Corresponding author: A. J. Prenni, Department of Atmospheric Science, Colorado State University, Campus Delivery 1371, Fort Collins, Colorado, 80523, USA. E-mail: (email@example.com)
 It has long been known that precipitation can impact atmospheric aerosol, altering number concentrations and size-dependent composition. Such effects result from competing mechanisms: precipitation can remove particles through wet deposition, or precipitation can lead to the emission of particles through mechanical ejection, biological processes, or re-suspension from associated wind gusts. These particles can feed back into the hydrologic cycle by serving as cloud nuclei. In this study, we investigated how precipitation at a forested site impacted the concentration and composition of ice nuclei (IN). We show that ground level IN concentrations were enhanced during rain events, with concentrations increasing by up to a factor of 40 during rain. We also show that a fraction of these IN were biological, with some of the IN identified using DNA sequencing. As these particles get entrained into the outflow of the storm, they may ultimately reach cloud levels, impacting precipitation of subsequent storms.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Ice nuclei (IN) are those partiles that catalyze ice nucleation in the atmosphere. Although typically present in very low concentrations, IN are presumed to be vital to the initiation of ice in mixed-phase clouds, a critical first step in precipitation formation. Here, we aim to better understand the influence of precipitation on IN concentrations. Previous studies have shown that IN concentrations may be enhanced during precipitation, particularly during convective storms. These studies, which include surface measurements in cold thunderstorm outflows in Colorado [Langer et al., 1979], show typical IN enhancements of a factor of 10 [Isaac and Douglas, 1973; Isono and Tanaka, 1966]. Measurements made aloft during convective activity also have shown enhancements in IN, although to a lesser extent [Wisniewski and Langer, 1980].
 In the studies noted above, the authors suggested a variety of potential sources for the IN, including enhanced concentrations of soil particles generated from increased winds, IN residues from evaporated hydrometeors, and transport of IN from aloft. A potential source not mentioned is biological particles, which can be released during precipitation [Huffman et al., 2012b; Jones and Harrison, 2004; Lindemann and Upper, 1985] and during periods of high humidity [Huffman et al., 2012a; Toprak and Schnaiter, 2012] through mechanisms such as spore release from fungi [Elbert et al., 2007], mechanical ejection of bacteria and spores from leaf surfaces [Constantinidou et al., 1990; Jones and Harrison, 2004], and pollen release and fragmentation during wet weather [Miguel et al., 2006; Pummer et al., 2012; Taylor et al., 2007]. Increased wind speeds associated with storms can also lead to the release of biological particles from soils and plant surfaces [Jones and Harrison, 2004]. The release of these biological particles likely impacts the IN population; indeed, Constantinidou et al.  have observed increases of airborne ice nucleation active (INA) bacteria near a soybean field following rain, although these authors inferred that the bacteria came from a source outside of their sampling location.
 Biological IN, which typically represent a small fraction of all primary biological particles, include ice nucleation active bacteria [Vali et al., 1976], fungi [Pouleur et al., 1992], pollen [Diehl et al., 2002], leaf litter [Schnell and Vali, 1976], and IN derived from decomposition of these particles [Fall and Wolber, 1995; Pummer et al., 2012]. The ubiquity of biological IN in fresh snow from mid- to high-latitude locations has been reported [Christner et al., 2008], but there is uncertainty regarding their number concentrations in the atmosphere [DeMott and Prenni, 2010], and their importance is not well-established on a global scale [Hoose et al., 2010]. Little work has been done to investigate how precipitation might impact the atmospheric abundance of biological IN. If the net result of precipitation is to replenish or increase atmospheric IN concentrations, this relationship implies an important positive feedback in the hydrologic cycle and a potential link between the biosphere, hydrosphere, and atmosphere. In this study, we report changes in IN concentrations during multiple rain events, observed at high time resolution during BEACHON-RoMBAS (Bio-hydro-atmosphere interactions of Energy, Aerosols, Carbon, H2O, Organics and Nitrogen—Rocky Mountain Biogenic Aerosol Study). We show that changes in IN are correlated with changes in airborne fluorescent particle concentrations during rain and present evidence that these IN have a biological component.
 Measurements were conducted in July and August 2011 in Manitou Experimental Forest in Colorado (latitude 39.10°N, longitude 105.10°W). This time period falls near the peak of the North American monsoon in Colorado, when moisture is transported to this region and convective storms are prevalent. The Experimental Forest is representative of the Central Rocky Mountains Montane zone and contains open canopy ponderosa pine, large areas of Douglas fir, aspen and open grassy areas [Levin et al., 2012]. Upwind ecosystems may also include these species as well as oak and spruce forests, and riparian willow zones [Kim et al., 2010].
 A ground-based version of the Colorado State University continuous flow diffusion chamber (CFDC) [Rogers et al., 2001] was employed for real-time measurements of IN concentrations. The CFDC permits observation of ice formation on a continuous stream of particles at controlled temperatures and humidities. For this study, measurements were made above water saturation (103–106% RH with respect to water) over a temperature range of -15 to −32.5°C. At these conditions, contributions are expected to the IN population from both dust and biological particles [Prenni et al., 2009]. Particles larger than 2.4 µm were removed from the sample flow prior to entering the CFDC using an impactor, and during certain sampling periods, particle concentrations were enhanced upstream using an MSP Corporation (Model 4240) aerosol concentrator. Measurements made using the concentrator were corrected to ambient concentrations based on the manufacturer's specifications for 1 µm particles, adjusted slightly for the sampling conditions at Manitou, as determined from direct measurements made approximately every other day. This resulted in enhanced aerosol number concentrations by a factor of 103, on average. Although the concentration factor is size dependent, the exact size of IN is not determined using the CFDC, and so the average enhancement factor was used for all results. IN number concentrations are reported at standard temperature and pressure (STP).
 Ice crystals activated as IN in the CFDC were collected via impaction at the CFDC outlet [Garcia et al., 2012; Prenni et al., 2009]. For electron microscopy analysis, ice crystals were collected on a TEM grid, and the residual IN were characterized using scanning electron microscopy (JEOL, JSM-6500F). For biological analysis, ice crystals were collected on a glass slide, which was coated with 5 μL of molecular grade mineral oil (Bio-Rad). DNA from the residual IN was then enzymatically extracted using Proteinase K. The extracted DNA was PCR amplified using the universal 515F and 1391R primers. The presence of biological IN was determined after PCR amplification via acrylamide gel electrophoresis. Staining levels indicate relative amounts of DNA in the original samples. Molecular grade mineral oil (Bio-Rad) and tissue grade deionized water (Sigma) were run as negative controls. PCR products were cloned into a plasmid vector using the TOPO TA Cloning Kit® for sequencing (Invitrogen). Each clone was sequenced (Sanger method) and identified via Blast search against the National Center for Biotechnology Information (NCBI) genome database [Garcia et al., 2012].
 Our ability to identify biological IN relies on removal of particles >2.4 µm upstream of the CFDC. While our method removes >95% of these large particles, any large particles that are not removed by the upstream impactors, including those that do not nucleate ice, may get collected. Although this potential artifact may impact our sequencing results, we note here that for 14 of the 17 biological IN collections during BEACHON, levels of DNA from the IN impactor were too low to be detected, and so influence of large particles escaping impaction is considered negligible. Only samples with enhanced IN concentrations had DNA in sufficient quantities for characterization.
 Size-resolved number concentrations of total (N) and fluorescent (FP) particles were determined using an Ultraviolet Aerodynamic Particle Sizer (UV-APS, TSI 3314; 0.5–20 µm) [Hairston et al., 1997]. A certain fraction of biological particles are detected as fluorescent in the UV-APS, and thus, FP concentrations provide a lower limit estimate of biological aerosol particles [Huffman et al., 2010]. Mineral dust particles can also fluoresce under certain conditions, but both field and laboratory data suggest that particles observed by the UV-APS are dominated by biological material in most cases [Pöhlker et al., 2012; Pöschl et al., 2010]. UV-APS data presented here are limited to particles <2.5 µm for consistency with the CFDC measurement. FP and total UV-APS particle concentrations are also reported at STP.
3 Results and Discussion
 IN data were collected during 11 rain events. Figure 1 shows an example of one event from 2 August, for IN measurements at −25 °C. At 11:15, the first sign of an approaching storm was observed, with surface winds increasing to ~2 m s−1 and winds at the top of the instrument tower (28 m) reaching 6 m s-1. By 11:30, there was little change in IN concentrations, suggesting that the subsequent change in IN was not solely wind driven. At 11:35, the first rain was observed, accompanied by a decrease in temperature and an increase in relative humidity. Rain continued to fall through 12:00, with total accumulated precipitation exceeding 13.5 mm. For the hour prior to the arrival of the storm, IN concentrations were quite low, less than 2 L−1. These values increased dramatically upon the arrival of the rain, reaching almost 200 L-1 at their peak. IN concentrations began to decrease immediately following the rain, but an hour after the storm, IN concentrations were still an order of magnitude greater than during the pre-storm period. From 13:30 until 15:00, the CFDC was taken offline. Measurements resumed at 15:00 during another event, and again IN concentrations were elevated, although to a lesser extent. A third rain event beginning at 16:30 yielded another increase in IN concentration.
 FP concentrations are also shown in Figure 1 for this same time period. The FP response to the rain events is nearly identical in shape and magnitude to that of the IN, and FP and IN concentrations are highly correlated throughout the day (correlation coefficient, R = 0.94), suggesting that the FP are serving as IN during the rain events. FP concentrations exceeded those of the IN by about a factor of four during the first rain event, so that if we assume that the bulk of the IN population was comprised of biological FP, ~25% of the biological FP served as IN at −25 °C. However, the UV-APS can greatly underestimate the total concentration of biological particles [Huffman et al., 2012b], and non-fluorescent particles such as dust may have contributed to the increased IN, so 25% serves as an upper limit for the fraction of total biological particles that may have served as IN. Indeed IN were also highly correlated with the total particle concentration in the size range of 0.5–2.5 µm (R = 0.92) on this day, although this relationship likely resulted, in part, from non-fluorescent biological particles [Huffman et al., 2012a; Huffman et al., 2012b]. The strong correspondence between the number concentrations of IN and of total particles in this size range has been noted previously [DeMott et al., 2010].
 IN measurements were made during 10 other rain events. Results are summarized in Table 1. IN concentrations are given for the time immediately preceding (up to 1 h), during (up to 1 h and 40 min), and after (up to 1 h) the rain events. Times given in the table correspond to the extent of the IN data set during these time periods. Correlation coefficients also are given relating IN to FP concentrations (0.5–2.5 µm) and to total particle concentrations (0.5–2.5 µm). Correlations are given for the entire measurement period, including data more than 1 h before and after the rain. As shown in the table, there was a clear increase in IN concentrations during and after all rain events for CFDC measurements at −25 °C and warmer, even for rain events with as little as 0.01 mm precipitation. These increases ranged from about a factor of two to a factor of forty. For all but one of the measurement periods at -25°C and warmer, IN showed good correlation with FP (R = 0.68–0.95, 10 August being the exception) and total particle concentrations (R = 0.63–0.96, 22 August being the exception). IN made up a significant fraction of the FP number throughout this temperature range, although a strong temperature dependence was observed. For example, for the rain event on 10 August, when the CFDC was operating at −15°C, IN made up only 2% of the total FP number concentration; this number increased approximately linearly with decreasing temperature, reaching 27% at −25 °C. These values represent upper limits for the fraction of biological particles that can serve as IN during the observed rain events for the size range noted.
Table 1. Summary of IN data during rain events. Concentrations are given for measurements up to 1 h before, during, and up to 1 h after the event. Measurement uncertainty is listed as 2 standard deviations of variability around the mean value. The associated times indicate the extent of the data collected during each of these time periods. NA values indicate that no IN were collected during the given time period. Correlation coefficients are given for the entire measurement period
 For CFDC measurements colder than −25 °C, we did not observe the same trends (Table 1). IN concentrations during/after rain were typically equal to or less than their pre-rain values for measurements at −27.5, −30, and −32.5°C, and IN were only weakly correlated (R = 0.06–0.45) to FP for three of these measurement periods. This is likely due to washout of particles, including IN-active particles, during rain events; for example, for measurements on July 30, both total particle and FP concentrations (0.5–2.5 µm) decreased during rain. Likewise, IN active particles at -32.5°C decreased during and after rain on 28 July, consistent with wet removal of IN. At colder temperatures, the measured IN reflect the cumulative numbers of particles that can activate at the measurement temperature or warmer. As has been shown previously [Prenni et al., 2009], the fractional contribution of mineral dust particles to total IN increases at colder temperatures, whereas at temperatures warmer than −25°C, mineral particles represent a smaller fraction of total IN. Thus, while washout of preexisting IN during rain dominates the signal in the colder temperature cases, at warmer temperatures, and especially in the absence of washout, the production of new IN according to the mechanisms described above is readily observed.
 The correlation between IN and FP at warmer temperatures (>−25°C) during rain suggests biological particles as a potential source of IN. To explore this further, we performed DNA sequencing on a subset of the residual IN samples. The samples collected for DNA sequencing correspond to precipitation events for 2 August only. Of all the samples collected during the study (N = 17), the samples from 2 August exhibited the most intense staining after PCR amplification, indicating the highest concentrations of biological IN. Two samples were collected: one which encompassed the first event shown in Figure 1 and a second which encompassed the two later events. Here we focus on results from the first event, when IN concentrations showed the most dramatic increase of the entire study. DNA sequencing results suggest that a diverse population of micro-organisms contributed to the IN population, with DNA from 11 bacteria and 11 fungi identified. Of the 11 identified bacteria, four have been shown to exhibit ice nucleation activity, including Bacillus [Nejad and Ramstedt, 2006], Curtobacterium [Nejad et al., 2004], Pedobacter [Nejad and Ramstedt, 2006], and Pseudomonas brenneri [Nejad and Ramstedt, 2006; Nejad et al., 2006]. Further, fungal spores of Cladosporium, which were also observed, have been shown to exhibit ice nucleation activity at colder temperatures [Iannone et al., 2011], although with low efficiency. These data are consistent with DNA sequencing from the total aerosol population at BEACHON-RoMBAS, for which potential ice nucleating bacteria and fungi were identified more frequently during rain events [Huffman et al., 2012b]. In addition to DNA analysis, we collected and analyzed IN using scanning electron microscopy (SEM). Figure 2 shows an SEM image of an IN collected after the rain event on 5 August (−20 °C). Such particles were observed both before and after the rain event but were enhanced in number after the rain. The size and morphology of the particle are suggestive of a fungal spore and are consistent with both the DNA analysis indicating biological IN and observations of IN-active fungi during the BEACHON-RoMBAS campaign [Huffman et al., 2012b].
4 Summary and Implications
 Real-time IN measurements during BEACHON-RoMBAS suggest that IN concentrations are enhanced after rain events by about an order of magnitude, particularly IN which are active from −15 to −25°C. Although the exact generation mechanism remains unknown, correlation between IN and FP during these events, coupled with compositional analysis of some of the residual IN, suggests that biological particles make up a significant portion of the rain-generated IN. These particles may have been transported to the site in the storm downdraft or may have a local source. In any case, if surface IN are entrained in the outflow of the storm and are transported to higher levels in the atmosphere, they may influence subsequent precipitation events. Such a feedback between biogenic cloud active particles and precipitation is consistent with previous observations from the Amazon Basin [Pöschl et al., 2010] and suggests a strong link between the biosphere, hydrosphere, and atmosphere.
 This work is funded by NSF (ATM-0919042, AGS-1036028, and ATM-0841602). The authors wish to thank Richard Oakes, the USFS, Doug Day, Jose Jimenez, Jim Smith, and NCAR for providing logistical support and access to the Manitou Experimental Forest field site. The authors also wish to acknowledge Andrew Turnipseed for providing the meteorological data. Y.T. acknowledges the JSPS Postdoctoral Fellowships for Research Abroad. J.A.H., C.P. and U.P. acknowledge support from the Max Planck Society (MPG), the Max Planck Graduate Center with the Johannes Gutenberg University Mainz (MPGC), and the Geocycles Cluster Mainz (LEC Rheinland-Pfalz). J.A.H. acknowledges internal faculty funding from the University of Denver. The authors gratefully acknowledge support by B. Schmer and M.O. Andreae.