the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
The representativeness of observed Δt in calculating formation and growth rates of newly formed atmospheric particles – hypothesis and examination
Abstract. New particle formation (NPF) is commonly characterized by particle formation rates (FR) and growth rates (GR) derived from fixed-site Eulerian observations. However, particles detected at different times in Eulerian observations do not necessarily represent the evolution of the same air parcel, so the observational time intervals used to derive FR and GR may differ from the corresponding in-situ formation and growth timescales. Here, we examine the representativeness of the observational time interval (Δtobserve) in the calculation and interpretation of FR and GR using synchronous high-time-resolution measurements from multiple particle instruments in a semi-enclosed indoor environment and ambient air in southeastern Hainan, China. Nine short-term NPF cases were analyzed to determine the observational intervals associated with particle number increase and modal-size evolution. The results show that substantial increases in newly formed particle number concentration and particle growth to tens of nanometers can occur within several to tens of minutes, and that the derived FR of 2.7–21 cm-3 s-1 and GR of 22–186 nm h-1 were highly sensitive to the selected Δtobserve. Comparison of the FR-NMINP (net maximum increase in the nucleation-mode particle number concentration) and GR-Dpgmax (maximum geometric median diameter of the grown new particles) relationships between the NPF events in this study and previously reported regional NPF events further revealed substantially lower slopes (by approximately 1–2 orders of magnitude) for events characterized by longer Δtobserve, suggesting that such intervals may not adequately represent the true formation and growth timescales of newly formed particles.
- Preprint
(1201 KB) - Metadata XML
-
Supplement
(839 KB) - BibTeX
- EndNote
Status: open (until 11 Dec 2026)
-
CC1: 'Comment on egusphere-2026-5714', Simo Hakala, 02 Nov 2026
reply
-
AC1: 'Reply on CC1', Xiaohong Yao, 05 Nov 2026
reply
We appreciate Simo Hakala's interest in our preprint and his prompt comment. We consider this an important point and are pleased to address it here. We agree that the studies by Kivekäs et al. (2016) and Hakala et al. (2023) provide important insights into the effects of spatial heterogeneity and atmospheric transport on fixed-site NPF observations. Specifically, (1) Kivekäs et al. (2016) demonstrated through model simulations how spatial variations in particle formation and growth, combined with air-mass advection, can influence the observed evolution of particle size distributions; and (2) Hakala et al. (2023) further demonstrated, using field observations and Lagrangian particle-growth modeling, that apparent particle shrinkage can be explained by the sequential transport of particles with different growth histories, without requiring evaporation. Accordingly, these important references will be incorporated into the Introduction and relevant discussion in the revised manuscript.
We would also like to elaborate further on this point. Our study addresses a related but distinct aspect of this issue by highlighting the importance of observational timescales in interpreting FR and GR derived from fixed-site measurements, particularly whether these timescales adequately represent the underlying particle formation and growth processes. In addition to the transport-related effects highlighted in these two studies, our study provides complementary insights into how observational timescales may affect the interpretation and quantification of NPF processes. Taken together, these studies address complementary aspects of a common question: how spatial heterogeneity, atmospheric transport, and observational timescales influence the quantitative interpretation of NPF processes from fixed-site measurements.
Furthermore, we will revise the statement in the Conclusions to avoid overgeneralization regarding previously reported FR and GR values. We again thank Simo for this comment, which has helped us improve the scientific context and balance of our manuscript.
Zihan Kong (PhD candidate and first author), on behalf of all authors
Citation: https://doi.org/10.5194/egusphere-2026-5714-AC1
-
AC1: 'Reply on CC1', Xiaohong Yao, 05 Nov 2026
reply
-
RC1: 'Comment on egusphere-2026-5714', Anonymous Referee #1, 18 Nov 2026
reply
General Comments
Kong et al. investigate an important yet often overlooked issue in characterizing NPF events: due to the continuous movements of airflows, the observational time-interval at fixed measurement sites may differ from the actual timescales of particle evolution within a given air parcel. This discrepancy can potentially bias the calculation and interpretation of formation rates (FR) and growth rates (GR). Based on high-time-resolution CPC and SMPS measurements indoors and outdoors in Hainan, the authors analyzed nine short-term NPF events and derived apparent FRs and GRs that differ by 1–2 orders of magnitude from previously reported regional values.
However, the current manuscript requires major revisions before it can meet the publication standards as a research article in ACP. My primary concerns are as follows:
First, the authors attribute the high GRs calculated from short time windows to genuine rapid growth, but fail to rule out artifacts caused by SMPS scan-smearing and ground on unjustified assumptions. Deriving an exceptionally high GR of 186 nm/h from only 3 scans within a 10-minute window, based on unverified assumptions about initial particle sizes, seems problematic.
Claiming that particles grow rapidly "from initial clusters or extremely small particles" to the nucleation/Aitken mode relies on the assumption that the initial size is ~2 nm. As the SMPS lower detection limit is 10 nm, using an assumed starting size to calculate an extremely high GR, and then using that GR to prove "rapid growth," sounds unreasonable. Furthermore, the NPF event identification approach lacks a standardized methodology; subjectively determining the start and end times directly alters Δt.
Second, the data processing methodology requires clarification. Constraining the intercept to zero when correlating SMPS and CPC measurements appears unreasonable given their distinct lower size detection limits (10 nm vs. 4 nm).
Finally, the presentation quality needs improvement. The case-by-case analysis of the nine short-term NPF events is overly lengthy and descriptive. These cases should be grouped by similarities and differences. Furthermore, the connection between the PMF bias analysis (Categories 1 and 2) and the FR/GR calculations for the 9 cases is unclear.
Specific Comments
- Line 19-26: The abstract contains long and complex sentences laden with clauses and brackets. Please break this down into shorter, more direct sentences. In addition, I was wondering why compare the slopes of the FR-NMINP and GR- Dpg,max relationships instead of directly comparing FR and GR? A slope that is 1–2 orders of magnitude higher could result from either an overestimated numerator (FR/GR) or an underestimated denominator (NMINP/Dpg,max). Please make this more concise and direct.
- Line 48-54: Key references regarding the influence of transport on NPF are missing. Please refer to Zhang et al. (2026, ES&T: Investigating Ultrafine Aerosol Turbulent Fluxes during Atmospheric New Particle Formation Events”) and Cai et al. (2018, ACP: “Estimating the influence of transport on aerosol size distributions during new particle formation events”).
- Line 71-72: Previous studies (e.g., Kulmala et al., 2004; Chu et al., 2019 as the authors mentioned in Line 65) have implicitly considered these comparisons. Please revise this exaggerated statement accordingly.
- Line 82-90: Could the authors elaborate on the physical relationship between the defined “States” and the classic NPF event types? Specifically, I was wondering why Class I NPF events are broadly classified in to State 3, where Δt_observe is presumed to diverge from Δt_in-situ. If a regional Class I NPF event occurs, air parcels across a vast area are generally homogeneous and undergo concurrent NPF processes. Under these conditions, transport primarily advects similar air masses past the fixed site, meaning Eulerian observations should closely approximate the intrinsic formation and growth rates (Δt_observe ≈ Δt_in-situ).
- Introduction: The extensive use of inline mathematical symbols and expressions disrupts the reading flow. Please extract key variables and expressions and present them as standalone numbered equations. Additionally, the detailed descriptions of “t-transport” do not seem necessary since it is not addressed in later analysis and discussions.
- Line 144-156: Please explain the rationale for using the CCN counter data and explicitly state the criteria used to determine “stable operating conditions”. Why was 240 s used for the 0.21% SS level while 180 s was used for others? The CCN data seems not to be adequately discussed in the text.
- Line 219 and 226: Manually constraining the regression intercept to zero is physically unreasonable because the CPC (~4 nm) and SMPS (~10 nm) have different lower detection limits. The 23% systematic underestimation by the SMPS is likely due to this 4-10 nm measurement gap rather than “the loss of nucleation mode particles” resolved by PMF. In line 238, when dealing with Category 2b, a non-zero intercept was used in the regression analysis. Suggest leaving out the zero intercept constraints and keep consistency throughout the manuscript.
- Fig 2(c)-(d): The SMPS and CPC have different time resolutions (3.5 min vs. 1 s). Please explicitly describe the data alignment or averaging methodology used to generate these scatter plots.
- Line 245: Please clearly state the criteria used to classify the different cases into specific “States”.
- Line 273-275: Please provide a plot showing the evolution of Dpg,1 from 15:06 to 15:38 to support this description.
- Line 286-290: Could the authors provide some explanations on why Dpg,1 remained constant for ~5 min when NPF was first detected, and why total particle concentrations decreased between 09:44 and 09:48 while Dpg,1 continued to grow (is this due to coagulation?). Additionally, claiming that a GR of 33 nm/h is “reasonably consistent” with 22 nm/h is inaccurate, as it represents a 50% increase.
- Line 302-309: First, please provide references for the expected κ values of particles consisting of H2SO4 and organics, and compare them with the authors’ observations before concluding the event was “driven by the combined condensation of H2SO4 and organic vapors”. Moreover, as indicated by the authors in Line 198, κ is a hygroscopic parameter influenced by particle size, chemical composition and mixing state, linking it directly to specific condensing precursors requires caution. Same goes for relevant expressions in the following three paragraphs.
- In addition, the study reports exceptionally high GRs (e.g., 186 nm/h for Case 7). Condensational growth from ~2 nm to 33 nm in 10 minutes requires an immense concentration of condensable vapors. To validate that this is true in-situ growth and not simply the advection of an older, larger air mass passing the sensor (a spatial gradient masquerading as temporal growth), the authors should calculate the theoretical vapor concentration required to sustain a GR of 186 nm/h. If this calculated concentration exceeds physically realistic atmospheric levels, it strongly implies the observation is a transport/spatial heterogeneity artifact rather than genuine growth.
- Another small note, during a 7-minute NPF period, a 3.5-min resolution SMPS can only complete two full scans; I was wondering why does Figure S5f show three data points?
- Line 310-343: For Case 7, calculating a GR of 186 nm/h relies on the assumption that particles grew from ~2 nm to 33 nm in 10 minutes. Since the 2–10 nm range is unmeasured, the initial size is an arbitrary assumption (e.g., assuming an initial size of 5 nm or 8 nm would yield drastically different GRs, such as 168 nm/h or 150 nm/h). How do the authors distinguish a newly formed mode from the transport of an existing regional mode? For Case 6, the window used to calculate GR=33 nm/h is unclear. Please provide the starting Dpg,1, time windows and the uncertainties of the estimated GR.
- Line 333 (Case 8): The text mentions "corresponding κ values suggested that this event was likely driven mainly by H2SO4...". Given the high time resolution of the NPF burst, does the CCNC data (which operates on a 5-minute per SS cycle) possess sufficient temporal resolution to accurately attribute chemical drivers to an 11-minute event?
- Line 345-354: The conclusion that classic banana-shaped events “do not necessarily represent the true in-situ growth” is too strong and unsupported by the presented evidence. The 9 short-term cases presented here (some with highly dubious GR calculations from only 2-3 scans) are insufficient to overthrow the physical interpretations of massive historical datasets. Please soften this tone. Additionally, the claim regarding sensitivity to “spatial scale” lacks discussions and direct evidence in this study.
- Line 362-374: The terminology seems contradictory. Short-window GRs are referred to as both “apparent” and “intrinsic”. Please unify terms (e.g., short-window GR ≈ intrinsic; long-window GR = apparent average if I understand correctly). Furthermore, the authors assume that short-window GRs are “true” and long-window GRs are “underestimated”, without recognizing that short windows may severely overestimate GR due to noisy signals or scan-smearing. For example, if a rapid burst of particles reaches the detector while the DMA is scanning the 25 nm bin, the inversion algorithm will report a number concentration peak at 25 nm, potentially falsely interpreting this concentration spike as a size shift. This artifact may therefore masquerade as rapid particle growth. Calculating a GR from the "modal diameter" shift between just 2 or 3 highly transient and potentially smeared scans is technically unreliable. The authors should address this fundamental instrumental limitation and discuss the uncertainty bounds it places on their ultra-high GR estimates. One the other hand, did the authors compare the GR value from the “modal-fitting” and the “appearance time” method? Without independent verification (e.g., Lagrangian tracking or dense multi-site observations), I suppose the authors cannot conclusively prove that short-window GRs represent true intrinsic growth.
- PMF analysis in the SI: The CPC measures down to ~4 nm, while the SMPS measures down to 10 nm. During NPF, this 4~10 nm gap contains massive amounts of newly formed particles. This discrepancy is a physical measurement limit by design, not a complex source profile error or losses. Furthermore, the PMF methodology lacks basic details: which software was used? How many factors were tested? Were constraints applied? How was the temporal mismatch (hourly averages vs. 3.5-min scans) handled, given that hourly averaging smooths out the very short-term NPF bursts the authors identified?
Technical Corrections
- Line 121: suggest changing “maintain minimal internal solar irradiance” to “minimize indoor solar irradiance”.
- Suggest placing Figure 1 after it is first mentioned in the text.
- Line 176-177, the lower detection limit of SMPS at 10 nm sounds awkward. Suggest changing to “The nucleation mode was partially captured during NPF events due to the 10 nm lower detection limit of the SMPS”.
- Line 183, Equation 3: Please provide the assumed parameters or the explicit formula used to calculate the Coagulation Sink (CoagS) by the pre-existing particles.
- Line 186: suggest changing “coagulation sink of the preexisting particle population” to “coagulation sink by the preexisting particle population”.
- Line 193-195: Describing the integration limits and process entirely in text makes the definition of Dd confusing and hard to follow. Please rewrite this sentence for clarity and add the mathematical equation representing the integration from Dmax to Dd that yields NCCN.
- Line 215-217: “whereas overestimation occurred less frequently” is confusing. It seems describing overestimation in Category 2 is less frequent than those in Category 1, however, the results show that there is almost no overestimation in Category 1 (Line 213-214). Suggest remove or revise this sentence.
- Line 241: suggest changing “genuine probabilistic events” to “genuine physical events”.
- Line 263: please increase the resolution of Figure 3, as it becomes blurred when zooming in.
- Line 383-386: The phrase “intrinsic rate of new particle formation” sounds awkward. Suggest changing to “the intrinsic (or in-situ) new particle formation rate within an air parcel”.
Citation: https://doi.org/10.5194/egusphere-2026-5714-RC1
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 320 | 84 | 27 | 431 | 49 | 28 | 32 |
- HTML: 320
- PDF: 84
- XML: 27
- Total: 431
- Supplement: 49
- BibTeX: 28
- EndNote: 32
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This preprint seems to be lacking some critical references that should be acknowledged:
The issue discussed in this manuscript was (first) identified and clearly demonstrated by modeling already in:
Kivekäs, N., Carpman, J., Roldin, P., Leppä, J., O’Connor, E., Kristensson, A., & Asmi, E. (2016). Coupling an aerosol box model with one-dimensional flow: a tool for understanding observations of new particle formation events. Tellus B: Chemical and Physical Meteorology, 68(1). https://doi.org/10.3402/tellusb.v68.29706
and it has also been in the focus of explaining previous observations:
Hakala, S., Vakkari, V., Lihavainen, H., Hyvärinen, A.-P., Neitola, K., Kontkanen, J., Kerminen, V.-M., Kulmala, M., Petäjä, T., Hussein, T., Khoder, M. I., Alghamdi, M. A., and Paasonen, P.: Explaining apparent particle shrinkage related to new particle formation events in western Saudi Arabia does not require evaporation, Atmos. Chem. Phys., 23, 9287–9321, https://doi.org/10.5194/acp-23-9287-2023, 2023.
Overly strong statements, such as: "Overall, this study strongly suggested that all FR and GR previously reported in the literature should be re-examined..." in the conclusions need to be modified accordingly.