Benoît Crouzy

Knowledge gaps in the source and parameterization of ice-nucleating particles (INPs) remain a major uncertainty in quantifying the properties and climate impacts of mixed-phase clouds (MPCs). Bioaerosols are increasingly recognized as important INPs for MPCs, yet it is unclear whether their contribution is critical through modelling studies. We investigate this using field observations at a semi-rural site in the central Europe, combining INP and aerosol measurements, remote sensing, and air-mass source analysis. We synergically use the results of in-situ and offline measurements to identify INPs originated from different sources, including biological aerosols, dust, and biomass burning aerosols, and to quantify their abundance and relative contributions to total INPs. More than 85% of immersion-mode INPs (>−24°C) are heat-labile and significantly correlated with fluorescent biological aerosols particles and pollen, while heat-resistant INPs (<~10%) are likely mineral dust, while biomass burning is an insignificant source. The proposed bioaerosol-ware INP parameterization reproduces observations across multi-regional datasets, showing that neglecting bioaerosol-INPs results in an average ~50% (~32%) bias in predictions of immersion-mode INPs active warmer than −15°C (−24°C), with frequent deviations of up to an order of magnitude from observations.

Knowledge gaps in the source and parameterization of ice-nucleating particles (INPs) remain a major uncertainty in quantifying the properties and climate impacts of mixed-phase clouds (MPCs). Bioaerosols are increasingly recognized as important INPs for MPCs, yet it is unclear whether their contribution is critical through modelling studies. We investigate this using field observations at a semi-rural site in the central Europe, combining INP and aerosol measurements, remote sensing, and air-mass source analysis. We synergically use the results of in-situ and offline measurements to identify INPs originated from different sources, including biological aerosols, dust, and biomass burning aerosols, and to quantify their abundance and relative contributions to total INPs. More than 85% of immersion-mode INPs (>−24°C) are heat-labile and significantly correlated with fluorescent biological aerosols particles and pollen, while heat-resistant INPs (<~10%) are likely mineral dust, while biomass burning is an insignificant source. The proposed bioaerosol-ware INP parameterization reproduces observations across multi-regional datasets, showing that neglecting bioaerosol-INPs results in an average ~50% (~32%) bias in predictions of immersion-mode INPs active warmer than −15°C (−24°C), with frequent deviations of up to an order of magnitude from observations.

Characterizing atmospheric bioaerosols is important for understanding their potential impacts on the environment and public health. In this study, we developed a novel flow cytometry-based approach to determine the low nucleic acid (LNA), high nucleic acid (HNA), dead, and intact bioaerosol populations in samples collected with a wet cyclone at Payerne, Switzerland, during spring and summer 2024. We found that the average bioaerosol number concentration reached (2.47 ± 3.35)×104 m−3. The HNA and intact populations were the most abundant populations, representing the largest fraction of total bioaerosols within 82% and 92% of the samples, respectively. Our results show that the LNA can be composed of dead bioaerosols, which correlated strongly with atmospheric particulate mass. Quantitative PCR and metagenomic analysis reveal significant correlations (Spearman, PERMANOVA, and Mantel) and associations between the different kingdoms analyzed, reflecting complex ecological interactions in the atmosphere among the communities. Despite this complexity, the LNA and HNA populations were mainly composed of bacteria, archaea, and fungi. Pollen abundance was positively correlated with temperature and negatively correlated with relative humidity (RH) and pollution (NOx and NO2), as these conditions promote the formation of sub-pollen particles (pollen fragments) through osmotic (bursting) and oxidative stress. Factor analysis indicates a seasonal dynamics transition from plant-associated bioaerosols in the spring season, to other bioaerosol types to be co-emitted during summer. Overall, the integration of flow cytometry with molecular analysis provides a framework to characterize and quantify bioaerosols and provides new insights into the ecological structure, variability, and sources of the atmospheric microbiome.

Knowledge gaps in the source and parameterization of ice-nucleating particles (INPs) remain a major uncertainty in quantifying the properties and climate impacts of mixed-phase clouds (MPCs). Bioaerosols are increasingly recognized as important INPs for MPCs, yet it is unclear whether their contribution is critical through modelling studies. We investigate this using field observations at a semi-rural site in the central Europe, combining INP and aerosol measurements, remote sensing, and air-mass source analysis. We synergically use the results of in-situ and offline measurements to identify INPs originated from different sources, including biological aerosols, dust, and biomass burning aerosols, and to quantify their abundance and relative contributions to total INPs. More than 85% of immersion-mode INPs (>−24°C) are heat-labile and significantly correlated with fluorescent biological aerosols particles and pollen, while heat-resistant INPs (<~10%) are likely mineral dust, while biomass burning is an insignificant source. The proposed bioaerosol-ware INP parameterization reproduces observations across multi-regional datasets, showing that neglecting bioaerosol-INPs results in an average ~50% (~32%) bias in predictions of immersion-mode INPs active warmer than −15°C (−24°C), with frequent deviations of up to an order of magnitude from observations.

Knowledge gaps in the source and parameterization of ice-nucleating particles (INPs) remain a major uncertainty in quantifying the properties and climate impacts of mixed-phase clouds (MPCs). Bioaerosols are increasingly recognized as important INPs for MPCs, yet it is unclear whether their contribution is critical through modelling studies. We investigate this using field observations at a semi-rural site in the central Europe, combining INP and aerosol measurements, remote sensing, and air-mass source analysis. We synergically use the results of in-situ and offline measurements to identify INPs originated from different sources, including biological aerosols, dust, and biomass burning aerosols, and to quantify their abundance and relative contributions to total INPs. More than 85% of immersion-mode INPs (>−24°C) are heat-labile and significantly correlated with fluorescent biological aerosols particles and pollen, while heat-resistant INPs (<~10%) are likely mineral dust, while biomass burning is an insignificant source. The proposed bioaerosol-ware INP parameterization reproduces observations across multi-regional datasets, showing that neglecting bioaerosol-INPs results in an average ~50% (~32%) bias in predictions of immersion-mode INPs active warmer than −15°C (−24°C), with frequent deviations of up to an order of magnitude from observations.

Advances in automatic bioaerosol monitoring require updated approaches to evaluate particle classification algorithms. We present a training and evaluation framework based on three metrics: (1) Kendall's Tau correlation between predicted and manual concentrations, (2) scaling factor, to assess identification efficiency, and (3) off-season noise ratio, quantifying off-season false predictions. Metrics are computed per class across confidence thresholds and five stations, and visualised in graphs revealing overfitting, station-specific biases, and sensitivity–specificity trade-offs. We provide optimal ranges for each metric respectively calculated from correlations on co-located manual measurements, worst-case scenario off-season noise ratio, and physical sampling limits constraining acceptable scaling factor. The evaluation framework was applied to seven deep-learning classifiers trained on holography and fluorescence data from SwisensPoleno devices, and compared with the 2022 holography-only classifier. Classifier performances are compared through visualisation methods, helping identifying over-training, misclassification between morphologically similar taxa or between pollen and non-pollen particles. This methodology allows a transparent and reproducible comparison of classification algorithms, independent of classifier architecture and device. Its adoption could help standardise performance reporting across the research community, even more so when evaluation datasets are standardised across different regions.

Advances in automatic bioaerosol monitoring require updated approaches to evaluate particle classification algorithms. We present a training and evaluation framework based on three metrics: (1) Kendall’s Tau correlation between predicted and manual concentrations, (2) scaling factor, to assess identification efficiency, and (3) off-season noise ratio, quantifying off-season false predictions. Metrics are computed per class across confidence thresholds and five stations stations, and visualised in graphs revealing overfitting, station-specific biases, and sensitivity–specificity trade-offs. We provide optimal ranges for each metric respectively calculated from correlations on co-located manual measurements, worst-case scenario off-season noise ratio, and physical sampling limits constraining acceptable scaling factor. The evaluation framework was applied to seven deep-learning classifiers trained on holography and fluorescence data from SwisensPoleno devices, and compared with the 2022 holography-only classifier. Classifier performances are compared through visualisation methods, helping identifying over-training, misclassification between morphologically similar taxa or between pollen and non-pollen particles. This methodology allows a transparent and reproducible comparison of classification algorithms, independent of classifier architecture and device. Its adoption could help standardise performance reporting across the research community, even more so when evaluation datasets are standardised across different regions.

Advances in automatic bioaerosol monitoring require updated approaches to evaluate particle classification algorithms. We present a training and evaluation framework based on three metrics: (1) Kendall’s Tau correlation between predicted and manual concentrations, (2) scaling factor, to assess identification efficiency, and (3) off-season noise ratio, quantifying off-season false predictions. Metrics are computed per class across confidence thresholds and five stations stations, and visualised in graphs revealing overfitting, station-specific biases, and sensitivity–specificity trade-offs. We provide optimal ranges for each metric respectively calculated from correlations on co-located manual measurements, worst-case scenario off-season noise ratio, and physical sampling limits constraining acceptable scaling factor. The evaluation framework was applied to seven deep-learning classifiers trained on holography and fluorescence data from SwisensPoleno devices, and compared with the 2022 holography-only classifier. Classifier performances are compared through visualisation methods, helping identifying over-training, misclassification between morphologically similar taxa or between pollen and non-pollen particles. This methodology allows a transparent and reproducible comparison of classification algorithms, independent of classifier architecture and device. Its adoption could help standardise performance reporting across the research community, even more so when evaluation datasets are standardised across different regions.

Advances in automatic bioaerosol monitoring require updated approaches to evaluate particle classification algorithms. We present a training and evaluation framework based on three metrics: (1) Kendall’s Tau correlation between predicted and manual concentrations, (2) scaling factor, to assess identification efficiency, and (3) off-season noise ratio, quantifying off-season false predictions. Metrics are computed per class across confidence thresholds and five stations stations, and visualised in graphs revealing overfitting, station-specific biases, and sensitivity–specificity trade-offs. We provide optimal ranges for each metric respectively calculated from correlations on co-located manual measurements, worst-case scenario off-season noise ratio, and physical sampling limits constraining acceptable scaling factor. The evaluation framework was applied to seven deep-learning classifiers trained on holography and fluorescence data from SwisensPoleno devices, and compared with the 2022 holography-only classifier. Classifier performances are compared through visualisation methods, helping identifying over-training, misclassification between morphologically similar taxa or between pollen and non-pollen particles. This methodology allows a transparent and reproducible comparison of classification algorithms, independent of classifier architecture and device. Its adoption could help standardise performance reporting across the research community, even more so when evaluation datasets are standardised across different regions.