1. Introduction
Hot-air balloons float in the air and travel with the wind. Global Navigation Satellite System (GNSS) data acquired during these flights provide a displacement during a time interval that is a measure of the airspeed. This wind information is obtained in the atmospheric boundary layer (ABL), where in general few observations are present.
In the Netherlands there are approximately 500 registered balloons and on a yearly basis between 8000 and 9000 flights are made. The measurements consist of recorded GNSS positions provided courtesy of professional balloonists. A hot-air balloon is a passive moving platform, and in meteorology moving platforms are used for collecting data like, for example, radiosondes. These balloons are filled with helium gas and have an ascent speed of approximately 5 m s−1, because they have to sample the troposphere and a part of the stratosphere in a certain time slot. Further, controlled meteorological balloons (Voss et al. 2013) can repeatedly take soundings and also offer an attractive possibility for upper-air observations. Other observations from moving platforms are obtained from commercial aircraft like Aircraft Meteorological Data Relay (AMDAR; WMO 2003) and Mode-S Enhanced Surveillance (de Haan 2011). In addition research aircraft also collect a good deal of atmospheric data. For example, unmanned aerial vehicles (UAVs) are employed for vertical profiling of the ABL (e.g., Jonassen et al. 2015) and even for continental-scale observations (Intrieri et al. 2014). These data are received from moving platforms, which have their own means of propulsion. The wind speed is calculated by subtracting the airspeed from the ground speed and therefore the airspeed has to be measured accurately. For hot-air balloons it is simpler, because they travel with the wind within a Lagrangian framework, and therefore the displacement is closely related to the wind. Thus only positions (latitude, longitude, and altitude) and accurate time stamps are needed to obtain the wind components.
Balloon flights usually last about 2 h and generally take place after sunrise and before sunset. During the day, when thermals develop, ballooning can become dangerous, because the up- and downdrafts can deform the shape of the balloon, which affects the buoyancy, resulting in dangerous drops.
The transition from stable to neutral and then unstable conditions and vice versa offer insights into a research field of interest. In the morning there is increasing turbulence and in the evening the turbulence dies out. During the Boundary Layer Late Afternoon and Sunset Turbulence (BLLAST) campaign (Lothon et al. 2014), the evening transition was studied in southern France. Observations from a wide range of platforms have been applied to record this typical ABL regime during several days in late June 2011. Beare et al. (2006) studied the same phenomenon, but from a large eddy simulation (LES) model’s perspective, and in Beare (2008) the focus was on the morning transition. In the GEWEX Atmospheric Boundary Layer Study 3 (GABLS3; Bosveld et al. 2014), a diurnal cycle with an evening and a morning transition was studied in one location, namely Cabauw, in the central Netherlands.
The trajectories of hot-air balloons can be regarded as airborne wind observations that sample the ABL in the above-mentioned regimes on different locations. Balloons have been widely applied in atmospheric research; for example, Businger et al. (2006) conducted experiments with balloons that sampled at low altitudes the marine boundary layer with advanced instruments. These balloons were filled with helium and designed to operate autonomously. Laakso et al. (2007) used a hot-air balloon as a measuring platform. They studied particle and cluster formations and measured profiles of aerosols. From a Lagrangian perspective, they benefited from the effect that the balloon was carried along by the wind, and thus the effects of advection and heterogeneity did not play a major role in the measurements.
In this study the focus is on the wind in the ABL over land, and instead of data from a measuring campaign, we use data from a leisure activity. The data can be applied for process studies and model validation. Furthermore they can be useful as an extra wind observation in the lower troposphere for the operational forecaster.
In this paper, we start by pointing out the principle of measuring wind from hot-air balloon flight tracks. We address the dynamics of the hot-air balloon and the interaction with the drag forces. In the next section we describe briefly the observations that have been used for intercomparisons. We also introduce the HIRLAM–ALADIN Research on Mesoscale Operational NWP in Europe (HARMONIE) model, which provides model winds that have been used for comparison. Further, HARMONIE is also necessary to gain insight into the meteorological background. Subsequently, in section 3 we present four case studies during 2013 with interesting meteorological features on different scales. In the final sections we discuss the results followed by conclusions and outlook.
2. Characteristics of a hot-air balloon
a. Hot-air balloons as wind measuring device
In meteorology, radiosonde balloon observations are a well-known measuring method that is used for determining wind, temperature, and humidity at various altitudes. The radiosonde balloon is tracked with GNSS, and from two consecutive positions, the wind speed and wind direction can be derived using the time delay between these points. The radiosonde or sounding balloon has an initial content of 1.5 m3, and its purpose is to measure profiles of wind, temperature, and humidity. A sounding balloon has an elastic envelope, is filled with helium gas, and ascends until it bursts in the stratosphere.
A hot-air balloon (Fig. 1) has a nonelastic envelope, and the content varies from 3000 to 8000 m3. In contrast to the sounding balloon, a hot-air balloon remains on a more or less constant level and stays in the lower troposphere. The ceiling of a flight is usually not higher than 2000 m. The energy to heat the air in the balloon is supplied from a propane heating device. The buoyancy is dependent on the temperature excess between the balloon and the surrounding air. To determine the wind speed and direction, we make use of the movement of the hot-air balloon. A hot-air balloon is about 30 m high, and the payload has a mass of about 500 kg. The total mass of the balloon varies between 4000 and 10 000 kg. For navigation, the balloonist uses a GNSS receiver. In this paper the three-dimensional positions were recorded at a maximum rate of 4 Hz. Two successive positions in combination with the time interval deliver the velocity of the air in which the balloon is traveling. The flying height is determined by the balloonist, while the vertical displacement is influenced by buoyancy, turbulence, and other external factors. The accuracy of the measured position depends mainly on the constellation of the satellites. Typical values for the standard deviation in the horizontal and vertical plane are 2.5 and 30 m, respectively. In the vertical this is quite inaccurate, and therefore some dedicated GNSS receivers are equipped with a barometer, which reduces the error. The balloon flight is a leisure activity, but the recorded GNSS data conceal useful meteorological information. In fact the data are a kind of by-product and require some organization to obtain them, but there are no substantial costs involved. Finally, it should be noted that a hot-air balloon is not a rigid body and that it will deform easily. For this reason hot-air ballooning takes place only in light wind conditions with moderate turbulence and without intense up- and downdrafts. Practically speaking, the wind speed should be less than 6 m s−1, and during takeoff the gustiness should not be excessive.
Photograph of a hot-air balloon.
Citation: Weather and Forecasting 31, 5; 10.1175/WAF-D-16-0028.1
b. Drag and response time
During the launch the balloon experiences a certain drag force, but as soon as it takes off, the drag force decreases. During the flight the balloon feels the drag as soon as the relative speed is no longer zero, for instance, when the balloon enters a layer with vertical wind shear.
In Fig. 2 we present the relative wind speed as a function of time. It is seen that it takes 300 s before the initial speed difference of 2 m s−1 is reduced to 10%. In the beginning the response is fast and levels off as the speed difference diminishes. So it takes a couple of minutes before the hot-air balloon eventually travels with the ambient air velocity. This makes sense because a hot-air balloon is a large body of substantial weight. Because of the inertia of the hot-air balloon, its displacement does not capture the small-scale variations in the wind pattern. The wind observations represent an average in space and time.
Response time of a hot-air balloon due to a step-wise changing wind.
Citation: Weather and Forecasting 31, 5; 10.1175/WAF-D-16-0028.1
3. Intercomparison data
In this study we have applied data from the KNMI observation network. It should be noted that some of the balloon flights (see Fig. 3b) took place in areas where wind observations in the lower atmosphere were not present, and therefore we also used a numerical weather prediction (NWP) model for comparison (see Fig. 3a). Further, we have used the model to study the meteorological circumstances of the balloon flights. We have chosen the HARMONIE model because this NWP scheme represents the wind quite well, especially in the ABL (Baas et al. 2015).
(a) HARMONIE domain (2.5-km grid resolution, 800 × 800 grid points). (b) Hot-air balloon tracks and the nearby observations, which are used for the intercomparison. The balloon tracks are presented as wind flags; a full barb corresponds with 5 m s−1; the color is a measure of the height.
Citation: Weather and Forecasting 31, 5; 10.1175/WAF-D-16-0028.1
a. Routine observational network
The KNMI network consists of 33 automatic weather stations on land, 15 wind poles in coastal areas, and 13 automatic weather stations on North Sea platforms. In this study the focus was on the wind, which is sampled every 12 s and becomes available as 10-min averages. For the upper air, a radiosonde launch is available but is only launched at midnight. Other upper-air observations are supplied by a RASS wind profiler and tower observations. Each observation has its own characteristics: that is, radiosondes deliver profiles of wind, temperature, and humidity, but they are not frequently sampled; aircraft observations deliver wind and temperature information with a high sampling rate, but their locations are in small corridors above the tropopause. Further, they also measure vertical profile data from the takeoff and landing areas. Recently, new aircraft sensors have also begun delivering humidity observations, albeit in small numbers. Surface synoptic observations (SYNOPs) give observations on the surface level every hour, but they are representative for the local scale only. In our study we also use Cabauw data, because two of the flights took place in the area surroundings the KNMI observatory. This observational site is located in a rural area, 20 km southwest of the city of Utrecht. There is a 213-m-high tower with wind sensors at 10, 40, 80, 140, and 200 m, and the data are available as 12- and 600-s averages. At the site a whole range of instruments is deployed and among them there is a RASS wind profiler. The wind profiler measures wind profiles up to 4000 m, dependent on the concentration of small particles.
b. NWP
In this study we have used data from HARMONIE (cycle 38; see Fig. 3a), which has been implemented at the computer system of the ECMWF. HARMONIE is a nonhydrostatic model (Seity et al. 2011) with a horizontal resolution of 2.5 km. In the vertical, there are 65 layers defined, and 10 layers are positioned in the lower 2 km. HARMONIE covers an area as large as western Europe and receives lateral boundary condition (LBC) information from the global ECMWF model. HARMONIE runs every 3 h and a three-dimensional variational data assimilation (3DVAR) scheme is used with a time window of 1.5 h. The following surface observations are assimilated: SYNOPs, along with buoy and ship data, for ground level; in the upper air, radiosondes and aircraft observations are used. The analysis tries to optimally combine the information based on the previous model run (first guess) and the observations.
Relevant parts of the model for ABL processes are the turbulence and the land surface scheme. The turbulence scheme describes the transport of momentum, heat, and humidity, from the earth’s surface to the atmosphere and vice versa on the basis of the turbulent kinetic energy (TKE) equation. The turbulence and convection schemes work together within the framework of an eddy-diffusivity mass-flux (EDMF) scheme (Siebesma et al. 2007), which adequately describes up- and downdrafts. In the land surface scheme the energy balance is solved. The soil moisture content, air temperature, humidity, and wind speed, as well as the surface albedo, roughness lengths, soil temperature profile, and the heat conductivity of dry soil, affect how the net amount of radiation is distributed over sensible, latent, and soil heat fluxes. The roughness lengths, vegetation, and land use are important parameters that have a clear impact on the wind profile. The surface data are derived from the ECOCLIMAP database.
4. Case studies
Now, we present four case studies of hot-air balloon flights during the summer of 2013. The first case concerns a large-scale baroclinic phenomenon that is well captured by KNMI’s observation network.
a. Occluded front, 28 May 2013
Although balloon flights usually took place in fair weather conditions, this flight from Amersfoort to Utrecht (see Fig. 4a) commenced when thunderstorms were developing in the southern part of the Netherlands. These thunderstorms belonged to an occluded front that moved slowly in a northeasterly direction and approached the area where the balloon flight took place. Just ahead of the occluded front, a convergence line in the wind field could be recognized (see Fig. 4b). This was the area where two air masses collided. Despite the adverse weather forecast, the balloonist decided to take off and during the flight the wind direction was steady around 90°, the wind speed was more variable, and then increased slightly when the balloon went up to higher levels (see Fig. 5a). When the balloonist approached the city of Utrecht, the balloonist realized that he would fly over an urban area with limited possibilities to land. Therefore, he decided to land in the outskirts of Utrecht, which was still rural area. During the descent, the passage of the frontal system commenced, resulting in a wind direction change of 120°. The balloon was pushed toward an area more favorable for landing, but as reported by the balloonist the gustiness increased, which hindered his ability to land the balloon. The balloonist lost control and was forced to land in a ditch. This sudden wind change was also recognized at the nearby KNMI stations in De Bilt (5 km) and Cabauw (30 km), as depicted in Fig. 5b. Note that this frontal system arrived 1 h earlier at Cabauw and was also recognized at 200 m. All in all, the wind information from this balloon track confirmed the passage of the frontal zone and provided extra vertical wind information. In the next case, we present a very small phenomenon that was hardly visible in the KNMI observational network.
(a) Wind data based upon a hot-air balloon flight from Amersfoort to Utrecht during the evening of 28 May 2013. (b) Weather map at 1800 UTC 28 May 2013 with an occluded front (magenta) and a convergence line (red) shown.
Citation: Weather and Forecasting 31, 5; 10.1175/WAF-D-16-0028.1
(a) Height, wind direction, and wind speed during 28 May 2013 along the trajectory of the hot-air balloon. (b) Wind observations at Cabauw and De Bilt during 28 May 2013. Note that the wind shift arrives 1 h later at De Bilt.
Citation: Weather and Forecasting 31, 5; 10.1175/WAF-D-16-0028.1
b. Major wind shifts, 14 June 2013
In this case a sea-breeze circulation pattern developed and at the same time a depression over the Atlantic Ocean deepened and moved slowly eastward. The pressure gradient was small and there was a gradual transition from a southerly to a westerly flow. In this transition zone, substantial wind shifts developed that impacted the balloon’s track. From Figs. 6a and 7a it may be seen that the balloon remained more or less at the same altitude and described two circles in a time span of less than 300 s. These features are not artifacts from the GNSS system, because the balloonist has confirmed (D. Kleinlugtenbeld 2013, personal communication) that two circular tracks were flown. The balloonist managed to escape from the unsettled conditions by ascending to a higher altitude where the flow was less variable.
(a) Circular wind patterns recorded from a balloon’s track between 1820 and 1830 UTC 14 Jun 2013. Note that the wind data are multiplied by a factor 2. (b) HARMONIE +3-h forecast at model level 55 valid at 1800 UTC 14 Jun 2013. Distance between the two wind flags is 5 km. Note that model level 55 corresponds to 200 m AGL, which is the altitude of the balloon at the beginning of the flight.
Citation: Weather and Forecasting 31, 5; 10.1175/WAF-D-16-0028.1
(a) Height, wind direction, and wind speed during 14 Jun 2013 along the trajectory of a hot-air balloon. (b) Wind observations (10-m height) at nearby stations during 14 Jun 2013.
Citation: Weather and Forecasting 31, 5; 10.1175/WAF-D-16-0028.1
Interestingly, the observations in Lelystad and Marknesse in Fig. 7b show a significant wind change during the flight of the hot-air balloon. This change arrived later in Marknesse, which indicates that a convergence zone is passing over. The NWP output confirms this finding. The observations are measured at 10-m height and are averaged over 600 s. The first circle takes 300 s, and the second one takes only 180 s, with diameters of a few hundred meters. Note that the height varies slightly. The sampling rate of the KNMI network cannot represent this small-scale phenomenon. A higher sampling rate is required to capture the details of this wind shift. Moreover, the measurements were taken at 10-m height, while the wind shifts were observed at approximately 200-m height.
The small-scale phenomena are for the same reasons not represented by the NWP model (see Fig. 6b), because the horizontal and the temporal resolutions are too coarse; that is, the grid size is 2.5 km and the output frequency is 1 h. It is evident that the wind shifts are a subgrid-scale feature that is not resolved by the model. Possibly a nested run with a very high resolution or an LES model simulation might reveal these wind patterns. On the other hand, because of the balloon’s inertia (see section 2b) not all variations in the wind are captured, so the accuracy of the balloon-based wind measurements is questionable, especially in this case. At the end of the flight at heights below 200 m, the balloon again met variable winds, resulting in a curved trajectory (not shown). Without further experiments an explanation of this event is rather speculative, but it is possible that a convergence line with embedded updrafts had passed over the area where the hot-air balloon flight took place.
c. Wind shear during a hot summer day, 18 June 2013
During this case the trajectory of the balloon was curved because the balloon went up and down in an ABL with considerable wind shear (Fig. 8a). It was a hot summer day with northeasterly winds and there was a distinct temperature contrast over the Netherlands. In a barotropic atmosphere the wind changes with height as a result of the vertical gradient in stress between the top of the surface layer and the free atmosphere. As a result, the wind veers with height (clockwise turning), which is the so-called Ekman spiral. However, the wind can also change with height because of baroclinic effects. If the horizontal temperature gradient in a certain layer is positive, the wind veers with height; conversely, if the horizontal temperature gradient is negative, the wind backs (anticlockwise turning) with height. Using NWP output, we have calculated the geostrophic wind based on the pressure gradient and the Coriolis force and found that the geostrophic wind changed with height, which confirms the presence of the thermal wind.
(a) Wind data derived from a hot-air balloon flight from Utrecht to Cabauw during 1840–1950 UTC 18 Jun 2013; in addition, the tower observations at Cabauw at 60 and 200 m are depicted at 1930 UTC. (b) Height, wind direction, and wind speed during 18 Jun 2013 along the trajectory of the hot-air balloon.
Citation: Weather and Forecasting 31, 5; 10.1175/WAF-D-16-0028.1
The balloon took off from a city park and went in southerly direction. During the ascent the balloon’s trajectory veered with height (Fig. 8b) and eventually went northwest. At 1910 UTC the balloon descended slightly, and as a result of baroclinic and frictional constraints the wind backed. The balloon went up, and the wind veered again. At 1934 UTC the balloon prepared for landing and remained below 200 m for 10 min. The balloon again went in a southerly direction, and the wind turned farther toward the northwesterly direction. The wind was steadier, with speeds of 4 m s−1, which is in contrast with the variable winds at the beginning of the flight.
In Fig. 9 we compare the balloon wind data during the last 20 min of the flight with HARMONIE data and with the observations taken at Cabauw. The HARMONIE data consist of the +1- and +2-h forecasts starting from the analysis at 1800 UTC. The wind profiler is located at Cabauw, and the data are available as a 30-min average. Because of the radar reflections from the nearby obstacles (i.e., tower), the wind-profiler data are unreliable below 400 m. In addition, the averaged wind data from anemometers from the mast are shown. There is some wind shear in the model, but it is underestimated. The wind-profiler and tower observations, as well as the balloon-derived winds, reveal more of a gradient. Perhaps the turbulence scheme is not able to represent this gradient because of its parameterization and because of insufficient vertical resolution. Another reason might be the representativity of the roughness length in the grid box of the model. Further, it should be noted that there are slight timing and location mismatches between the moving balloon and the observations at Cabauw.
Intercomparison of wind observations at Cabauw (tower, profiler, and balloon) and HARMONIE data during 1930–2000 UTC 18 Jun 2013.
Citation: Weather and Forecasting 31, 5; 10.1175/WAF-D-16-0028.1
All in all, the balloon-derived winds are closer to the other observations than to the HARMONIE model results. It is obvious that this trajectory reveals an interesting Lagrangian representation of an atmospheric flow in baroclinic conditions with veering and backing winds.
d. Low-level jet, 28 September 2013
Now, we present the final case with a typical ABL and a baroclinic effect in an area where no other observations are present. This flight, depicted in Fig. 10a, took place early in the morning in a stable atmosphere with clear-sky conditions. The low-level jet (LLJ) usually occurs on top of the stable boundary layer when the turbulent mixing ceases as the ABL stabilizes as a result of longwave radiational cooling (see Baas et al. 2009). The frictional effects are reduced in a shallow layer above the top of the surface inversion and within the residual layer from the previous convective mixed layer. The ageostrophic wind in the residual layer was in balance with the frictional forces in the late afternoon and went through an inertial oscillation that had a period of around 15 h. This flight took place at the end of this oscillation during the morning transition.
(a) Balloon wind data from a morning flight during 0610–0736 UTC 28 Sep 2013. (b) Height, wind speed, and wind direction during 0610–0736 UTC 28 Sep 2013.
Citation: Weather and Forecasting 31, 5; 10.1175/WAF-D-16-0028.1
As soon as the balloon took off, the wind speed increased from 0 to 12 m s−1, and it is clear that the balloon had entered the LLJ. This jet is located in a small vertical zone not higher than 500 m (see Baas et al. 2009) and also occurs in the HARMONIE simulation (see Fig. 11). The sharp gradient in the wind speed is recognized in the model and in wind data from the hot-air balloon track. Note that the first 10 min are shown, and the model and observations are in good accordance. However, the observations show some noise, possibly caused by the unsettled conditions during takeoff.
HARMONIE profile for 0300 UTC 28 Sep 2013 + 3 h valid at 0600 UTC and balloon-derived wind data, just after takeoff.
Citation: Weather and Forecasting 31, 5; 10.1175/WAF-D-16-0028.1
At 0625 UTC (see Fig. 10b) the balloon descended about 100 m and left the jet immediately. Subsequently, the balloon rose gradually and crossed the LLJ again, arriving at a height of 1500 m, which is also the ceiling of the flight. The change in the wind speed and direction is clearly present in the recorded data. At 1500 m the wind speed decreased to values of 4 m s−1, but it is remarkable that the wind direction changed from an east to a northeasterly direction. It is seen that during the ascent the wind direction changed counterclockwise. This change in wind direction is caused by the advection of cold air, which is confirmed by HARMONIE (see Fig. 11, right). While descending to lower levels the balloon again encountered the wind direction change and the LLJ, and it is clear that the flow had been quite stationary.
5. Discussion
In this paper we have shown that balloon-derived winds are an interesting new type of upper-air observations; however, there are some remarks that have to be made. At low altitudes the movement of the balloon is also influenced by other forces like surface friction when the balloon is dragged over the ground. For this reason, the observations lower than 10 m are excluded from the dataset. An automatic quality control algorithm as developed by Houchi et al. (2015) would have been useful. Such an algorithm does the preprocessing and creates a reliable dataset with reduced errors. Since the dataset consists of four case studies only, we have inspected the data visually and removed outliers by hand.
The short-term fluctuations in the wind vector are also not properly measured, as a result of the inertia as discussed in section 2b. Moreover, we have neglected the acceleration term in the derivation for the wind speed. For the application in data assimilation it is essential that the observations are in agreement with the dimensions of the grid box. Because of the response time of the balloon, the small scales are inherently filtered out. A spectral analysis of the hot-air balloon data would reveal which scales are present, and this a subject for future research.
Finally, we address the inaccurate vertical coordinate, which is caused by instrument noise and by the coordinate system. The local geoid can deviate from the World Geodetic System (WGS) 84 ellipsoid as applied in the GNSS system, but this error is not substantial.
6. Conclusions and outlook
Hot-air balloon tracks revealed interesting meteorological features ranging from the meso- to the microscale and can be useful for process studies and for validating an NWP model. Process studies presented in this paper comprise the scope of turbulence, shallow convection, and air–surface interaction.
Hot-air balloon data are obtained without large investments or without the presence of an operator. However, the data are available during a limited period of the day and only during fair weather conditions. The data were collected with simple instrumentation by frequently sampled positions and time stamps from a GNSS receiver. The wind speed and direction were derived by simple time differencing. We showed that a hot-air balloon responds quickly on a step-wise changing wind in the beginning and the velocity difference decays slowly asymptotically. This can be expressed in terms of a response length. We presented four interesting trajectories that reveal valuable wind information of the lower atmosphere.
This study is based on offline GNSS navigation data that are received from collaborating balloonists. This new observation type can also be used in operational weather services as an ABL wind observation and might be useful for guidance of other balloonists.
For that to happen, an infrastructure designed and able to collect the data in a timely fashion is necessary. There are two ways to proceed. First, data can be collected via air traffic control (ATC), in which case the hot-air balloons should be equipped with a transponder. There are technical constraints because a transponder requires a power supply and also navigational data should be provided.
Second, citizen technology can be applied. This means that the data are collected by balloonists and passengers who carry a smart phone or mobile device with a dedicated app. At KNMI, work is in progress to realize this possibility and the first results are encouraging.
Acknowledgments
The balloonists Peter Barlo, Dick Kleinlugtenbeld, and Willem Hijink are gratefully thanked for providing their GNSS navigation data of the balloon flights in this paper. We are indebted to Henk Klein Baltink for supplying the wind profiler data at Cabauw. We are also thankful to Wouter Lablans for his critical comments on a draft of the manuscript, and finally we would like to acknowledge the reviewers for their positive criticisms, which have improved this paper.
REFERENCES
Baas, P., Bosveld F. C. , Klein Baltink H. , and Holtslag A. A. M. , 2009: A climatology of nocturnal low-level jets at Cabauw. J. Appl. Meteor. Climatol., 48, 1627–1642, doi:10.1175/2009JAMC1965.1.
Baas, P., Bosveld F. C. , and Burgers G. , 2015: The impact of atmospheric stability on the near-surface wind over sea in storm conditions. Wind Energy, 19, 187–198, doi:10.1002/we.1825.
Beare, R. J., 2008: The role of shear in the morning transition boundary layer. Bound.-Layer Meteor., 129, 395–410, doi:10.1007/s10546-008-9324-8.
Beare, R. J., Edwards J. M. , and Lapworth A. J. , 2006: Simulation of the observed evening transition and nocturnal boundary layers: Large-eddy simulation. Quart. J. Roy. Meteor. Soc., 132, 81–99, doi:10.1256/qj.05.64.
Bosveld, F. C., and Coauthors, 2014: The GABLS Third Intercomparison Case for Boundary Layer Model Evaluation. Part B: Results and process understanding. Bound.-Layer Meteor., 152, 157–187, doi:10.1007/s10546-014-9919-1.
Businger, S., Johnson R. , and Talbot R. , 2006: Scientific insights from four generations of Lagrangian smart balloons in atmospheric research. Bull. Amer. Meteor. Soc., 87, 1539–1554, doi:10.1175/BAMS-87-11-1539.
de Haan, S., 2011: High-resolution wind and temperature observations from aircraft tracked by Mode-S air traffic control radar. J. Geophys. Res., 116, 2156–2202, doi:10.1029/2010JD015264.
Houchi, K., Stoffelen A. , Marseille G. , and de Kloe J. , 2015: Statistical quality control of high-resolution winds of different radiosonde types for climatology analysis. J. Atmos. Qceanic Technol., 32, 1796–1811, doi:10.1175/JTECH-D-14-00160.1.
Intrieri, J., and Coauthors, 2014: Global Hawk dropsonde observations of the Arctic atmosphere obtained during Winter Storms and Pacific Atmospheric Rivers (WISPAR) field campaign. Atmos. Meas. Tech., 7, 3917–3926, doi:10.5194/amt-7-3917-2014.
Johnson, J. C., 1954: Physical Meteorology. J. Wiley and Sons–M.I.T. Technology Press, 393 pp.
Jonassen, M. O., Tisler P. , Altstädter B. , Scholtz A. , Vihma T. , Lampert A. , König-Langlo G. , and Lüpkes C. , 2015: Application of remotely piloted aircraft systems in observing the atmospheric boundary layer over Antarctic sea ice. Polar Res., 34, 25 651, doi:10.3402/polar.v34.25651.
Kristensen, L., 1998: Cup anemometer behavior in turbulent environments. J. Atmos. Oceanic Technol., 15, 5–17, doi:10.1175/1520-0426(1998)015<0005:CABITE>2.0.CO;2.
Laakso, L., and Coauthors, 2007: Hot-air balloon as a platform for boundary layer profile measurements during particle formation. Boreal Environ. Res., 12, 279–294.
Lothon, M., and Coauthors, 2014: The BLLAST field experiment: Boundary-Layer Late Afternoon and Sunset Turbulence. Atmos. Chem. Phys., 14, 10 931–10 960, doi:10.5194/acp-14-10931-2014.
Munson, B. R., Young D. F. , and Okiishi T. , 1990: Fundamentals of Fluid Mechanics. J. Wiley and Sons, 843 pp.
Seity, Y., Brousseau P. , Malardel S. , Hello G. , Bénard P. , Bouttier F. , Lac C. , and Masson V. , 2011: The AROME-France convective-scale operational model. Mon. Wea. Rev., 139, 976–991, doi:10.1175/2010MWR3425.1.
Siebesma, A. P., Soares P. M. M. , and Teixeira J. , 2007: A combined eddy-diffusivity mass-flux approach for the convective boundary layer. J. Atmos. Sci., 64, 1230–1248, doi:10.1175/JAS3888.1.
Voss, P. B., Hole L. R. , Helbling E. F. , and Roberts T. G. , 2013: Continuous in-situ soundings in the Artic boundary layer: A new atmospheric measurement technique using controlled meteorological balloons. J. Intell. Rob. Syst., 70, 609–617, doi:10.1007/s10846-012-9758-6.
WMO, 2003: Aircraft Meteorological Data Relay (AMDAR) reference manual. WMO-958, 80 pp. [Available online at https://www.wmo.int/pages/prog/www/GOS/ABO/AMDAR/publications/AMDAR_Reference_Manual_2003.pdf.]
