Visibility during Blowing Snow Events over Arctic Sea Ice

Qiang Huang Centre for Earth Observation Science, University of Manitoba, Winnipeg, Manitoba, Canada

Search for other papers by Qiang Huang in
Current site
Google Scholar
PubMed
Close
,
John Hanesiak Centre for Earth Observation Science, University of Manitoba, Winnipeg, Manitoba, Canada

Search for other papers by John Hanesiak in
Current site
Google Scholar
PubMed
Close
,
Sergiy Savelyev Centre for Research in Earth and Space Science, York University, Toronto, Ontario, Canada

Search for other papers by Sergiy Savelyev in
Current site
Google Scholar
PubMed
Close
,
Tim Papakyriakou Centre for Earth Observation Science, University of Manitoba, Winnipeg, Manitoba, Canada

Search for other papers by Tim Papakyriakou in
Current site
Google Scholar
PubMed
Close
, and
Peter A. Taylor Centre for Research in Earth and Space Science, York University, Toronto, Ontario, Canada

Search for other papers by Peter A. Taylor in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

A field study on visibility during Arctic blowing snow events over sea ice in Franklin Bay, Northwest Territories, Canada, was carried out from mid-January to early April 2004 during the Canadian Arctic Shelf Exchange Study (CASES) 2003–04 expedition. Visibilities at two heights, wind and temperature profiles, plus blowing and drifting snow particle flux at several heights were monitored continually during the study period. Good relations between visibility and wind speed were found in individual events of ground blowing snow with coefficients of determination >0.9. Regression equations relating 1.5-m height visibility to 10-m wind speed can be used for predicting visibility with a mean relative error in the range of 19%–32%. Similar regression functions obtained from the data for observed visibility of less than 1 km could predict visibilities more accurately for more extreme visibility reductions and wind speeds (>9.5 m s−1) with mean relative error ranging from 15% to 26%. For the event of ground blowing snow, a simple power law relationship between wind speed and visibility is sufficient for operational purposes. A poorer relationship was observed in the event of blowing snow with concurrent precipitating snow. A theoretical visibility model developed by Pomeroy and Male fit well with observed visibilities if using a mean radius of 50 μm and an alpha value of 10. The predicted visibility had a mean relative error of 30.5% and root-mean-square error of 1.3 km. The observed visibility at 1.5 m had a strong relation with particle counter readings, with an R2 of 0.92, and was consistent among all events.

Corresponding author address: John Hanesiak, Centre for Earth Observation Science, University of Manitoba, Winnipeg, MB R3T 2N2, Canada. Email: john_hanesiak@umanitoba.ca

Abstract

A field study on visibility during Arctic blowing snow events over sea ice in Franklin Bay, Northwest Territories, Canada, was carried out from mid-January to early April 2004 during the Canadian Arctic Shelf Exchange Study (CASES) 2003–04 expedition. Visibilities at two heights, wind and temperature profiles, plus blowing and drifting snow particle flux at several heights were monitored continually during the study period. Good relations between visibility and wind speed were found in individual events of ground blowing snow with coefficients of determination >0.9. Regression equations relating 1.5-m height visibility to 10-m wind speed can be used for predicting visibility with a mean relative error in the range of 19%–32%. Similar regression functions obtained from the data for observed visibility of less than 1 km could predict visibilities more accurately for more extreme visibility reductions and wind speeds (>9.5 m s−1) with mean relative error ranging from 15% to 26%. For the event of ground blowing snow, a simple power law relationship between wind speed and visibility is sufficient for operational purposes. A poorer relationship was observed in the event of blowing snow with concurrent precipitating snow. A theoretical visibility model developed by Pomeroy and Male fit well with observed visibilities if using a mean radius of 50 μm and an alpha value of 10. The predicted visibility had a mean relative error of 30.5% and root-mean-square error of 1.3 km. The observed visibility at 1.5 m had a strong relation with particle counter readings, with an R2 of 0.92, and was consistent among all events.

Corresponding author address: John Hanesiak, Centre for Earth Observation Science, University of Manitoba, Winnipeg, MB R3T 2N2, Canada. Email: john_hanesiak@umanitoba.ca

1. Introduction

Blowing snow is a common weather phenomenon in the Canadian Arctic and prairies. It occurs when the wind is strong enough to raise the snow particles to sufficient heights above ground that horizontal visibility (meteorological optical range, MOR; also described as Vis thereafter) is reduced to 9.7 km (6 mi) or less (Atmospheric Environment Service 1977). It can be a hazard to public safety and transportation since the visibility can be significantly reduced in some blowing snow events (Baggaley and Hanesiak 2005; Tabler 1979; Schmidt 1979). To mitigate this hazard, it is necessary to accurately forecast the occurrence of blowing snow and its visibility (Baggaley and Hanesiak 2005).

According to Koschmieder’s relationship (Koschmieder 1924), the visibility is related to the extinction coefficient as
i1520-0434-23-4-741-e1
in which Vis is visibility (m) and μ is the extinction coefficient for visible light (m−1) (Middleton 1952). The constant 3.912 comes from −ln(Ce), where Ce is the threshold value of the luminal contrast below which an observer cannot discern an object against the background. Usually, Ce is between 0.01 and 0.03 and was assumed to be 0.02 according to Mellor (1966). Pomeroy and Male (1988) used standard theory of electromagnetic attenuation, and an assumed two-parameter gamma distribution for particle sizes, to calculate the extinction coefficient for blowing snow and obtained a visibility formula as
i1520-0434-23-4-741-e2
where
i1520-0434-23-4-741-e3
is a factor associated with the ratio of the total particle cross-sectional area to the total particle volume density for the gamma distribution assumed. Here, r (=αβ, where β is the second parameter in the gamma distribution) is the mean radius of snow particles (m), ρp is the density of snow particles (kg m−3), ρd is the drift density of blowing snow (kg m−3), α is the shape parameter of the particle size distribution, and Qext is the extinction efficiency and varied with particle size and incident wavelength. For the event of blowing snow, because the snow particles are much larger compared with visible light wavelengths, Qext can be approximated as a constant equal to 2 (Pomeroy and Male 1988). From Eq. (2), we can see that the visibility is determined by the mean particle size, the shape of the particle size distribution, and the blowing snow density. Pomeroy and Male (1988) conducted a sensitivity test for Eq. (2) and concluded that the alpha parameter and mean particle radius of the gamma distribution of particle radii are very important in visual range modeling. They showed that the extinction coefficient varies by 19% as the shape parameter α varies from 5 to 15, and a reduction in the mean particle radius from 90 to 40 μm results in an approximate doubling of the extinction coefficient for a constant blowing snow density. Note that shape parameters and the mean radius may be functions of height above the ground.

Although the above physically based model indicates that three factors have effects on visibility, several other studies have shown that visibility in blowing snow could be inferred from some easily measured parameters. Budd et al. (1966) confirmed Liljequist’s (1957) conclusion that visibility is inversely proportional to the blowing snow density assuming a constant particle size distribution, and developed an empirical model to relate visibility to the blowing snow density at 2-m height. Mellor (1966) established an empirical formula that related visibility to snow accumulation rate. Schmidt (1979) developed a system to measure the particle frequency, mean particle diameter, and wind speed, and related these variables to visibility.

Among the above empirical models, blowing snow density is a common and important factor for inferring visibility. Mean particle size and the shape of the particle size distribution were not taken into account in some of these formulas. Nevertheless, these kinds of empirical formula can still be useful for operational meteorology or simple parameterizations in various models.

Wind speed is an important factor in determining the blowing snow density and therefore visibility. Past studies have revealed that the transport rate of snow in the suspension layer of blowing snow has a power or exponential law profile with height (Takahashi 1985; Budd et al. 1966; Lister 1960; Kobayashi 1978). The threshold wind speed for transporting snow is a critical parameter for any attempts to describe blowing snow processes. It is determined by snow particle bonding, cohesion, and kinetic friction, which are highly related to snow surface temperature. It increases nonlinearly with ambient temperature above −25°C and inversely proportional below −25°C (Li and Pomeroy 1997). Few studies have shown the relationship between visibility and wind speed. Baggaley and Hanesiak (2005) developed an empirical scheme to discriminate conditions that produce significantly reduced visibility in blowing snow using wind speed, air temperature, and time since last snowfall as predictors. This technique is more practical than other visibility models and achieved a critical success index as high as 66%.

In our present research, visibility, snow particle drift density, and standard meteorological data over Arctic sea ice were collected during blowing snow events to investigate the relationship between visibility, snow particle counter readings, and wind speed. The purpose is to provide an empirical set of equations that relate wind speed to visibility over Arctic sea ice as a simplified method to predicting reduced visibilities.

2. Data and methods

a. Research area and instrumentation

Field data used in this study were collected during the Canadian Arctic Shelf Exchange Study (CASES) in Franklin Bay, Northwest Territories, Canada, from 28 January to 6 April 2004. The study site was located at 70.03°N, 126.30°W, on smooth first-year sea ice (fast ice) with at least 20 km of fetch in all directions (Fig. 1). Snow cover varied with time but was considered homogeneous in space; except for some rough ice several hundred meters to the NW. Wind profiles indicated a roughness length, z0, of approximately 0.001 m. Average snow depth was 7 cm at the beginning of the campaign and increased to 17 cm toward the end. Snow density in the first 8 cm of snow cover varied from 80 to 450 kg m−3 with a median value of 240 kg m−3 (Savelyev et al. 2006). Instrumentation relevant to this study included a 10-m tower outfitted with one R. M. Young wind monitor (model No. 05103) at 10-m height and two Gill cup anemometers (model No. 12102), mounted at 4 and 1 m. Another Gill cup anemometer was mounted at 2 m on the particle counter post. The suite of wind measurements provided wind profiles in the first 10 m above the ice surface.

Four particle counters were mounted on a standoff post with initial heights of 0.1, 0.2, 1.0, and 2.0 m above the ice surface, respectively. The particle counters are based on the design of Brown and Pomeroy (1989), partly redesigned and assembled at York University. Detection of particles is based on the drop in voltage generated by a photodiode illuminated by an infrared beam when a particle gets into its path. The distance that light travels in the sampled volume of air was 2 cm and was the same for all units. A cover on the photodiode has a 0.15-mm-diameter hole that allows part of the light beam to reach the active surface of the diode. Additional discussion of the particle counters is provided in Savelyev et al. (2006).

Two visibility sensors (manufactured by Sentry) were mounted on separate posts at heights of 1.5 and 3.0 m. The sensor emits a narrow beam of 880-nm-wavelength light, some of which is forward scattered into a north-facing, narrow admittance angle detector. The output depends on the amount of light forward scattered from any particle. The sensor output has a maximum meteorological optical range (MOR) of 16 km. Calibration was performed before installation. It is the output from these sensors that we will use as our measurement of visibility (Vis).

The snow depth was continually measured by an ultrasonic distance ranger (model No. SR50, Campbell Scientific, Edmonton, Alberta, Canada). These data were used to correct the heights of all measurements above the snow surface when required.

Wind sensors were interrogated every 2 s and the other sensors were interrogated every second. Individual measurements were averaged over 5-min intervals and recorded into datalogger memory (Campbell Scientific Inc. CR23X and CR10X).

b. Data description

Data were collected from 15 January to 6 April 2004 (Julian days 15–96), while the visibility data recorded by two visibility sensors (at 1.5- and 3.0-m heights, respectively) were available from 28 January to 6 April 2004 (Julian days 28–96). From 28 January to 6 April, there were eight extended blowing snow events—three of which were associated with ground blowing snow (Table 1), all of which are used in this study. Two other blowing snow events (11–12 February and 4–6 April; not used here) were created by an east wind and were discarded from this study. Savelyev et al. (2006) found differences between visibility data collected during easterly wind events and other situations and attributed these differences to the location of the visibility sensor housing that obstructed the wind and blowing snow particles in easterly winds with the sensor oriented N–S as recommended by the manufacturer. We have not used easterly wind cases in the present analysis.

According to the definition by Environment Canada, snow lifted from the earth’s surface by the wind to a height of 2 m or more is called blowing snow, while less than 2 m is referred to as drifting snow. In this paper, all events were classified as blowing snow. In addition, standard meteorological observations were performed at all times during the events used in this study.

3. Results and discussion

a. Relation between observed visibility and wind speed

In this study, empirical models were established to relate visibility to wind speed. The model is expressed (after Pomeroy and Male 1988; Li and Pomeroy 1997) as
i1520-0434-23-4-741-e4a
or, equivalently, using Eq. (1),
i1520-0434-23-4-741-e4b
where Vis is the 1.5-m height visibility (in km), μ is the equivalent extinction coefficient (in km−1), U is the 10-m height wind speed (in m s−1), and Ut is a 10-m height threshold wind speed (in m s−1). The constant, 16, is obtained from the maximum MOR indicated by the visibility sensor. Based on the observed visibility (Vis) and wind speed (U), and using the least squares technique, the parameters, b and Ut, can be found. When assessing the goodness of fit of the established models to the observed visibility, we employed four parameters, which are the coefficient of determination (R2), mean relative error (MRE), root-mean-square error of the visibility [RMSE(Vis)], and root-mean-square error of the extinction coefficient [RMSE(μ)]. RMSE(Vis) is expressed as
i1520-0434-23-4-741-e5
where Visobs and Vismod represent observed the visibility and modeled visibility, respectively, and n is the number of data points used for the analysis. Replacing visibility data with extinction coefficient data in the above formula, the RMSE(μ) will be obtained. The rationale for RMSE(μ) is that we anticipate that it will provide a better representation of errors since it will be the extinction coefficient [see Eq. (1)] that is more directly related to wind speed and particle counts. MRE signifies mean relative error and is expressed as
i1520-0434-23-4-741-e6

During the study period, some of the blowing snow events occurred with concurrent falling snow while others did not. These two kinds of blowing snow events have different relationships between visibility and wind speed. For the events of blowing snow without concurrent falling snow (ground blowing snow), the observed visibility had a good relationship with wind speed (Fig. 2) with an R2 greater than 0.89 for each individual event. The R2, RMSE(Vis), RMSE(μ), and MRE values when combining all events are 0.75, 0.8 km, 4.5 km−1, and 37%, respectively, while the RMSE(Vis), RMSE(μ), and MRE ranges lie between 0.39 and 0.43 km, 1.94 and 4.44 km−1, and 19% and 33% for the individual ground blowing snow events (only the data for visibilities of less than 16 km were included in these calculations). Therefore, a better relation between visibility and wind speed is obtained for each individual ground blowing snow event. The threshold 10-m wind speed (wind speed required to produce a blowing snow event) for ground blowing snow was calculated to range from 5.48 to 7.78 m s−1.

Figure 2 shows that the relationship between visibility and wind speed was slightly different among three ground blowing snow events. The event of 20–21 March obviously had a higher visibility than the other two events at the same wind speed, and the calculated threshold wind speed is greater than the other two events. This event had the highest measured top layer snow densities (250–370 kg m−3) of all three cases examined here (see Table 3). The 1–3 March event had the coldest temperatures and relatively less dense snow at the uppermost layer (150–250 kg m−3), which resulted in weaker winds needed to create extreme visibility reductions, while the 22–24 March event had snow densities in the upper few centimeters between 170 and 310 kg m−3 (Table 3). This result is similar to those of Baggaley and Hanesiak (2005), where blowing snow occurred most readily at air temperatures near −34°C regardless of the age of the snowpack. Also, the 20–21 March event had a shorter time since the previous snowfall and a higher top-layer snow density than that of 1–3 March, and this indicated that air temperature played a more important role for increasing snow density than other processes. It should be noted that the 22–24 March event was only 1 h after the latest snowfall (see Table 3); however, accumulations were not significant.

Li and Pomeroy (1997) developed an empirical function relating threshold wind speed to ambient air temperature. The threshold wind speed is lowest around −25°C and increases slightly as temperatures increase or decrease away from −25°C. However, in this study, the lowest threshold wind speed occurred between −30° and −35°C, which is slightly colder than the suggested temperature range of Li and Pomeroy.

For blowing snow with concurrent falling snow (Fig. 3), more scatter was observed in the plot of observed visibility versus wind speed and much smaller R2 values than those of ground blowing snow events were obtained. For each individual case of ground blowing snow, wind speed was the main factor affecting visibility and can be a good indicator for visibility in those kinds of events. In the event of blowing snow accompanied by falling snow, the visibility was influenced by both wind speed and snowfall and cannot be predicted by wind speed alone. Note the different ranges of wind speed in Figs. 2 and 3 and the low visibility (of order 100 m) in some high wind speed blowing snow cases.

b. Extreme visibility reductions

The public, and operational meteorology personnel, are especially concerned with blowing snow events that cause extreme visibility reductions due to their impact on transportation and public safety. In this section, we will investigate more extreme visibility reductions (visibilities of less than 1 km) for three ground blowing snow events.

If the regression functions from the previous section of each ground blowing snow event are used for predicting the visibilities of less than 1 km, R2 would range between 0.77 and 0.82; the RMSE(Vis) and RMSE(μ) between 0.11 and 0.25 km and 2.28 and 5.19 km−1, respectively; and the MRE between 18% and 35%. When new regression functions are generated using only the data with MOR between 0 and 1 km for each event, R2 increased to 0.83–0.86, and RMSE(Vis) and MRE would be reduced to 0.07–0.15 km and 15%–25%, respectively (Fig. 4). In general, the new regression functions would be better predictors of visibility even though the RMSE(μ) is slightly higher for the event of 20–21 March over the limited visibility range compared to using the original regression equations over the full range. Wind speeds at 10 m associated with ground blowing snow events with visibilities less than 1 km were all >8 m s−1. The regression function obtained when combining all data with visibilities less than 1 km for all ground blowing snow events is
i1520-0434-23-4-741-e7
With R2 = 0.65 and an MRE of 43%, this is not as robust as the individual event regression equations (Fig. 3). However, it still may be used as a tool for predicting extreme low visibilities over smooth Arctic sea ice, provided the user is knowledgeable about the inherent errors cited here.

c. Relation between observed visibility at different levels

Although we have focused on the 1.5-m MOR, we also have visibility measurements at 3 m. Figure 5 shows the relationship between measurements at the two heights for the three extended occurrences of blowing snow used in Figs. 2 and 4. In these log–log plots, we can see a clear power law relationship, with a slope close to 1.0, especially in the low visibility cases. Fitting these data points gives a relationship close to MOR3m = 1.5 MOR1.5m.

d. Relation between observed visibility and particle counter data

Figure 6 shows the relationship between visibility and particle counts (PCs) at 1.5 m. The particle counter values for 1.5 m used here are interpolated logarithmically between 1- and 2-m counters, implying a power law relationship: PC = AzB, where A and B are constants. Logarithmic interpolation is warranted based upon historical turbulent transfer studies over surfaces such as those experienced in this study. The nature of the wind flow near the surface during the events lacked gustiness, which created a more consistent vertical distribution of particle sizes with height on average. Generally, the observed visibility had a strong power law relationship with particle counter readings with an R2 of 0.92, regardless of whether the blowing snow event is with or without concurrent precipitation and the particle counters that were used in this research could in principle be used for predicting visibility during blowing snow. However, as stated in the introduction, the visibility in blowing snow is determined by drift and blowing snow density, particle size distribution, and mean particle radius (Pomeroy and Male 1988). The particle counter readings reflect a product of particle number densities and wind speed (i.e., a particle flux), not particle size distribution and mean particle radius. Thus, the prediction function that converts particle counter readings to visibility would not account for the particle size and its distribution and, therefore, cannot relate directly to visibility. We can obtain an estimate of the particle number density if we divide the counts by the cross-sectional area of the beam and the wind speed, interpolated to the level of the counter. This was done and the results are shown in Fig. 7 as particle density (interpolated to 1.5 m) versus wind speed.

When the relation between observed visibility and particle counter readings was investigated in each individual case, the same strength of the relationship between visibility and particle counter readings was obtained for most cases, and the distribution of the data in the plot of visibility versus particle counter readings followed a similar regression line (not shown) to that in Fig. 6. Since the properties of snow and environmental conditions (e.g., wind speed, temperature, snow age, etc.) are different from event to event, the particle size distribution and mean particle radius may also be different from event to event. However, results suggest that the variation in particle size distribution and mean particle radius from event to event may not have a strong influence on visibility in this study. As there should be an obvious difference in particle size distribution and mean particle radius between blowing snow events with and without concurrent precipitation, the relationship between visibility and particle counter readings should be different between ground blowing snow and the blowing snow with concurrent precipitation. However, this difference was not observed here, which suggests that the difference between particle size distribution and mean particle radius may not contribute significantly to the variations in visibility possibly due to the nature of the precipitation during the blowing snow events in this study (i.e., light precipitation at cold temperatures that may have easily broken apart into small particles in the wind). We emphasize that no measurements were available to prove this hypothesis; hence, the explanations are purely theoretical. Further experiments to test the particle size sensitivity of visibility in the visible range are required.

e. Comparison of modeled and observed visibilities

We have revised Pomeroy and Male’s (1988) visibility model (PM88N) to relate visibility to the particle number density, N (m−3), as
i1520-0434-23-4-741-e8
where, for a gamma distribution, the cross-sectional area-weighted mean particle radius is
i1520-0434-23-4-741-e9
The shape parameter, α, and the mean particle size during a blowing snow event may vary with observation height and also wind speed (Budd et al. 1966) and it is not technically appropriate to assume constant parameters of the particle size distribution and mean particle size for all blowing snow events. However, for testing our visibility model we have assumed this to be the case for z = 1.5 m. In the second alternative in Eq. (8) above, Vis depends only on ra and N.

There are various reports on the mean particle size of blowing snow. Budd et al. (1966) reported the mean diameter of blowing snow particles to be 132 μm at the surface and 87 μm at 2 m. Schmidt (1986) reported 60–120 μm at 25-cm height, and Brown and Pomeroy (1989) suggested a radius of 40 μm at 2-m height. For the blowing snow events, Budd et al. (1966) argued that α increases from the surface to higher levels. Schmidt (1982, 1986) measured α values of 14 at 1-m height and 2 ∼ 4 at 25-cm height. Mann et al. (2000) reported a constant α of 2 from 0.1- to 4-m height. As long as α > 2, it could be argued that r and ra are approximately equal, within the range of uncertainties associated with blowing snow.

A mean radius, ra, of 50 μm and an α of 10 were used as the best guesses in the PM88N model to simulate visibility and the results are shown in Fig. 8. Results with ra = 25 and 100 μm are also shown. The PM88N-predicted visibilities fit reasonably well with observed values with R2 of 0.83, RMSE of 1.3 km, and MRE of 30.54%. Results with ra = 25 and 100 μm lie above and below the observed values, suggesting that, for these data at least, the mean particle size is reasonably close to 50 μm at 1.5 m.

4. Conclusions

During blowing snow events, visibility is reduced by drifting and blowing snow, and can depend on number density, mean particle size, and parameters of the snow particle size distribution (Pomeroy and Male 1988). Drifting and blowing snow densities can be affected by many factors including wind speed, snowpack conditions, and air temperature (Baggaley and Hanesiak 2005). These parameters may be different from case to case, making it difficult to precisely predict visibility using wind speed alone. However, our results have shown that regression equation correlations between wind speed and visibility for ground blowing snow events can be as high as 0.93 with a reasonable MRE (19%–32%), which can be useful for operational purposes. The threshold wind speed for producing ground blowing snow events was found to range between 5.5 and 7.9 m s−1, with T < −15°C. The regression equations developed for visibility of less than 1 km are recommended for use during more extreme visibility reductions and wind speeds (>9.5 m s−1), with MRE reduced to 15%–25%. As might be expected, poorer relationships were observed in the event of blowing snow with concurrent precipitation. The ground blowing snow relations derived are valid for events over flat, snow-covered terrain, with roughness lengths of order 10−3 m and, given the temperature ranges involved in most events, high liquid water equivalent ratios (i.e., near 10:1 or dry snow). Since no liquid water equivalent precipitation measurements were available, there is no way of providing a range over which our relations apply in this regard; however, given the time of year and air temperatures for the events presented, dry snow is a valid assumption. Wetter snow regimes and rougher surfaces would affect the results from this study but quantification of this is not possible. Future work will examine different terrain roughness lengths.

The Pomeroy and Male (1988) visibility model can be revised to relate the 1.5-m visibility during blowing snow to the particle number density. The model visibilities had a good relation with the observed visibilities with an R2 of 0.83, RMSE of 1.3 km, and MRE of 30.54%.

The observed visibility had a strong relationship with particle counter readings with an R2 of 0.92 in general. This relationship can be used to predict visibility from particle counter readings. However, the predicted visibility could have relative errors due to counter characteristics leading to inaccuracies in N, nonspherical particle shapes, and other factors. Some fine particles may not have been registered by the particle counter due to the high wind speeds but this also did not appear to significantly contribute to the relationship between particle counts and reduction in visibility during the blowing snow events observed in this study.

There is no obvious difference in the relationship of observed visibility with particle counter readings for most blowing snow events and between blowing snow events with and without concurrent precipitation. This result may suggest that the snow particle size and its distribution did not change significantly from event to event and did not have a large effect on visibility; however, no measurements were available to corroborate this hypothesis.

Acknowledgments

This work has been funded by the Canadian Foundation for Climate and Atmospheric Sciences (CFCAS) through a project grant (Drifting and blowing snow: Measurements and modelling) to PT, JH, and TP, and the Canadian Arctic Shelf Exchange Study (CASES) and ArcticNet Network Centres of Excellence (NCE) grants to JH. The authors also wish to thank the Hydrometeorology and Arctic Lab (HAL) and the Prairie and Arctic Storm Prediction Centre (PASPC) of the Meteorological Service of Canada for in-kind and financial support during the CASES project. We also acknowledge the assistance of several students (Teresa Fisico, Jim Butler, and Alex Langlois from the University of Manitoba; Mark Gordon from York University, and Rob Pierson from the University of Alberta), as well as the CCGS Amundsen research icebreaker crew, for their support during CASES.

REFERENCES

  • Atmospheric Environment Service, 1977: Manual of Surface Weather Observations (MANOBS). 7th ed. Environment Canada, 332 pp.

  • Baggaley, D. G., and Hanesiak J. M. , 2005: An empirical blowing snow forecast technique for the Canadian Arctic and the Prairie Provinces. Wea. Forecasting, 20 , 5162.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brown, T., and Pomeroy J. W. , 1989: A blowing snow particle detector. Cold Regions Sci. Technol., 16 , 167174.

  • Budd, W. F., Dingle R. J. , and Radok U. , 1966: The Byrd Snow Drift Project: Outline and basic results. Studies in Antarctic Meteorology, M. J. Rubin, Ed., Antarctic Research Series, Vol. 9, Amer. Geophys. Union, 71–134.

    • Search Google Scholar
    • Export Citation
  • Kobayashi, D., 1978: Snow transport by katabatic winds in Mizuho camp area, East Antarctica. J. Meteor. Soc. Japan, 56 , 130139.

  • Koschmieder, H., 1924: Theorie der horizontalen Sechtweite (Theory of horizontal visibility). Beitr. Phys. Freien Atmos., 12 , 3353.

  • Li, L., and Pomeroy J. W. , 1997: Estimates of threshold wind speeds for snow transport using meteorology data. J. Appl. Meteor., 36 , 205213.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liljequist, G., 1957: Energy exchange of an Antarctic snowfield. Norwegian–British–Swedish Antarctic Expeditions, 1949–1952: Scientific Results, Vol. II, Norwegian Polar Institute, Part 1C.

    • Search Google Scholar
    • Export Citation
  • Lister, H., 1960: Glaciology (1): Solid precipitation and drift snow. Trans-Antarctic Expedition 1955–1958 Sci. Rep. 5, Royal Society London.

  • Mann, G. W., Anderson P. S. , and Mobbs S. D. , 2000: Profile measurements of blowing snow at Halley, Antarctica. J. Geophys. Res., 105 , 2449124508.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mellor, M., 1966: Light scattering and particle aggregation in snow-storms. J. Glaciol., 6 , 237248.

  • Middleton, W. E. K., 1952: Vision through the Atmosphere. University of Toronto Press, 250 pp.

  • Pomeroy, J. W., and Male D. H. , 1988: Optical properties of blowing snow. J. Glaciol., 34 , 310.

  • Savelyev, S. A., Gordon M. , Hanesiak J. , Papakyriakou T. , and Taylor P. A. , 2006: Blowing snow studies in the Canadian Arctic Shelf Exchange Study, 2003–2004. Hydrol. Processes, 20 , 817827.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schmidt, R. A., 1979: Measuring visibility in blowing snow. Proc. Second Int. Symp. on Snow Removal and Ice Control Research, Hanover, NH, National Academy of Sciences/Transportation Research Board, 208–214.

  • Schmidt, R. A., 1982: Vertical profiles of wind speed, snow concentration, and humidity in blowing snow. Bound.-Layer Meteor., 23 , 223246.

  • Schmidt, R. A., 1986: Transport rate of drifting snow and the mean wind speed profile. Bound.-Layer Meteor., 34 , 213241.

  • Tabler, R. D., 1979: Visibility in blowing snow and applications in traffic operations. Proc. Second Int. Symp. on Snow Removal and Ice Control Research, Hanover, NH, National Academy of Sciences/Transportation Research Board, 200–207.

  • Takahashi, S., 1985: Characteristics of drifting snow at Mizuho station, Antarctica. Ann. Glaciol., 6 , 7175.

Fig. 1.
Fig. 1.

(a) General view of the study site and (b) location of the study area.

Citation: Weather and Forecasting 23, 4; 10.1175/2008WAF2007015.1

Fig. 2.
Fig. 2.

The plot of the 1.5-m-height observed visibility vs the 10-m-height wind speed using the data obtained in three major ground blowing snow events listed in Table 2.

Citation: Weather and Forecasting 23, 4; 10.1175/2008WAF2007015.1

Fig. 3.
Fig. 3.

The plot of the 1.5-m-height observed visibility vs the 10-m-height wind speed using the data obtained in three events of blowing snow with concurrent falling snow listed in Table 4.

Citation: Weather and Forecasting 23, 4; 10.1175/2008WAF2007015.1

Fig. 4.
Fig. 4.

Observed visibility at 1.5- vs 10-m-height wind speed in three major ground blowing snow events using only data with visibility <1 km listed in Table 5.

Citation: Weather and Forecasting 23, 4; 10.1175/2008WAF2007015.1

Fig. 5.
Fig. 5.

Comparisons of MOR at 1.5 and 3 m for the three major ground blowing snow events. Symbols are the same as in Figs. 2 and 4: (a) full range and (b) low visibility cases (<1 km) only.

Citation: Weather and Forecasting 23, 4; 10.1175/2008WAF2007015.1

Fig. 6.
Fig. 6.

The plot of 1.5-m-height observed visibility vs 1.5-m-height (interpolated) particle counter readings using the data obtained in all events of blowing snow (from a westerly direction). Only every fifth data point is shown, but best fit is to all 3300 data points. Power law fit, Vis(1.5m) = 3.23 (counts)−0.86 km, R2 = 0.92, RMSE = 3.04 km, and MRE = 32.3%

Citation: Weather and Forecasting 23, 4; 10.1175/2008WAF2007015.1

Fig. 7.
Fig. 7.

Particle number density at 1.5 m, N (m−3) vs U10 (m s−1) for all events listed in Table 1. Power law fit, N = 1.06 × 10−5 U4.2510 and R2 = 0.74.

Citation: Weather and Forecasting 23, 4; 10.1175/2008WAF2007015.1

Fig. 8.
Fig. 8.

Comparison of observed visibility (pluses) with model predictions for three values of mean radius ra as functions of N. Black line is with ra = 50 μm. Gray lines above and below are for ra values of 25 and 100 μm, respectively.

Citation: Weather and Forecasting 23, 4; 10.1175/2008WAF2007015.1

Table 1.

The events of blowing snow in this study; BS is ground blowing snow and FS is precipitating (falling) snow.

Table 1.
Table 2.

The date, symbol, regression equation, correlation (R2), root mean square error (RMSE; km), and mean relative error (MRE; %) for each of the three major ground blowing snow events used in Fig. 2.

Table 2.
Table 3.

The snowpack age, air temperature, and uppermost snow-layer densities of the three ground blowing snow events.

Table 3.
Table 4.

The date, symbol, regression equation, correlation (R2), root mean square error (RMSE; km), and mean relative error (MRE; %) for each of the three blowing snow with concurrent falling snow events used in Fig. 3.

Table 4.
Table 5.

The date, symbol, regression equation, correlation (R2), root mean square error (RMSE; km), and mean relative error (MRE; %) for the three major ground blowing snow events with visibility <1 km used in Fig. 4.

Table 5.
Save
  • Atmospheric Environment Service, 1977: Manual of Surface Weather Observations (MANOBS). 7th ed. Environment Canada, 332 pp.

  • Baggaley, D. G., and Hanesiak J. M. , 2005: An empirical blowing snow forecast technique for the Canadian Arctic and the Prairie Provinces. Wea. Forecasting, 20 , 5162.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Brown, T., and Pomeroy J. W. , 1989: A blowing snow particle detector. Cold Regions Sci. Technol., 16 , 167174.

  • Budd, W. F., Dingle R. J. , and Radok U. , 1966: The Byrd Snow Drift Project: Outline and basic results. Studies in Antarctic Meteorology, M. J. Rubin, Ed., Antarctic Research Series, Vol. 9, Amer. Geophys. Union, 71–134.

    • Search Google Scholar
    • Export Citation
  • Kobayashi, D., 1978: Snow transport by katabatic winds in Mizuho camp area, East Antarctica. J. Meteor. Soc. Japan, 56 , 130139.

  • Koschmieder, H., 1924: Theorie der horizontalen Sechtweite (Theory of horizontal visibility). Beitr. Phys. Freien Atmos., 12 , 3353.

  • Li, L., and Pomeroy J. W. , 1997: Estimates of threshold wind speeds for snow transport using meteorology data. J. Appl. Meteor., 36 , 205213.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Liljequist, G., 1957: Energy exchange of an Antarctic snowfield. Norwegian–British–Swedish Antarctic Expeditions, 1949–1952: Scientific Results, Vol. II, Norwegian Polar Institute, Part 1C.

    • Search Google Scholar
    • Export Citation
  • Lister, H., 1960: Glaciology (1): Solid precipitation and drift snow. Trans-Antarctic Expedition 1955–1958 Sci. Rep. 5, Royal Society London.

  • Mann, G. W., Anderson P. S. , and Mobbs S. D. , 2000: Profile measurements of blowing snow at Halley, Antarctica. J. Geophys. Res., 105 , 2449124508.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mellor, M., 1966: Light scattering and particle aggregation in snow-storms. J. Glaciol., 6 , 237248.

  • Middleton, W. E. K., 1952: Vision through the Atmosphere. University of Toronto Press, 250 pp.

  • Pomeroy, J. W., and Male D. H. , 1988: Optical properties of blowing snow. J. Glaciol., 34 , 310.

  • Savelyev, S. A., Gordon M. , Hanesiak J. , Papakyriakou T. , and Taylor P. A. , 2006: Blowing snow studies in the Canadian Arctic Shelf Exchange Study, 2003–2004. Hydrol. Processes, 20 , 817827.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schmidt, R. A., 1979: Measuring visibility in blowing snow. Proc. Second Int. Symp. on Snow Removal and Ice Control Research, Hanover, NH, National Academy of Sciences/Transportation Research Board, 208–214.

  • Schmidt, R. A., 1982: Vertical profiles of wind speed, snow concentration, and humidity in blowing snow. Bound.-Layer Meteor., 23 , 223246.

  • Schmidt, R. A., 1986: Transport rate of drifting snow and the mean wind speed profile. Bound.-Layer Meteor., 34 , 213241.

  • Tabler, R. D., 1979: Visibility in blowing snow and applications in traffic operations. Proc. Second Int. Symp. on Snow Removal and Ice Control Research, Hanover, NH, National Academy of Sciences/Transportation Research Board, 200–207.

  • Takahashi, S., 1985: Characteristics of drifting snow at Mizuho station, Antarctica. Ann. Glaciol., 6 , 7175.

  • Fig. 1.

    (a) General view of the study site and (b) location of the study area.

  • Fig. 2.

    The plot of the 1.5-m-height observed visibility vs the 10-m-height wind speed using the data obtained in three major ground blowing snow events listed in Table 2.

  • Fig. 3.

    The plot of the 1.5-m-height observed visibility vs the 10-m-height wind speed using the data obtained in three events of blowing snow with concurrent falling snow listed in Table 4.

  • Fig. 4.

    Observed visibility at 1.5- vs 10-m-height wind speed in three major ground blowing snow events using only data with visibility <1 km listed in Table 5.

  • Fig. 5.

    Comparisons of MOR at 1.5 and 3 m for the three major ground blowing snow events. Symbols are the same as in Figs. 2 and 4: (a) full range and (b) low visibility cases (<1 km) only.

  • Fig. 6.

    The plot of 1.5-m-height observed visibility vs 1.5-m-height (interpolated) particle counter readings using the data obtained in all events of blowing snow (from a westerly direction). Only every fifth data point is shown, but best fit is to all 3300 data points. Power law fit, Vis(1.5m) = 3.23 (counts)−0.86 km, R2 = 0.92, RMSE = 3.04 km, and MRE = 32.3%

  • Fig. 7.

    Particle number density at 1.5 m, N (m−3) vs U10 (m s−1) for all events listed in Table 1. Power law fit, N = 1.06 × 10−5 U4.2510 and R2 = 0.74.

  • Fig. 8.

    Comparison of observed visibility (pluses) with model predictions for three values of mean radius ra as functions of N. Black line is with ra = 50 μm. Gray lines above and below are for ra values of 25 and 100 μm, respectively.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 311 104 24
PDF Downloads 217 68 15