Measurement Report: Impact of African Aerosol Particles on Cloud Evolution in a Tropical Montane Cloud Forest in the Caribbean

African aerosol particles, traveling thousands of kilometers before reaching the Americas and the Caribbean, directly scatter and absorb solar radiation and indirectly impact climate by serving as cloud condensation nuclei (CCN) 10 or ice nuclei (IN) that form clouds. These particles can also affect the water budget by altering precipitation patterns that subsequently affect ecosystems. As part of the NSF-funded Luquillo Critical Zone Observatory, field campaigns were conducted during the summers of 2013, 2014, and 2015 at Pico del Este, a site in a tropical montane cloud forest on the Caribbean island of Puerto Rico. Cloud microphysical properties, which included liquid water content, droplet number concentration, and droplet size, were measured. Using products from models and satellites, as well as in-situ 15 measurements of aerosol optical properties, periods of high and low dust influence were identified. The results from this study suggest that meteorology and air mass history have a more important effect on cloud processes than aerosols transported from Africa. In contrast, air masses that arrived after passing over the inhabited islands to the southeast led to clouds with much higher droplet concentrations, presumably due to aerosols formed from anthropogenic

The meteorological parameters that were derived along the trajectories were temperature, pressure, relative humidity, and rainfall. The air mass origins are classified according to where the air was located seven days before 130 reaching the site.
The metrics selected for assessing the environmental effects of these air masses of differing origin were the scattering and absorption coefficients with derived Ångström exponents to characterize the aerosol properties, the droplet size distribution, and liquid water content to describe the cloud microphysical properties and the concentration 135 of water-soluble ions to characterize the cloud water chemistry. The hourly latitude, longitude, altitude, and precipitation along the back trajectories are the metrics evaluated for characterizing the air mass properties.

Data Quality Assurance
The BCP data were recorded every second and reduced to 10-minute averages. Beswick et al. (2013) provide 140 a detailed error analysis of the BCP measurements, in particular the uncertainty associated with deriving the size from the light scattering signal. For this study, the BCP was installed in a wind tunnel with constant airflow and orientation towards the prevailing wind direction. Laboratory tests of different wind speeds measured inside the tunnel showed that this implementation enhanced the outside wind speed by a factor of 1.4 inside the tunnel. The average corrected airspeed was 11.2 ± 1.8 m s -1 .

145
To ensure a sampling uncertainty of 10% or less, a population of 100 particles per sampling interval is needed.
Coincidence errors become significant, > 10% undercounting, when droplet concentrations exceed 500 cm -3 (Beswick et al., 2013); however, the PDE cloud concentrations rarely exceeded 500 cm -3 . The overall estimated uncertainties in number concentration, EOD and LWC,using error propagation,are 15%,20%,and 40%,respectively. 150 At CSJ, nephelometer and CLAP data were also gathered at one-second intervals and reduced to sixminutes and hourly averages. The nephelometer data were corrected for the truncation error (Anderson and Ogren, 1998), and the CLAP data were corrected for enhanced absorption artifacts (Bond et al., 1999;Ogren, 2010). Data from both instruments were adjusted to standard temperature and pressure.

Statistical Analysis
The directly measured and derived parameters were evaluated to determine if their values were Gaussian distributed in order to select parametric or non-parametric statistical methods to use in testing the null hypothesis. This was done using the Shapiro-Wilk methodology (Shapiro and Wilk, 1965). Given that the majority of the metrics had 160 frequency distributions that were not Gaussian, the Mann-Whitney, two-tailed test was used to determine if the average values of the selected parameters were statistically different with respect to the air mass origins.

Aerosol Properties
Based upon the criteria described previously, using the light scattering coefficients at a wavelength of 550 nm, three periods of high dust (H1 -H3) and four of low dust (L1 -L4) were identified for 2013, one period of high dust (H4) in 2014 and twelve high dust periods (H5 -H16) and five low dust periods (L5 -L9) were identified in 2015. Figure 2a-c shows the time series for the scattering coefficients at three wavelengths (colored curves) and the 170 Ångström scattering exponents derived from these coefficients (black). The orange boxes identify the periods of light scattering exceeding the selected threshold to designate dust. The blue boxes encompass low light scattering events.
The solid arrows in Fig. 2c mark the days selected for a more in-depth analysis of the air mass origins on those days and the associated cloud properties, as will be discussed in later sections.

175
There are no regular trends in the optical properties that can be linked to daily variations in local emissions, indicating that the observed fluctuations are related to larger-scale changes in the air masses. The trends are irregular and marked by sharp increases and decreases, such as those seen in 2013 on days 165 (increase) and 167 (decrease).
The durations of the events are also quite variable, several lasting more than 24 hours in 2013 and 2014, while in 2015 there are many events that are only several hours long. It is possible that some of the short events that are close to each 180 other are actually related, i.e., coming from the same dust event but interrupted by brief shifts in the local circulation patterns before returning to the broader scale flows that are bringing the dust.  time series identify the periods whose cloud properties were analyzed in detail when the air origins were from the African Sahel (red), Saharan Desert (orange), Atlantic Ocean (blue) and Southeast of Puerto Rico (black). These periods correspond to those evaluated in the back trajectory analysis.
190 Figure 3 shows box plots for the parameters studied for high and low dust periods. Both the scattering and absorption coefficients were considerably higher for high dust periods. The SAE was lower for high dust periods, which indicates the presence of larger particles than in low dust periods. The AAE for high dust periods was usually above two, while on low dust periods were usually below two. Both the low SAE and high AAE values confirm the 195 presence of dust in high dust periods (Cazorla et al., 2013).
Non-sea salt calcium (nss-Ca 2+ ) in cloud water samples, collected in periods with high dust, was in considerably higher concentration than those taken in low dust periods (average of 83 vs. 0.97 µeq/L), validating the presence of minerals in periods designated as high dust. A similar finding was previously reported in several other  All the metric variables showed statistically significant differences between high and low dust periods (p < 210 0.001; α = 0.05) Table 1.1 shows the averages and standard deviations of the metric variables for high and low dust periods.

Air Mass Origin Analysis
An analysis of back trajectories from the PDE site was used to track the histories of air masses that were 220 arriving during periods of cloud at the site. Back trajectories were computed every two hours. Those time periods when six or more consecutive trajectories of air had been over the same general region, seven days previously, are used to identify air whose aerosol particles were in all likelihood distinctly different in physical and chemical properties. The four regions that were identified in this manner are: 1) Atlantic Ocean (AO), 2) Southeast (SE), 3) Saharan Desert (SD) and 4) African Sahel (AS). The days that were selected were those during which there was 225 already a cloud present at the PDE.  highlights the primary difference between the AO and SE airmasses. They both are over the ocean seven days prior, in the mid-latitudes and more than 1500 km from the nearest land mass; however, the AO air arrives at the PDE from the northeast whereas the SE air first travels south of Puerto Rico's latitude before turning back to the northwest. This latter trajectory brings it over the inhabited islands to the east and south of Puerto Rico.

235
Although the SD and AS trajectories are over the African continent seven days prior to arriving at PDE, the SD air masses were in the region over the desert while the AS air is more tropical and over the region known as the African Sahel. The other metric that clearly distinguishes the air masses from one another is the altitude history that is documented in Fig. 4b. Of the four types, the SE air masses are the only ones that have traveled much of their sevenday trajectory near the surface, especially the three days before reaching the island. Although their histories vary, as 240 illustrated by the vertical bars that mark the standard deviations about the mean, the AO, SD and AS air masses are all descending as they reach PDE. This is an important observation due to the different forcing mechanisms that lead to cloud formation, a factor that will be discussed below.
The altitude histories are also important indicators of the possible sources of aerosols that arrive at PDE and 245 may influence the cloud properties. For example, SE air masses spent more than three days of their travel below 1000 m and would presumably be influenced by mixtures of sea spray and anthropogenic emissions as they passed over the inhabited islands east and southeast of Puerto Rico. On the other hand, neither the AO or AS trajectories took them close to the surface, suggesting that the aerosols in these air masses were either introduced from the surface at some point even earlier in the air mass history than seven days or, alternatively, convective activity could transport aerosols SD airmass trajectory, many of those air masses were near the surface seven days previously, so that desert dust lofted from the surface would be a major source of the aerosols that arrived at PDE.
The amount of precipitation that was derived with the HYSPLIT model is shown for the four air mass types 255 in Fig. 4c, represented here as total accumulated rainfall along the track. The trends in accumulated precipitation reveal a number of interesting details about meteorological events along the trajectories: 1) the majority of AO, SD, and AS precipitation events occurred more than five days before arriving at PDE, 2) there was little precipitation along the SE air mass until 30 hours before arriving at PDE, 3) the AS air masses experienced the most precipitation, the majority of which occurred within a day after leaving the African continent and 4) the AO and SD air masses were generally 260 very dry, i.e. very little precipitation events transpired over the seven-day journey. The significance of these precipitation events on the subsequent impact of these air mass is discussed below.
One point to mention is that there were no volcanic eruptions or ash plumes reported for the Soufriere Hills

Cloud Analysis
The hypothesis that drives the analysis in this study is that the microphysical properties of the mountaintop 280 clouds in Puerto Rico are not strongly dependent on the air mass histories. In order to test this hypothesis, we have taken a case study approach whereby four periods were selected during which clouds were already present on the mountaintop when the air with different histories arrived. The cloud properties are evaluated as this new air arrives over a period lasting anywhere from 12 to 20 hours, the time periods when the back trajectory analysis has shown that the air is arriving from the same source.

285
An alternative, and perhaps preferable, analysis method would be to identify periods when clouds were formed under conditions related to the four air mass types and then evaluate the subsequent evolution; however, such conditions could not be unambiguously identified for all four types with the current data set. Hence, the analysis evaluates changes in the size distributions relative to the cloud properties that existed at the time when the new air 290 mass arrived. This is analogous to what is done during cloud seeding programs, where the material is introduced to an existing cloud in order to alter the subsequent evolution of cloud properties.    have documented that clouds can form at any time of the day and last for many hours or even days (Allan et al., 2008).
Note, all times reported are in UTC.

315
There are changes in the cloud properties after the new air is introduced in all four cases; however, the magnitude and trends differ. Clear differences are seen in the shapes of the size distributions over time following the introduction of air with new time histories. The AO air mass arrives at 1400 Local Standard Time (LST) as a cloud is forming, and the Nc increases rapidly during the first three to four hours, remains more or less constant over the next 10 hours, and then begins fluctuating by ±40 cm -3 over the last six hours that the air mass is coming from the AO. The 320 average number concentration and MVD during the 18 hours of AO were 80 cm -3 and 11 µm, respectively.
The SD air arrived at two o'clock in the morning, LST, to an already formed cloud and continued until ten o'clock in the evening. At the time this air arrived, the cloud Nc and MVD had average values of 80 cm -3 and 10 µm, respectively. The Nc decreased to 60 cm -3 as the new air arrived and stayed at that value for the next eight hours before 325 becoming highly variable during the remaining 12 hours. The MVD stayed constant for six hours after the air mass arrival but then began decreasing throughout the remainder of the air masses lifetime at PDE until reaching a minimum of 5 µm.
The cloud that was already formed at PDE displayed no changes in the microphysical properties when the 330 air mass with the AS history arrived at 2000 LST, and these properties remain almost constant over the next three hours with an average Nc of 60 cm -3 and MVD of 10 µm. At that time, the Nc decreased to 40 cm -3 , but the MVD remained constant. The cloud persisted another 12 hours with very little change in Nc, LWC or MVD at which time it dissipated while the AS air was still present for another two hours.

335
The SE air mass arrived at 1000 LST during a period of sparse cloud that remains very low in concentration with average Nc < 10 cm -3 for five hours, at which time the number concentration began to increase, reaching a maximum of 160 cm -3 at the end of the air mass's time over the PDE. The cloud then rapidly dissipated in the hours following the end of this air mass of SE history. Of note also is that over the six-hour period of increasing concentration, the average MVD also increased by a factor of three, from 5 to 15 µm. of these parameters -100% from the average because the initial cloud concentration that the air encountered was very low. The initial increase in the ΔMVD is not reflected in the ΔNc or ΔLWC because of the low concentrations. Once 390 the cloud begins developing, however, the ΔNc and ΔLWC continue increasing, along with the ΔMVD until the end of the SE residency.
To encapsulate in a few sentences the physics behind the trends that we see in the deviations of the cloud microphysical parameters: 1) when ΔNc increases, new droplets are activating, 2) when ΔLWC is increasing it can be 395 due to increases in ΔNc and ΔMVD (droplets growing in size by condensation or coalescence), 3) when ΔNc decreases, droplets are being removed by evaporation or precipitation and 4) when ΔLWC is decreasing, it may be a result of decreasing Nc and MVD.
In this context, the impact of the air masses with four different histories is:

400
• AO -Activation and growth of new droplets followed by eventual evaporation due to entrainment and mixing of subsaturated air; • SD -Initial entrainment of subsaturated air that led to removal of droplets by evaporation then new activation of CCN, either the residuals of the evaporated droplets or those residuals in addition to CCN introduced by the SD air mass, then further entrainment of subsaturated air that led to droplet evaporation and the dispersion 405 of the cloud; • AS -The impact of this air mass is largely to remove the existing droplet by evaporation with only a brief period when it appears that some new activation occurred; • SE -The arrival of the new air eventually resulted in the activation of droplets in much higher concentrations than seen in the clouds under the influence of the other air masses and these CCN are being continuously 410 activated throughout the period, suggesting that the supersaturation is continuing to increase and smaller sized CCN are being activated. At the same time, the droplets continue to grow, as evidenced by the increasing MVD throughout the residency of this air at PDE. activity. In addition, in both these cases, the clouds persisted well into the evening, even after these particular air masses were no longer in residence. Hence, the clouds that were already on the mountain during the SD and AS incursions were also most probably those that had formed earlier in the day by convective activities. So before finalizing our conclusions about how the clouds in the four cases were altered by the changing air mass histories, it is 435 important to evaluate if there were obvious differences in the winds or thermodynamic variables. The thermodynamic history of the air masses, i.e., the temperature and relative humidity (RH), tells a different story as is illustrated in Fig. 9 where the back-trajectory analysis summarizes how these two parameters evolved in

455
As expected, the temperature histories follow those of altitude that were drawn in Fig. 4b, i.e., the African air masses, SD and AS, had been at lower altitudes and warmer temperatures seven days prior to reaching while the AO and SE over the Atlantic were much higher and colder. All of the trajectories converge on an average temperature of 18 o ± 2 o , 24 hours prior to arrival at PDE so that there are no differences that might impact the subsequent 460 temperature of the cloud when mixed with this air.
The RH histories of the air masses are distinctly different from one another (Fig. 9b) and reflect to some degree both their temperature and altitude trends, in particular the air mass with the SE history that was within 1000 m of the ocean surface throughout most of its seven-day travel. The proximity to this water vapor source kept its RH above 80% from three days back until reaching PDE. The decrease in RH at 20 hours is a result of its ascent as it 465 approached the island.
The RH trends of the AO, SD and AS air masses are more puzzling and would require an analysis beyond the scope of the current study to explain the rapid increase in the RH of these air mass beginning at 30 hours that leads to RH values above 90% when these parcels reach the measurement site. If the RH had been much lower than 90% when arriving at PDE, mixing it into the existing cloud would have led to more rapid evaporation of the droplets and 470 dissipation of the cloud. However, as the time series of the size distributions shows, the dissipation happened hours after the arrival of the air masses, so mixing will decrease the concentration, not due to removal of droplets by evaporation, but just by mixing in droplet free air.
It is important to acknowledge that we have imposed conditions on the analysis that might have biased the results towards the higher RH air at the time of arrival or one hour before; however, as the time traces show in Fig.   475 9b, there was already a strong, positive trend of humidification many hours prior to the air mass arrivals.
To conclude the discussion, we return to the analysis of the accumulation of precipitation along the air mass trajectories that was first introduced in Fig. 4c. The rainfall that is reported from the HYSPLIT model is an estimate based on the humidity fields and atmospheric cooling derived from the meteorological fields that are used when the model is run. Although these are not observed values, previously published comparisons between in situ and satellite 480 measurements have shown that it is a reasonable approximation that can be used to evaluate precipitation along the trajectories. In the present study, we are interested in how precipitation affects the population of CCN in the air mass that will eventually reach Puerto Rico. The model predicts that the majority of rainfall is happening in the SD and AS air masses more than five days prior to their arrival at PDE. This suggests that some fraction of the dust and biomass burning aerosols that might have been picked up over Africa will be removed by precipitation, well before they will 485 have a chance to influence the mountaintop clouds in Puerto Rico. On the other hand, if this precipitation is a result of deep convection, while rainfall might remove some CCN, the vertical air motions that lead to the formation of clouds and precipitations will also transport CCN that are not washed out. This latter mechanism of aerosol transport is particularly important for those air masses with AO and AS histories whose altitudes from seven days forward never brought them near the surface as was seen with the SD and SE air masses.
from the boundary layer that subsequently led to higher concentrations of droplets and that the primary influence of 505 the SD and AS air is to dilute and eventually dissipate the clouds with no new droplets activated. .

Summary and Conclusions
In an effort to expand the current database of measurements related to the interaction of aerosol particles transported from the African continent to the Caribbean, and their subsequent impact on cloud formation and 510 evolution, we conducted measurement campaigns on the Caribbean island of Puerto Rico during the summers of 2013, 2014, and 2015. On frequent occasions, the island was clearly inundated with air masses from the Saharan desert and African Sahel. Measurements were made at a coastal location of the aerosol optical properties and during these same periods, cloud microphysical properties were measured at a mountain site that was downwind from the coastal site.
The wind velocity, temperature, RH, and rain rate were measured at both sites. Based on the averaged wind directions,

515
neither of these sites were influenced by anthropogenic emissions from local sources.
The aerosol optical properties and particle chemical composition measured at the coastal site were used to identify sixteen high dust and nine low dust events of African dust. During high dust periods, the scattering and absorption coefficients, AAE, and nss-Ca 2+ ion concentrations were higher, and the SAE were lower, than for low 520 dust periods.
The HYSPLIT back trajectory model was used to identify air masses with four distinctly different characteristics based on their geographic locations prior to arriving at the Pico de Este (PDE) measurement site: midlatitude Atlantic Ocean (AO), Sahara Desert (SD), African Sahel (AS) and arrival from southeast of Puerto Rico (SE).

525
Time periods were selected for analysis during which these air masses were arriving continuously over durations ranging from 12-20 hours when there was a cloud present on the mountain peak. The cloud droplet size distributions were measured with a droplet spectrometer over the size range from 5-90 µm, from which the microphysical properties, total number concentration, Nc, liquid water content, LWC, and median volume diameters, MVD, were derived as a function of time.

530
Based on the atmospheric dynamics and thermodynamics, and the trends in the microphysical properties, we arrive at the following conclusions regarding the impact of these air masses with four different histories on cloud evolution: • The AO air masses influence the reactivation of existing CCN as a result of the mixing of high humidity air 535 with the resident cloud and not due to the introduction of new CCN.
• The initial entrainment of SD air parcels leads to removal of droplets by evaporation then new activation of CCN, either the residuals of the evaporated droplets or those residuals in addition to CCN introduced by the SD air mass, then further entrainment of subsaturated air that leads to droplet evaporation and the dispersion of the cloud.
• The impact of air masses with AS history is largely to remove the existing droplets by evaporation with only a brief period when it appears that some new droplet activation occurred, most likely on CCN that were residuals of evaporated droplets.
• The arrival of the new air along the SE trajectories eventually resulted in the activation of droplets in much higher concentrations than seen in the clouds under the influence of the other air masses and these CCN are 545 being continuously activated throughout the period, suggesting that the supersaturation is continuing to increase and smaller sized CCN are being activated. At the same time, the droplets continue to grow, as evidenced by the increasing MVD throughout the residency of this air at PDE.
These conclusions do not support those by Spiegel et al. (2014) who concluded that microphysical properties of