The intercomparison campaign of hygroscopic aerosol properties using remote sensing and in situ techniques (HYGRO19 campaign) was carried out from June to December 2019 at the MeteoSwiss aerological station in Payerne, Switzerland (46.82° N, 6.95° E, 49 m above sea level (a.s.l.)), which forms part of ACTRIS (Aerosol, Clouds and Trace Gases Research Infrastructure), EARLINET (European Aerosol Research Lidar Network), E-PROFILE (European Profiling Network), and GRUAN (GCOS Reference Upper-Air Network). The HYGRO19 campaign focused on the characterisation of hygroscopic aerosol growth under ambient atmospheric conditions using three complementary approaches: remote sensing measurements from the RALMO Raman lidar, ground-based in situ measurements from tandem nephelometers, and vertically resolved in situ measurements from COBALD sondes operated alongside RS41 radiosondes (COBALD-RS41). A central objective was to assess the consistency and limitations of the hygroscopicity parameters obtained with these techniques, accounting for their different optical properties, wavelengths, and sampling conditions. Continuous RALMO measurements of aerosol backscatter, temperature, and water vapour mixing ratio also enabled the temporal and vertical variability of aerosol hygroscopicity to be investigated within selected well-mixed layers. Additional aerosol and meteorological observations were used to characterise the atmospheric conditions and support the interpretation of the observed hygroscopic behaviour. The experimental station is located in a rural area of the Swiss Plateau, between the Jura Mountains (25 km to the Northwest) and the Alpine foothills (20 km to the Southeast). The site is characterised by a mid latitude continental climate with some influence of long-range transported pollution .

The RAman Lidar for Meteorological Observations (RALMO) system is a custom-designed instrument developed by the Swiss Federal Institute of Technology (EPFL) in collaboration with MeteoSwiss. It has been operating in the Payerne station since August 2008 and joined EARLINET the same year. The RALMO system uses a Nd:YAG laser source that emits pulses at 355 nm (400 mJ), with a repetition rate of 30 Hz and a pulse duration of 8 ns. Before the laser emits to the atmosphere it is expanded to a diameter of 14 cm to reduce beam divergence (to 0.1 mrad). The receiving system consists of four telescopes with 30 cm parabolic mirrors symmetrically arranged with the vertically mounted beam expander to collect the backscattered photons, giving a field of views (FOV) of 0.2 mrad. These characteristics enable RALMO to operate during both daytime and nighttime. Optical fibres connect the telescope mirrors to two grating polychromators, which isolate the rotational-vibrational (VR) and pure-rotational (RR) signals. A detailed description of RALMO can be found in .

The system optimises its operability in the incomplete overlap region because the Raman backscatter is excited by a single monochromatic beam and the designed reflective-type lidar telescopes minimise chromatic aberrations. Moreover, the fibre-coupled diffraction grating-based polychromator further reduces uncertainties induced by range-dependent optical imperfections of the lidar receiver to a negligible level . The water vapour mixing ratio is computed from VR measurements of nitrogen and water vapour at wavelengths of 386.7 and 407.5 nm. The accuracy of RALMO water vapour measurements has been validated by , showing good agreement with radiosonde observations and integrated water vapour retrieved from the Global Positioning System (GPS) and MWR.

Temperature measurements with RALMO are obtained using four sections of the RR Raman spectrum around 355 nm, which relies on the rotational Raman scattering of oxygen and nitrogen molecules. The RR signal is separated from the elastic (Rayleigh and Mie) scattering using a two-stage temperature polychromator, which isolates specific rotational Raman lines while guiding the elastic signal through optical fibres to the detector. High-efficiency parabolic mirrors collect the signals, which are then filtered, fibre-coupled, and diffracted by a reflective holographic grating to resolve Stokes and anti-Stokes lines from different rotational quantum numbers. This configuration enables accurate and stable retrieval of atmospheric temperature profiles. Further details and a complete validation of the RALMO temperature channel are provided by . The system also registers the backscattered signal at 355 nm backscattered signal, which combined with one RR measurement allows the retrieval of independent aerosol backscatter and extinction measurements .

For water vapour, nitrogen, and oxygen, the interference filters are centred at 407.5, 386.7 and 375.4 nm, respectively, each with a full width at half maximum (FWHM) of 0.33 nm. This configuration ensures minimal temperature sensitivity of the Raman scattering signal, with a relative change in transmitted intensity of 0.4 % per 100 K , in agreement with values reported in the literature . It also minimises spectral differences between the laser emission and Raman channels .

COBALD is a lightweight (500 g) aerosol backscatter detector for balloon-borne measurements developed at the Swiss Federal Institute of Technology (ETH Zürich) based on the original prototype by . Using two light-emitting diodes (LEDs) as light sources and a photodiode detector with FOV of 6°, COBALD provides aerosol backscatter at 455 and 940 nm with an accuracy of 5 % and a precision of 1 % for both channels, and temporal and spatial resolutions of 1 s and 5 m, respectively . During the HYGRO19 field campaign, COBALD was combined with an additional meteorological radiosonde (RS41, Vaisala) to obtain measurements of temperature, pressure, and relative humidity. The RS41 sonde specifications for nominal measurement uncertainties, as provided by the manufacturer, are 0.3 °C for temperature below 16 km and 0.4 °C above that altitude, 0.01 hPa (decreasing with altitude) for pressure, and 4 % RH .

The COBALD measurements are typically expressed as the backscatter ratio (BSR, Eq. ), defined as the ratio of the total measured signal (normalised to the altitude-dependent LED power) to its molecular contribution . The molecular contribution is computed following using the atmospheric number density of molecules derived from the collocated RS . A molecular extinction-to-backscatter ratio of 8π/3 sr is assumed. 1BSR=βaer(λ)+βmol(λ)βmol(λ)

During the HYGRO19 field campaign, the COBALD–RS41 sonde was used to obtain vertically resolved in situ observations of aerosol hygroscopicity parameters, providing the reference measurements for direct comparison with those retrieved from RALMO. However, the COBALD and RALMO systems differ in receiver FOV (0.01° for RALMO), which results in systematic differences in backscatter measurements that are corrected using Mie scattering simulations .

The tandem nephelometers consisted of two integrating nephelometers operating in parallel and measuring the same air sample. One instrument operated at RH lower than 40 % (dry conditions), whereas the second was equipped with a humidification system allowing measurements at different RH conditions from 40 % to 90 % (wet conditions) . The dry scattering coefficients (σspdry) were measured using an Ecotech Aurora 3000 nephelometer at three wavelengths (450, 525, and 635 nm) with a 1 min sampling interval. The scattering coefficients under wet conditions were measured using a TSI 3563 integrating nephelometer at 450, 550, and 700 nm, recording data every 1 min. The nephelometer data were corrected to account for angular truncation errors and instrument non-idealities, following the method of for the TSI instrument and for the Ecotech nephelometer. The setup consisted of a 30 min cycle beginning with a baseline (zero) measurement, followed by gradual increases of RH up to approximately 85 %–90 %. For the computations of the hygroscopicity parameters, the dry scattering coefficient measured at 525 nm by the Ecotech instrument was corrected to 550 nm using the Ångström exponent .

During the HYGRO19 campaign, complementary instruments were also employed to provide a comprehensive characterisation of aerosols. Continuous measurements of aerosol optical depth (AOD) and column-integrated extinction Ångström exponent (AOD-AE) were obtained from an AERONET sun photometer (SP) operating at Payerne since 2000 Level 2 Version 3, . The station is also part of the European E‑PROFILE network since 2008 and operates a CHM15k ceilometer (Lufft) emitting at 1064 nm, which continuously records aerosol attenuated backscatter. Furthermore, within the framework of E‑PROFILE, a Doppler wind lidar profiler operates to measure horizontal and vertical wind components every 10 min, typically covering altitudes from 150 m to about 5 km above ground level (a.g.l.) .

Ground-based in situ measurements included an Aethalometer (AE31), operated by the Centre for Energy and Environmental Sciences at the Paul Scherrer Institute (PSI), capable of providing aerosol absorption coefficients (σab) at seven wavelengths (from 370 to 950 nm). The Aethalometer can also estimate equivalent black carbon (EBC) concentrations using wavelength-dependent mass absorption cross sections . Data processing followed Global Atmosphere Watch (GAW)/ACTRIS recommendations, including corrections for filter-loading and multiple-scattering effects and was harmonised as described in .

Additional in situ measurements of aerosol properties were conducted by the Swiss Federal Laboratories for Materials Science and Technology (EMPA) as part of the NABEL (National Air Pollution Monitoring Network) programme . Particle mass concentrations of PM_2.5 and PM₁₀ were determined in real time with a Fidas 200 optical aerosol spectrometer and corrected versus gravimetric reference data . Elemental carbon (EC) in PM_2.5 was obtained in the laboratory from PM_2.5 filter samples using a thermal optical EC analyser (Sunset Laboratory Inc.) following the EUSAAR2 protocol . The concentrations of sulphate (SO42-) and nitrate (NO3-) in PM₁₀ were determined from PM₁₀ filter samples using ion chromatography . Ammonium (NH4+) was estimated assuming complete neutralisation of sulphate and nitrate , and organic matter (OM) was derived from PM mass closure (PM = inorganic species + black carbon (BC) + OM). Furthermore, total carbon (TC) concentrations were obtained using the novel FATCAT (Fast Thermal Carbon Totalizer) system, developed by the University of Applied Sciences and Arts Northwestern Switzerland (FHNW), with a detection limit of 0.16 µg-C (micrograms of carbon) for a sampling interval of 2 h . Additionally, aerosol size distributions in the range 10 to 487 nm, were measured using a scanning mobility particle sizer (SMPS, TSI 3034) PSI,.

The combined use of elastic and Raman backscattered signals allows the independent calculation of the aerosol backscatter coefficient (βaer) as described by and : 2βaer(z,λL)=βmol(z,λL)⋅k⋅P(z,λL)P(z,λi)-1 where the sub-index i indicates the Raman molecular species, βmol (z,λL) is the molecular backscatter coefficient, k is the calibration constant, and P (z,λ) is the backscattered signal from range z at either the elastic (λL) or the Raman shifted λi wavelengths. Stokes and anti-Stokes portions of the pure-rotational spectrum with opposite temperature dependence are summed, giving in good approximation a temperature-independent inelastic signal . The RALMO system provide accurate aerosol backscattered and extinction coefficients at 355 nm with a bias of −2 % ± 37 % (-0.018±0.237 Mm⁻¹ sr⁻¹) compared with COBALD at 455 nm .

Water vapour mixing ratio (r) is defined as the ratio of the mass of water vapour to the mass of dry air in a sample of the atmosphere . Consequently, the ratio NH2O (z) / NN2 (z) is proportional to the water vapour mixing ratio . Assuming identical overlap factors and the range-independent Raman backscatter cross sections for the two signals, this ratio can be expressed as : 3NH2O(z)NN2(z)=P(z,λH2O)P(z,λN2)⋅FN2[T(z)]FH2O[T(z)]⋅KN2σN2KH2OσH2O⋅exp⁡∫0z[α(z′,λH2O)-α(z′,λN2)]dz′ and thus, 4r(z)=P(z,λH2O)P(z,λN2)⋅FN2[T(z)]FH2O[T(z)]⋅K⋅exp⁡∫0z[α(z′,λH2O)-α(z′,λN2)]dz′ where the term P(z,λH2O)/P(z,λN2) represents the backscattered signal ratio, FN2[T(z)]/FH2O[T(z)] is the ratio of the temperature-dependent functions for the Raman nitrogen and water vapour channels, and K is the instrument calibration constant. For RALMO the calibration constant was determined by combining absolute calibration using collocated radiosondes with relative calibration based on the solar background . The passband filter employed in RALMO allows the ratio FMR[T(z)]/FH2O[T(z)] to be reasonably approximated as 1 (Sect. ), resulting in essentially temperature-independent measurements .

The exponential term (Eq. ) accounts for the difference in atmospheric transmission between the nitrogen and water vapour Raman wavelengths, mainly due to the λ-4 dependence of Rayleigh scattering by molecules and an additional contribution from wavelength-dependent Mie scattering by aerosols. Molecular extinction can be estimated assuming US Standard Atmosphere, while the differential aerosol extinction is generally negligible under typical Payerne conditions , though it can reach up to 10 % in hazy conditions .

Temperature measurements with RALMO are obtained using RR Raman lidar signals , and combining these with water vapour mixing ratio measurements, it is possible to obtain RH profiles following : 5RH=e(z)ew(z)⋅100 where e(z) is the water vapour pressure, and ew(z) denotes the saturation vapour pressure. The water vapour pressure is related to the water vapour mixing ratio through: 6e(z)=p(z)⋅r(z)0.622+r(z) where p(z) represents the air pressure profile, estimated from the closest RS41 operational at Payerne. The saturation vapour pressure ew(z) is computed using the empirical formulation: 7ew(z)=6.107⋅exp⁡MA[T(z)-273]MB+[T(z)-273] where the constants MA and MB are 17.08 and 234.2, for temperature above 273 K . RALMO relative humidity profiles have been validated against collocated RS41 radiosondes , demonstrating strong performance. These capabilities make RALMO one of the few instruments worldwide capable of providing continuous daytime and nighttime profiles of water vapour mixing ratio, temperature, and relative humidity.

Aerosol hygroscopicity refers to the equilibrium water content of water-soluble particulate matter at RH below 100 % and, thereby altering particle size, shape and optical properties. Particularly, the impact of hygroscopic growth on aerosol optical properties is commonly studied using the enhancement factor, fζ (RH), defined by : 8fζ(RH)=ζ(RH)ζ(RHref) where ζ(λ,RH) denotes an optical or microphysical aerosol property at RH, while the reference value, RH_ref, corresponds to the lowest value within the analysed layer or time interval.

In the case of laboratory measurements, when RH can be controlled, the scattering enhancement factor can be determined using two instruments that measure the same aerosol optical property, such as the aerosol scattering coefficient. One instrument operates at RH below 40 % (dry conditions), whereas the second is equipped with a humidification system that enables measurements over an RH range from 40 % to 90 % (wet conditions). This configuration, known as a tandem nephelometer , allows the scattering enhancement factor to be determined using the following two-parameter equation : 9fσ(RH)=a(1-RH)-γσ where a is the value at RH = 0 % and γσ is the Hänel parameter characterising the magnitude of hygroscopic growth for the scattering enhancement. In the case of laboratory measurements, different RH ranges can be controlled, allowing measurements over a wide RH range from 40 % to 90 %. This enables a more robust determination of a through linear fitting. Moreover, the use of the same measurement approach at different sites allows a direct intercomparison of the scattering enhancement factor fσ (λ,85%/<40 %). Hereafter, fσ(λ) for simplicity.

For lidar measurements, the aerosol optical parameter used to study aerosol hygroscopicity is the aerosol backscatter coefficient (βaer) and the backscatter enhancement factor is described as : 10fβ(RH)=βaer(RH)βaer(RHref)=1-RH1-RHref-γβ In the particular case of the RALMO system, the Hänel parameter (γβ) can only be obtained at 355 nm. Since ambient RH cannot be controlled, it must be ensured that changes in aerosol optical properties are solely related to changes in RH. This is achieved by assuming a well-mixed atmospheric layer that typically occurs when water vapour mixing ratio and potential temperature profiles remain nearly constant with altitude . Additionally, following , potential changes in air-mass origin with altitude should be assessed using backward trajectories as complementary information to the thermodynamic profiles. When these conditions are fulfilled, RH_ref is selected as the point with the lowest RH, typically corresponding to the lowest altitude, as RH and βaer increase simultaneously with altitude . Consequently, humidograms (fζ(RH) versus RH) can be computed, and the Hänel parameter is obtained as the slope of the linear fit between the natural logarithms of βaer and RH. Note that, for comparison with tandem nephelometers and previous studies, the backscatter enhancement factor is extrapolated or interpolated to the 40 %–85 % range (fβ (λ,85%/40 %)), depending on the measured RH range. In the following, it is denoted as fβ(λ).

It should be highlighted that the Hänel parameter may depend on the aerosol optical property for its computation. Consequently, γ values derived from aerosol backscatter measurements are not directly comparable to those obtained from total scattering measurements from tandem nephelometers. Moreover, the spectral dependence of hygroscopic growth can vary with particle size, as shown by Mie simulations in previous studies . Therefore, the enhancement factor for the same RH may differ between wavelengths and particle sizes, which further complicates comparisons between the different techniques. In addition, uncertainties associated with the retrieval of aerosol hygroscopicity parameters should also be considered. In the case of lidar-derived parameters, these uncertainties can reach up to 20 % for RH uncertainties below 10 % .

Continuous measurements of AOD and extinction AE (AOD-AE) obtained from the AERONET sun photometer during the HYGRO19 campaign in Payerne were analysed to characterise the columnar aerosol properties (Fig. ). Two clearly different periods were identified: the first, from June to August, characterised by relatively high aerosol loading, and the second, from September to December, dominated by low-pressure systems, persistent cloudiness, and lower aerosol concentrations . During the first period, a large number of measurements were available, with a median AOD (500 nm) of 0.24 ± 0.09, where the uncertainty is expressed as half of the interquartile range (IQR/2 = (Q3–Q1)/2)). Fine-mode particles predominated, as indicated by the median AOD-AE (440–870 nm) value of 1.50 ± 0.19. Backward trajectory analyses using the HYSPLIT model (not shown) revealed that most air masses originated over mainland Europe, a major source of anthropogenic aerosols. However, on 26 and 27 June, an event influenced by North African air masses was observed, with high aerosol load (AOD (500 nm) up to 0.3) and low AOD-AE (reaching 0.4), typical of long-range transport of Saharan dust. In contrast, fewer measurements were available during the second period. The median AOD (500 nm) and AOD-AE (440–870 nm) values were 0.07 ± 0.02 and 1.15 ± 0.13, respectively, characteristic of continental background aerosol.

Figure shows the temporal evolution of in situ ground-based measurements of aerosol scattering (σsp, a) and absorption (σab, c) coefficients at different wavelengths, scattering Ångström exponent (SAE, b), absorption Ångström exponent (AAE, d), single scattering albedo (SSA, e), particulate matter (f), and chemical composition (g–h). Data are hourly values, except for TC concentrations and chemical compositions that correspond to 2-hourly and daily measurements, respectively (panels g and h). Unfortunately, during the HYGRO19 campaign, data from the Aurora dry nephelometer and the Aethalometer AE31 were available only since 6–8 August (a–e, g), whereas FATCAT data date since 27 July 2019 (g). The main statistical parameter for the entire campaign, as well as for the first and second periods separately, are summarised in Table .

During the campaign, the temporal evolution of the scattering coefficient (Fig. a) shows remarkable day-to-day variability, with values ranging from 0.97 to 206 Mm⁻¹, and median values of 28 ± 21 Mm⁻¹ and 19 ± 14 Mm⁻¹ at 450 and 550 nm, respectively. Several intense events with higher scattering coefficients values were observed between 21 August and 21 September, 17 October and 1, 9, 26, 30 November and 7 December 2019, coinciding with increases in aerosol loading (Fig. ) and particulate matter (Fig. f). Concurrently, the analysis of the SAE (450–635 nm, Fig. b) suggests a predominance of fine particles (SAE =1.64±0.26), although some outliers are observed. Separately, the AERONET AOD-AE measurements (Fig. b) also suggest a general predominance of fine particles in the atmospheric column, with a median AOD-AE (440–870 nm) value of 1.46 ± 0.21 for the entire campaign. These observations refer to different temporal coverage and sampling conditions and should not be interpreted as a quantitative agreement between the two datasets. The predominance of fine particles near the surface is also supported by the small difference observed between PM₁₀ and PM_2.5 (PM10-2.5=3.0±1.6 µg m⁻³, Fig. f). The low PM concentrations (PM10=13.2 µg m⁻³ and PM2.5=9.2 µg m⁻³, ), characterised by a predominance of fine-mode (PM_2.5) over coarse-mode (PM_10−2.5) mass, are typical of a rural site such as Payerne. Previous studies have reported that Payerne is usually influenced by local and Swiss Plateau emissions, as well as by the transport of anthropogenic pollution from mainland Europe . This description provides background information on the site and does not constitute a source attribution for individual episodes during HYGRO19.

The analysis of the aerosol absorption coefficient (Fig. c) exhibits values ranging from 0.06 to 26 Mm-1, with a median of 3.4 ± 1.9 Mm-1 at 520 nm, indicating substantial temporal variability but generally low absorption levels compared with values typically reported for suburban, urban, and traffic sites . This is consistent with the single scattering albedo values (SSA (520 nm) =0.87±0.05, Fig. e). However, differences were observed between the two periods, with larger presence of absorbing particles during the second (σab=3.5±2.1, SSA =0.85±0.05 and AAE =1.09±0.10) than in the first period (σab=2.7±1.3, SSA =0.90±0.02 and AAE =0.92±0.11). The analysis of carbonaceous measurements further suggests that these changes can be associated with a higher frequency of local wood burning (Fig. g–h). Regarding chemical composition (Fig. h), inorganic species (sulphate, ammonium, and nitrate) show an important contribution to particulate matter, accounting for 11 %. However, the TC values compared with measured inorganic species and PM_2.5 support that organic matter is the main fraction (83 %), which is typically the largest fraction in summer (OM =87 %), compared with the second part of the campaign (OM =81 %). The predominance of organic aerosol and its larger contribution during summer (55 %–60 %) than in winter (30 %–40 %) have also been reported at Payerne .

Figure shows the temporal evolution of vertical profiles of the aerosol backscatter coefficient at 355 nm, temperature, water vapour mixing ratio, relative humidity, and the backscatter enhancement factor (fβ (355 nm)) obtained with the RALMO system during the HYGRO19 campaign. For simplicity, the data are shown only in the lower troposphere (0–6 km a.g.l.). The temporal and spatial resolutions were 30 min and 30 m, respectively. Note that the white regions in the figure correspond to areas where retrievals were not possible due to cloudy conditions.

The key result from Fig. is the capability of RALMO to provide continuous daytime and nighttime observations with high temporal and vertical resolution, enabling continuous monitoring of atmospheric vertical profiles and aerosol hygroscopic parameters (panel e). Temperature and water vapour mixing ratio (Fig. b–c) show a similar temporal evolution over the campaign period, with higher values in summer and lower values towards winter, and generally decreasing with altitude. Both parameters exhibited higher values during late June to early August, reaching maxima of 305.6 K and 15.2 g kg⁻¹ below 1.5 km a.g.l., respectively. Later, both temperature and water vapour mixing ratio decrease, reaching their minimum values in December. These behaviours correspond to the typical seasonal evolution of atmospheric conditions in Payerne . The median and half of the interquartile range of these parameters for the entire period, as well as for the aforementioned sub-periods, are summarised in Table . The values are separated below 1.5 km a.g.l. and above that, because aerosol, temperature, and water vapour profiles (Fig. a–c) reveal the highest values within the first 1.5 km a.g.l. corresponding approximately to the mean atmospheric boundary layer (a.b.l.) height .

The analysis of βaer (355 nm) (Fig. a) reveals again two very distinct periods. The first corresponds to June–August, when the highest values are observed, while the second period is from September–December and exhibits considerably lower values. This is consistent with the analysis of AERONET measurements (Fig. a). During the first period, aerosol is frequently present throughout the entire ABL and even within the free troposphere (FT). However, on some days, aerosol layers are observed above and decoupled from the ABL (e.g. 2–5 and 17–19 July). In the second period, aerosol backscatter is generally lower (Table ), and aerosol is typically confined below 1.0 km a.g.l., consistent with a lower ABL height . However, an episode of relatively enhanced aerosol backscatter extending into the FT is observed around 1 December.

The backscatter enhancement factor retrieved with RALMO (fβ (355 nm)) during the campaign was computed within the layers fulfilling the criteria described in Sect. . To assess potential changes in air-mass origin with altitude, as reported by , 5 d (120 h) backward trajectories arriving over Payerne were calculated for each investigated hygroscopic aerosol layer using the Hybrid Single-Particle Lagrangian Integrated Trajectory HYSPLIT model 5.4.0;. The trajectory analysis considered arrival times corresponding to the RALMO observations and arrival heights at the lower and upper boundaries of the investigated layers. A detailed comparison of the trajectories arriving at both boundaries is provided in the Supplement (Sect. S1 and Fig. S1). For comparison with parameters derived with others instruments, fβ (355 nm) was interpolated or extrapolated to the 40 %–85 % RH range (Fig. e). A total of 227 hygroscopic aerosol growth cases were recorded, with fβ (355 nm) values ranging from 1.2 to 7.2 and a median of 2.64 ± 0.64, corresponding to γβ (355 nm) of 0.70 ± 0.18. Most hygroscopic layers were observed in summer (177 hygroscopic layers, 78 %), whereas only a few cases were detected during October–December, not permitting a robust statistical comparison between the two periods.

Frequency histograms of the fβ (355 nm) and γβ are shown in Fig. . To study differences with altitude, two layers were distinguished: one below 1.5 km a.g.l. (orange line in Fig. ), typically below the ABL, where most aerosol particles are found, and the other one above 1.5 km a.g.l. (red line), generally associated with the FT at the Payerne station. To evaluate whether the differences with altitude are statistically significant, the non-parametric Kruskal-Wallis and Mann–Whitney U tests were applied. The Kruskal-Wallis test determines whether three or more independent groups differ significantly based on ranked data, while the Mann–Whitney U test assesses differences between two independent groups based on ranked observations using the sum of ranks .

For the backscatter enhancement factor (Fig. a), no significant differences were observed between all data and the different layer base heights (Kruskal-Wallis p-value =0.99 and Mann–Whitney U p-values ≈1.0; at the 95 % confidence level), with the frequency distribution skewed towards values lower than 4.0. The Hänel parameter (Fig. b), however, follows an approximately normal frequency distribution, and the differences between layers are small and not statistically significant (Kruskal-Wallis p-value =0.98). In general, the retrieved values indicated significant aerosol hygroscopicity, with 64 % of the cases exhibiting fβ (355 nm) values between 2.0 and 4.0, and γβ (355 nm) values in the range 0.50–0.99. The most frequent hygroscopic aerosol layers were observed below 1.5 km a.g.l., with 162 layers (71.4 %) and retrieved hygroscopicity parameters of fβ (355 nm) =2.64±0.59 and γβ (355 nm) =0.70±0.16. Layers above 1.5 km a.g.l. were less frequent but exhibited similar hygroscopicity (fβ (355 nm) =2.72±0.71 and γβ (355 nm) =0.72±0.20).

The large database of aerosol hygroscopic cases during the HYGRO19 campaign allows for an investigation of fβ (355 nm) and γβ (355 nm) with height. In this context, Fig. shows box–whisker plots of these parameters for different height ranges, which were determined by optimising the number of bins in the vertical histograms using established statistical rules for histogram bin selection (Doane’s rule , Rice’s rule , and Scott’s normal reference rule ). These methods account for different characteristics of the data distribution, including the number of observations, variability, and skewness. The analyses are also presented separately for the convective boundary layer (CBL), the residual layer (RL), and the FT, using the ABL operational products at Payerne station developed by . Additionally, the statistical parameters and the frequency of occurrence (rel. freq.) derived are summarised in Table .

Results from Fig. a, c and Table show the vertical variability of aerosol hygroscopicity parameters. In general, some vertical variability is observed, with the highest hygroscopicity parameters found between 606–912 m a.g.l. (fβ (355 nm) =3.22± 0.75 and γβ (355 nm) =0.84± 0.18), and between 1830 and 2136 m a.g.l. (fβ (355 nm) =3.17± 0.77 and γβ (355 nm) =0.83± 0.19). Some height ranges show appreciable differences (e.g., between 606–912 m a.g.l., and between 1218–1524 m a.g.l.), for which the Mann–Whitney U test yielded a p-value of 0.01. However, the low number of data, particularly for the highest layers, and the range of variability of median values are below 20 %, which is within the uncertainties of the methodology. The most important point to remark is the large variability of the hygroscopicity parameters, as reflected by the large standard deviations and percentile ranges. Grouping fβ (355 nm) and γβ (355 nm) into the CBL, RL and FT (Fig. b, d) increases the number of data and makes the statistics more robust. Nevertheless, the analysis did not reveal any significant differences, as the mean, median and percentile values are very similar. Particularly, slightly higher hygroscopicity was observed in the FT (fβ (355 nm) =2.75±0.59 and γβ (355 nm) =0.73± 0.16), lower values were obtained in the CBL (fβ (355 nm) =2.56± 0.60 and γβ (355 nm) =0.68± 0.17) and the highest frequency of occurrence was found in the RL. Overall, these results suggest no clear vertical dependence of aerosol hygroscopicity, and the observed height-dependent patterns should be interpreted with caution, since apparent differences may be amplified by the limited number of observations per layer or by inherent methodological limitations. For instance, the different RH ranges used to compute fβ (355 nm) and γβ (355 nm) in the humidograms (not all experimental cases cover the same minimum-to-maximum RH range), along with uncertainties in RH measurements, can affect the retrieval .

To further investigate temporal changes in aerosol hygroscopicity, Fig. shows the monthly evolutions of fβ (355 nm) and γβ (355 nm) during the HYGRO19 campaign. Results are presented as box-whisker plots, with the most important statistical parameters summarised in Table , which also includes surface chemical composition (Sect. ). Note that chemical composition data are available at daily resolution and correspond to surface measurements, whereas hygroscopicity parameters are derived from RALMO measurements at higher temporal resolution and at different altitudes. These hygroscopicity studies require well-mixed conditions within the analysed layers, which may be decoupled from the surface, particularly when located in the free troposphere. Therefore, thermodynamic homogeneity within a selected layer does not establish that it is mixed with the surface or that its aerosol chemical composition is represented by surface measurements. Additionally, these conditions are met during specific periods and may not persist throughout the day to which the surface chemical composition measurements correspond. Therefore, the surface chemical measurements are used to characterise near-surface aerosol composition, rather than to attribute the hygroscopicity of individual elevated layers to specific chemical components. Such attribution would require vertically resolved chemical measurements, and the uncertainty associated with extrapolating surface composition cannot be quantified here.

The largest number of cases for studying aerosol hygroscopicity was in July (104 layers, 45.8 %), followed by August (55 layers, 24.3 %) and September (40 layers, 17.6 %), whereas fewer cases were recorded from October to December (Table ). The most hygroscopic layers occurred in June (fβ (355 nm) =3.65± 0.70 and γβ (355 nm) =0.94± 0.15), while lower hygroscopicity was observed in October and December (fβ (355 nm) =2.36± 0.26 and γβ (355 nm) =0.60± 0.08), as well as in July (fβ (355 nm) =2.47± 0.49 and γβ (355 nm) =0.65± 0.15). Consistent with these differences, the Mann–Whitney U test revealed statistically significant differences between June and July (p-value =0.001). High values were also recorded in September, which likewise showed statistically significant differences compared with July. High hygroscopicity was also observed in November, although only three cases were identified (Fig. and Table ).

Results from Tables and show that organic matter was the dominant component of the near-surface aerosol (4.9 ± 2.5 µg m⁻³ (83 %)), whereas inorganic species accounted for approximately 11 % of the aerosol composition. The inorganic fraction was mainly composed of sulphate (0.30 ± 0.13 µg m⁻³ (4.4 %)), nitrate (0.11 ± 0.08 µg m⁻³ (4.1 %)) and ammonium (0.17 ± 0.07 µg m⁻³ (2.8 %)). Elemental carbon showed a low concentration (0.28 ± 0.13 µg m⁻³ (4.6 %)). Previous studies have also reported an important contribution of organic aerosol at Payerne e.g.,. During the campaign, this contribution was particularly high in summer, especially in June (Tables and ), and may partly reflect enhanced photochemical activity during this season . In contrast, during autumn and winter, black carbon concentrations increase due to greater residential wood burning and a shallower boundary layer in the region , and these particles are typically less hygroscopic . The predominance of scattering and fine particles, together with lower concentrations of carbonaceous species during summer, may be associated with the higher hygroscopicity observed during this period compared with the second part of the campaign (Fig. and Table ). However, the surface composition does not establish the chemical composition of the layers analysed with RALMO and cannot, by itself, explain their retrieved hygroscopicity.

Changes in the light-scattering and backscatter due to hygroscopic water uptake at a given RH also depend on particle size. This effect was assessed using the extinction AE from sun photometer measurements. The relationship between particle size and the enhancement factor shows no clear dependence on particle size, with differences not being statistically significant. Fine particles yielded fβ (355 nm) =2.60± 0.49 and γβ (355 nm) =0.70± 0.14, compared to fβ (355 nm) =2.30± 0.80 and γβ (355 nm) =0.60± 0.22 for larger particles. The corresponding analysis is shown in Fig. S3 in the Supplement. Previous studies have also reported no clear dependence of aerosol hygroscopicity on particle size, with different size-dependent hygroscopic behaviours observed depending on aerosol properties and atmospheric conditions .

Wind speed and wind direction allow investigating the relationship between air-mass origin and aerosol hygroscopic behaviour. The profile data were obtained using a Doppler wind lidar. Figure illustrates the temporal evolution of wind direction for the detected hygroscopic aerosol layers (a), monthly wind-rose diagrams as a function of wind speed (b) and the corresponding fβ (355 nm) (c). Data from Fig. show that during summer more wind profiles could be monitored due to lower cloud cover, allowing the observation of distinct vertical wind structures. Near the surface, Northeast (NE) winds were frequently observed, while above that winds predominantly originated from the Southwest (SW) and West (W). Wind speeds were around 5 to 8 m s⁻¹ (inset in Fig. a), with a median value of 4.2 ± 1.9 m s⁻¹, corresponding to a moderate breeze . These conditions are typical of high-pressure systems, which occur more frequently in summer over central Europe . In contrast, during autumn and winter months, layers were mostly monitored below 1.5 km a.g.l., predominantly due to instrumental limitations under cloudy conditions associated with more frequent presence of low-pressure systems . Southwestern and Western air masses, originating in the Atlantic, travel through France before reaching Payerne, suggesting a possible mixture of particles (backward trajectories computed with HYSPLIT are not shown for clarity). Similar results were found by at the Jungfraujoch alpine site (Switzerland), where higher hygroscopicity was observed for winds from the SW or W and lower for other wind directions. Layers with fβ (355 nm) >3 in July–September also showed SW–W winds above 5 m s⁻¹. Lower hygroscopicity in late June and July can be associated to the presence of dust particles, due to the influence of air masses from Southern Europe and North Africa, consistent with other studies in Switzerland . Northeast winds are observed mostly at surface level, with wind speeds of 3–6 m s⁻¹, suggesting the influence of Central European urban and industrial emissions, containing less hygroscopic black carbon and organic matter.

Table summarises some of the most relevant studies on aerosol hygroscopicity based on aerosol optical properties. When reported, aerosol chemical composition is provided; otherwise, site classification (e.g., urban, rural, marine) is used as a proxy for aerosol type. These studies are also classified according to whether they are based on remote sensing (lidar and ceilometer) or in situ (tandem nephelometers) techniques. Note that RALMO hygroscopicity parameters are provided at 355 nm for the 40 %–85 % RH range, whereas for some other studies, values are provided at different wavelengths and RH ranges due to differences in system configurations and experimental conditions. The main advance provided by the HYGRO19 campaign is the investigation of the temporal and vertical variability of the hygroscopicity parameters. Long-term analyses using remote sensing techniques have been reported in only a few studies e.g.,, whereas most studies are limited to a small number of case studies or were conducted during specific field campaigns e.g.,. On the other hand, for tandem nephelometers (fσ and γσ), the results of this work are also comparable to those reported in previous studies based on in situ techniques e.g.,, with the differences being within the uncertainties associated with each technique.

Hygroscopic growth retrievals depend on both the measurement wavelength and the aerosol optical property, while chemical composition may also play an important role . Consequently, fζ and γζ values derived from aerosol backscatter measurements (e.g. lidar or COBALD) are not directly comparable to those obtained from total scattering measurements using tandem nephelometers. In addition, previous studies based on Mie simulations have shown that the spectral dependence of hygroscopic growth can depend on particle size . Therefore, enhancement factors for the same RH may differ between wavelengths and particle sizes, and values obtained from different techniques are expected to correlate in their variability, whereas their absolute values are not expected to be directly comparable.