Factors Influencing Snow Crystal Riming at Storm Peak, Steamboat Springs, Colorado

Laura Betker

Department of Atmospheric and Oceanic Sciences, Undergraduate, University of Wisconsin – Madison

Abstract: Drawing on past research and research techniques, the author of this paper along with a co-researcher collected and analyzed rime ice and snow at Storm Peak Laboratory in Steamboat Springs, Colorado. The collection times are paired with air trajectories from the same period to examine the possible effects of anthropogenic emissions on cloud condensation nuclei concentrations, which are a large factor in determining riming levels on snow crystals. A model derived for Lohmann et al. (2004) shows the cloud condensation nuclei concentration is affected by anthropogenic emissions. This paper will evaluate research done by direct observation and not on a model run to support or disprove the Lohmann et al. model (2004). Research techniques are adapted from these papers and are fully described.

Introduction sizes are small enough that the collision- coalescence process is stopped, which Often in winter storm clouds, occurs when cloud droplets fall through liquid cloud droplets exist in addition to the cloud to collide and merge with other ice crystals. Rime is defined as “a white droplets combining to form larger or milky granular deposit of ice formed droplets. ( Meteorology Today , 2000). by the rapid freezing of supercooled The CCN concentrations are important water droplets as they come in contact for determining the size of the cloud with an object in below freezing air” droplets. This is shown in a model (Meteorology Today , 2000). The objects produced for Lohmann et al. (2004) and on which rime accumulates may be in also examined in a Berg et al. study the form of ice crystals suspended within (1990) that agreed with the later the cloud or large surface objects such as developed model findings. vegetation and buildings. The riming The importance of the study of effects and efficiencies are dependant on the riming process is presented by cloud condensation nuclei (CCN) of Hindman et al. (1983) as he presents the which the concentrations within the frequency of riming events is high and cloud determine the size of the cloud therefore important to the overall droplets and thus the ability for ice hydrological cycle and can lead to crystals to collect the rime within the differences in snow chemistry, density cloud through the Bergeron process, and fall amounts. The study also which describes how a crystal grows at discusses the efficiency and collection the expense of the liquid cloud droplets processes of riming events. The findings, surrounding it. Borys et al. (2003) states also supported by Hindman and Grant that within a polluted cloud, the riming (1981) show that rime ice deposition is essentially shut down when droplet occurred during most snowfall events in Betker - 1 - high elevations during a study period of As previously mentioned, the February to April 1981 at Vail and riming effects and efficiencies are Steamboat Springs, Colorado. These affected by pollution as CCN in the results as presented in Hindman et al cloud. Borys et al (2003) states that the (1983) show a rime ice occurrence of riming rate as well as the snowfall rate is 75% in high elevation winter storms at initially increased with the increase of Steamboat Springs as well as other pollutants. If the pollutant research locations not dicussed here. concentrations and therefore CCN Also presented in the Hindman et concentrations increase to a level at al (1983) study is the spatial situation of which the cloud droplet sizes become rime ice collection. It is observed that too small for the collision-coalescence on average the uppermost 400 meters of process to work properly (diameter equal Storm Peak are enveloped in a cloud. to or less than 10 µm) the riming process As winds move the cloud and the is shut off. The pure snowfall rate then droplets suspended within it over an area is eventually increased due to the rime ice is deposited onto all collection accretion of snow crystals with drizzle- surfaces present at the time. Through a size drops within the polluted cloud. series of calculations involving the rate With this idea in mind that an of liquid water passage across Storm increase in CCN concentrations directly Peak, temperature, wind speed and cloud affects the riming process, Lohmann et droplet radius, it is determined by al. (2004) studies the effects on Hindman et al. (1983) that rime is increased CCN concentration due to deposited on the order of 10 4 g/hr. anthropogenic emissions. A model Also through observation it is known as the ECHAM4 GCM was documented that the riming rates are up created for the study to show the riming to seven times greater than the effects of pollution introduced into the precipitation rates (Hindman et al., atmosphere. When sulfate, black carbon 1983). In comparing precipitation and and organic carbon emission rates were riming rates as well as the frequency of set to zero in the model run, only natural rime ice events it is concluded by emissions from forests were left to Hindman et al. (1983) that the riming influence the riming rates of a cloud. process contributes as much as 60% of The model concluded a decrease in the annual snowpack water content, riming rates with the decrease in CCN showing that the study of rime ice events diameter was due to increased is important to furthering understanding anthropogenic aerosols from pollution. the hydrological cycle as well as storm It should be noted that in this study all processes. As water is an important observations were taken from a model issue in the western United States and runs and natural occurrences of riming very regulated by the government, the may not be accurately portrayed and study of rime can contribute to the should be further investigated in the rationing of water and prediction of natural environment. The study liquid equivalent precipitation amounts performed by Betker and Eagan at Storm available for public use. Rime ice on Peak Laboratory will attempt to create a high elevation trees is said to account for less model-dependant database for the 10% while 50% is attributed to the comparison of pollution levels to riming riming of ice crystals within the cloud. levels through CCN concentrations.

Betker - 2 - Model air trajectory runs using the sample size differences are inescapable Hybrid Single-Particle Lagrangian due to the differing chemistry of Integrated Trajectory (HYSPLIT) were individual rime events at each location ran for the purpose of this paper’s (Berg et al., 1998). The repetition of research to estimate pollution levels as sampling sought to reduce the errors that high or low depending on origin and will may have occurred. be discussed later. Many factors can affect the Prior to the development of the analysis of the influence of rime, as ECHAM4 GCM, a Berg et al. study presented in Berg et al, (1990). The (1990) took observations of rime rime collection efficiency increases as deposits also to determine the effects of the size of supercooled droplets and pollution on rime. Berg et al. (1990) wind speed increase. The shapes of the looked at ion concentrations within the collector, such as the varying shapes of rime deposits to determine the effects of ice crystals also affect the efficiency. pollution. Betker and Eagan will also The mere fact that riming occurs on look at sample conductivity as well as snow crystals may somewhat alter the pH levels for similar comparisons and chemistry of the snow by mixing with attempt to connect this data with that of the chemistry of the rime in some natural the HYSPLIT and other observations. snow events. The chemistry of the rime The cloud acidity and chemistry of rime may also be altered due to sublimation ice and snow can and will be addressed that may have occurred between the time by Betker and Eagan to analyze the of deposit and the time of collection. extent to which the pollution has These factors that have the potential to affected the cloud processes. Berg et al. alter the chemical and physical analysis (1988) and Berg et al. (1991) studies of rime ice as well as snow were mention that pollutant concentrations are presented by the authors Berg et al in significantly higher in rime deposits than their 1990 study. in snow. Higher conductivity and lower Each of the previous studies pH indicate more riming on the snow mentioned investigates different aspects crystals. The melted rime samples at of riming. The topics of pollution, each of five studied sites were found frequency of events, rate of riming and have larger ion concentrations than the chemistry were mainly developed melted snow samples at each site. For throughout each of the papers. Betker winter storms the ion concentration and Eagan continue the study of rime in differences were found to be the Rocky Mountains at Storm Peak significantly greater than 0. This Laboratory in Steamboat Springs, difference of conductivity and pH of Colorado, attempting to integrate all rimed and unrimed snow shows that high previous research into a single concise pollution levels act to decrease riming study. It has been determined that the due to decreasing the size of the cloud effects of pollution due to anthropogenic droplets within a polluted cloud. It is sources impact the effects, efficiency noted that small sample sizes could alter and chemistry of riming due to the the impact of this data. It may be alteration of CCN concentrations and the possible in that larger samples could introduction and alteration of ions into produce less dramatic results but also the atmosphere. The study of rime was that the differences that may occur with shown to be valuable as its effects on the

Betker - 3 - environment with respect to annual bags attached with rubber bands to de- snowpack water content and pollution iced wind-vane-equipped cylinders rates. All data and observations taken by (shown in figure 1b). The wind vane Betker and Eagan are supportive of the allowed for the plastic bag openings to previous data mentioned and will be be continually directed into the wind for further discussed throughout the maximum collection volumes. Later following paper. samples were collected in a secondary collection device of plastic bags rubber- Methodology banded to a cardboard tube and the tube then was anchored to the roof hatch All data was collected and opening and faced into the wind (figure processed over a weeklong period from 1c) similar to the prior collection device. 11-18 March 2006 at the Storm Peak Eagan is shown in figure 1d attaching a Laboratory in Steamboat Springs, plastic bag to the secondary collection Colorado. Data was taken in device. Neoprene gloves were worn at collaboration with Bridgette Eagan, also all collection times so as not to an Undergraduate in the department of contaminate the collected rime or snow. Atmospheric and Oceanic Sciences at The cloud sieve and snow collection the University of Wisconsin – Madison. bags were mounted simultaneously and Samples of pure rime were collected were allowed to collect samples for a using a Cloud Sieve during snow events. period of approximately 18 minutes. The cloud sieve was mounted on the Each time the sieve and snow collection roof of the Storm Peak Laboratory bags were mounted and retrieved, three perpendicular to the mean wind wind measurements were taken and direction, determined by the main wind averaged using a hand-held wind vane shown in figure 1a. This allowed measurement gauge. By means of a for maximum collection of rime during DMT SPP-100 droplet-sizing probe, each event. During the same events and cloud condensation nuclei were also in close proximity to the Cloud Sieve, counted and sized continually snow was collected using clean plastic throughout the collection period.

Figure 1a: Main Wind vane determining Figure 1b: Snow collection wind vane. wind direction.

Betker - 4 -

Figure 1c: Secondary Snow collection Figure 1d: Eagan attaching plastic snow device. collection bag to secondary Snow collection device.

Samples were assessed during contamination. The plastic bags each sample collection time using a containing samples of rime and snow, black felt-covered block and a respectively, were weighed using a microscope to determine levels of riming sliding scale. The weight of a single on the snow crystals. Snow was allowed empty bag was noted to be 4.6 grams. to fall on the felt block and then was Samples were then transferred to dry, immediately placed under a microscope clean beakers, covered with plastic wrap to analyze the crystals. Riming to avoid evaporation and allowed to melt classification techniques followed the at room temperature. Conductivity and five-degree system proposed by Mitchell pH meters were next used to assess the and Lamb in 1989. The degree of conductivity and acidity of each sample. riming was “classified into five These meters were each calibrated categories according to the fraction of separately to insure accuracy. Acidity of the crystal surface covered by accreted the pure rime will be compared to the drops: ‘light’ (less than about 1/5), acidity of the snow to evaluate how ‘light-to-moderate’ (1/5 to 2/5), riming affects the acidity and ‘moderate’ (2/5 to 3/5), ‘moderate-to- conductivity of the snow later in this heavy’ (3/5 to 4/5), and ‘heavy’ (greater paper. than 4/5, including graupel)” (Mitchell Finally, melted samples were and Lamb, 1989). It must be noted that transferred to a small cylinder held in a there is no universal classification for test tube clamp attached to a ring stand, riming levels, and this is a qualitative rinsed with distilled water, to collect analysis. conductivity and pH measurements. The Samples of rime were scraped off conductivity probe was first rinsed with the cloud sieve using a plastic scraping distilled water to prevent sample card onto a clean plastic sheet. The contamination then inserted into the loose solid rime ice was then transferred melted sample and the meter reading to a clean plastic bag using a soft- was recorded. The probe was then re- bristled paint brush, here also neoprene rinsed with the distilled water for the gloves were used to avoid sample same reason. Next the rinsed pH probe

Betker - 5 - was inserted into the cylinder containing Results the melted sample and the reading was recorded. The pH probe as well as the A total of eight sample periods holding cylinder was then rinsed again were completed. Data from three of with distilled water and the process was these periods is taken as representative repeated with the next sample. This was and is analyzed to create results for this repeated until all samples of melted rime paper. Quantitative values of this data ice and snow were analyzed are presented in table 1. The data covers For increased accuracy in all high and low riming levels and thus a collection both researchers were variety of values for all other variables. involved in the collection process. This As the riming level is shown to included collecting samples as well as increase from level one to level five, the properly storing and maintaining them. number of CCN decreases from 306.9 Betker specifically handled conductivity #/cc to 23.0 #/cc while the cloud droplet tests while Eagan conducted pH tests. diameter increases from 21.6 µm to 10.0 Both researchers were responsible for µm. It is also noted that as more riming snow crystal analysis to determine occurs on the snow crystals the mass of riming degrees. Dr. Randolph D. Borys snow collected increases from 4.55 aided the researchers in the education of grams at the lowest riming level to 12.89 operating equipment and general grams at the highest riming level. This processes. is intuitive as more riming increases the Also, for the times of each volume of each snow crystal, leading to analyzed snow event and the 120 hours a larger mass. Conductivity prior, air trajectories were created using measurements also increase with riming the HYSPLIT model to assess and levels while snow pH measurements analyze the effects of anthropogenic decrease indicating that more ions are forcings on the CCN and thus the riming present in more rimed snow leading to a levels. more acidic chemical makeup of heavily rimed snow compared to lightly rimed snow.

Riming level CCN conc. Number Conductivity Snow pH Collected (#/cc) mass (amps/volt) Snow Mass (grams) diameter (µm) 1 306.9 10.0 4.90 7.40 4.55 3 47.7 15.6 7.40 5.12 6.25 5 23.0 21.6 9.65 5.05 12.89 Table 1

The HYSPLIT data, later shown shows that air passing over southern in figure 4, is from a model run taken for California as well as northern Mexico. the period where the riming level was This indicates large levels of one. The trajectory, run for the 120 anthropogenic emissions present in the hours prior to the level one riming event, air leading to large CCN concentrations

Betker - 6 - and small droplet diameters. This The observations and data outcome is consistent with the data records showing that as conductivity manually taken for the level one riming increased, pH decreased and riming on period at the Storm Peak Laboratory. snow crystals increased (figure 3) HYSPLIT air trajectories run for the supports the conclusions of Berg et al. other two riming levels produced results (1988) and Berg et al. (1991), presented that also supported the data taken above. The snow pH (blue line) manually at the Storm Peak Laboratory. decreases and the conductivity (green line) increases as riming level increases Discussion as outlined in the data in table 1.

25

20 15

10 5 Numerical value Numerical 0 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

Riming Level

Droplet diameter Conductivity Snow pH Snow mass

Figure 3: Data taken at Storm Peak Laboratory during the week of 11-18 March, 2006.

This increased riming also led to As the cloud droplet diameter increases increased snow mass (figure 3, purple from 10.0 µm to 21.6 µm the riming level line). This is reasoned by noting that an increases from one to five. This is also increase in the riming from level one to in concurrence with the previously five adds bulk to the snow crystal itself mentioned research of Borys et al. and therefore increased that mass from (2003). The cloud droplet diameter was 4.55 grams to 12.89 grams by increasing ultimately affected by the CCN the number of cloud droplets riming onto concentration within the cloud. As CCN a single crystal. The sum of all cloud concentrations decreased (figure 4) droplets combined onto a given crystal is cloud droplet diameters increased (figure positively correlated with the mass of the 3) as shown when figure 3 is compared rimed crystal due to added volume. with figure 4. Logically, as more CCN The increased riming levels can are present there is less liquid water be explained by an examination of the available for each aerosol, thus each cloud droplet diameter. Also, shown in droplet is smaller than if there were less the red line of figure 3, is this positive CCN and more liquid water for each correlation of increased cloud droplet aerosol. diameter with increased riming levels.

Betker - 7 - CCN Concentrations vs. Riming Level

350 300 250 200 150 100 50 CCN Concentrations (#/ Concentrations CCN 0 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

Riming Level

Figure 3: Data taken at Storm Peak Laboratory during the week of 11-18 March, 2006.

The changes in the CCN also noted that during this trajectory concentrations are explained by an period there was no precipitation examination of the pollution levels. This recorded although for a portion of the is achieved by using the HYSPLIT period high relative humidity values model showing air trajectories. The were recorded. Because there was no trajectory in figure 4 shows the air precipitation there was no opportunity to present during the level one riming rid the air of the aerosols from the period previously passed over southern anthropogenic emissions collected in California and then northern Mexico southern California and northern before reaching Storm Peak. Southern Mexico. This is supportive of the California is a highly populated area findings from Lohmann et al. (2004), leading to high levels of pollution in that which states that increased area in general. Northern Mexico is also anthropogenic emissions leads to highly polluted because there is an increased CCN concentrations and extreme lack of regulation of emissions. decreased riming levels. Combined with This trajectory explains the increased the Borys et al. (2003) study, this is number of CCN (306.9 #/cc) during this explained by a shut down of the level one riming period compared to a collision-coalescence process with lower CCN concentration of 23.0 #/cc decreased droplet diameters. during a level five riming period. It is

Betker - 8 -

Figure 4: Air trajectory for 120 hours prior to the level one riming event using the HYSPLIT model.

Betker - 9 - Conclusion stem directly from changes in CCN concentrations with the increase in The data taken by Betker and anthropogenic emissions and need to be Eagan at Storm Peak Laboratory in addressed for the benefit and Steamboat Springs, Colorado over the understanding of the environment we week of 11-18 March, 2006 has aided in depend on. the support of previous research done on the topic of rime ice deposition and References and Acknowledgements named in the introduction section of this Ahrens, Donald C. (1991), Meteorology paper. The research of Betker and Today: An Introduction to Weather, Eagan combines model data with Climate, observations to provide conclusions and the Environment , West Publishing directly connecting previous research in Company. New York, NY. one study instead of using individual data from separate papers that may Berg, Neil, Paul Dunn, and Mark Fenn provide varying information. The data (1991), Spatial and Temporal Variability suggests a negative correlation between of Rime anthropogenic emissions and riming Ice and Snow Chemistry at Five Sites in levels. Possible complications may have California. Atmos. Environ. , 25(A), 5/6, occurred due to the differing techniques 915-926. of snow collection that was necessary, however it is assumed for the purpose of Borys, Randolph D., Douglas H. this research that the effects were Lowenthal, Stephen A. Cohn, and minimal. Further research is suggested William O.J. Brown (2003), on the effects of riming to the Mountaintop and Radar Measurements environment and precipitation rates. It is of Anthropogenic Aerosol Effects on also suggested that direct observations of Snow Growth and Snowfall Rate. wind and emission concentrations be Geophys. Res. Let. , 30, 10. gathered and composed in a way to create an air trajectory that is not derived Hindman, Edward E., Randolph D. from model data. As anthropogenic Borys, and Paul J. DeMott (1983), emissions continue to influence and Hydrometeorological Significance of change the amounts and types of Rime Ice Deposits in the Colorado precipitation around the world, research Rockies. Wat. Res. Bul. , 19, 4, 619-624. that discovers relationships between these factors is important for Hindman, Edward E., Mechel A. understanding and determining the Campbell, and Randolph D. Borys necessary actions to take to resolve (1994), A Ten-Winter Record of Cloud- newly developed problems as well and Droplet Physical and Chemical prepare for future changes in the Properties at a Mountaintop Site in precipitation cycle that may potentially Colorado. AMS , July 1994, 797-807. occur. Further research will also help forecasters more accurately predict Lohmann, U. (2004), Can precipitation amounts by analyzing Anthropogenic Aerosols Decrease the emission levels and the effects of riming Snowfall Rate? AMS , 15 October 2004, on snow crystals. These changes will 2457-2468.

Betker - 10 -

Mitchell, David L. and Dennis Lamb, (1989), Influence of Riming on the Chemical Composition of Snow in Winter Orographic Storms. Journal of Geophysical Research. 94(D12), 14831- 14840.

Rauber, Robert M., John E. Walsh, and Donna J. Charlevoix (2002), Severe and HazardousWeather , 17pp., Kendall/Hunt Publishing Company. Dubuque, IA.

Snider, Jefferson R., Derek C. Montague, and Gabor Vali (1992), Hydrogen Peroxide Retention in Rime Ice. J. of Geophys. Res. , 97(D7), 7569- 7578.

Snider, Jefferson R., and Jun Huang (1998), Factors Influencing the Retention of Hydrogen Peroxide and Molecular Oxygen in Rime Ice. J. of Geophys. Res. , 103(D1), 1405-1415.

Acknowledgements also to Bridgettet A. Eagan, Dr. Greg Tripoli and Dr. Randolph D. Borys for aid in research and data interpretation.

Betker - 11 -