the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 02 Oct 2026
Nonlinear Effects of Relative Humidity on the Light Absorption of Water-Soluble Brown Carbon: Molecular-Level Constraints under Different Pollution Levels
Abstract. Brown carbon (BrC) plays an important role in aerosol light absorption, yet its relative humidity (RH) dependent optical evolution remains poorly constrained. Based on field observations and ultra-high-resolution mass spectrometry, this study reveals that the light absorption of water-soluble BrC is nonlinearly regulated by the interplay between pollution level and RH. The mass absorption efficiency at 365 nm shows a biphasic response to RH. It increases with RH, reaching a maximum in the range of approximately 50%–60% RH, and then decreases when RH exceeds this threshold. Molecular-level analysis reveals that RH may alter the chemical fate of water-soluble BrC chromophores, steering their formation and attenuation pathways while concurrently reshaping their optical absorption profiles. In the low-to-moderate RH regime, liquid-phase chemical processing drives the accumulation of chromophores. Beyond the inflection point, increasing aerosol liquid water activates several synergistic suppression pathways, including oxidative bleaching, hydrolytic cleavage of N/S-containing chromophores, and physical dilution. High pollution levels may partially offset the RH-driven suppression effects. These findings challenge static parameterizations of BrC in models and highlight the necessity of incorporating RH-driven molecular restructuring into radiative forcing assessments.
- Preprint
(1489 KB) - Metadata XML
-
Supplement
(1069 KB) - BibTeX
- EndNote
Status: open (until 13 Nov 2026)
- RC1: 'Comment on egusphere-2026-5498', Anonymous Referee #1, 15 Oct 2026 reply
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 284 | 82 | 17 | 383 | 38 | 22 | 25 |
- HTML: 284
- PDF: 82
- XML: 17
- Total: 383
- Supplement: 38
- BibTeX: 22
- EndNote: 25
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Overarching comments:
This is an interesting manuscript with the potential of being an improtant contribution to the field of WS-Brc research. The combination of optical measurements and molecular composition has the potential to make a useful contribution to ACP. However, the principal conclusions currently go beyond what this short observational campaign can establish. In particular, the manuscript attributes differences among air masses and emission episodes to humidity-driven chemical transformations without adequately separating source variability, atmospheric processing, and sampling effects.
The main contribution should be framed around the observed covariation of water-soluble brown carbon absorption and molecular composition with relative humidity and pollution conditions. A causal transition from chromophore formation to oxidative and hydrolytic destruction requires stronger evidence. The analysis of a humidity threshold, the interpretation of mass-normalized absorption, and the structural assignments from molecular formulas all need substantial revision. The regional literature also needs to be broadened, including the directly relevant field constraints of e.g., Dasari et al. (2019) and other South Asian studies.
Major Comments:
1 Separate the humidity association from changing emission sources
Sections 2.1, 3.1 and 3.4; lines 84-92, 145-165 and 365-398. The campaign includes 32 day/night filters over 4-19 February 2021, with molecular analysis of 11 selected samples. Different humidity regimes occur on different dates and under different source conditions. Similar PM2.5 mass does not make two samples equivalent in source composition, age, oxidant exposure or transport history. Consequently, plotting MAE against RH does not by itself isolate a humidity effect.
Please provide a complete time series and a sample-level table linking sampling interval, RH and temperature distributions, meteorology, optical properties, source markers and molecular analyses. Identify the number of independent filters and distinct episodes in each group. Assess the RH association using a small, defensible set of available source and meteorological covariates, together with separate day/night analyses and sensitivity tests that exclude the fireworks episode and individual pollution episodes. Avoid a heavily parameterized model that this dataset cannot support.
The fireworks episode introduces a different source mixture and is not independent validation of the proposed RH mechanism. Elevated levoglucosan in Fig. S5 also warrants consideration of concurrent biomass-burning emissions. Without source-resolved or controlled evidence, present this episode as an illustrative mixed-source case. If RH effects cannot be separated statistically, retain the observational relationships and qualify their causal interpretation.
2 Quantify the evidence for a humidity threshold
Lines 171-178, Figs. 1b and 4, and the conclusions. The central threshold is inferred from separately fitted RH ranges. These fits do not establish that a change point has been estimated from the data or that the segmented relationship outperforms a simpler alternative. Please compare parsimonious linear, curved and segmented descriptions, report the fitting procedure and uncertainty of any estimated turning point, and show the individual observations with sampling-period or source information. Assess sensitivity to influential samples and to the chosen boundaries.
The molecular samples span 40%-60% RH and greater than 80% RH according to Fig. 4, leaving the intervening range unmeasured by FT-ICR-MS. The molecular data therefore cannot locate a transition near 60% or demonstrate the proposed sequence through 60%-80%. Discuss this gap explicitly. Reconcile the RH categories in the Fig. 1b legend with the ranges used in the text and molecular analysis. A maximum or turning point is also different mathematically from an inflection point.
3 Correct the explanation based on physical water dilution
Lines 346-349 and the abstract. The claim that adding particle water directly lowers absorption efficiency per unit carbon mass is not established for the reported measurement. MAE is obtained by dividing absorption by WSOC. For a chemically unchanged solution obeying Beer-Lambert behavior, dilution reduces absorbance and carbon concentration proportionally, leaving their ratio unchanged.
Moreover, the measurements are made on extracted material rather than on intact particles at controlled ambient RH. Please distinguish chemical changes retained after extraction from reversible effects of particle water on size, refractive index, aggregation or optical geometry. An increase in nonabsorbing organic carbon could lower bulk MAE, but that would be a compositional change and needs its own evidence. Any proposed concentration-dependent association or protonation effect also requires demonstration.
Provide the absorption and MAE equations, carbon-mass basis, blank and baseline treatment, extraction volumes and pH, and measurement uncertainties. Check whether differences in extract pH or ionic strength could affect the spectra. Remove the simple water-dilution explanation unless an appropriate measurement-specific mechanism is demonstrated.
4 Distinguish molecular formulas from identified chromophore structures
Lines 215-256, 323-345 and 378-390. FT-ICR-MS elemental formulas and a DBE screening criterion identify candidate formula classes; they do not uniquely identify nitroaromatics, organosulfates, nitrooxy organosulfates, polymers, thiosulfinates or polysulfides. CHON formulas may represent several nitrogen functionalities with different chemistry and absorptivity. High O/N or DBE alone does not establish a nitro functional group, its position, or the extent of conjugation.
Please distinguish measured formulas, candidate structural interpretations and directly identified compounds throughout. Where standards, fragmentation spectra or chromatographic optical information are unavailable, use appropriately tentative assignments and avoid assigning class-specific molar absorptivities to unidentified mixtures. Likewise, fluorescence and E2/E3 provide indirect compositional information rather than unique molecular weights or structures.
Table S2 needs particular attention. The haze-specific CHON subset has lower mean O/N and modified aromaticity than the clean-specific subset: O/N is 2.60 versus 4.66 and AImod is 0.18 versus 0.46. The corresponding whole-pool averages give a much smaller O/N difference and no increase in mean AImod. Explain which population supports the structural interpretation, and provide a valid basis for the asserted statistical significance. These summaries do not independently establish that nitroaromatics drive the optical enhancement
5 Strengthen molecular quantification and statistical reporting
Sections 2.4 and 3.2-3.3; Figs. 3-6 and S3-S9. Explain how the 11 molecular samples were selected and whether the selection was made independently of optical outcomes. Provide sample identifiers, actual RH values, sample loading, SPE carbon recovery, blank results, mass accuracy, formula-assignment constraints, detection thresholds and normalization procedures. The different filtration and extraction treatments used for spectroscopy and FT-ICR-MS also need to be considered when comparing the two datasets.
Formula counts and normalized ESI intensities cannot be treated as carbon mass, molecular concentration or absorption contributions without calibration. Variable ionization efficiency, matrix suppression, SPE selectivity and detection thresholds can alter both counts and relative abundances. In particular, loss of a formula from a spectrum is not equivalent to atmospheric destruction of that compound. Define every percentage and specify whether it refers to formula number, summed ion intensity or carbon mass.
Figs. S3 and S6 appear to correlate five regime means. State whether the regressions use those means or the individual molecular samples, and report n, confidence intervals and p values. As a conditional check, a Pearson r of 0.76 with only five independent observations has a two-sided p value of approximately 0.14. This calculation is not a reanalysis of unavailable raw data, but illustrates why the regression unit matters. For formula-by-formula RH correlations, specify thresholds, treatment of nondetections and correction for multiple testing. Six low/moderate-RH samples and five high-RH samples offer limited independent replication regardless of the number of detected formulas.
6 Recast oxidation and hydrolysis as hypotheses unless directly constrained
Lines 289-364. Cross-sectional changes in formula distributions, estimated volatility and relative abundance do not establish parent-product relationships or reaction rates. Source changes, selective scavenging, gas-particle partitioning and analytical recovery could also produce differences among these samples. A negative abundance-RH relationship cannot demonstrate that a hydrolytic loss rate exceeds a formation rate.
For oxidation, explain why the proposed fragmentation mechanism is favored over those alternatives. The reported shift toward lower mean carbon oxidation state is not a unique signature of oxidative bleaching; it requires a chemically explicit explanation, including selective loss or formation of products if relevant. For hydrolysis, identify a plausible hydrolysable functional group and discuss its lifetime under the estimated conditions. Nitro groups and organic nitrate groups should not be treated interchangeably.
If existing data cannot distinguish these pathways, present them as possible explanations and revise the repeated claims of direct evidence and validation. Additional experiments or structural analyses would strengthen a mechanism-specific claim, but a carefully qualified observational paper can be revised without demanding every proposed process be measured anew. Also check the applicability of the cited radical-production literature: Lelieveld et al. (2021) concerns epithelial lining fluid and does not directly demonstrate radical production in atmospheric particles.
7 Document aerosol water and pH calculations and verify the fog classification
Lines 99-106, 145-153 and 269-278. The description of ISORROPIA-II forward-mode calculations lists particle ions, temperature and RH, but does not explain the gas-phase constraints. Forward-mode calculations require total gas-plus-particle inputs for the relevant semivolatile systems. State whether NH3, HNO3 and HCl were measured, estimated or omitted, and assess the consequences for pH and water. Hennigan et al. (2015) provides relevant guidance on this issue.
Clarify whether the reported ALWC includes organic-associated water and distinguish aerosol water from actual fog-droplet water. Explain whether the model was run at the time resolution of RH measurements or using filter-period averages; averaging RH before applying a nonlinear water-uptake calculation may alter the result. Because RH is itself an input to modeled ALWC, an ALWC-RH relationship is not an independent validation of the proposed chemical mechanism.
The stated visibility criterion of less than 10 km does not establish fog. The WMO definition requires surface horizontal visibility below 1 km. Provide the actual visibility record and evidence of fog, and describe how the PM2.5 inlet treated droplets and interstitial aerosol. If these observations cannot distinguish fog from humid haze or mist, revise the classification. Also document the gas measurements used for NOR, SOR and Ox, and the EC-tracer assumptions and uncertainty used to estimate SOC.
8 Broaden the regional literature and define the advance over previous work
Introduction, lines 35-80; bleaching discussion, lines 310-355; model implications, lines 418-427. Dasari et al. (2019), Science Advances, is absent from the reference list despite being directly relevant. The authors should cite and discuss this work because it provides regional field constraints on the loss of WS-BrC absorptivity during atmospheric transport and develops an aging framework relevant to dynamic optical parameterization. It reports an ambient first-order bleaching rate of 0.20 +/- 0.05 day-1 during over-ocean transport in the South Asian outflow. This study does not establish the RH threshold proposed here, and should not be cited as if it does.
The discussion should also incorporate relevant field evidence from other polluted regions, particularly South Asia. Satish et al. (2017) provides high-time-resolution winter observations in Kanpur that are useful for evaluating temporal changes and competing explanations. Mukherjee et al. (2020) links BrC and HULIS optical and compositional properties to sources and processing in the eastern Indo-Gangetic Plain. Rastogi et al. (2021) provides diurnal spectral and source information from Delhi. Discuss what these studies imply for source variability, bleaching and the transferability of the present findings. The contribution here should be distinguished from previous evidence that BrC absorption evolves during atmospheric aging.
9 Match the model implications and data access to the evidence
Abstract, conclusions and data availability. The present analysis does not establish a transferable RH-only parameterization or quantify a radiative-forcing bias. Its findings may motivate models that couple source composition, water and atmospheric processing, but the location and universality of a threshold remain uncertain. Either provide a quantitative relationship with uncertainty and a stated domain of applicability, or moderate the model claims and identify the observations needed to evaluate them.
Deposit the sample-level optical and chemical data, gas and meteorological records, molecular formula and intensity matrices, model inputs and analysis code in an accessible repository. This is particularly important for assessing the regression units, episode sensitivity, molecular selection and numerical discrepancies identified above. Provide source data for all figures and define the weighting and uncertainty used for averages.
Minor comments
Lines 155-165 and 369-371. The shared ordering asserted for absorption and MAE conflicts with the later statement that fireworks MAE is slightly below clean-day MAE. Give separate rankings and reconcile Fig. 1a with the text.
Lines 201-204 and Fig. S2a. The SOR values assigned to fog and fireworks in the text appear inconsistent with the regime positions of the maxima in the radar plot. Check the labels and values against the underlying sample data.
Line 350 and Fig. S2b. The claim that fog corresponds to the peak ALWC appears inconsistent with the plotted maximum during fireworks. Reconcile the text, plot labels and averaging procedure.
Lines 351-352 and Fig. 1a. Clarify whether the quoted clean/fog MAE means and uncertainties describe all filters or only a selected subset. State whether error bars are standard deviations, standard errors or analytical uncertainties.
Lines 236-240. Lower levoglucosan supports a relative change in biomass-burning influence; it does not by itself establish negligible total combustion influence or constrain coal contributions.
Lines 199-205 and 274-278. NOR and SOR are bulk conversion metrics, not direct measurements of the dominant oxidation route or nitration-precursor supply. Avoid using them alone to establish a specific mechanism.
Figs. 3-6 and Tables S1-S2. Define all abundance denominators, weighting, sample counts and symbols, including RA, Xn, Xc and X/C. Variability among formulas must be distinguished from variability among independent filters.
Lines 132 and 139. Clarify how the algorithm described for D-labeled spectra was applied here, and correct DEB to DBE. Supply the actual formula-assignment and screening settings.
Figs. 1d, 4 and 6. Define the integration and units in Fig. 1d; show actual RH values and sample identifiers in Fig. 4; give volatility-class boundaries, parameterization equations and uncertainty in Fig. 6. Estimated volatility is not an independent experimental measure of bond cleavage.
References suggested for the revision
The following references support the specific comparisons and methodological comments above. Izhar et al. (2020) and Rawat et al. (2022) are already present in the manuscript and should be integrated into the expanded discussion.
Dasari, S., et al. (2019). Photochemical degradation affects the light absorption of water-soluble brown carbon in the South Asian outflow. Science Advances, 5, eaau8066.
https://doi.org/10.1126/sciadv.aau8066
Satish, R., Shamjad, P., Thamban, N., Tripathi, S., and Rastogi, N. (2017). Temporal Characteristics of Brown Carbon over the Central Indo-Gangetic Plain. Environmental Science & Technology, 51, 6765-6772.
https://doi.org/10.1021/acs.est.7b00734
Mukherjee, A., Dey, S., Rana, A., Jia, S., Banerjee, S., and Sarkar, S. (2020). Sources and atmospheric processing of brown carbon and HULIS in the Indo-Gangetic Plain: Insights from compositional analysis. Environmental Pollution, 267, 115440.
https://doi.org/10.1016/j.envpol.2020.115440
Rastogi, N., et al. (2021). Diurnal variability in the spectral characteristics and sources of water-soluble brown carbon aerosols over Delhi. Science of the Total Environment, 794, 148589.
https://doi.org/10.1016/j.scitotenv.2021.148589