater than that near the top of Cu clouds. For Cu clouds, the occurrence probability, mean base height, and δof the elevated ducts are 14%, 730m, and 10.9-units, respectively. For Sc clouds, they are 22%, 1511m, and 12.5-units, respectively.

If there are no clouds, the structural feature of the MBL is the typical mixed boundary layer and is similar to that under Cu clouds. When Sc clouds are coupled, long wave radiation and entrainment atop Sc clouds form a strong temperature inversion and humidity gradient, which results in strong elevated ducts. When Sc clouds are decoupled, there may be weak temperature inversion and humidity gradient between the surface mixed layer and subcloud layer, leading to a weak elevated duct near themixbesides the top height of Sc clouds.

Acknowledgements

This research work is supported by the National Natural Science Foundation of China (No. 41975008). The authors would like to thank the Department of Atmospheric Science of the University of Wyoming for providing island-based GPS sounding data. We are also grateful to the European Centre for Medium-Range Weather Forecasts (ECMWF) and the Meteorological Satellite Center (MSC) of the Japan Meteorological Agency (JMA) for providing ERA-interim dataset and Himawari-8 satellite data.

Alappattu, D. P., Wang, Q., and Kalogiros, J., 2016. Anomalous propagation conditions over eastern Pacific Ocean derived from MAGIC data., 51 (7): 1142-1156, https://doi.org/10.1002/2016RS005994.

Bean, B. R., and Dutton, E. J., 1968.. Dover Publication, New York, 435pp.

Bessho, K., Date, K., Hayashi, M., Ikeda, A., Inoue, H., Kumagai, Y., Miyakawa, T., Murata, H., Ohno, T., Okuyama, A., Oyama, R., Sasaki, Y., Shimazu, Y., Shimoji, K., and Sumida, Y., 2016. An introduction to Himawari-8/9–Japan’s new-generation geostationary meteorological satellites., 94 (2): 151-183, https://doi.org/10.2151/jmsj.2016-009.

Carrillo, J., Guerra, J. C., Cuevas, E., and Barrancos, J., 2015. Characterization of the marine boundary layer and the trade-wind inversion over the sub-tropical North Atlantic., 158 (2): 1-20, https://doi.org/10.1007/s10546-015-0081-1.

Chen, S. S., and Houze, R. A., 1997. Interannual variability of deep convection over the tropical warm pool., 102 (22): 25783-25795, https://doi.org/10.1029/97JD02238.

Cheng, Y. H., Zhou, S. Q., Wang, D. X., Lu, Y. Z., Huang, K., Yao, J. L., and You, X. B., 2016. Observed characteristics of atmospheric ducts over the South China Sea in autumn., 34 (3): 1-10, https://doi.org/10.1007/s00343-016-4275-2.

Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, P., Kobayashi, S., Andrae, U., Balmaseda, G., Bauer, P., Bechtold, P., Beljaars, A. C. M., van de Berg, L., Bidlot, J., Bormann, N., Delsol, C., Dragani, R., Fuentes, M., Geer, A. J., Haimberger, L., Healy, S. B., Hersbach, H., Hólm, E. V., Isaksen, L., Kållberg, P., Köhler, M., Matricardi, M., McNally, A. P., Monge-Sanz, B. M., Morcrette, J. J., Park, B. K., Peubey, C., de Rosnay, P., Tavolato, C., Thépaut, J. N., and Vitart, F., 2011. The ERA-Interim reanalysis: Configuration and performance of the data assimilation system., 137 (656): 553-597, https://doi.org/10.1002/qj.828.

Dinc, E., and Akan, O., 2014. Beyond-line-of-sight communications with ducting layer., 52 (10): 37-43, https://doi.org/10.1109/MCOM.2014.6917399.

Ding, J. L., Fei, J. F., Huang, X. G., and Cheng, X. P., 2013. Observational occurrence of tropical cyclone ducts from GPS dropsonde data., 52 (5): 1221-1236, https://doi.org/10.1175/JAMC-D-11-0256.1.

Freehafer, J. E., and Kerr, D. E., 1988.. Peninsula Publishing, Gallipoli, Turkey, 162-163.

Guo, X. M., Zhang, D. L., Hao, X. J., Lin, L., Zhang, Y. S., and Kang, S. F., 2016.. Gulin, 213-215.

Haack, T., Wang, C., Garrett, S., Glazer, A., Mailhot, J., and Marshall, R., 2010. Mesoscale modeling of boundary layer refractivity and atmospheric ducting., 49 (12): 2437-2457, https://doi.org/10.1175/2010JAMC2415.1.

Hermann, J. A., Kulessa, A. S., Lucas, C., Vincent, R. A., Hacker, J. M., and Ewenz, C. M., 2002. Impact of elevated atmospheric structures upon radio-refractivity and propagation.. Leura, Australia, 20-22.

Horn, G. L., Ouwersloot, H. G., Arellano, J. V. G. D., and Sikma, M., 2015. Cloud shading effects on characteristic boundary-layer length scales., 157 (2): 237-263, https://doi.org/10.1007/s10546-015-0054-4.

Johansson, C., Hennemuth, B., Bösenberg, J., Linné, H., and Smedman, A. S., 2005. Double-layer structure in the boundary layer over the Baltic Sea., 114 (2): 389-412, https://doi.org/10.1007/s10546-004-1671-5.

Katzwinkel, J., Siebert, H., and Shaw, R. A., 2012. Observation of a self-limiting, shear-induced turbulent inversion layer above marine stratocumulus., 145 (1): 131-143, https://doi.org/10.1007/s10546-011-9683-4.

Kim, C. K., and Yum, S. S., 2011. Marine boundary layer structure for the sea fog formation off the west coast of the Korean Peninsula., 169 (5-6): 1121-1135, https://doi.org/10.1007/s00024-011-0325-z.

Laing, A., and Evans, J. L., 2011. An Introduction to Tropical Meteorology. In:Second edition,University Corporation for Atmospheric Research, Boulder, Colo.

Liu, S. Y., and Liang, X. Z., 2010. Observed diurnal cycle climatology of planetary boundary layer height., 23 (21): 5790-5809, https://doi.org/10.1175/2010JCLI3552.1.

Lopez, P., 2009. A 5-yr 40-km-resolution global climatology of superrefraction for ground-based weather radars., 48 (1): 89-110, https://doi.org/10.1175/2008jamc1961.1.

MacKellar, M. C., McGowan, H. A., Phinn, S. R., and Soderholm, J. S., 2013. Observations of surface energy fluxes and boundary-layer structure over Heron Reef, Great Barrier Reef, Australia., 146 (2): 319-340, https://doi.org/10.1007/s10546-012-9767-9.

Manjula, G., Raman, M. R., Ratnam, M. V., Chandrasekhar, A. V., and Rao, S. V. B., 2016. Diurnal variation of ducts observed over a tropical station, Gadanki, using high resolution GPS radiosonde observations., 51 (4): 247-258, https://doi.org/10.1002/2015RS005814.

Medeiros, B., Hall, A., and Stevens, B., 2005. What controls the mean depth of the PBL?, 18 (16): 3157-3172, https://doi.org/10.1175/JCLI3417.1.

Nicholls, S., and Leighton, J., 1986. An observational study of the structure of stratiform cloud sheets: Part I. Structure., 112 (472): 431-460, https://doi.org/10.1002/qj.49711247209.

Prtenjak,, M. T., Horvat, I., Tomažić, I., Kvakić, M., Viher, M., and Grisogono, B., 2015. Impact of mesoscale meteorological processes on anomalous radar propagation conditions over the northern Adriatic area., 120: 8759-8782, https://doi.org/10.1002/2014JD022626.

Schubert, W. H., Ciesielski, P. E., Lu, C., and Johnson, R. H., 1995. Dynamical adjustment of the trade wind inversion layer., 52 (16): 2941-2952, https://doi.org/10.1175/1520-0469(1995)0522.0.CO;2.

Shupe, M. D., Persson, P. O. G., Brooks, I. M., Tjernström, M., Sedlar, J., Mauritsen, T., Sjogren, S., and Leck, C., 2013. Cloud and boundary layer interactions over the Arctic sea ice in late summer., 13 (18): 9379-9399, https://doi.org/10.5194/acpd-13-13191-2013.

Sinclair, V. A., Belcher, S. E., and Gray, S. L., 2010. Synoptic controls on boundary-layer characteristics., 134 (3): 387-409, https://doi.org.10.1007/s10546-009-9455-6.

Song, X. Z., and Yu, L. S., 2013. How much net surface heat flux should go into the Western Pacific Warm Pool?, 118 (7): 3569-3585, https://doi.org/10.1002/jgrc.20246.

Stevens, B., 2005. Atmospheric moist convection.,33: 605-643, https://doi.org/10.1146/annurev.earth.33.092203.122658.

Stevens, B., Moeng, C. H., and Sullivan, P. P., 1999. Large-eddy simulations of radiatively driven convection: Sensitivities to the representation of small scales., 56 (23): 3963-3984, https://doi.org/10.1175/1520-0469(1999)056<3963:lesord>2.0.co;2.

Turton, J. D., Bennetts, D. A., and Farmer, S. F. G., 1988. An introduction to radio ducting., 117: 245-254.

Viher, M., Prtenjak, M. T., and Grisogono, B., 2013. A multi-year study of the anomalous propagation conditions along the coast of the Adriatic Sea., 97: 75-84, https://doi.org/10.1016/j.jastp.2013.01.014.

Wang, Q., Wendt, R. T., Wang, S., Cherrett, R. C., Rogers, T., and Yardim, C., 2016. Variability of index of refraction from Large Eddy Simulations.. Fajardo, Puerto Rico, 109-110.

Wood, R., 2012. Stratocumulus clouds., 140 (8): 2373-2423, https://doi.org/10.1175/MWR-D-11-00121.1.

Wyant, M. C., Bretherton, C. S., Rand, H. A., and Stevens, D. E., 1997. Numerical simulations and a conceptual model of the stratocumulus to trade cumulus transition., 54 (1): 168-192, https://doi.org/10.1175/1520-0469(1997)0542.0.CO;2.

Yu, L., and Weller, R. A., 2007. Objectively analyzed air-sea heat fluxes for the global ice-free oceans (1981–2005)., 88 (4): 527-540, https://doi.org/10.1175/bams-88-4-527.

Zhang, W. C., Guo, J. P., Miao, Y. C., Liu, H., Song, Y., Fang, Z., He, J., Lou, M. Y., Yan, Y., Li, Y., and Zhai, P. M., 2017. On the summer-time planetary boundary layer with different thermodynamic stability in China: A radiosonde perspective., 31 (4): 1451-1465, https://doi.org/10.1175/jcli-d-17-0231.1.

Zhao, X. F., Wang, D. X., Huang, S. X., Huang, K., and Chen, J., 2013. Statistical estimations of atmospheric duct over the South China Sea and the tropical eastern Indian Ocean., 58 (23): 2794-2797, https://doi.org/10.1007/s11434-013-5942-8.

Zhu, M., and Atkinson, B. W., 2005. Simulated climatology of atmospheric ducts over the Persian Gulf., 115 (3): 433-452, https://doi.org/10.1007/s10546-004-1428-1.

Ziemba, D. A., 2013. Ducting conditions for electromagnetic wave propagation in tropical disturbances from GPS dropsonde data. Master thesis. Department of Meteorology, Naval Postgraduate School.

February 28, 2020;

September 1, 2020;

December 14, 2020

© Ocean University of China, Science Press and Springer-Verlag GmbH Germany 2021

. E-mail: shenglf@ouc.edu.cn

(Edited by Xie Jun)