The Dornier 128 with the call sign D-IBUF (Do128-6) is a twin-engine propeller aircraft which had been deployed by TU Braunschweig for atmospheric research from 1986 until 2021. It was operated at an air speed of 60–65 m s⁻¹. The aircraft is described in detail in . An overview of meteorological measurement campaigns with this research aircraft is given in .

The Cessna F406 with the call sign D-ILAB is a twin-engine propeller aircraft which has been chosen as the successor for the successful aircraft Dornier 128. It has similar specific properties which are required for operation in the atmospheric boundary layer. The non-pressurised cabin allows for the installation of equipment in the fuselage looking through openings in the roof and on the bottom. It can fly at a low cruising speed of 70 m s⁻¹ for high-resolution measurements. The two engines enable operations at low altitudes as required for the flights at hub height above the North Sea. The aircraft provides space for two pilots and two to three operators depending on the installed measurement equipment in the cabin.

The state of the art data acquisition system consists of a core computer with several interfaces (such as different digital and analogue input and output signals, including the format ARINC429, and serial and ethernet ports). This computer receives the sensor data in real time, provides time stamps to the data packages and stores the raw data items on an internal hard drive. It provides the raw data via ethernet to a second computer, which makes real-time calculations for, for example, wind speed in all three dimensions. The second computer performs unit conversions and applies calibration and correction factors where required. At the time of the campaign, it provided a text-based user interface with tables and columns for several raw and calculated parameters. Now it also includes a graphical user interface with maps, primary flight display, system surveillance and protocol. The pre-selected parameters are recorded on a removable hard drive which allows for quick data transfers after a measurement flight. The raw data on the first computer remain in the aircraft as backup and are overwritten after a couple of flights. This approach ensures reliable data handling and data security. A third computer is used to provide online graphics and quicklooks of different parameters since the beginning of the recording. This feature displays time series and vertical profiles and can be used for on-board decision-making, for example, to determine the most relevant flight altitude for a specific mission. The overall layout of the data acquisition and recording system allows for a flexible integration of additional sensors for synchronous recording. The user interface for monitoring of the flight and mission parameters as well as an interaction with the system is installed on two different mission consoles. In addition, a wireless access point is implemented to stream the real-time data inside the cabin, which allows more scientists or observers to monitor the acquired parameters on a mobile device as well.

Figure  shows a front view of the aircraft with the nose boom. Another sensor below the aircraft nose next to the nose boom is a forward-looking camera. The location of the measurement head in front of the aircraft is a compromise between structural requirements of the aircraft frame and distance in front of the aircraft to perform measurements in the undisturbed airflow. The vibrations of the measurement head were measured during the initial flight tests and found to be negligible. So, a correction is not required.

The data acquisition system can record so-called “events”, i.e. specific points in time when one of the cockpit crew members or one of the scientists/observers has detected a noticeable event during the flight and triggered the “event button”. Such an event increases an internal counter and writes a short protocol note into the logfile, complemented by a photo of the forward-looking camera. This feature is very useful during the process of data evaluation.

The concept of using two complementary sensors for temperature has been implemented for both aircraft. In the data processing, the data from a highly accurate temperature sensor, 102DB1AG (Rosemount, USA), and a fast temperature sensor, 102E4AL (Rosemount, USA), can be combined or used for quality check. Further, the effect of self-heating induced by the air speed is taken into account in the data processing . The method of calculating the best possible result based on a slow but more accurate and a fast but drifting sensor is done by complementary filtering, as described in detail in .

Humidity is measured with three sensors based on different measurement principles: in the nose boom of the Dornier 128, there is a TP 3-S dew point mirror (Meteolabor, Switzerland), a Humicap H233 capacitive sensor (Vaisala, Finland) and an L-6/HMS-2 Lyman-alpha sensor (Buck Research, USA). The dew point mirror provides the most precise measurements at a temporal resolution of 1 Hz. The Lyman-alpha, based on optical absorption in the ultraviolet wavelength range, delivers data at a high temporal resolution of 100 Hz but with a drifting signal.

In the nose boom of the Cessna F406, there are a dew point mirror and capacitive sensor of the same type and a KH-20 optical sensor (Campbell Scientific, USA) based on absorption in the ultraviolet wavelength range for fast water vapour fluctuations.

Determining the wind vector is based on the principle of subtracting the air speed vector from the ground speed vector for both aircraft. The ground speed vector is derived from the high-precision measurements of the aircraft position and attitude using an integrated system of inertial measurement unit (IMU) and Global Navigation Satellite System (GNSS). For the air speed vector, measuring the airflow angles, the angle of attack (angle between the velocity vector of the aircraft relative to the air and the aircraft longitudinal axis, describing the longitudinal component of the aircraft velocity) and the angle of sideslip (angle between the velocity vector and the projection of the aircraft longitudinal axis, which describes whether there is a lateral component to the aircraft velocity) are required. This is done via a five-hole probe at the front of the nose boom and needs fusing the data with high-precision attitude data . The static and dynamic pressure as well as the pressure differences occurring at the five-hole probe (Rosemount, USA) are measured with pressure transducers of Setra, USA. The pressure transducers are located right behind the probe to minimise the length of the required tubes. While on-board the Dornier 128 the inertial data is measured with the IMU iNAV-RQH-1003 (iMAR, Germany) and the GNSS data with an external GNSS receiver OEM6 (NovAtel, Canada), on-board the Cessna F406, the GNSS receiver OEM6 (NovAtel, Canada) is already integrated into the successor IMU iNAT-RQT-4001 (iMAR, Germany). Accuracy and resolution of the systems used are compared in Table . The accuracy of the horizontal wind speed components of the D-IBUF is better than 0.5 m s⁻¹ and of the vertical wind speed component better than 0.1 m s⁻¹ . As pressure measurements are the bottleneck for wind measurement accuracy, and the D-ILAB deploys the same pressure sensor types as the D-IBUF, no differences in wind speed accuracy between the two aircraft is expected, which was also the observation during calibration flights.

The surface temperature of both aircraft is measured with an infrared radiation thermometer (KT15.82D in the Dornier 128 and KT19.85 in the Cessna F406, both by former Heimann, now Heitronics, Germany), with an accuracy of ±1.2 K at 20 °C surface temperature and a temporal resolution of 20 Hz. As the infrared radiation passes through the atmosphere between the surface and the aircraft, with an unknown concentration of humidity, particles or other factors that potentially influence the infrared radiation, the measured surface temperature depends on the flight altitude with a linear factor of typically between 0.1 and 0.15 K per 100 m.

The VZ-1000 laser scanner (Riegl, Austria) is included on board the Dornier 128 to measure sea state properties through an opening in the fuselage. With this setup, it is possible to measure the significant wave height and derive wave spectra along the flight path. The laser scanner covers a line of sight range of up to 450 m.

During these flights, no laser scanner was available on board the Cessna F406.

In the Dornier 128, two downward-looking cameras are integrated: the MV1-D1312-G2 (Photonfocus, Switzerland) for the visible wavelength range and the A35SC (FLIR, Germany) for the infrared wavelength range. In the Cessna F406, nadir-looking cameras had not yet been installed.

Only in the Cessna F406, sensors for measuring broadband upward and downward solar and terrestrial irradiance are included: two CMP22 pyranometers (Kipp and Zonen, The Netherlands) for measuring solar irradiance in the wavelength range 210–3600 nm and two CGR4 pyrgeometers (Kipp and Zonen, The Netherlands) for measuring terrestrial irradiance in the wavelength range 4.5–42 µm. The sensors' response time is given as <1.7 s for the solar radiation and <18 s for the terrestrial radiation, which is quite low for the high ground speed of the aircraft and cloud structures typically varying on the scale from a few hundred metres to 1 km for shallow convective clouds . Radiation measurements have been included to complement the atmospheric measurements and as an indicator to identify cloudiness.

To investigate the interaction between several wind parks or even wind park clusters and quantify wake effects, the flight pattern “meander” was applied (Fig. a). The wind speed reduction in wakes can be on the order of up to 30 % . Transects perpendicular to the mean wind direction were flown across the interacting wind park wakes. The distance between the legs was adapted to the atmospheric stability, as very long wakes occur mostly for stable conditions . For stable conditions, the spacing between legs was typically 10 km. If possible, flight legs were performed at hub height upstream, between wind parks and downstream. A safety distance of 500 m was kept to the closest row of wind turbines.

Further, vertical profiles were performed at different locations and times from an altitude of 15 m up to 1 km, if permitted by clouds, to document atmospheric stability and wind shear. In this case, the aircraft climbed continuously with a typical vertical speed of 5 m s⁻¹, and descended again, with the same vertical speed, corresponding to a horizontal translation of around 12 km per profile.

For the occasions of two aircraft, the second aircraft performed flight legs upstream of the wind park clusters or above the wind parks. The meander flight pattern was deployed during 17 out of the 49 flights. Results using the meander flight patterns have been published in and .

The erection of large wind park clusters provides an obstacle for the flow, with the consequence that the air tends to flow partly around and above the wind parks instead of passing straight through the wind turbines. This so-called blockage effect was investigated with a particular pattern (“blocking”) upstream of the wind park clusters. The expected order of magnitude of the effect is much lower than the wake effects and in the range of up to 4 % .

To investigate the flow deflection around wind parks, flights were performed directly upstream of wind parks (Fig. b). The flight permission allowed us to approach a minimum distance of 500 m to the closest wind turbines. As the effect is expected to only be small, in the range of few percent, flights were performed as close to the wind parks as possible, with upstream distances between 500 m and 2 km and distances between legs from 100 to 500 m. The blocking flight pattern was deployed during 16 out of the 49 flights. An example of the measurements of the blocking effect is shown in Sect. . As the effect is very small, it is difficult to distinguish it from the spatial and temporal variability, and so far no results using the blocking flight patterns have been published.

The phenomenon of the coastal transition was another target of the measurement flights. The coastal effect means that the abrupt changes in the surface properties at the coast, with lower surface roughness and different heat capacity, lead to modifications of the wind field and thermal stratification , which is particularly challenging to include in simulations . The increase or decrease in wind speed due to synoptic-scale changes and coastal effect can be on the order of several metres per second e.g..

To investigate changes induced by coastal effects on spatial scales of 50 km, the flight pattern “coast” was applied, which consists of meander patterns along the coastline and perpendicular to the prevailing wind direction at hub height (which is 90 m for the N4 cluster close to the eastern coast of the German Bight and 120 m for the N2 and N3 clusters close to the southern coast of the German Bight). Additionally, offshore vertical profiles were performed at the beginning and end of each leg to document the changes in stability. For the combined flights with two aircraft (flight 42 to flight 48), the Cessna F406 repeated the same legs closest to the coast and the Dornier 128 performed meander flight legs from the coast to the open water to be able to separate the effects of temporal changes in the inflow conditions and spatial effects induced by the coast.

The “coast” flight pattern was deployed during 14 out of the 49 flights.

Results using the coast flight patterns have been published in and .

Finally, flights were not only performed upstream or downstream of the wind parks, but also above the wind parks to investigate changes in the boundary layer. This was done with the flight pattern “above” (Sect. ).

For the flights with two aircraft dedicated to investigating wakes, the Dornier 128 performed the meander pattern downstream of the wind parks, while the Cessna F406 sampled upwind conditions and performed additional flight legs above the wind parks to investigate the effect above rotor height. The flights above the wind parks were done at an altitude of 75 m above the top of the rotor blades, which corresponds to an altitude of 255 m. Flight legs above wind parks were conducted with the Cessna F406 during four out of the seven joint measurement flights. So far, no results using the above flight pattern have been published.

As the flights were performed during different seasons (earliest flight of the year on 14 March and the last flight on 25 September), the temperature is highly variable between the flights. The temperature near the surface was between 4 and 19 °C (see Fig. ). Generally, temperature decreased with altitude. A slight temperature inversion up to an altitude of 50 m is visible in the mean profile, and, on average, there is a temperature inversion at an altitude of around 850 m. The whole flights were taken into account; therefore, temperature values during takeoff and landing are included, which may differ significantly from temperature values above the North Sea. The grey scale illustrates that the flights were performed as blocks, and not distributed continuously over time: lower temperatures, with near-surface air temperature in the range from 4 to 5 °C, occurred during the frequent measurements in early spring (March and April), but most flights were performed during the summer season with relatively warm near-surface air temperature in the range of 15 to 19 °C (see Tables  and ). Therefore, the atmospheric conditions during the flights do not represent climatologically relevant statistics.

Figure  shows the distribution of the lapse rate as an indicator of stability for all flights. Values near zero indicate neutral conditions. Most measurements were performed for neutral and slightly stable conditions. For wake measurements and coastal effect, stable conditions were selected on purpose for the measurement flights, as the strongest effects are associated with stable conditions. For investigating the blockage effect, both stable and unstable stratification were probed. During spring and summer, wind directions from land are typically associated with warmer air masses during the day than the sea surface, leading to stable stratification. X-Wakes investigated not only the extreme cases of stability, but also the interaction of wind park wakes for different stability regimes.

What is notable is the enhanced stability at the altitude of the lowest 100 m. On average, there is a strong increase in stability within the lowermost 100 m. This is associated with the altitude of the temperature inversion (see Fig. ) and has important consequences for the development of the wake and if there is a warming or cooling effect downstream of the wind park .

Figure  shows an overview of the wind speed conditions encountered during the measurement flights. The flights were performed under different wind speed conditions. At hub height, the wind speed varied between around 3 and 17 m s⁻¹, but most measurements were performed in the wind speed range between the cut-in speed of 4 m s⁻¹, where the wind turbines start turning, and the rated wind speed of 12 m s⁻¹, where the wind turbines reach their maximum power output. In this wind speed range, the power output cubically depends on the wind speed; therefore, reductions in wind speed have a direct impact on the energy yield.

Figure  shows a wind rose diagram of the wind speed and wind direction that was observed during all measurement flights. The measurement flights covered average wind directions between 90 and 350° at hub height.

There are several pronounced features: wind from the south-west sector was frequently associated with a wind speed between 5 and 15 m s⁻¹. This corresponds to the most frequent wind direction for the North Sea , and is typically associated with stable conditions in spring and summer . Further, conditions with wind directions from the east were measured, which is also typically associated with stable conditions. Several flights were performed during relatively low wind speed from the north-west, when typically neutral or unstable atmospheric conditions are expected.

Figure  shows the frequency of the occurrence of different values of relative humidity during all measurement flights. The profiles of relative humidity can be very different depending on atmospheric stability. On average, the relative humidity increases with altitude, which is caused by the decreasing temperature with altitude, as shown in Fig. . The decreasing relative humidity within the lowermost 100 m is in agreement with the increasing temperature in that altitude range.

Figure a shows an example of wakes downstream of several wind parks during Flight 12 in the afternoon on 29 June 2020, potentially superposed by spatial and temporal gradients of the wind speed. Despite the coverage with low-level stratocumulus clouds, the atmospheric conditions were stable, as the temperature of the air masses advected from land was higher than the water surface temperature of the North Sea. The main wind direction at hub height was 230° (south-west), and the mean wind speed was 15 m s⁻¹. The changes in the colour from red/orange to blue indicate a reduction in the wind speed to below 13 m s⁻¹. This can be observed north-east (downstream) of the wind park clusters N3 and N4. Further, the wind speed is reduced at the south-eastern edge of the flight pattern, which may be due to the proximity of the coast or downstream of the island of Heligoland. Wind park cluster N2 was partly not in operation on that day according to the flight protocol, which may explain that no pronounced reduction in wind speed is visible downstream.

Figure b shows an example of measurements upstream of the wind park cluster N4 during flight 13 in the morning on 30 June 2020. The main wind direction at hub height was 250° (west), and the mean wind speed was 17 m s⁻¹. A spatial and temporal gradient of the wind field is also obvious here, with higher wind speed toward the north and changes in wind speed between the individual flight legs, as can be seen by the different colours indicating a wind speed range between 13 m s⁻¹ and 15 m s⁻¹ in the southern part of the flight, where no influence of the blockage effect is expected. The variability in the wind speed for the time period of one flight leg (around 10 min) and for different flight legs is higher than the expected effect of wind speed reduction in the blockage effect in the range of a few percent. Therefore, continuously available datasets, e.g. from a scanning wind lidar, are more suitable to identify and quantify the blockage effect .

Figure a illustrates the coastal effect, i.e. the spatial variability in the wind field close to and induced by the coast transition from higher to lower surface roughness and from higher surface temperature variability due to lower heat capacity to more constant surface temperatures due to the large heat capacity of the waterbody. At the edges of the individual legs, vertical profiles were performed to document the vertical distribution of wind, temperature and humidity, with enhanced wind speed for higher altitudes.

In this case, for wind from the south, the wind speed increases with the distance from the coast along the wind direction, the so-called fetch length. This is not only the more frequent case, but a systematic decrease in wind speed with distance from the coast is also possible, as described in . Further, a high variability in the wind speed along the coast is observed as well. It may be influenced by local orography, with increased wind speed between the different islands and reduced wind speed downstream, as described in detail in .

Figure b illustrates the modification of the wind speed during a transect above the wind park N3 at a constant altitude of 250 m. The wind speed is enhanced at the edges of the wind park and reduced directly above. Similar investigations using airborne measurements above wind parks obtained during the project WIPAFF are presented in and .

Figure  shows an example of the variation in the measured radiation with time when underpassing a cloud. As expected, the solar downward irradiance decreases under the cloud, which is visible here as the reduction from around 600 to 300 W m⁻². At the same time, the solar upward (reflected) irradiance decreases as well as less radiation reaches the surface. The terrestrial upward irradiance decreases slightly (by less than 1 W m⁻²) along the flight leg, probably due to, on average, slightly decreasing surface temperature (lowest panel of Fig. ; see trend line) and independently of the cloud, as the water surface temperature does not rapidly change with cloud cover. The terrestrial downward irradiance increases slightly (around 1 W m⁻²) due to the enhanced temperature of the cloud. The unfiltered raw data show spikes on the order of <0.5 W m⁻², but the causes have not been identified in more detail yet. Even if the sensor response time provided by the manufacturer is quite high and despite the hemispheric field of view of the sensors, suitable algorithms can help to estimate the cloud situation during the flights.

Figure  shows measured sea surface temperatures in the area of the Jade stream close to the coast for several observation times within one day. For comparison, sea surface temperature values of the two relevant grid areas of the ERA5 model are also shown in the diagrams. Depending on the tides and the exact location, the sea surface temperature varies around 2 °C during the same day. The agreement between ERA5 sea surface temperature and measured surface temperature was very close for the time of around 14:00 UTC. For other times of the day, the difference between ERA5 sea surface temperature and the measured temperature was in the range of 1 °C. This indicates that the ERA5 model is not capable of resolving the temporal changes in sea surface temperature associated with tides very well. The gathered data have the potential to deliver contributions for a better understanding of the complex stream and surface temperature situation in coastal areas.