Dataset for "CALISHTO campaign dataset for the publication On the Drivers of Ice Nucleating Particle Diurnal Variability in Eastern Mediterranean Clouds" by Gao et al. (2024)

npj | climate and atmospheric science Article Published in partnership with CECCR at King Abdulaziz University https://doi.org/10.1038/s41612-024-00817-9 On the drivers of ice nucleating particle diurnal variability in Eastern Mediterranean clouds Check for updates Kunfeng Gao 1,6 , Franziska Vogel 2,7, Romanos Foskinis1,3,4, Stergios Vratolis5, Maria I. Gini 5, Konstantinos Granakis5, Olga Zografou5, Prodromos Fetfatzis5, Alexandros Papayannis1,3, Ottmar Möhler2, Konstantinos Eleftheriadis 5&AthanasiosNenes 1,4 We report the drivers of spatiotemporal variability of ice nucleating particles (INPs) for mixed-phase orographic clouds (~−25 °C) in the Eastern Mediterranean. In the planetary boundary layer, pronounced INP diurnal periodicity is observed, which is mainly driven by biological (and to a lesser extent, dust) particles but not aerosols from biomass burning. The comparison of size-resolved and fluorescence-discriminated aerosol particle properties with INPs reveals the primary role of fluorescent bioaerosol. The presence of Saharan dust increases INPs during nighttime more than daytime, because of lower boundary layer height during nighttime which decreases the contribution of aerosols (including bioaerosols) from the boundary layer. INP diurnal periodicity is absent in the free troposphere, although levels are driven by the availability of bioaerosol and dust particles. Given the effective ice nucleation ability of bioaerosols and subsequent effects from ice multiplication at warm temperatures, the lack of such cycles in models points to important and overlooked drivers of cloud formation and precipitation in mountainous regions. Atmospheric ice nucleation plays a vitalroleincloudformationandcloud microphysical properties, which considerably influences regional and global precipitation, hydrological cycle1, atmospheric radiative forcing2and the Earth’senergybalance 3,4.Fortemperature(T)lowerthan−38 °C, atmospheric ice formation can occur spontaneously via homogeneous freezing5. However, for warmer temperatures, the initiation of cloud ice formation in most clouds necessitates ice nucleating particles (INPs) that heterogeneously freeze6. Considering the strong impacts that INPs can have on cloud properties, though a minor fraction of total particles7,INPscanbear large impacts on the hydrological cycle and the climate8. The ice formation ability and the abundance of INPs depend on temperature, particle types and their degree of atmospheric aging9. It is well known that dust particles from desert and agricultural lands7,biological particles, and soil dust particles containing biological materials9constitute major sources of INPs for warmer mixed-phase clouds (MPCs), along with regionallysourcedblackcarbonandorganicparticlesfrombiomassburning emissions for colder cirrus clouds4,6. Moreover, airmass transport10,11 and atmospheric aging processes12,13 can modulate INP concentrations and characteristics14. The large uncertainty in the spatiotemporal variability of INPs7,15 (both abundance and distribution), together with the uncertainty in subsequent cloud processes such as ice multiplication16–18,leadstonotfully constrained effects of INPs in regional and global weather, climate, and Earth system models4,9. Therefore, there is a significant need to improve the predictability of INPs in models8. Periodic (seasonal and diurnal) solar radiation and anthropogenic activities may induce corresponding cycles on aerosol sources19, hence INPs20,21. Previous studies reported a seasonal periodicity of INPs in different regions22–24; studies on INP variabilities within a day are still scarce, owing to the insufficient time resolution of offline INP spectrometers, short duration of observations, and incomplete attribution of INP sources. Diurnal variability of INPs can be an especially important driver for ice formation in orographic cloud systems, given their dynamic nature. 1Laboratory of Atmospheric Processes and Their Impacts, School of Architecture, Civil and Environmental Engineering, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland. 2Institute of Meteorology and Climate Research, Karlsruhe Institute of Technology, Karlsruhe, Germany. 3Laser Remote Sensing Unit (LRSU), Physics Department, National Technical University of Athens, Zografou, Greece. 4Centre for the Study of Air Quality and Climate Change, Institute of Chemical Engineering Sciences, Foundation for Research and Technology Hellas, Patras, Greece. 5ENvironmental Radioactivity & Aerosol Technology for atmospheric & Climate ImpacT Lab, INRASTES, NCSR Demokritos, Attica, Greece. 6 Present address: Institute for Atmospheric and Climate Science, ETH Zurich, Zurich, Switzerland. 7 Present address: Institute of Atmospheric Sciences and Climate (ISAC), National Research Council (CNR), Bologna, Italy. e-mail: kunfeng.gao@epfl.ch;athanasios.nenes@epfl.ch npj Climate and Atmospheric Science | (2025) 8:160 1 1234567890():,; 1234567890():,; Rosinski et al.25 found that INPs (at −15 and −20 °C) show concentration maxima at 06:00 and 18:00 local time of the day. Despite the short duration and low time resolution of the data (4 h), the results from Rosinski et al.25 suggested an INP diurnal cycle that subsequent studies supported11,23,albeit with limited insights on the implications. Wieder et al.11 observed INPs at the Weissfluhjoch mountaintop (2693 m a.sl., Alps) at Davos, Switzerland every 2 h using an offline droplet freezing assay and found a diurnal cycle of INPs showing a minimum in the morning and a peak after sunset. Night time observations, however, were missing11. Not all studies exhibit a diurnal cycle. Isono et al.26 reported no appreciable diurnal variations for INPs (at −15 °C) observed at the Manuna Loa observatory at ~3400 m above sea level (a.s.l.) (Hawaii, US) using an offline static cloud chamber to measure INPs every ~6 h. The availability of automated online INP spectrometers27,28 with high temporal resolution facilitates investigations on the INP diurnal periodicity. Using an automated continuous flow diffusion chamber, Brunner et al.23 performed year-long continuous INP observations at a high altitude observatory at Jungfraujoch (~3580 m a.s.l., Alps, Switzerland) and reported the diurnal periodicity of INPs for days only with planetary boundary layer (PBL) air mass intrusions without Saharan dust events. Detailed sources responsible for these diurnal cycles were still underexplored23. Temporal changes of INPs also depend on the airmass origin at the site, given that the aerosol diurnal cycles differ between the PBL and free tropospheric (FT)29,30 aerosols. Contrast to these studies, Wieder et al.11 found no diurnal cycles of INPs at Wolfgangpass (1632 m a.s.l., Alps, at Davos) which tends to reside in the PBL during the observation period, different from another in parallel observation site near Weissfluhjoch ( ~ 1.0 km higher) at the mountaintop close to the FT. Therefore, determining the atmospheric condition of the observation site is necessary for investigating the INP abundance and its diurnal periodicity. In mountainous regions, local or regional biological particle20 sources in the PBL exhibiting a diurnal cycle may drive INP variability; the diurnal changes of PBLH may also regulate the availability of aerosol particles from the PBL31. Of particular interest is the importance of bioaerosols, since most often forested areas are in mountainous regions. If not forested, arid mountains could also be sources of dust, given that mountain flows in general can generate high velocities that are capable of lifting large amounts of coarse mode particles. Such processes are poorly described or lacking in models, thus it is critical to assess their importance for INP diurnal variability. This is because when combined with the effects of ice multiplication, these INP sources can rapidly glaciate clouds and generate heavy snowfall and extreme precipitation close to the ground during winter storms, as demonstrated by Georgakaki et al.18. In this study, we present the drivers for diurnal cycles of INPs under different atmospheric conditions. The observations were collected during a field campaign, the Cloud-AerosoL InteractionS in the Helmos background TropOsphere (CALISHTO, https://calishto.panacea-ri.gr/) between October and November, 2021. CALISHTO was conducted at the Helmos Hellenic Atmospheric Aerosol and Climate Change (in short as (HAC)2 hereafter) station (37.9843° N, 22.1963° E, 2314 m a.s.l.) close to the summit of Mt. Helmos in the Pelloponnese. Using the PBL characteristics and INP source apportionment in companion studies32,33, we determine the diurnal cycle of INPs for a variety of atmospheric states represented by the relative position of the (HAC)2with respect to the PBL height (PBLH), airmass origin and the effect of Saharan dust events. The distinct contributions of bioaerosol and dust to the diurnal cycle are determined for different locations in the atmospheric column. Results Determination of atmospheric conditions at (HAC)2 (HAC)2contributes data to the Global Atmospheric Watch and ACTRIS programs since 2016 and is located near the summit of Mt. Helmos at the heart of the Peloponnese in Greece. (HAC)2, frequently situated in the FT or at the FT/PBL interface31, is at a cross-road of different airmasses, each of which is characterized as different INP sources32. The observational setup of CALISHTO is presented in the SupplementaryFig. S1, andalso elsewhere in detail32, including high resolution measurements of in-situ INPs (~6‒ 7 min), microphysical and chemical aerosol properties at (HAC)2,remote sensing measurements conducted at Vathia Lakka (VL), a site located lower ~0.5 km than (HAC)2, and back trajectory analysis for calculating the origin of airmasses sampled at (HAC)2. Timeseries of measurement results are presented and introduced in Fig. S2 and Text S2 in the Supplementary Material.ThePBLHisdeterminedbyawindlidar atVLandexpressedasthe vertical distance between VL and the FT/PBL interface31.ThePBLHisused to determine whether (HAC)2is in the PBL, given that (HAC)2is located 500 m above VL (illustrated in Supplementary Fig. S1). Measurements of aerosolpropertiesandairmasscharacteristicsat the (HAC)2canalsobe used to determine whether the site is in the PBL or the FT when PBLH from lidar remote sensing measurements is not available. For example, the site is in the PBL (for 82.5% of the period) if the concentration of particles between 95 nm and 800 nm (SMPS >95nm,<800nm , measured by a TSI Inc., model 3938 scanning mobility particle sizer; SMPS) exceeds a threshold value of 100 std cm−332,34. A thorough evaluation against other metrics and the PBLH31,35 further confirms the universality of the 100 std cm−3threshold, likely because aerosols in the FT are highly diluted after transporting from far-away PBL sources31 and generally are much lower in concentration compared to PBL-influenced airmasses. Depending on the (HAC)2position with respect to the FT/PBL interface (i.e., its relation to the PBLH), we classify a number of atmospheric state scenarios including: (HAC)2in FT throughout the day (Fig. 1bfor PBLH < 0.5 km, 5 days), (HAC)2in PBL throughout the day (Fig. 1cfor PBLH > 0.5 km, 9 days), and (HAC)2partially in the PBL throughout the day (Fig. 1f, 30 days), which is termed (HAC)2~ PBL means (HAC)2fluctuates around PBL/FT interface. Continuous dust events were recorded between November 5 and 9 (Supplementary Fig. S2). Thus, we further divide the scenario of (HAC)2in the PBL throughout the day into two cases, i.e., without (Fig. 1d, 4 days) and with (Fig. 1e, 5 days) dust events –to investigate the effect of Saharan dust events on INP variability. The presence of dust is characterized by significant increases in coarse-sized particle (>2.5 μm) concentrations, low Ångström exponents ( < 1)32 from optical scattering measurement and the spatiotemporal dust mass distribution predicted by modeling experiments (Supplementary Fig. S2c, d, f respectively). For the scenario of (HAC)2in FT throughout the day (Fig. 1b), no appreciable diurnal cycle is seen for both PBLH and SMPS >95nm,<800nm , while a clear diurnal cycle is observed for days when the (HAC)2is in the PBL (Fig. 1c), showing the maximum in the afternoon. The drivers of PBLH diurnal cycles are discussed in Supplementary Text S3 based on results in Figs. S3 to S6. We note that larger SMPS >95nm,<800nm concentrations of nondust days (Fig. 1d) compared with those of dust days (Fig. 1e) are because of differences in aerosol sources. On non-dust days, (HAC)2is influenced by continental aerosols enriched in fine mode particles (generally <500 nm, see Figs. S7 and S8), whereas the presence of dust events on dust days tends to shift the size distribution of aerosols to larger sizes with a depletion in fine mode particles according to parallel studies on the aerosol sources32 and PBLH36 at (HAC)2during CALISHTO campaign. INP diurnal cycles under different atmospheric conditions The INP number concentration at (HAC)2was determined by a portable ice nucleation experiment (PINE) at T=−25.2 ± 1.4 °C in the mixed phase cloud regime (Fig. S2b). PINE samples aerosols from an omnidirectional total inlet and tests INPs in all freezing mode by addressing supersaturated conditions with respect to water32.Analysisoftheirconcentrationover24h periods (e.g., Fig. 2) reveals diurnal periodicities depending on the PBL condition. As shown in Fig. 2a, median INP concentration overall increases from 3.0 std L−1during daybreak to a maximum of 12.0 std L−1in the early afternoon (12:00‒15:00) and subsequently decreases to approximately 3.0 std L−1. Then it remains a fairly constant level of 3.0‒4.0 std L−1 throughout the evening until the early morning (at 3:00), after which the concentration occasionally spikes up to ~10.0 std L−1(Fig. 2a). Figure 2 shows that the INP diurnal patterns for days influenced by PBL airmasses (Fig. 2c–f) are generally analogous to the overall INP diurnal periodicity https://doi.org/10.1038/s41612-024-00817-9 Article npj Climate and Atmospheric Science | (2025) 8:160 2 presented in Fig. 2a but show different maximum and minimum values. INPsfordaysonlyintheFT (Fig. 2b), however, do not exhibit such a diurnal cycle but generally show median concentration values less than 3.0 std L−1 throughout the day. The above results suggest that the source of INPs observed at (HAC)2originates primarily from the PBL, and that the diurnal cycle of INPs is driven by the influx of aerosol particles from the PBL to the site. This is consistent with Brunner et al.23, who found an absence of INP diurnal cycles in the FT, but a clear diurnal cycle is observed when PBL airmasses are available. The dependence of INP diurnal cycles on aerosol particle size To study the correlations between the diurnal variabilities of INPs and aerosol particle sizes, we compare the INP diurnal cycles with diurnal changes of the number concentration of total aerosol particles in different size ranges measured by a SMPS (10‒800 nm, electrical mobility diameter) and an aerodynamicparticlesizer(APS,0.5‒20 μm, aerodynamic diameter). Particles in different size ranges used to compare with INPs include SMPS+APS total (0.01‒20μm,Fig.S7),APS >0.5μm, total (0.5‒20μm,Fig.2),SMPS <500nm (Fig.S8), APS >0.5μm,<1.0μm (Fig.S9),APS >1.0μm,<1.5μm (Fig. S10), APS >1.5μm,<2.0μm (Fig. S11), APS >2.0μm,<2.5μm (Fig. S12) and APS >2.5μm (Fig. 3). The definition of particlesineachsizerangeisprovided in the footnote of Table 1.SMPS +APS total is superimposed by both SMPS and APS results32,37. Additionally, scatter plots comparing INP number concentrations with meteorological parameters (ambient T, i.e., T ambient , horizontal wind velocity and direction) and different aerosol properties (fluorescent and optical properties, as well as eBC mass concentration) under different PBL conditions are provided in Fig. S13 in Supplementary S5. Also, the correlation between INP diurnal cycles and eBC mass concentration and aerosol optical properties are also examined respectively (Figs. S14 to S16 in Supplementary S6). Given the size dependence of INPs6,38,APS >0.5μm, total particles are assumed to be major INP contributors in the literature whereas smaller-sized particles (<0.5 μm), which are even not included in some INP parameterizations7,39,40, are assumed to be insignificant INPs. Coarse mode particles (e.g., APS >2.5μm ) are more relevant for dust and bioaerosols and consideredwithahigherprobability of serving as INPs6,7.Here, we evaluate the contribution of APS >0.5μm, total and APS >2.5μm particles to the observed INPs Fig. 1 | Diurnal cycles of PBLH measured by wind lidar31 at the VL site (on the left axis) and corresponding diurnal cycles of the number concentration of particles with diameter between 95 nm and 800 nm (SMPS >95nm, <800nm ) measured at (HAC)2(on the right axis). Solid lines indicate median values and the shading area around the median line shows the range between 25th and 75th quartiles. The horizontal dashed red lineindicates both thealtitude difference between (HAC)2and the lidar (~0.5 km) and the threshold particle number value of 100 std cm−331,35. Different (HAC)2atmospheric conditions are classified in different panels. aAll observations during the campaign. b(HAC)2in the FTthroughout theday. c(HAC)2 inthePBLthroughouttheday.d(HAC)2inthe PBL throughout the daywithoutdust event influence. e(HAC)2in the PBL throughout the day with dust event influence. f(HAC)2for days with both PBL and FT influences. The data points of each (HAC)2 position scenario are resampled for every 20 min and each panel shows period cycles of 24 h starting at 00:00 UTC +2 (local time) of the day. https://doi.org/10.1038/s41612-024-00817-9 Article npj Climate and Atmospheric Science | (2025) 8:160 3 under different atmospheric conditions (Figs. 2and 3) and also discuss the importance of particles in other size ranges (Table 1). The APS >0.5μm, total (Fig. 2a) and APS >2.5μm (Fig.3a) concentration cycles fromall observations show an overalldiurnalcycle similartothatof INPs, respectively. Overall concentration diurnal cycles and significant correlations with INPs are also observed for particles in other size ranges presented in Table 1and Figs. S7a to S12a. Analogous to the INP data for days only in the FT, none of sizeresolvedparticlespresents adiurnalcycle(e.g.,Figs. 2band3b),butallshows the lowest median value at the same hours compared to the other scenarios influenced by PBL air masses. Notably, Fig. 2b shows an overall constant APS >0.5μm, total to INP concentration ratio (for median values) of approximately 250, while Fig. 3b presents that thedifference between APS >2.5μm and INP medians is generally within a factor of 10. Table 1shows that particles having a size range larger than 0.5 μm have significant (p< 0.05) and positive correlations with INPs throughout the day in the FT whereas particles smaller than 0.5 μm(SMPS <500nm )showaninsignificant role, which highlights the importance of particles larger than 0.5 μmandis consistent with the literature7,39,40.ThecaseofSMPS+APS total in the FT (Table 1and Fig. S7b) presents similar results to that of SMPS <500nm ,given that SMPS <500nm takesamajorfractionofSMPS+APS total 32. Fordayswhenthe(HAC) 2is only in the PBL, Table 1shows that both APS >0.5μm, total and SMPS+APS total present significant contributions (R= 0.35 and 0.51 respectively) to the observed INP diurnal cycles (also Fig. 2c and Fig. S7c). It also shows that only particles within a size range smaller than 1.0 μmsignificantly contribute to the INP diurnal cycle, whereas particles with a larger size range (>1.0 μm) are weakly linked to INP variabilities. This may be because fine mode particles have much higher number concentrations than coarse mode particles32 and also because particles from various sources that span different size ranges are responsible for the observed INPs in the PBL32. Thus, a higher particle number concentration including smaller-sized particles is more associated with INP variability than low number concentrations of larger-sized particles, when aerosol sources affecting the particle population are many. Only when INP sources are further determined, can the importance of larger-sized particles for INP variability be determined, which isdemonstratedby theresultsinTable1for thecaseof(HAC) 2in PBL with and without dust events. It shows that only particles in size ranges larger than 0.5 μm have significant correlations with INP diurnal medians while smaller-sized particles (SMPS <500nm and SMPS +APS total ) are insignificant. This is consistent with the case of (HAC)2in FT (Table 1) and the literature7,39,40. Fig. 2 | Diurnal cycles of INP (tested at T=−25.2 ± 1.4 °C, on the left axis) and APS >0.5μm, total (the total number concentration of particles between 0.5 and 20 μm measured by an Aerodynamic Particle Sizer, on the right axis) measured at (HAC)2under different atmospheric conditions. Solid lines indicate median values and the shading area around the median line shows the range between 25th and 75th quartiles. Different (HAC)2atmospheric conditions are classified in different panels. aAll observations during the campaign. bFor days only in the FT. cFor days only in the PBL. dDays in the PBL without dust events. eDays in the PBL with dust events. fDays not exclusively in the PBL or FT. The data points of each scenario are resampled for every 20 min and each panel shows a cycle period of 24 h starting at 00:00 UTC +2 (local time) of the day. The Pearson correlation coefficient (R) and Spearman’s rank coefficient (ρ), as well as corresponding pvalues, are provided to evaluate the correlation between INP concentration and APS >0.5μm, total . The Pvalue is the probability of obtaining an R(ρ) value no smaller than the true R (ρ) value if thereis noliner correlation between INP andAPS >0.5μm, total . The number of data points (n) for each case of above statistical analysis is 72. https://doi.org/10.1038/s41612-024-00817-9 Article npj Climate and Atmospheric Science | (2025) 8:160 4 Inaddition, therearedifferentsizedependences for INPvariability observedbetweencasesof (HAC)2in PBL with and without dust events. Table 1showsthatwithanincreasing size range forparticleslarger than 0.5 μm (from a range of 0.5‒1.0 μm to a range of 2.5‒20 μm), the significance of larger-sized particles become less and less pronounced for non-dust days. Again, this highlights a more important role of high number concentrations particles in regulating INPs in the PBL with continental aerosols32, given that continental aerosols generally contain fine particles without dust events. Differently, the case of (HAC)2in PBL with dust events (Table 1) shows the least significant contributions of particles between 0.5 and 1.0 μm, compared to particles in larger size ranges (>1.0 μm), presenting the importance of larger-sized dust particles in regulating INP variabilities. Notably, particles in size ranges larger than 1.0 μm show similar correlation (R≥0.70) and significance levels (<0.01) with INP variabilities (Table 1). This indicates dust particles enrich the observed INPs from a size limit of 1.0 μm, which is in agreement with Gao et al.32 (a parallel study) who reported a size distribution inflectionpointat1.0μm for aerosols observed at (HAC)2 in the PBL during dust events. For the case of (HAC)2~ PBL, it shows similar results to the case of all observations for the correlations between INPs and particles indifferent size ranges (Table 1). Particles from different size ranges play a significant role in regulating the INP diurnal cycle. It also presents a more important role for particlesinsizerangeslargerthan0.5μm(R≥0.51) compared with smallersized particles (R= 0.27). In brief, we conclude that particles larger than 0.5 μm are generally important contributors to the INP diurnal periodicity for the (HAC)2in the FT or PBL in the absence or presence of dust events. When a mixture of varying INP sources is relevant, the contribution from smaller-sized particles (<0.5 μm) is non-negligible. Also, the different size dependence of INPs observed for (HAC)2inPBLwithandwithoutdust events implies different abundance of INP types in each aerosol source. The influence of biomass burning particles on INP diurnal variability APS >0.5μm, total observed on days when the (HAC)2is only in the PBL without dust influence (Fig. 2d) shows an increase in the evening (~20:00‒ 22:00), however, such an aerosol particle increase does not lead to increased INPs. Note that such a particle concentration spark is only present for Fig. 3 | Diurnal cycles of INP (tested at T=−25.2 ± 1.4 °C, on the left axis) and the concentration of aerosol particles between 2.5 and 20 μm (APS >2.5μm , on the right axis) measured at (HAC)2under different atmospheric conditions. Solid lines indicate median values and the shading area around the median line shows the rangebetween25thand75thquartiles. Different (HAC)2atmosphericconditions are classified in different panels. aAll observations during the campaign. bFor days only intheFT.cFordaysonlyin the PBL.dDays inthe PBLwithoutdust events. eDaysin the PBL with dust events. fDays not exclusively in the PBL or FT. The data points of each scenario are resampled for every 20 min and each panel shows a cycle period of 24 h starting at 00:00 UTC +2 (local time) of the day. The Pearson correlation coefficient(R) and Spearman’s rank coefficient(ρ),aswellas corresponding pvalues, are provided to evaluate the correlation between INP concentration and APS >2.5μm . The pvalue is the probability of obtaining an R(ρ) value no smaller than the true R (ρ) value if there is no liner correlation between INPs and APS >2.5μm . The number of data points (n) for each case of above statistical analysis is 72. https://doi.org/10.1038/s41612-024-00817-9 Article npj Climate and Atmospheric Science | (2025) 8:160 5 particles in a range smaller than 1.5 μm (Figs. S8d to S10d). The presence of those increased particles coincides with the peak values of elemental black carbon (eBC) mass concentration (measured by an aethalometer, AE31, Magee Scientific, US) in the evening (Fig. S14d in Supplementary S6), probably due to increased use of fossil fuels for heating during the cold days at the nearby village of Kalavryta (Supplementary Fig. S4d for low T ambient ). This is also supported by a positive and significant correlation between eBC mass concentration and the number concentration of total (observed by both SMPS and APS) particles smaller than 1.5 μm(R=0.44 (p= 8.31E −6) and ρ=0.68(p= 4.63E −14)) for the (HAC)2in the PBL without dust events but a negative and insignificant correlation with largersized (>1.5 μm) particles (R=−0.11 (p= 0.28) and ρ=−0.09 (p= 0.39)). Particles containing eBC masses or black carbon particles are poor INPs in the MPC regime41. Thus, those eBC containing particles (<1.5 μm) do not lead to INP increases. This is also similar to the results reported by Brunner et al.23 for Jungfraujoch. Therefore, contributions from eBC-containing particles are negligible to INPs tested at T=−25.2 ± 1.4 °C in this study. The dependence of INP diurnal cycles on fluorescent biological aerosol particles in different size ranges Particles of biological origin (even if not all INP-active) are key contributors to INPs in the MPC regime, particularly for T>−15 °C42,43.Also,dust particles, as important contributors to INPs for T<−15 °C42,43,areoften found to be associated with biological material44, like airborne bacteria coexisting with Saharan dust particles45 and soil dust rich in biological materials at source46. Fluorescence is an important property indicating particles carrying biological materials, although some particles of nonbiological origin can also fluoresce21,47. Fluorescent particles have been frequently viewed as a lower-limit proxy for biological particles48,called fluorescent biological aerosol particles (FBAPs)47. Here, a wideband integrated bioaerosol sensor (WIBS, WIBS-5/NEO, Droplet Measurement Technologies, LLC. US) was used to measure the number concentration of FBAPs at (HAC)2and record the optical size of the particle (0.5‒30 μm)32,47. Particles fluorescing in any one of the three WIBS fluorescent channels is classified as Fluo WIBS particles32 which includes all FBAPs detectable by a WIBS. FBAPs are demonstrated to be significant INP predictors during the CALISHTO32 and other field campaigns in other regions24,39. Hence, we compare the diurnal cycles of both INPs and total FBAPs, represented by Fluo WIBS>0.5μm, total , to investigate the role of bioaerosols in INP variabilities (Fig. 4). Also, we investigate the contribution of FABPs in different size ranges to the observed INP diurnal cycles, including Fluo WIBS>0.5μm,<1.0 μm (Fig.S17),Fluo WIBS>1.0μm,<1.5μm (Fig. S18), Fluo WIBS>1.5μm,<2.0μm (Fig. S19), Fluo WIBS>2.0μm,<2.5μm (Fig. S20) and Fluo WIBS>2.5μm (Fig. S21). The calculated correlation coefficients between diurnal medians INPs and Fluo WIBS particles in different size ranges are summarized in Table 2. Table 2showsanoverallmediumandsignificant correlations between the diurnal cycles of INPs and FABPs in all calculated size ranges (the columnofallobservations),suggestingFBAPsinallsizerangescontributeto the observed INPs. Of all size ranges, Fluo WIBS>1.5μm, <2.0μm and Fluo WIBS>2.5μm are more important INP contributors, showing stronger correlations with INPs (R≥0.64) than other size ranges. Fluo WIBS>1.5μm,<2.0μm may coincide with large-sized bacteria or small-sized fungal spores (<2.0 μm), and Fluo WIBS>2.5μm is likely related to fungal spores or larger-sized pollen fragments49,50, all of which are often found as effective Table 1 | The Pearson correlation coefficient (R) and Spearman’s rank coefficient (ρ) for the relationship evaluation between diurnal INP median number concentration and the median number concentration of aerosol particles with different sizes under different atmospheric conditions Scenarios All observations (HAC)2in FT (above PBL) (HAC)2in PBL (HAC)2in PBL without dust events (HAC)2in PBL with dust events (HAC)2~ PBL top R (p) ρ (p) R (p) ρ (p) R (p) ρ (p) R (p) ρ (p) R (p) ρ (p) R (p) ρ (p) APS >0.5μm, total a0.69 0.52 0.40 0.45 0.35 0.39 0.46 0.53 0.73 0.65 0.65 0.54 <0.01 0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 SMPS+APS total b0.45 0.29 0.13 0.03 0.51 0.46 0.05 0.11 0.15 0.23 0.33 0.25 <0.01 0.01 0.27 0.81 <0.01 <0.01 0.70 0.37 0.20 0.06 0.05 0.04 SMPS <500nm c0.40 0.23 0.12 0.01 0.51 0.47 0.04 0.10 0.12 0.18 0.27 0.20 <0.01 0.05 0.31 0.90 <0.01 <0.01 0.73 0.40 0.33 0.13 0.02 0.09 APS >0.5μm,<1.0μm d0.59 0.43 0.33 0.43 0.58 0.55 0.47 0.53 0.44 0.40 0.62 0.53 <0.01 <0.01 0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 APS >1.0μm,<1.5μm e0.71 0.62 0.43 0.42 0.09 0.08 0.37 0.34 0.77 0.77 0.63 0.55 <0.01 <0.01 <0.01 <0.01 0.47 0.48 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 APS >1.5μm,<2.0μm f0.60 0.53 0.44 0.38 0.08 0.09 0.28 0.32 0.76 0.77 0.57 0.54 <0.01 <0.01 <0.01 <0.01 0.49 0.46 0.02 0.01 <0.01 <0.01 <0.01 <0.01 APS >2.0μm,<2.5μm g0.58 0.53 0.43 0.35 0.10 0.09 0.24 0.31 0.75 0.75 0.51 0.46 <0.01 <0.01 <0.01 <0.01 0.41 0.44 0.05 0.01 <0.01 <0.01 <0.01 <0.01 APS >2.5μm h0.64 0.54 0.46 0.40 0.15 0.14 0.30 0.34 0.70 0.65 0.57 0.49 <0.01 <0.01 <0.01 <0.01 0.19 0.23 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 A critical pvalue of 0.05 from F-test for Rand ρis used to assess the significance level of the relationship. A pvalue smaller than 0.05 suggests that the probability of obtaining an R(ρ) value no smaller than the true R(ρ) value is less than 5% if there is actually no liner correlation between INPs and the given parameter, thusthe calculated R(ρ) is of statistical significance. Evaluated significant relationships are indicated in bold. The correlation coefficients are also provided in Figs. 2and 3and Figs. S7 to S12 in Supplement S4. aTotal particle (0.5–20 μm) number concentration measured APS, bTotal particle (0.01–20 μm) number concentration measured by both SMPS and APS, cThe number concentration of particles smaller than 500 nm measured by SMPS, dThe number concentration of particles between 0.5 and 1.0 μm measured by APS, eThe number concentration of particles between 1.0 and 1.5 μm measured by APS, fThe number concentration of particles between 1.5 and 2.0 μm measured by APS, gThe number concentration of particles between 2.0 and 2.5 μm measured by APS, hThe number concentration of particles larger than 2.5 μm measured by APS. https://doi.org/10.1038/s41612-024-00817-9 Article npj Climate and Atmospheric Science | (2025) 8:160 6 INPs51–53.Forthecaseof(HAC) 2in FT, FBAPs in all size ranges show weak (both Rand ρ< 0.2) and insignificant correlations with the observed INPs (Table 2). Nevertheless, the INP diurnal medians in the FT are approximate to those of Fluo WIBS>0.5μm, total (within a factor of 3, see Fig. 4b), particularly similar to those of Fluo WIBS>2.5μm (in Fig. S21b). This suggests scarce FABPs in the FT are non-negligible INP sources, supported by a strong correlation between the scatter plots of all data points of INPs and Fluo WIBS>0.5μm, total (ρ= 0.63, p= 6.45E-10, see Fig. S13h). When the (HAC)2is located in the PBL, FBAPs in all calculated size ranges show medium or stronger (e.g., Fluo WIBS>2.0μm,<2.5μm ) correlations with INP diurnal median values (Table 2). Compared with the combined casefor (HAC)2in PBLregardlessoftheeffectofdustevents,FABPs onnondust days in each size range generally show much stronger (R and ρ≥0.6) correlations with INP diurnal medians, while they show less pronounced correlations (Rand ρ< 0.5) during dust days. It suggests a non-negligible role of FBAPs as INP contributors during Saharan dust events. Notably, this indicates a more important role of FABPs in contributing INPs when continental and local aerosols are more dominant INP sources at the (HAC)2. Altogether, this suggests that FBAPs relevant for INPs generally originate from PBL aerosols in the absence of dust; during dust events, only FABPs between 1.0 and 2.5 μm are moderately and significantly correlated with the observed INPs, likely because smaller-sized FBAPs have a longer residence time and are easier to be mixed with transported dust aerosols. However, much smaller-sized Fluo WIBS>0.5μm, <1.0 μm particles show a weak and insignificant correlation with INPs during dust days (Table 2and Fig. S17e) but a strong correlation during non-dust days (Fig. S 17d). Likely, thosesmall-sizedparticlesrelevant for INPsmay be relatedto ice-nucleating bacteria (later discussed with results presented in Table 3and Supplementary S8) –as local and continental aerosols may contain more bacteria whereas transported Saharan dust across less polluted Mediterranean may carry relatively less. Additionally, we note that eBC-containing particles as interfering FBAPs47 mayberesponsibleforthesmallpeakofFluo WIBS>0.5μm, total at 20:00-22:00 h in Fig. 4d, which is only pronounced for FBAPs smaller than 1.5 μm (Figs. S17d and 18d). Again, the results in the case of Fig. 4 | Diurnal cycles of INP (tested at T=−25.2 ± 1.4 °C, on the left axis) and Fluo WIBS>0.5μm, total (the number concentration of particles between 0.5 and 30 μm showing fluorescence in any of three channels of a wideband integrated bioaerosol sensor (WIBS), on the right axis) measured at (HAC)2under different atmospheric conditions. Solid lines indicate median values and the shading area around the median line shows the range between 25th and 75th quartiles. Different (HAC)2atmospheric conditions are classified in different panels. aAll observations during the campaign. bFor days only in the FT. cFor days only in the PBL. dDays in the PBL without dust events. eDays in the PBL with dust events. fDays not exclusively in the PBL or FT. The data points of scenario are resampled for every 20 min and each panel shows a cycle period of 24 h starting at 00:00 UTC +2 (local time) of the day. The Pearson correlation coefficient (R) and Spearman’s rank coefficient (ρ), as well as corresponding pvalues, are provided to evaluate the correlation between INP concentration and Fluo WIBS>0.5μm, total . The pvalue is the probability of obtaining an R(ρ) value no smaller than the true R(ρ) value if there is no liner correlation between INPs and Fluo WIBS>0.5μm, total . The number of data points (n) for each case of above statistical analysis is 72. https://doi.org/10.1038/s41612-024-00817-9 Article npj Climate and Atmospheric Science | (2025) 8:160 7 Table 2 | The Pearson correlation coefficient (R) and Spearman’s rank coefficient (ρ) for the relationship evaluation between diurnal INP median number concentration and the median number concentration of fluorescent biological aerosol particles (FBAPs) with different sizes under different atmospheric conditions Scenarios All observations (HAC)2in FT (above PBL) (HAC)2in PBL (HAC)2in PBL without dust events (HAC)2in PBL with dust events (HAC)2~ PBL top R (p) ρ (p) R (p) ρ (p) R (p) ρ (p) R (p) ρ (p) R (p) ρ (p) R (p) ρ (p) Fluo WIBS>0.5μm, total a0.67 0.55 0.21 0.18 0.62 0.50 0.72 0.73 0.26 0.20 0.65 0.53 <0.01 <0.01 0.08 0.13 <0.01 <0.01 <0.01 <0.01 0.03 0.09 <0.01 <0.01 Fluo WIBS>0.5μm, <1.0μm b 0.43 0.39 −0.06 −0.05 0.59 0.46 0.70 0.71 0.17 0.21 0.34 0.26 <0.01 <0.01 0.62 0.68 <0.01 <0.01 <0.01 <0.01 0.17 0.07 <0.01 0.03 Fluo WIBS>1.0μm, <1.5μm c 0.42 0.42 −0.06 −0.05 0.62 0.56 0.58 0.62 0.43 0.53 0.38 0.41 <0.01 <0.01 0.62 0.68 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 Fluo WIBS>1.5μm, <2.0μm d 0.64 0.60 0.07 0.01 0.68 0.61 0.67 0.67 0.39 0.49 0.42 0.46 <0.01 <0.01 0.56 0.95 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 Fluo WIBS>2.0μm, <2.5μm e 0.41 0.43 0.02 −0.03 0.70 0.59 0.75 0.69 0.39 0.41 0.29 0.30 <0.01 <0.01 0.88 0.78 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 0.01 0.01 Fluo WIBS>2.5μm f0.65 0.58 0.20 0.18 0.42 0.38 0.74 0.70 0.21 0.17 0.61 0.56 <0.01 <0.01 0.09 0.13 <0.01 <0.01 <0.01 <0.01 0.07 0.16 <0.01 <0.01 A critical pvalue of 0.05 from F-test for R(ρ) is used to assess the significance level of the relationship. A pvalue smaller than 0.05 suggests that the probability of obtaining an R(ρ) value no smaller than the true R(ρ) value is less than 5% if there is actually no liner correlation between INPs and the given parameter, thus the calculated R(ρ) is of statistical significance. Evaluated significant relationships are indicated in bold. The correlation coefficients are also provided in Fig. 4in the main text and Figs. S17 to S21 in Supplement S7. aTotal particle (0.5–30 μm) number concentration measured by WIBS, bThe number concentration of particles between 0.5 and 1.0 μm measured by WIBS, cThe number concentration of particles between 1.0 and 1.5 μm measured by WIBS, dThe number concentration of particles between 1.5 and 2.0 μm measured by WIBS, eThe number concentration of particles between 2.0 and 2.5 μm measured by WIBS, fThe number concentration of particles larger than 2.5 μm measured by WIBS. Table 3 | The Pearson correlation coefficient (R) and Spearman’s rank coefficient (ρ) for the relationship evaluation between diurnal INP median number concentration and the median number concentration of different types of fluorescent biological aerosol particles (FBAPs) under different atmospheric conditions Scenarios All observations (HAC)2in FT (above PBL) (HAC)2in PBL (HAC)2in PBL without dust events (HAC)2in PBL with dust events (HAC)2~ PBL top R (p) ρ (p) R (p) ρ (p) R (p) ρ (p) R (p) ρ (p) R (p) ρ (p) R (p) ρ (p) A WIBS a0.65 0.67 0.22 0.24 0.68 0.61 0.81 0.77 0.40 0.43 0.44 0.50 <0.01 <0.01 0.08 0.05 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 B WIBS b0.51 0.39 −0.10 −0.05 0.64 0.47 0.63 0.66 0.29 0.33 0.36 0.25 <0.01 <0.01 0.40 0.67 <0.01 <0.01 <0.01 <0.01 0.01 0.01 <0.01 0.04 C WIBS c0.53 0.54 0.15 0.08 0.53 0.49 0.70 0.69 0.32 0.28 0.28 0.28 <0.01 <0.01 0.20 0.52 <0.01 <0.01 <0.01 <0.01 <0.01 0.02 0.02 0.02 AB WIBS d0.45 0.70 −0.09 −0.18 0.59 0.44 0.65 0.68 0.32 0.36 0.41 0.65 <0.01 <0.01 0.53 0.19 <0.01 <0.01 <0.01 <0.01 0.01 <0.01 <0.01 <0.01 AC WIBS e0.38 0.51 −0.07 0.12 0.33 0.47 0.51 0.71 0.28 0.22 0.12 0.22 <0.01 <0.01 0.72 0.57 0.01 <0.01 <0.01 <0.01 0.02 0.06 0.32 0.06 BC WIBS f0.58 0.51 0.10 0.05 0.55 0.42 0.70 0.72 0.16 0.15 0.49 0.44 <0.01 <0.01 0.40 0.71 <0.01 <0.01 <0.01 <0.01 0.18 0.22 <0.01 <0.01 ABC WIBS g0.52 0.51 0.08 0.09 0.67 0.55 0.78 0.72 0.40 0.40 0.44 0.52 <0.01 <0.01 0.48 0.47 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 <0.01 A critical pvalue of 0.05 from F-test for R(ρ) is used to assess the significance level of the relationship. A pvalue smaller than 0.05 suggests that the probability of obtaining an R(ρ) value no smaller than the true R(ρ) value is less than 5% if there is actually no liner correlation between INPs and the given parameter, thus the calculated R(ρ) is of statistical significance. Evaluated significant relationships are indicated in bold. The correlation coefficients are also provided in Fig. 5in the main text and Figs. S22 to S27 in Supplement S8. aThe total number concentration of particles showing fluoresce only in WIBS FL1 channel, bThe total number concentration of particles showing fluoresce only in WIBS FL2 channel, cThe total number concentration of particles showing fluoresce only in WIBS FL3 channel, dThe total number concentration of particles showing fluoresce only in both WIBS FL1 and FL2 channels, eThe total number concentration of particles showing fluoresce only in both WIBS FL1 and FL3 channels, fThe total number concentration of particles showing fluoresce only in both WIBS FL2 and FL3 channels, gThe total number concentration of particles showing fluoresce only in all WIBS FL1, FL2 and FL3 channels. https://doi.org/10.1038/s41612-024-00817-9 Article npj Climate and Atmospheric Science | (2025) 8:160 8 (HAC)2~ PBL (i.e., (HAC)2fluctuates around PB/FT interface) are similar to the case of all observations, possibly because both cases contain a mixture of varying types of FABPs. In summary, Table 2shows that both FBAPs and dust particles are important sources of INPs, but INP variabilities on nondust days are more dependent on FBAPs whereas FBAPs are less important (butstillsignificant,e.g.,Fluo WIBS between1.0and2.5μm)INP contributors during dust events overwhelmed by dust particles. The dependence of INP diurnal cycles on different types of fluorescent biological aerosol particles Different types of biological particles may exhibit distinct fluorescence characteristics in the WIBS47. When considered in our analysis (together with size ranges expected for each population), this can help further understand the specific contribution of each biological type to the observed INPs. A detailed FABP type classification is provided in Table 3.Bacteria showfluorescence dominantly in WIBS FL1 channel and can be recorded as A WIBS (dominant), AB WIBS ,andABC WIBS particles around 1.0 μmbut generally less than 2.0 μm49,50. Fungal spores and fungi are also frequently detected as A WIBS (dominant), AB WIBS ,andABC WIBS particleslargerthan 2.0 μm47,50. Intact and fragmented pollen grains are often recognized as C WIBS ,BC WIBS ,andABC WIBS (dominant) particles larger than 2.0 μm47,50. ABC WIBS particles are reported as FBAPs measured by a WIBS, with the highest probability of being biological particles47,50 and minimal interference from non-biological particles like black carbon and dust21,47. Table 3showsthatalltypesofFBAPsaremoderatelyorevenstrongly correlated with overall INP diurnal cycles, likely suggesting that different types of biological particles are relevant contributors to the observed INPs throughout the campaign. Similar results, but with weaker correlations, were observed for the case of (HAC)2fluctuating around the PBL-FT interface.Fordayswhen(HAC) 2is exclusively in the FT, the diurnal cycle of INPs is not correlated with any types of FBAPs. However, the median number concentration values of INPs are generally within a factor of 3 compared to those of A WIBS ,B WIBS ,C WIBS ,BC WIBS and ABC WIBS particles (Fig. 5b and Supplement S8), suggesting that biological particles in FT airmasses are non-negligible INP sources. For days when (HAC)2resides exclusively in the PBL (Fig. 5and Supplement S8), the diurnal cycle of all different types of FBAPs is significantly correlated with the corresponding INP cycle (e.g., ABC WIBS R=0.67andp<0.01inFig.5c), which is stronger onnon-dust days(e.g.,ABC WIBS withR=0.78inFig.5d)butweakerondust days (e.g., ABC WIBS with R=0.40inFig.5e). The generally much stronger correlations between the diurnal cycles of INPs and all different types of FBAPs for non-dust days in the PBL compared with those of dust days suggest the overall enrichment of different types of biological particles for non-dust days. Also, it indicates that continental and local aerosols are the primary sources of biological particles but not the transported dust plume, which is consistent with the aerosol source apportionment conducted in a parallel study32 and results in Table 2. The above results demonstrate FBAPs are primary drivers of the INP diurnal cycles observed at the (HAC)2in the PBL. The influence of dust events on aerosol property and INP diurnal variability Remotely-transported dust plumes modulate aerosols and INPs at the (HAC)2in the PBL. We present the influences of dust events on aerosol properties observed at the (HAC)2by comparing the diurnal cycles of number concentrations of size-resolved total aerosol particles (Fig. S28) and FBAPs (Fig. S29), as well as different types of FBAP (Fig. 6), on non-dust days with dust days. Further discussions on the correlations between aerosol propertychanges and INPvariabilitiespresentedin Figs.2to5forbothcases are conducted. The results in Fig. S28 show that total aerosol particles with size smaller than 0.5 μm during non-dust days are more than those on dust days by at least a factor of 2 (panel c). Dust events enrich particles in size ranges larger than 1.0 μm (panel e to h) for dust days by a factor of 5 ~ 10 compared with those of non-dust days. Similarly, comparing with dust days, Fig. S29 shows that non-dust days containmore FBAPs of size below 2.0 μm (panel b, c and d) but less FBAPs larger than 2.5 μm (panel f). These results suggest continental aerosols enriched with fine mode aerosols are dominate INP sources during non-dust days and exhibit substantially different size and contribution of INPs in comparison with dust episode days. Figure 6generally shows that dust days have much higher number concentration of total FBAPs indicated by Fluo WIBS>0.5µm, total in the morning from 00:00 to 08:00 h. The enriched FABPs are primarily attributed to C WIBS and BC WIBS particles (from 00:00 to 08:00 h in Fig. 6e, h) which are often associated with dust events and biological particles47.The enriched C WIBS and BC WIBS may be large-sized (3–10 μmasseeninFig.6e from Gao et al.32) dust-containing particles that are more effective INP (Fig. S29f) than seen for smaller-sized particles, explaining higher INP concentrationson dustdaysinthemorning (Figs.2eto5e)comparedwiththose on non-dust days (Figs. 2dto5d). From 08:00 to 16:00 h, Fluo WIBS>0.5µm, total particles on non-dust days majorly consist of ABC WIBS ,B WIBS ,andBC WIBS particles, while Fluo WIBS>0.5µm, total particles on dust days mostly include ABC WIBS ,A WIBS , C WIBS ,andBC WIBS particles (Fig. 6). The higher concentrations of ABC WIBS (by a factor of ~2) and B WIBS particles (by a factor of ~4) on non-dust days may explain their higher number concentrations of INPs (by ~50%) compared with those on dust days, given that ABC WIBS are of the highest probability of being biological particles32,47 which are active INPs at warm temperatures. B WIBS particles on non-dust days are likely from continental aerosols smaller than 2.0 μm(seeFig.6c in Gao et al.32) and related to bacteria that may be IN active53. Differently, the enriched A WIBS and C WIBS particles on dust days are probably of sizes between 3 and 10 μm(seeFig. 6a, d in Gao et al.32), which can be attributed to fungal spores and/or fragmentedpollengrains of similarsizes47,50 orsmallbacteriacombinedwith large-sized dust particles45.Additionally,thehigherconcentrationsof FABPs for both non-dust and dust days between 08:00 and 16:00 h than the other periods of the day coincide with the noon peaks of INPs for both cases (Figs. 2to 5). Notably, the peak of INPs on non-dust days at 12:00 h overall showsanoverlapwiththatofABC WIBS (Figs. 5dand6b) but not with APS >2.5μm particles(showingapeakat14:00 hinFig.3d),suggestingthat the contribution of FBAPs to the observed INPs is more important than that of total coarse-sized ( > 2.5 μm) aerosol particles. In contrast, the coincided overlaps of INPs and APS >2.5μm particles on dust days as shown in Fig. 3e indicate the significant role of coarse-sized dust particles as INP contributors. From 16:00 h to the end of the day, Fluo WIBS>0.5µm, total particles on non-dust days are up to twofold higher than those for dust days (Fig. 6a). Those enriched FABPs on non-dust days are generally contributed by ABC WIBS ,B WIBS and BC WIBS particles (Fig. 6b, d, h). In particular, the elevated concentrations of FABPs (ABC WIBS ,B WIBS and BC WIBS )coincide with the spark of INP evening peak (by 50%, from 20:00 to 21:00 in Figs. 2d to 5d) on non-dust days, suggesting the contribution of FBAPs to the increase in INPs. In contrast, a relatively constant FABP median concentration (e.g., Fluo WIBS>0.5µm, total in Fig. 6a and the other types) corresponds to stable INP concentrations on dust days during the same period (Figs. 2eto5e). Additionally, the increased B WIBS and BC WIBS particles with a peak around 20:00 h on non-dust days may be attributed to the elevated eBC particles (Fig. S14d). In summary, FBAPs and dust particles are the key drivers for the INP diurnal cycles observed at (HAC)2.ABC WIBS FBAPs are the most important particles regulating INPs in the PBL without dust events. Dust events can substantially enrich larger-sized dust containing FABPs (A WIBS and C WIBS ) during noon time when PBL is high (suggesting that airmasses in PBL may also be of the sources of A WIBS and C WIBS FBAPs), while they supply more large-sized C WIBS and BC WIBS FABPs when PBLH is lower during nighttime, which also contributes to the observed INPs. Vertical INP distributions with respect to PBL/FT interface for each atmospheric classification Figure 7illustrates INP concentration distributions with PBLH under different atmospheric conditions. In general, INPs show a positive and https://doi.org/10.1038/s41612-024-00817-9 Article npj Climate and Atmospheric Science | (2025) 8:160 9