The aim of the Pallas Cloud Experiment (PaCE) was the characterisation of aerosols and clouds in the vertical column at high resolution around the Pallas-Sodankylä Global Atmospheric Watch (GAW) Sammaltunturi station in Finnish Lapland, approximately 160 km north of the Arctic Circle . The field campaign took place from 15 September to 15 December 2022 and was organised by the Finnish Meteorological Institute (FMI). Besides the Max Planck Institute for Dynamics and Self-Organization (MPI-DS) and the FMI, the Swiss Federal Institute of Technology Lausanne, the Karlsruhe Institute of Technology, the University of Hertfordshire, and the Vienna University of Technology participated in this campaign. The measurements were taken via fixed-wing and multi-copter drones, tethered balloon systems, mountain-top stations, ground stations and remote sensing instruments. A general overview within this Special Issue is given in . In this paper we present data obtained with the MPCK⁺ via in-situ measurements aboard the tethered balloon system MPCK platform between 19 and 26 September 2022.

The site from which the balloons with the instruments were launched is within the Pallas-Yllästunturi National Park in Finnish Lapland at the western shoreline of lake Pallasjärvi, approximately 280 m above mean sea level (a.m.s.l.)

For location details, visit https://en.ilmatieteenlaitos.fi/pallas-atmosphere-ecosystem-supersite (last access: 2 August 2026).

In this section, we introduce the Max Planck CloudKite platform for airborne atmospheric measurements, abbreviated MPCK platform. It consists of one or more helium-filled tethered Helikites (type Desert Star Helikite by Allsopp Helikites Ltd.) and the accessories needed for flight operation (ground anchor, pulley, motor winch with main tether, etc.). The MPCK platform can carry different scientific payloads, such as the MPCK⁺ or the WinDarts .

A Helikite is a balloon-kite combination whose main advantage over an ordinary balloon is the stable position in windy conditions. While ordinary balloons are being pushed towards the ground, a Helikite is stabilised by keel and sail underneath the bubble at a 45° angle to the vertical. The flight altitude, typically up to 2 km, is controlled by a motorised winch reeling in/out the main tether. By changing the tether length, the flight altitude is regulated. The maximum recommended wind speed for use is 25 ms-1. During PaCE 2022, the wind speed was always below 20 ms-1. A photo from a test site next to the MPI-DS shows the MPCK platform with one 250 m³ Helikite (Fig. ). During most of the PaCE 2022 flights, we operated a tandem configuration with one 250 m³ Helikite as the lower balloon and one 34 m³ Helikite as the higher balloon. This tandem configuration increases the static lift by 10 %, and, due to the stabilising effect by the small Helikite, take-off and landing of the big Helikite can be performed under much safer conditions, in particular in gusty winds. Flight strategies with the MPCK platform include constant altitude flights, vertical profiles, and staircase flights. The static and dynamic lift generated by the Helikites was sufficient to reach altitudes of more than 1200 m above ground with a 100 kg payload.

While the MPCK platform has been used for flying different scientific payloads during PaCE 2022, we focus on the MPCK⁺ instrument in this paper. An overview of the MPCK⁺ installation on the main tether is shown in Fig. . This type of installation was chosen in order to minimise the influence of the balloons on the dynamics measured by the MPCK⁺ instrument. A particular challenge is the measurement of atmospheric variables under ambient conditions, i.e. the measurements not being influenced by the instrument deployment or instrument geometry. Here, the MPCK⁺ instrument is hanging on the main tether at least 50 to 100 m below the 250 m³ Helikite, which is the equivalent vertical distance of 5 to 10 balloon diameters. This mounting option is called tether-mount in . One side effect of this setup is the non-linear relationship between line length and instrument altitude due to the catenary, which means that the gain of altitude decreases with increasing line length.

The MPCK⁺ consists of nine devices to measure different platform and atmospheric properties. A summary of the instrument specifications is given in Table . The core unit consists of a one-dimensional (1D) pitot static system (Model type PSS8, Simtec AG) for measurement of air speed, barometric altitude and temperature, an inertial navigation system (Model Ellipse D, SBG Systems) for measurement of the platform position, orientation, velocity, angular velocity, acceleration and angular acceleration, a humidity and temperature sensor unit (labeled “rht” in Table ) with an AM2315 sensor (Guangzhou Aosong Electronic Co., Ltd.) and a nearby HMP7 sensor (Vaisala), a backup global navigation satellite system (GNSS), model NEO-M8U (uBlox), and a thermal anemometry unit, which comprises a 54T42 mini-CTA (Dantec Dynamics) and a T7 Data Acquisition unit (LabJack Corporation). Particle size and concentration in 1D are measured by a Fast Cloud Droplet Probe (FCDP) from Stratton Park Engineering Company (SPEC Inc.). The basic variables are measured by multiple sensors to allow for redundancy of essential variables and in-situ instrument intercomparison. For measurement of 3D particle position, size and shape, an in-house-built digital inline holography unit is used. Particle velocity in the stream-wise-vertical plane is measured with an in-house-built Particle Imaging Velocimetry (PIV) unit. Inline holography and PIV have a partially overlapping sample volume. In addition, every fifth hologram (75 Hz recording rate) has a corresponding PIV image pair (15 Hz recording rate) at exactly the same time. This unique design allows in-situ inter-instrument calibration as cloud droplets are detected by both instruments in the overlapping volume. The low air speed together with the high image acquisition rate yields inter-hologram distances on the order of 5 to 6 cm. One feature of the MPCK⁺ control software is that particles detected by the FCDP above a given size and concentration threshold can be used to automatically trigger the holography and PIV unit. Data acquired by the two imaging units are not part of the published data set, but the flight table provides information about availability of these data.

Our measurements provide valuable insights into clouds on spatial scales typically smaller than 10 m, depending on wind speed and acquisition rate. Also, the thermodynamic and dynamic properties of the atmospheric boundary layer can be studied using these data, for example the role of wind shear and convection in different meteorological conditions. The details of the MPCK⁺ are summarised in another paper .

Besides the MPCK⁺ instrument, additional scientific instruments were operating simultaneously: the WinDarts see, and the FishBox, which was developed by the Finnish Meteorological Institute (FMI) team and comprises several aerosol particle size spectrometers, a GPS and meteorological instrumentation seefor details. Along with the MPCK platform and its scientific payloads, we operated a ground weather station next to the launch site. This weather station was equipped with a WS500 UMB combined weather monitor (Lufft) (for static air pressure, temperature, relative humidity and 2D wind), a uSonic Class A-MP 3D sonic anemometer (for high-resolution 3D wind and temperature), a LI-7500DS optical trace gas analyser (for high resolution data of static air pressure, temperature, water vapour and CO₂ concentration), and a CS 110 field mill (for the vertical component of the atmospheric electric field). In addition to these instruments, we operated a LD 250 lightning detector and a pingStation Mode S and ADS-B receiver for safety reasons (uAvionix), i.e. to monitor the risk of thunderstorms and lightning and to monitor the airspace. More details about the ground monitoring and a schematic of the setup are presented in .

Between 19 and 26 September 2022, we performed eight flights with the MPCK⁺ with a total flight time of 19 h and 37 min. The longest flight took 3 h and 12 min, the shortest flight was 1 h and 32 min long. While the lowest flight had a maximum altitude of 240 m above ground, the highest flight was at maximum 1210 m above ground. A summary of all the flights is given in Table . The flight identifier is the date of departure, followed by the take-off UTC time (hours and minutes).

Figure  shows the barometric altitude time series records for flights 20220919.1236, 20220921.1721, 20220923.0711, 20220923.0914, 20220923.1622, 20220925.0839, 20220926.1209 and 20220926.1712. The first six flights were done with the tandem configuration whereas the flights on 26 September 2022 were done with just the 250 m³ Helikite. Flights 20220923.0711, 20220925.0839 and 20220926.1209 had at least one WinDart measuring simultaneously with the MPCK⁺ at a different altitude. These data are published in .

The data are provided in NetCDF format, following a common file naming structure. This structure is as follows: MPDS.MPCKplus.b1.yyyymmdd.hhmm.nc, where:

MPDS is the institute identifier (Max Planck Institute for Dynamics and Self-Organization).

Each data file contains 50 attributes (institution, contact information, standard name vocabulary, campaign time, campaign location, flight number and many more) and one group, which is labeled “Level_1”. This group contains 10 subgroups and 51 attributes. Each subgroup represents one particular measurement device. The device names are summarised in Table . An overview of the data structure for the different devices is given in Table . Each group and data set contains attributes that describe the type of data, the calibration factors applied, and the physical units.

Not all devices mentioned in Table are included in the data published in this paper. The additional groups, which are excluded from the published data, are labjack, holo, piv and serialcameras due to ongoing QC/QA (quality control/quality assurance) procedures.

Within the comprehensive list of variables contained in the published NetCDF files, there is a subset of core variables that have been used to do the analyses and generate the figures presented in this paper. Table lists and describes the most important variables, which are usually enough for most users. For generating these variables, the calibration polynomials either from the manufacturer or an in-house calibration have been applied to the raw data. Especially for the true air velocity values, which are also required for the calculation of the droplet concentration, corrections of the air velocity values are necessary due to the geometry of the MPCK⁺, more on this is provided in Sect. . For relative humidity, temperature and dew point, we recommend using the data in group corrected_rht as described in Table .

At this point, it needs to be mentioned that there are two more correction factors. These correction factors result from tests with the MPCK⁺ in our wind tunnel to determine the disparity between undisturbed and measured air speed at the pitot tube and other locations, e.g. at the probing volume of the FCDP for calculation of droplet concentration. The calibrated air speed from the pss8 has already been corrected to air speed at infinity by applying the correction factor 1/0.647. At the sample volume of the FCDP, the actual air speed is the air speed at infinity multiplied by 1.09. The second correction factor is included in the sample Python code linked to this document to ensure that users are given clear instructions on how to calculate the concentration and apply various corrections, and to facilitate the use of data.

Different devices and microcontrollers acquire the data of the various instruments of the MPCK⁺. This requires time synchronisation of all instruments against one master clock. The synchronisation is the very first step in the data processing from raw data to Level 1 processed data. After synchronisation, the data are checked for physical plausibility to detect obvious issues due to a broken sensor, a loose connection or other reasons. In the next step, the calibration polynomials are applied to the raw data; they are either from a calibration certificate of the manufacturer or based on an in-house calibration. Data sets of a missing sensor have all values above 9.9×1036. Anything larger than 264 is considered invalid data or missing data.

The accuracy of temperature and relative humidity sensors given by the manufacturer is summarised in Table . However, these values are valid under laboratory conditions. To clarify the accuracy of humidity measurements, we chose flights that had clouds at the instrument altitude. For this reason we selected flights 20220925.0839 and 20220926.1209 as examples. Starting with Flight 20220925.0839 as shown in Fig. , we see a strong positive deviation of the pss8 temperature data compared to the rht and hmp7 units of about +3.5 K. The hmp7 dew point and temperature data suggest high humidity during the entire period while the FCDP detected particles in a few short events during the flight. As the deviation between the rht and hmp7 units in temperature is typically smaller than 200 mK, we trust both sensors but prefer the rht temperature data over the hmp7 temperature data due to the fact that the rht sensor is not heated. However, the rht dew point is too low in those sections where the FCDP detected cloud droplets. Therefore, we prefer the hmp7 dew point over the rht dew point.

A different example is shown in Fig.  for Flight 20220926.1209. Here, the instrument was flying in clouds for most of the flight time. Again, we see little deviation in the temperature data between rht and hmp7. The strong deviation in the dew point temperature is about the same as in the previous example, and also here, the dew point from the hmp7 sensor is more trustworthy than the dew point data from the rht sensor. Based on the results shown in Figs. and , we decided to show only rht temperature, hmp7 dew point and ground station temperature in the following graphs instead of plotting data for each individual sensor.

In this section we present example time series data from the different sensors of MPCK⁺ during two selected flights. One of them was without clouds at the instrument altitude, during the other one the instrument was inside the cloud throughout most of the flight time.

One example of data from a flight without clouds and precipitation (Flight 20220921.1721) is presented in Fig. . The barometric altitude and GPS altitude in the top panel have little deviation before 18:10 and after 18:50 UTC. In between, there is a deviation of 15 to 20 m between the sbg altitude and ublox altitude, which is at least partly due to the fact that the sbg has a dual-band GNSS. The second row of Fig.  shows the pitch and roll angle of the MPCK⁺. At lower air speed (around 10 ms-1 visible in the third row of Fig. ), the instrument orientation is almost ideal (pitch angle close to zero, roll angle slightly positive). With stronger wind, the pitch angle becomes negative, but no more than 10°. At these angles, the velocity measurement is not impaired by non-ideal platform orientation. In comparison to the wind speed at ground level (blue line in the third row of Fig. ), the upper air wind speed is between 8 and 15 ms-1 higher. The second to last panel in Fig.  shows temperature and dew point as defined in the previous section along with the temperature measured by the ground station. The bottom panel in Fig.  shows the particle size and concentration measured by the FCDP as a function of time. There are particle counts at low concentration between 3 and 15 µm particle diameter, which most likely correspond to a few large aerosol particles. These particles were present below 800 m a.m.s.l.

In contrast to Flight 20220921.1721, Flight 20220926.1209 was mostly inside clouds as shown in Fig. . This is indicated by much higher particle concentration in the FCDP data. The cloud base was reached only about 70 m above ground during this flight, and particle data from the FCDP indicate a bimodal cloud droplet size distribution (CDSD) throughout the sampled cloud portion. The primary mode is around 5 µm, while the secondary mode is around 20–25 µm. There is also a strong signal of particles larger than 40 µm diameter, which may be attributable to those droplets that turn into drizzle by collision and coalescence. The analysis of the entire CDSD of this cloud, including droplets larger than 50 µm, is subject to a follow-up study with additional data from the inline holography and PIV units. The dew point stays slightly above the temperature during the entire time series of this flight, which indicates water-saturated to water-supersaturated conditions. However, the trend in air speed measured by the pss8 around 13:00 UTC is artificial as the tubing of the system got clogged by cloud water. Such events need to be identified manually as there is no sensor within the system which could automatically detect water inside the tubing. The instrument orientation was very stable in pitch and roll angle, but the yaw angle is characterised by sudden changes of more than 30° within a few seconds. Vertical wind shear and convection may partly explain this trend, as the balloon might have been exposed to a different wind speed and direction than the MPCK⁺.

In addition to the time series data, the vertical profile data of temperature and dew point allow some conclusions about the meteorological situation during the measurements. For all eight flights, these profiles are summarised in Fig. . In the data of Flight 20220921.1721 we see an almost isothermal layer from the ground to 800 m altitude above mean sea level with an inversion layer between 800 and 900 m altitude. While the FCDP data in Fig.  indicate few particles below 800 m a.m.s.l. (which is about 500 m above ground), there are no particles detected at higher altitude. So we can conclude that the aerosol particles in the boundary layer did not reach the free atmosphere during this measurement. Two more contrasting examples are Flight 20220923.0914 and Flight 20220925.0839 as the temperature profile decreases with altitude and the dew point data indicate water-saturated conditions at some altitude. During these two days, there was cumulus convection and a few clouds were sampled with the MPCK⁺. The trend in temperature and dew point during Flight 20220926.1209 is quite interesting as the data indicate a rapid cooling around 270 m above ground. However, for flights 20220926.1209 and 20220926.1712, the dew point temperature is about 0.5 K higher than the air temperature, which is unphysical. This discrepancy can be attributed to measurement uncertainties arising from the combination of two sensors used to derive and correct these quantities, leading to a systematic bias in these two flights, both conducted in near-saturated conditions.

As already mentioned in the data description, the published data set does not contain holograms or processed holograms from the MPCK⁺ holography unit. Nevertheless, we would like to show an example of the particle data that can be derived from a single image. Details about the hologram processing and data extraction are presented in . Figure displays the position of each droplet with colour-coded diameter in the top panel, and the logarithmically weighted droplet size distribution in the bottom panel. Each dot in the top panel represents one droplet. This allows a detailed look into the spatial distribution of droplets to examine droplet clustering, the presence of vortices or other phenomena which are not accessible by one-dimensional data from e.g. FCDP measurements. The droplet size distribution in the bottom panel has its maximum close to 20 µm and another local maximum around 40 µm, which is in agreement with the droplet size distribution measured by the FCDP (compare the red and the blue curve in Fig. ). The primary mode around 5 µm droplet diameter in the FCDP data is too small to be measured by the holography unit, as its effective pixel pitch is 3 µm and objects smaller than 2 pixels in diameter are not distinguishable from background noise. As found in previous studies, the droplet concentration measured with the holography unit is in the same range as the FCDP data for particles with at least 11 µm diameter. On the other hand, we see a statistical effect when comparing the particle concentration around 40 µm diameter: There was one droplet in the holographic sample volume of 8.6 cm³ within this size bin whereas the FCDP detected two droplets in a volume of less than 1.5 cm³. The size distribution statistics from the holography unit are more robust compared to the FCDP data, thanks to the sample volume, which is about 6 times larger. The effect of counting statistics is also visible in the shaded area and in the FCDP median concentration in the bottom panel of Fig. .

The orientation of the MPCK⁺ measured with the primary (sbg) and backup (powerinterlock) INS is summarised in Fig.  and Table . Flight 20220919.1236 was excluded from the analysis as the repeated power outages precluded a GNSS fix of both INS. The median pitch angle measured by the sbg device was between -1.6 and +5.9° with the higher values in the last three flights. However, we see the exact opposite trend in the roll angle statistics with higher angles in the first five flights. The agreement between sbg and powerinterlock is better than 1° for the median pitch and roll angle. The highest variability, as expressed by the standard deviation, is seen in the yaw angle statistics. As the yaw angle is primarily determined by the course of the Helikite, wind shear in the boundary layer may contribute to the high variability. The histograms in Fig.  show a relatively narrow angular distribution with a full width at half maximum (FWHM) of 3–6° with a maximum close to zero. We conclude that the orientation of the MPCK⁺ was stable and nearly horizontal.