Whiteface Mountain (WFM) Atmospheric Research Observatory, located in the remote high peaks regions of upstate New York, northeastern United States, is an exceptional site for investigating cloud water chemistry. It is situated at an elevation of 1483 m (4867 ft) and spends 20 %–60 % of the time during summer immersed in clouds . This mountain top site is the 5th highest peak of the Adirondack Mountains and downwind of diverse natural and anthropogenic emission sources from Canada, the Great Lakes, and the western United States. Since the 1970s, WFM has been providing critical insights into atmospheric chemistry, influencing CAA regulatory actions in the 1990s aimed at reducing acid deposition .

Spanning from June through September each year, the dataset captures warm cloud events during the late spring (June), summer (July–August), and early fall (September). These variations are influenced by biogenic activity during the plant growth season in the late spring and meteorological shifts across the sampling period. Additionally, the dataset includes periods of episodic wildfire events, which commonly occur during the summer months. Fall season is also influenced by biogenic emissions due to leaf foliage. By including both regular seasonal cycles and intermittent high-impact smoke as well as heatwave events, this study offers a comprehensive view of the factors influencing LMWOA dynamics in cloud water chemistry. Furthermore, the remote location of WFM minimizes direct anthropogenic influence, with minimal local pollution interference, enabling the observation of broader atmospheric trends. This makes WFM highly relevant for studying the transport and transformation of air masses influenced by regional and long-range emissions, making it a one-of-a-kind high-elevation site to investigate cloud water chemistry. It should be noted that this study does not include wintertime measurements due to operational limitations at the site during freezing conditions from late fall to May. Sampling ends by the first week of October to prevent problems associated with riming. Therefore, the dataset focuses exclusively on warm cloud measurements, capturing the most dynamic period of cloud water chemistry.

Cloud water samples were collected at WFM using an automated Mohnen Omni-directional passive cloud water collector equipped with Teflon strings (0.035–0.04 mm diameter) with a 50 % droplet cut-off diameter of 2–5 µm at typical summit wind speeds of 2–10 ms-1 . Sampling occured under specific meteorological conditions: Liquid Water Content (LWC) >0.05 gm-3 (measured by a Gerber Particle Volume Monitor), temperature >2 °C, no rain (detected by a CAPMoN rain sensor), and wind speed >2 ms-1 (measured by an RM Young anemometer). When these conditions were met for at least one minute, the collector exposed Teflon strings to capture cloud water from the passing airflow . Samples were collected every 12 h into a refrigerated 1 L ISCO bottle connected to a 12 L high-density polyethylene (HDPE) accumulator vessel that is continuously weighed to monitor accumulated cloud water sample volume. The collected water was filtered through an in-line 0.4 µm track-etched polycarbonate filter using a peristaltic pump to prevent microbial degradation, and filters were replaced twice a week. From 16 July 2024, we started using 0.2 µm filters for microbial analysis . Each sample was then transferred to a 250 mL bottle and kept frozen until chemical analysis. While LMWOA measurements have been reported for specific cloud events at various locations, including WFM, routine bulk sampling at a remote site required careful sample handling to prevent degradation or volatile losses. All samples were filtered during collection and stored at 2 °C to prevent microbial degradation during the three-day holding time until delivery to the lab for chemical analysis. Chloroform (CHCl3) was not added to the cloud water samples since controlled administration of CHCl3 concentrations based on collected sample volume is not yet a feature of the automated cloud water collection system. It should be noted that used CHCl3 and 0.4 µm filters prior to organic acid analysis of cloud water samples collected on an hourly basis for a short-term study at WFM in 1987. Although 0.2 µm filters are widely regarded as sufficient for sterilization purposes, isolated a wide variety of viable bacteria from 0.2 µm filtered freshwater samples, suggesting that even with this level of filtration, some microbes may end up in the filtered cloud water samples. Routine blank and rinse samples of the cloud water collection system using Type 1 ultrapure water (18 MΩcm resistivity) were collected multiple times each year to assess for contamination. A table showing measurement statistics for these samples is added to the Supplement (Table S3).

Detailed descriptions of the measurement methods for each year and each analyte are presented in the Supplement following some of the procedures described in previous studies . In each year of the study, quality control samples were prepared fresh daily and monitored to assess column performance. Lab duplicates were analyzed by randomly selecting one sample from that day's batch. Results within a relative percent difference (RPD) acceptance criterion of ±10 % were accepted; otherwise, more duplicates were analyzed or the entire batch was reanalyzed.

This study presents a seven-year study of three major LMWOA (formate, acetate, and oxalate). Although acetate was suspected to co-elute with glycolate in a few instances, the peaks for acetate were resolvable, and glycolate concentrations were consistently below the detection limit. Additionally, measurements for other organic acids were limited due to analytical limitations. Hence, the focus of this paper is on the three most abundant LMWOA. Method detection limits (MDL) for each organic acid analyte are provided in Table S1, along with the corresponding laboratories where the measurements were conducted for each year that samples were collected. More details about methods for inorganic analytes can be found in .

Analyte concentrations measured by ion chromatography (mgL-1) are converted to molL-1 by dividing by molar mass and then converted to equivalent concentrations (eqL-1) by multiplying by ionic charge. The total concentration of measured cations (∑Cat) and anions (∑An) are then calculated according to the following equations, with concentrations for all analytes in units of eqL-1: 1∑Cat=[NH4+]+[Ca2+]+[Mg2+]+[Na+]+[K+]+[H+]bulk2∑An=[SO42-]+[NO3-]+[Cl-]+∑LMWOA where ∑LMWOA is the sum of the measured formate, acetate, and oxalate concentrations. It should be noted that these are the maximum possible contributions of these LMWOA to the ion balance since these values do not account for the deprotonation state of those anions (except as calculated in Eqs.  and , and reported in Figs. –).

Then we infer the concentrations of missing anions by assuming that electrochemical charge balance was sustained in these bulk aqueous solutions, but assuming that the anions listed in Eq. () were insufficient to characterize all the anions present. Inherent to Eqs. () and () is the assumption that any cations other than those listed in Eq. () are present in negligible concentrations (Fig. S7 in the Supplement). 3missingAn=∑Cat-∑An4∑Ions=missingAn+∑Cat+∑An

The measured ionic concentrations of each analyte are then divided by the total ion concentration (∑Ions) to determine ionic fractions, and charge imbalance is calculated from 100%×(missing An/∑Ions)

Following the procedure of , we estimate acidity for the vast majority of cloud droplets as they existed in the atmosphere prior to collection using a “top down” approach based on the measured bulk cloud water acidity and measured concentrations of the two dominant non-volatile cations (Mg2+ and Ca2+), which are presumed to exist entirely within the coarse mode aerosol population, therefore representing a very small fraction of the particles present even when comprising a significant fraction of the aerosol mass loading, i.e. [H+]TD=[H+]bulk+[Ca2+]+[Mg2+], which is subsequently used to calculate pHTD. Similarly, a “bottom up” approach is used to estimate cloud droplet acidity based on measured concentrations of the dominant inorganic ions typically associated with the fine mode aerosol population, i.e., [H+]BU=([SO42-]+[NO3-])−[NH4+].

Since organic acids are also expected to exist primarily within the fine mode aerosol population , in this study, we extend the “bottom up” acidity calculation to include the measured LMWOA concentrations, thereby introducing a new variable: 5[H+]BU*=([SO42-]+[NO3-]+[H+]OA)−[NH4+] where [H+]OA=[HCOO-]+[CH3COO-]+[C2O42-], representing the dissociated fractions of the three LMWOAs calculated from their measured concentrations and pKa values at pHTD (our best estimate for cloud droplet pH), using equations from (used in Figs. –).

Similarly, the Missing Acid Fraction (MAF) concept introduced by was modified to include LMWOA, and the new variable MAF^* is introduced here: 6MAF*=([H+]TD−[H+]BU*)/[H+]TD For this study, negative values for [H+]BU and [H+]BU* are set to zero for calculating pHBU and pHBU*, respectively, as well as for calculating MAF and MAF^*.

Isoprene, a major BVOC at the site, was measured from grab samples collected in stainless steel whole air canisters from the base of WFM (610 m elevation) in the summer of 2022 and 2023. The local canopy at this elevation can be categorized as a deciduous hardwood forest, dominated by sugar maple, yellow birch and American beech trees. These trees are known to be strong monoterpene emitters with isoprene emissions depending on leaf age and heat stress . In this study, we report only isoprene, a short-lived, highly reactive BVOC that is also prone to wall loss in canisters, making it a challenging compound to measure accurately. Grab samples were analyzed using the NSF NCAR Trace Organic Gas Analyzer (TOGA), a gas chromatograph coupled with a Time-of-Flight mass spectrometer (GC-MS), following the method described by . At the WFM summit, trace gases are measured continuously throughout the year, including chemical species such as ozone (O3) and carbon monoxide (CO). Black carbon (BC) is measured by Aethalometer. Further details about the gas-phase dataset can be found in .

Formate and acetate are the most abundant monocarboxylic acids measured in WFM cloud water, with median concentrations of 3.95 and 3.75 µeqL−1 respectively, shown in Fig. . Oxalate and malonate are the next two abundant dicarboxylic acids, with median concentrations of 2.17 and 3.74 µeqL−1 respectively. Malonate was only measured in 2018 and may have a high bias, since co-elution with an unidentified peak precluded reliable measurements in the later years of the study. The rest of the organic acids measured had median concentrations <1 µeqL-1 and were frequently below detection limits, indicating their limited contributions to chemical properties in cloud water such as ion balance and acidity. For the remainder of the paper, the discussion of LMWOA focuses on formate (HCOO-), acetate (CH3COO-), and oxalate (C2O42-), given that these three organic acids were measured for the entire seven-year period of this study.

Monthly differences in LMWOA concentrations were observed for the three species of primary focus, as highlighted in Fig. S4. Formate, acetate, and oxalate concentrations exhibited the highest concentrations in the growing season (June and July) compared to August and September; here growing season corresponds to early summer, when canopy growth and leaf expansion are most active. Oxalate concentrations exhibited greater variability in July, when wildfire smoke events occurred more frequently, particularly in 2021 and 2023. These smoke events could provide additional precursors such as glyoxal, pyruvate, methylglyoxal, other LMWOA, further accelerating its dominant production pathway via in-cloud secondary processes, and aqueous phase reactions involving OH radical, as well as iron-catalyzed oxidation . The same wildfire smoke influence further explains both oxalate and DOC mean concentrations peaking in July. Also, a major fraction of in-cloud secondary formation of oxalate is known to originate from isoprene-derived compounds globally , which peaked in July at the site (Fig. ). This contrasts with formate and acetate, which peaked in June, likely due to direct photochemical oxidation of a broad mix of vegetation and soil emissions under maximum sunlight exposure, with possibly a higher morning OH from HONO photolysis and more persistent nighttime NO3 radicals that could accelerate the primary production of those in the aqueous phase . The decline in all LMWOA concentrations in September suggests reduced emission of precursors from the surrounding vegetation in the fall, which was also observed in isoprene measurements as shown in Fig. .

Oxalate generally accounts for ∼1 % of cloud water DOC concentrations. The measured oxalate to DOC ratio also exhibits an increase with measured ozone (O3) mixing ratios, especially apparent in cloud water samples collected in 2018. However, the relationship to O3 is less clear in subsequent years. Overall, for cloud water samples collected 2018–2024, the oxalate : DOC ratio increases only slightly from (1.0 ± 0.084) % to (1.4 ± 0.27) % for time periods with <50 and >50 ppbvO3, respectively, as shown in Fig. . Higher O3 levels can enhance secondary oxalate formation from its gas phase precursors such as glyoxal, methylglyoxal and pyruvic acid via BVOC oxidation pathways, as observed in an aerosol study from a forest at Mt. Fuji . However, higher DOC concentrations may alter oxidant mixing ratios, and extreme smoke events (typically associated with high DOC concentrations) can substantially reduce actinic flux , which can complicate the relationship between oxalate, DOC and O3, likely accounting for some of the variability observed in Fig. .

Figure  shows a comparison of isoprene concentrations measured at WFM in 1994–1995, reported in , overlaid with our 2022–2023 data at WFM shown in a box-whisker plot. The 2022–2023 data have a higher mean in July (∼7.31 ppbC) compared to (∼3.65 ppbC), possibly due to differences in sampling frequency, but also due to the inherent variability of isoprene due to its high reactivity. However, both datasets follow a similar seasonal trend, peaking in July and declining in fall, which aligns with the monthly trends of all three major LMWOA and DOC, suggesting that isoprene emissions play a key role in the production of LMWOA.

Ion fractions are averaged by year in Fig. . Major changes in cloud water ion fractions between 1987 and recent years include a decrease in sulfate ion and increases in ammonium and calcium. The three most abundant LMWOA account for 6 %–22 % of anions and 3 %–9 % of DOC on an annual basis (Figs.  and S3). On average, 14 %–40 % of anions remain unaccounted for, which could include other organic anions such as organosulfates, organonitrates, malonate, succinate, glutarate and other dicarboxylic acids, as well as other potential anions like halides and bicarbonate. High molecular weight multifunctional organics like Humic-Like Substances (HULIS), produced through aqueous phase reactions of small aldehydes with ammonium and amines during cloud processing and oligomerization of water-soluble organics during cloud processing of wildfire smoke , may also contribute to both the unidentified DOC species and ion imbalance. Assessing ion balance for all samples (2018–2024), the three major LMWOA increase the slope for measured anion versus cation concentrations from 0.61 to 0.72 over the seven years (Fig. S5).

In a growing fraction of cloud water samples, ammonium concentrations exceed the combined concentrations of sulfate and nitrate (i.e., [NH4+]>[SO42-]+[NO3-]). This shift is described by as a “New Chemical Regime” where the surplus NH4+ is no longer fully neutralized by strong inorganic acids, as also discussed by . Under this regime, we observe a strong positive correlation between surplus NH4+ and the sum of three major LMWOA such as formate, acetate, and oxalate (Fig. ). In contrast, samples with [NH4+]<([SO42-]+[NO3-]) show weak or no correlation between these species. This suggests that when surplus NH4+ is available, it interacts more with deprotonated organic acids, forming ammonium organic salts, especially in DOC-rich samples.

High surplus NH4+ concentrations associated with high LMWOA concentrations occurred during a heat wave in 2018 (Fig. ), which correlated with both high ozone mixing ratios and high BVOC concentrations, likely resulting in enhanced secondary formation of LMWOA . However, for DOC-rich samples collected in 2023 and 2024, the relationship between surplus NH4+ and LMWOA concentrations is much less clear, perhaps in-part due to differences in NH3 availability and nucleophilic addition reactions between NH3 and organic acids, which can produce diverse nitrogen-containing organic compounds (which are not included in our dataset) but may act as a sink for both NH4+ and some organic acids . The high DOC events from 2021 and 2023 are associated with two extreme wildfire smoke events, which likely follow chemically distinct aging pathways as compared to the 2018 pollution event modeled by . We discuss and contrast the chemical and meteorological conditions encountered during these two wildfire smoke events in Sect. 3.5 of this paper.

The significant contribution of organic acids to ion balance suggests that organic acids can also be important for cloud droplet acidity. used a “top down” vs. “bottom up” framework to estimate cloud droplet acidity based on the measured inorganic ions. By comparing pH from these two different calculations, a Missing Acid Fraction (MAF) was inferred, which was shown to be rapidly growing at WFM since ∼2009, with more than half of the inferred cloud droplet acidity no longer explained by the traditional inorganic ions SO42-, NO3- and NH4+ when that study concluded in 2021. They proposed that unmeasured LMWOA could be a dominant species reshaping the site’s chemical regime. The updated dataset presented here shows that the difference between pHTD (blue) and pHBU (orange) has continued to grow (Fig. ), resulting in a further increase in MAF since 2021 (Fig. ) and approaching MAF=1.0 in 2022–2024, with nearly all of the inferred cloud droplet acidity unaccounted for by the measured inorganic ions. Note that pHTD varied very little during the entire seven-year study, with annual median values averaging ∼4.4 ± 0.09 (1σ).

Adding the new measurements of LMWOA to the “bottom up” calculation for estimated cloud droplet pH decreases pHBU* by 0.1–0.45 units and explains 5 %–36 % of the cloud droplet acidity on an annual median basis. These findings are supported by previous studies where formic + acetic acids supplied up to 60 % of droplet acidity in remote and mountain sites . The contribution of LMWOA to droplet acidity was greatest in 2018, 2021 and 2023, as exhibited by a greater difference between MAF and MAF* in Fig. . The latter two of these years were heavily influenced by high primary LMWOA emissions from wildfire smoke events , with LMWOA contributing 30 %–33 % to cloud droplet acidity on an annual median basis in 2021 and 2023. The meteorological and chemical properties associated with wildfire smoke events in these two years are contrasted in Sect. 3.5 of this paper. However, the even stronger impact of LMWOA on cloud droplet acidity in 2018 likely resulted from different precursors and dominant reaction pathways than in 2021 and 2023, since the 2018 pollution event modeled by was associated with high temperatures, high O3 and high BVOC concentrations, with little influence from wildfire smoke.

The linear fit to annual median MAF* (which includes dissociated LMWOA) exhibits an upward trend with a 7.6 % increase per year and r2=0.59 over the seven years of measurements (Fig. ). The upper error bars represent MAF* with measured K+ concentrations added to the HBU* calculation to account for potential internal mixing of K+ in the fine aerosol mode. While K+ may be found in both the fine and coarse aerosol particles , showed that, in wildfire smoke, K+ is often internally mixed with fine organic aerosol. Figure  shows that the uncertainty in MAF* due to K+ mixing state is typically small compared to other sources of uncertainty. The lower error bar for MAF* represents the maximum possible contribution of measured LMWOA, assuming complete dissociation of these acids irrespective of pH. In the years 2021 and 2023 when wildfire events were more frequent, higher uncertainties are observed for the contribution of LMWOA to MAF* (Fig. ). Another source of uncertainty in the missing acid fraction calculation arises from the assumption that NO3- is entirely in fine mode aerosol, whereas in reality some fraction may exist in coarse mode aerosol . Note that if NO3- were alternatively found entirely within the coarse mode, the MAF* calculation would likely be minimally affected, since MAF*=1-(HBU*/HTD) and since both HTD and HBU* would be overestimated by the same degree, exactly equal to the measured bulk NO3- concentrations.

Wildfire smoke events are expected to grow more frequent with climate change . Smoke plumes are known to be highly enriched in organic matter and the associated particles can be acidic . While NH3 emissions are often thought to be associated with agriculture (e.g., fertilizers and animal waste), biomass burning plumes can also be enriched in NH3 . Thus, smoke-influenced cloud events present interesting case studies for more in-depth analysis of LMWOA, surplus NH4+ and MAF under high DOC loadings. Summertime smoke plumes over regions such as Saskatchewan, Manitoba, Ontario, and Quebec are dominated by emissions from wildland Boreal fires , frequently transported to WFM. In this section, two severe wildfire smoke episodes are compared to assess how smoke plume aging shapes the chemical composition of cloud water samples. Both of these smoke events were associated with Boreal forest fires in Canada, but the fires were located at different distances upwind from WFM, thereby impacting the transport time and transport path of the smoke plume to our site.

The first smoke event, which intercepted WFM on 19–22 July 2021, was detected across NY State with ground-based LIDAR measurements . reported record-breaking Boreal wildland forest fire emissions in 2021, with Manitoba, Saskatchewan, British Columbia, and Ontario provinces witnessing their largest total area burned in July 2021 , intensified by a Pacific Northwest heat dome. HYSPLIT back trajectories, launched at 12:00 UTC on 20 July 2021 from the WFM summit traced that plume back to south-central Canada with at least five days transport time from the source to the measurement site. Hence, we categorize this event as an “Aged Smoke” plume, and three cloud water samples were collected during this episode. We report LMWOA data from two of those samples since not enough sample volume was collected in the first sample to conduct all chemical analyses, since clouds were intercepted only at the very end of the 12 h sampling period (as indicated in yellow shading as the daytime period on 19 July in the LWC panel of Fig. ). Nevertheless, enough sample volume was collected to conduct all other measurements.

The second smoke event originated from lightning-caused wildfires from south-central Quebec . These fires produced a colossal smoke plume that caused severe air pollution and record-breaking daily mean PM_2.5 concentration in New York City (NYC) on 7 June, making it the city's worst air quality event in the last 50 years . The smoke plume arrived in New York City (Queens) in a mean of 47.5 h per the HYSPLIT forward trajectory, severely affecting the air quality of NYC from 6–9 June 2023 . Since WFM lies upwind from NYC, and showed elevated carbon monoxide (CO) and black carbon (BC) levels roughly 24 h before the plume reached the city, we estimate this smoke was about one day old, significantly fresher than the first event. Hence, it has been categorized as a “Fresh Smoke” event influencing two cloud water samples collected during that period. The woodsmoke contribution (obtained by subtracting the 370 nm absorption channel from the 880 nm absorption channel of the aethelometer used to measure BC, ) was markedly higher for this event than the “Aged Smoke” event. Note that, since BC can be scavenged in clouds, and cloud droplets >10 µm are not sampled by the aerosol inlet , the BC measurements during in-cloud periods should be treated as a lower bound of the actual BC mass loading since only interstitial aerosols and smaller cloud droplets containing BC are sampled by this instrument while in-cloud. Interestingly, the second sample from this “Fresh Smoke” event appeared to have higher BC and woodsmoke concentrations than the first sample, potentially indicating less efficient scavenging of smoke particles by cloud droplets, despite the high LWC, possibly due to a change in the smoke plume mixing-state and/or particle size distribution.

Elevated concentrations of biomass burning tracers, including CO, BC, DOC and K+ coincide with both elevated surplus NH4+ and LMWOA concentrations for these two smoke events, reinforcing the previously discussed chemical interactions between surplus NH4+ and LMWOA, but specifically for wildfire smoke plumes here. However, there are some marked differences between these two episodes. The “Aged Smoke” samples exhibited substantially higher surplus NH4+ than the “Fresh Smoke,” despite cloud droplets from both events exhibiting the same high acidity levels (pHTD∼3.9) and liquid water content (LWC), which both favor NH3 partitioning to the aqueous phase. This difference may be explained by the different transport paths and aging times of the smoke plumes. The “Aged smoke” plume was transported across the agriculturally intensive Upper Midwest (Wisconsin), where it appears to have picked up more NH3, resulting in surplus NH4+ concentrations nearly as high as the measured LMWOA concentrations. The “Fresh Smoke” plume passed over northern NY and Montreal, where there is much less agricultural activity, likely leading to the much lower surplus NH4+ concentrations observed in those samples, accounting for only about 10 % of the measured LMWOA concentrations. Samples from both events contained elevated levels of LMWOA, with “Fresh Smoke” samples richer in monocarboxylic acids (formate, acetate), which have been shown previously to be directly emitted by wildland fires .

The “Aged Smoke” samples exhibited lower LMWOA concentrations and a reduced LMWOA contribution to DOC compared to the “Fresh Smoke” samples. We hypothesize that for “Aged Smoke” samples, a major fraction of the freshly emitted monocarboxylic acids had undergone aqueous phase transformation as the plume aged, resulting in lower LMWOA concentrations several days later when the samples were intercepted at WFM. Conversely, DOC concentrations in the “Aged Smoke” samples were significantly higher than the “Fresh Smoke” samples. A simultaneous decrease in LMWOA and increase in DOC may suggest that oligomerization or accretion reactions dominate over fragmentation reactions during atmospheric aging , boosting DOC during transport and potentially forming CHON compounds , as also observed in a previous study on Boreal wildfire smoke plume aging . Functionalization reactions can further enhance DOC, since oxidation of insoluble BC and Brown Carbon (BrC, ) can also increase DOC loading upon aging , consistent with both the greater DOC and lower BC and woodsmoke signals observed in the “Aged Smoke” samples. CO mixing ratios were similar during both events, peaking at ∼500 ppbv, but generally in the 200–400 ppbv range.

When including LMWOA in the estimated cloud droplet pH calculation, 70 %–90 % of the Missing Acid Fraction (MAF*) was resolved in the “Fresh Smoke” samples (Fig. b, bottom panel), illustrating the importance of these measured compounds to cloud droplet acidity. However, for the “Aged Smoke” samples, the estimated cloud droplet acidity remained entirely unaccounted for, even after including LMWOA in the MAF* calculation (Fig. a, bottom panel). This difference indicates that cloud droplet acidity for the “Fresh Smoke” event is primarily driven by LMWOA (formate, acetate, oxalate), whereas cloud droplet acidity for the “Aged Smoke” event is dominated by other unmeasured anions, likely also organic-rich, including higher molecular weight, possibly multifunctional organic acids. Since these were smoke-influenced samples, the potential effect of K+ on pHBU was tested, with impact on MAF* represented by the upper error bars shown in bottom panels of Fig. . Again, we also evaluated the maximum possible contribution of LMWOA concentrations to pHBU by assuming full deprotonation regardless of pHTD, with impact on MAF* represented by the lower error bars shown in Fig. . It has been reported before that K+ participates in neutralizing cloud water acidity, potentially forming potassium organic salts .

Cloud droplet acidity was high for both of the highlighted smoke events, with pHTD<4.0 in most of these cloud water samples, matching median bulk cloud water pH from the 1990s prior to full implementation of the CAA amendments , as compared to pHTD values ≳4.5 in cloud water samples collected before and after these two smoke events. With the formation of higher molecular weight organic acids during the aging process, any new production of carboxylic acid functional groups could further lower pH. However, for the “Aged Smoke” event, cloud droplets acquired additional NH4+ through gas phase partitioning of NH3, neutralizing any added protons . This external buffering by ammonia uptake explains the higher surplus NH4+ in “Aged Smoke” samples at the same cloud droplet pH and LWC level as “Fresh Smoke” samples, a mechanism that is distinctly different in aerosols and clouds, as discussed by .

We further evaluate smoke-influence on measured LMWOA concentrations for the full seven-year dataset, using measured K+ concentrations >2.5 µeqL-1 as a proxy for smoke-influence, consistent with mean K+ concentrations for cloud water samples characterized as high probability of smoke-influence by . Figure  shows that LMWOA concentrations >50 µeqL-1 generally occur at pHTD values <4.7, coinciding with high K+ concentrations. This suggests that smoke-influenced cloud droplets are moderately acidic, often likely in part due to their high organic acid concentrations. A similar relationship between DOC and pHTD, fit to an exponential decay function, was reported by , with smoke-influenced samples from 2021 exhibiting the highest DOC values observed throughout that previous WFM study (note that cloud water samples from 2022–2024 were not included in that study). The review article by also characterized biomass burning aerosol as drivers of acidity in cloud droplets. A previous analysis of cloud water samples collected from WFM in 2014 showed an increase in O:C ratio as the bulk cloud water pH decreased, with two out of three of the smoke-influenced samples exhibiting the lowest bulk pH values, even more acidic than the urban-influenced cloud water samples selected in that study. They further reported that smoke-influenced samples contained more oxidized and chemically complex dissolved organic matter, including slightly higher average O:C ratios and double bond equivalent (DBE) compared to non-smoke-influenced samples (e.g., biogenic, urban, or mixed source). This previous analysis, which focused on higher molecular weight organic compounds, is consistent with the observed increase in LMWOA with cloud droplet acidity in the current study, suggesting that aqueous phase oxidation of both low molecular weight and high molecular weight compounds occurs simultaneously under acidic conditions.

While measured K+ concentrations in cloud water have been linked to smoke-influence, the available cloud water dataset cannot be used to directly ascertain the age of smoke plumes. The two extreme smoke events highlighted in this study suggest that atmospheric aging of smoke during transport over 1–5+ d can substantially alter the chemical properties of the smoke-influenced cloud water, consistent with the findings of , who found that aged smoke plumes were at risk of being misclassified due to the chemical transformations occurring during the aging process. Future work utilizing comprehensive backtrajectory analysis, in conjunction with satellite observations and potentially other in-situ chemical markers like NOx and NOy, is needed for robust assessment of smoke plume age on LMWOA concentrations and cloud droplet pH.