Surface atmospheric duct over Svalbard, Arctic, related to atmospheric and ocean conditions in winter

ABSTRACT Atmospheric ducts could disturb radar detection and radio communication in the Arctic and risk increasing human activities under Arctic warming. It is an urgent need to understand atmospheric ducts in the Arctic, with distinct atmospheric systems and environment. This study attempts to analyze the surface atmospheric ducts over Svalbard, as a critical region in the Arctic Ocean, and discuss the relation to atmospheric and oceanic conditions, based on observation and reanalysis data. The results show an occurrence of 12.6 percent, mean strength of 1.30 MU, and mean depth of 13 m for the surface duct in Svalbard winter from December 2018 to February 2020. The surface duct is characterized as two types: T-type ducts mainly constructed by temperature inversion and H-type ducts constructed by a humidity gradient. The T-type duct dominates the surface duct in Svalbard winter, with an appearance percentage of 80 percent, indicating an importance of temperature inversion in surface duct construction. The surface duct is closely related to atmospheric and oceanic conditions in Svalbard winter, because the T-type duct occurs in colder, drier, and northeasterly weather and the H-type duct occurs in warmer, wetter, and southeasterly weather with higher sea surface temperature.


Introduction
Atmospheric ducts are refractive layers in the atmosphere, trapping and guiding radio wave propagation along the curvature of Earth. Atmospheric ducts can greatly disturb radar detection with mispositioning and blind spots and also affect radio communication, resulting in low accuracy and efficiency (Mosczkowicz, Ciach, and Krajewski 1994;Borsum 1995;Mesnard and Sauvageot 2010). Surface ducts, an important category of atmospheric ducts, occurring in an atmospheric layer of 10 1 to 10 2 m thick on the ground can have great impacts on ground-based radar detection and radio communication (Mosczkowicz, Ciach, and Krajewski 1994;Borsum 1995;Brooks, Goroch, and Rogers 1999;Atkinson and Zhu 2006;Mesnard and Sauvageot 2010). Surface ducts in various regions worldwide have been detected and studied (Babin and Rowland 1992;Babin 1996;Brooks, Goroch, and Rogers 1999;Mentes and Kaymaz 2007;Mesnard and Sauvageot 2010;Viher, Prtenjak, and Grisogono 2013;Cheng et al. 2015;Kaissassou et al. 2015;Mai, Sheng, and Shi 2020). For example, Brooks, Goroch, and Rogers (1999) captured a thick and strong surface duct over the Persian Gulf in April 1996, with a depth up to 420 m and strength over 70 MU. Cheng et al. (2015) studied the surface duct over the South China Sea with more than ten years of radiosondes data onboard and found annual and winter mean occurrences up to 64 percent and 50 percent, respectively.
The Arctic is experiencing fast warming, with a warming rate more than twice the global mean, and this warming will continue in the twenty-first century (Symon, Arris, and Heal 2004;Berkryaev, Polyakov, and Alexeev 2010;Cohen et al. 2013Cohen et al. , 2014; Intergovernmental Panel on Climate Change 2013; Shepherd 2016; Davy and Outten 2020;Cai, You, and Wu 2021). As a consequence of Arctic warming, sea ice is melting rapidly in the Arctic Ocean, especially in coastal areas (Cosmiso, Parkinson, and Gersten 2008;Screen and Simmonds 2010;Gao et al. 2015;Arthun et al. 2021;Cai, Wang, and Beletsky 2021), resulting in more efficient and navigable Arctic shipping routes (Barnhart, Miller, and Overeem 2015;Chen et al. 2019;Yu et al. 2020;Cao et al. 2021). For instance, Yu et al. (2020) analyzed the sea ice conditions along the Northeast Passage by remote sensing data and found increased navigable days, especially in the Kara Strait, where the navigable days increased from less than 50 days/year before 1990 to more than 100 days/year after 2010. Barnhart, Miller, and Overeem (2015) estimated the navigable area in the Arctic Ocean for next century based on a numerical model and found a six-month navigability on all coastal waters by 2070. Melia, Haines, and Hawkins (2016) predicted trans-Arctic shipping routes in the twenty-first century based on a climate model and found the possibility of available routes across the central Arctic Ocean by mid-century.
The foreseen increasing navigability in the Arctic Ocean requires more attention to navigation safety, especially sea ice detection and emergence supported by radar navigation and radio communication (Cao et al. 2021;Li, Ringsberg, and Rita 2021). Surface atmospheric ducts over the Arctic Ocean, as an interference in radar and radio operation in Arctic navigation, is important to understand. Studies have been carried out on atmospheric ducts over the Arctic and provided evidence and general knowledge of atmospheric ducts (Engeln and Teixeira 2004;Lopez 2009;Hao, Li, and Guo 2018;Mai, Sheng, and Shi 2020). For example, Engeln and Teixeira (2004) and Lopez (2009) studied the duct distribution worldwide based on data analysis and found a winter occurrence of over 40 percent for all duct categories over Greenland and Siberia. Mai, Sheng, and Shi (2020) investigated atmospheric ducts over the Alaska based on twelve-station radiosonde observations and found a surface duct occurrence of about 10 percent on an annual basis and only 5 percent in winter.
The Svalbard Islands (78°13′ N, 15°39′ E) are an important Arctic archipelago, located in the North Atlantic sector of the Arctic Ocean, nearby the gateway of the Northeast Passage. Svalbard and its neighboring seas are a critical region in Arctic warming, with the highest warming rate and sea surface temperature (SST) in the Arctic and complicated regional atmospheric circulation systems (Screen and Simmonds 2010;Serreze, Barrett, and Cassano 2011;Dai et al. 2019;Tomczyk, Lupikasza, and Kendzierski 2019;Yamanouchi 2019). The atmospheric and ocean conditions in this region can cause atmospheric ducts different from those in other arctic regions. However, atmospheric ducts, especially surface ducts with more impact on navigation, over this region have not been studied, and their characteristics and possible mechanisms still remain unclear. This study attempts to analyze and discuss the features of the surface atmospheric duct over Svalbard and its neighboring seas and its relation to the atmospheric and ocean conditions based on observation and reanalysis data.

Data and methods
Observational radiosonde, cloud, and radiation data at 11.9° E), located on the coast of Kings Bay connected to the Greenland Sea, Svalbard, were used in this study. The surrounding sea surface is not enclosed by sea ice, with an average ice density about 0.38 in winter (Reynolds et al. 2002;Huang et al. 2021). The radiosonde data in winter months (December, January, and February) from December 2017 to February 2020 had 421 observed profiles of atmospheric pressure, temperature, relative humidity, and wind (Maturilli 2018a), with a vertical resolution of about 10 m, once a day at 12:00 UTC from December 2017 to February 2020, except in February 2018 and February 2020 when it was four times a day at 00:00, 06:00, 12:00 and 18:00 UTC. The downward/upward longwave radiation measured by a pyranometer at 2 m above the ground and the cloud base height measured by a ceilometer, at the same time as radiosonde observation, was also used in this study (Maturilli 2018b(Maturilli , 2020. ERA-5 reanalysis products from ECMWF (European Centre for Medium-Range Weather Forecasts), (Hersbach et al. 2020), with a horizontal resolution of 0.25°×0.25°, at the same time as radiosonde observation, were applied for atmospheric situation analysis. The average SST around Svalbard (11.25-11.75° E, 78.75-79.0° N) was calculated from the NOAA Optimum Interpolation Sea Surface Temperature V2 (NOAA_OI_SST_V2) with a horizontal resolution of 0.25°×0.25° for ocean condition analysis (Reynolds et al. 2002;Huang et al. 2021).
Analysis and diagnosis of atmospheric ducts are based on calculation and criteria of atmospheric refractivity distribution from pressure, temperature, and humidity profiles (Babin and Rowland 1992;Babin 1996;Brooks, Goroch, and Rogers 1999;Mentes and Kaymaz 2007;Cheng et al. 2015). Ducts with bottom touching the ground are identified as surface ducts. The occurrence of surface ducts is defined as the ratio of the number of surface duct observations to the total number of observations.
The modified refractivity M is obtained by where T, P, e, and z are the temperature, pressure, vapor pressure, and elevation, respectively. The vertical gradient of modified refractivity is obtained based on Equation (1) as @e @z þ 0:157 : (2) A layer with dM/dz <0 is considered an atmospheric duct. The strength of an atmospheric duct is defined as the absolute value of the difference between the modified refractivity at top and bottom of the duct layer. The duct strength is in units of 10-6, which is also called M-units (MU; Mentes and Kaymaz 2007;Mai et al., 2020). Figure 1a shows the monthly occurrence of surface ducts in Svalbard winter. In winter months of December 2017 to February 2020, surface ducts occurred sixty times, which is 12.6 percent of 421 observations at Svalbard. The monthly occurrence was a maximum of 26.7 percent in January 2019 and a minimum of 6.5 percent in December 2019, with large variability. The average monthly occurrence was 10.8 percent in winter 2017-2018, 14.5 percent in winter 2018-2019, and 12.4 percent in winter 2019-2020, with small variability year to year. Within the Arctic, the average occurrence of surface ducts in Svalbard winter was much higher than that in Alaska winter from 1998 to 2008, mostly below 3 percent (Mai, Sheng, and Shi 2020), although the winter occurrence was lower than those at lower latitudes in the same season (Mentes and Kaymaz 2007;Lenouo 2014;Cheng et al. 2015). Therefore, the winter occurrence of surface ducts over Svalbard is relatively high in the Arctic, with large monthly variation.  (Mai, Sheng, and Shi 2020) but lower than that at lower latitudes (Mentes and Kaymaz 2007;Lenouo 2014;Cheng et al. 2015). Therefore, the strength of surface ducts in Svalbard winter is similar to that in other arctic regions, with a small interannual variability. Figure 1c presents the monthly average depth of surface ducts in Svalbard winter. The average depth of surface ducts was 13 m over Svalbard in winter months from December 2017 to February 2020. The monthly average depth was a maximum of 22 m in January 2018 and a minimum of 7 m in December 2018. The average depth was 16 m, 11 m, and 12 m in winter 2017-2018, 2018-2019, and 2019-2020, respectively, with large interannual variability. The depth of surface ducts in Svalbard winter was much smaller than that (about 20 m) observed in Alaska winter from 1998 to 2008 (Mai, Sheng, and Shi 2020). Therefore, the depth of surface ducts in Svalbard winter was smaller than that in other arctic regions in the same season.

Surface duct in winter
Based on the above analysis and discussion, the occurrence of surface ducts in Svalbard winter was about 12.6 percent, higher than that in other Arctic regions, with large monthly variability. The strength of surface ducts in Svalbard winter was 1.26 MU, similar to that in Alaska winter, and the depth was 13 m, smaller than that in Alaska winter.

Surface duct characterization
The occurrence and development of surface ducts are determined by atmospheric properties, such as vertical pressure, temperature, and humidity structures (Bean and Dutton 1968); therefore, atmospheric ducts, described by refractivity gradients, are determined by gradients of pressure, temperature, and humidity in Equation (2). To understand atmospheric impacts on surface ducts, we could separately analyze the three gradient terms in Equation (2) and evaluate their contributions to the refractivity gradient. Here, the three gradient terms on the right-hand side of Equation (2) are pressure, temperature, and humidity terms, and were calculate their ratios to the refractivity gradient as contributions of pressure, temperature, and humidity structures to duct construction in Svalbard winter from December 2017 to February 2020. The contribution of the pressure term was in the range 5.4 percent to 22.7 percent, with an average of 14.9 percent; the contribution of the temperature term was −19.7 percent to 164.1 percent, with an average of 78.5 percent; and the contribution of the humidity term was −80.4 to 104.1 percent, with an average of 6.6 percent. Because the contribution of the pressure term was much smaller than that of the temperature or humidity terms in all ducts, its contribution was ignored, and only the contributions of temperature and humidity terms on surface ducts were considered (Gossard 1977;Zhanyu et al. 2000;Mentes and Kaymaz 2007;Viher, Prtenjak, and Grisogono 2013;Kaissassou et al. 2015).
All surface ducts observed in Svalbard winter from December 2017 to February 2020 are plotted in Figure 2 with percentage contributions of temperature and humidity terms. To identify the importance of temperature and humidity structures in constructing surface ducts, we defined two duct types, T and H, as the larger contribution of the temperature term to the refractivity gradient than the humidity term and vice versa, and the division of T-type and H-type is plotted as a dashed line in Figure 2. This definition determines whether the construction of surface ducts is mainly contributed by a temperature inversion or humidity gradient in lower atmospheric structure. In Figure 2, forty-eight ducts are identified as T-type, and twelve as H-type, as 80 percent and 20 percent of the total of sixty ducts. The total strength of T-type ducts was 60.96 MU and that of H-type ducts was 14.76 MU. Obviously, T-type surface ducts were the dominant winter duct from December 2017 to February 2020, with greater occurrence and highter strength than H-type ducts, and therefore the temperature inversion could be a main cause of surface ducts in Svalbard winter. T-type and H-type ducts in three winters are compared, and domination of T-type ducts in both numbers and total strengths was found in all winters (Table 1). There were eleven strong ducts with strength greater than 2.00 MU, including eight strong T-type ducts and three H-type ducts. Most of the strong T-type ducts were supported by a temperature term except one; however, all of the strong H-type ducts were mainly supported by a humidity term with weak support from the temperature term. Based on the above analysis, T-type ducts were dominant in surface ducts, indicating temperature inversion as the main cause of surface ducts in Svalbard winter.
Therefore, the surface duct in Svalbard winter could be characterized as two types according to contributions of temperature and humidity structures. Domination of T-type ducts in number and strength could indicate the importance of temperature inversion in construction of surface ducts in Svalbard winter.

Atmospheric and ocean conditions
As discussed above, the T-type surface duct is mainly attributed to a temperature inversion in the near-surface atmosphere, and H-type ducts are related to a vertical humidity gradient. Atmospheric and ocean conditions  benefiting constructions of two type ducts are analyzed and discussed in this section, based on ERA-5 reanalysis and observation data.

T-type surface duct
Average atmospheric situations for Svalbard winter, T-type duct days, and their difference are plotted in Figure 3, on a 1,000-hPa pressure surface, with geopotential height, temperature, and humidity distributions. On the winter average chart (Figure 3a), a high is located north of Greenland, over 90 m, and a low over the Norwegian Sea, below 0 m. The height contours run from northeast to southwest over Svalbard with a northeasterly geostrophic wind. On the T-type duct chart ( Figure 3b) the high is enhanced, over 110 m northeast of Greenland, and the low is decreased, below −30 m over the Norwegian Sea. With the enhanced pressure gradient between the high and low, the northeasterly over Svalbard is strengthened. The enhanced northeasterly could bring colder and drier air from the deep Arctic to Svalbard and result in a temperature decrease of more than 4°C and a humidity decrease of more than 0.4 g/kg over Svalbard on the difference charts (Figures 3c and 3f). The colder and drier air brought from the deep Arctic on T-type days could decrease the water vapor content and clouds over Svalbard, weaken downward longwave radiation from the atmosphere, and strengthen surface radiative cooling during polar nights without solar radiation (Ruckstuhl et al. 2007;Raddatz et al. 2013;Yeo et al. 2018;Yamanouchi 2019). The enhanced surface cooling could induce a strong temperature inversion in the near-surface atmosphere and support construction of T-type ducts (Yamanouchi and Kawaguchi 1984;Stull 1988;Bradley and Keimig 1992).
To prove this process, we retrieved and analyzed radiation, temperature, and cloud data from local observations at Ny-Ålesund. Low cloud days and column water vapor content were obviously decreased on T-type duct days. The frequency of cloudy days was decreased to 21.8 percent on T-type days from 63.4 percent on winter days. The column water vapor content was also decreased to 2.5 kg/m 2 on T-type days from 4.1 kg/m 2 on winter days. Consequently, the downward longwave radiation decreased to 185.7 W/m 2 on T-type days from 227.9 W/m 2 on winter days, and the net radiation decreased to −44.4 W/m 2 on T-type days from −30.4 W/m 2 on winter days, reflecting a more than 30 percent increase of surface radiative cooling. This enhanced surface cooling caused a 5.5°C decrease in surface air temperature to −14.6°C on T-type days from −9.1°C on winter days. The air temperature at 50 m was also decreased due to cold air invasion, but it was much weaker than on the surface. The temperature difference was −2.4°C between 2 m and 50 m aboveground on T-type days, whereas it was only −0.6°C on winter days. Therefore, the enhanced temperature inversion caused by increased surface cooling under the atmospheric conditions could be the main support for the more frequent and stronger T-type surface ducts over Svalbard in winter.

H-type surface duct
Similar to the analysis for T-type ducts, average atmospheric situations for winter days, H-type duct days, and their difference are plotted in Figure 4, on a 1,000-hPa pressure surface, with geopotential height, temperature, and humidity distributions. Because Figure 4a is same as Figure 3a, the discussion on atmospheric situation on winter days is the same as that for T-type ducts. On the H-type duct chart (Figure 4b), a high over 150 m is located over the Kara Sea and a low below 40 m on south of the Greenland Sea. The geopotential height contours lie from southeast to northwest over Svalbard, indicating southeasterly geostrophic wind over this region, associated with the high and low. The southeasterly on H-type days brings warmer and wetter air from Norwegian Sea, with temperature and humidity increases more than 3°C and 0.5 g/kg over Svalbard on the difference charts (Figures 4c and 4f).
Contrary to the T-type duct days, the atmospheric situation on H-type days could enhance water vapor content and low cloud with higher temperature and therefore increase downward longwave radiation to the surface (Ruckstuhl et al. 2007;Raddatz et al. 2013;Yeo et al. 2018;Yamanouchi 2019). The increased downward longwave radiation, as the only radiative heat source during polar nights, could warm the sea around Svalbard and lead to more evaporation on H-type days (Liu, Key, and Wang 2009;Eastman and Warren 2010;Monroe, Taylor, and Boisvert 2021). As result of the enhanced evaporation, the surface air humidity would be increased, and the vertical humidity gradient would be strengthened to support H-type ducts over Svalbard.
Observations at Ny-Ålesund and SST were investigated. The low cloud days and column water vapor content were obviously increased on H-type duct days, comparing to those on winter days. The frequency of low clouds was increased to 83.3 percent on H-type days from 63.4 percent on winter days, and the column water vapor content increased to 5.6 kg/m 2 from 4.1 kg/m 2 . Consequently, the downward longwave radiation was increased to 263.6 W/ m 2 on H-type days from 227.9 W/m 2 on winter days. As the sea heated, SST in the neighboring sea area of Ny-Ålesund was increased 1.2°C to 0.8°C on H-type days from −0.4°C on winter days. The enhanced SST brought a surface air humidity increase of 0.62 g/kg to 2.11 g/kg on H-type days from 1.49 g/kg on winter days and a humidity increase of 0.15 g/kg between 2 m and 50 m aboveground to 0.17 g/kg from 0.02 g/kg. Therefore, the enhanced humidity gradient caused by sea surface warming under the atmospheric condition could be the main support for the H-type surface ducts in Svalbard winter.
Based on the above analysis and discussion, the frequent and strong T-type surface ducts in Svalbard winter are related to atmospheric conditions with colder, drier, and northeasterly weather, and the temperature inversion caused by enhanced surface radiative cooling is the main support for duct construction. The H-type surface duct is related to atmospheric and ocean conditions with warmer, more humid, southeasterly weather and higher SST, and the humidity gradient formed by sea water evaporation is the main support for duct construction. The change in downward longwave radiation is the controlling process in surface duct construction in Svalbard winter.

Conclusions
In this study, surface ducts in Svalbard winter from December 2017 to February 2020 were analyzed and discussed based on observation and reanalysis data. The results are summarized as follows: (1) The average occurrence of surface ducts was 12.6 percent, with a mean depth of 13 m and strength of 1.30 MU, during the observational period in Svalbard winter. As part of the Arctic, surface ducts in Svalbard winter are more frequent than in other arctic regions, such as Alaska but less frequent and weaker than at lower latitudes.
(2) According to the contribution of atmospheric structure, surface ducts in Svalbard winter could be characterized to two types: T-type ducts mainly supported by temperature inversion and H-type ducts supported by a humidity gradient. T-type ducts were the dominant surface duct in Svalbard winter, comprising 80 percent of all duct cases, indicating the importance of temperature inversion in constructing surface ducts.
(3) Surface ducts in Svalbard winter are closely related to atmospheric conditions. The temperature inversion caused by enhanced surface radiative cooling could be the main support for T-type surface ducts under atmospheric conditions with colder, drier, and northeasterly weather, and the humidity gradient formed by increased evaporation on surrounding water could be the main support for H-type ducts under atmospheric and ocean conditions with warmer, wetter, and southeasterly weather and higher SST.