Introduction
Characterizing rain-on-snow events and their impacts
Arctic rain-on-snow (ROS) events occur when liquid precipitation, in the form of rain or freezing rain, falls on an existing snowpack (Grenfell and Putkonen, Reference Grenfell and Putkonen2008; Rennert and others, Reference Rennert, Roe, Putkonen and Bitz2009; Bieniek and others, Reference Bieniek2018; Serreze and others, Reference Serreze2021). In general, research on Alaskan ROS events noted that these conditions are most likely to occur from October through April – with some studies narrowing it further to November through March – when conditions are favorable and a snowpack is present (Bieniek and others, Reference Bieniek2018; Pan and others, Reference Pan, Kirchner, Kimball, Kim and Du2018; Crawford and others, Reference Crawford, Alley, Cooke and Serreze2020). North American and Eurasian ROS events may also occur during this seasonal window (Cohen and others, Reference Cohen, Ye and Jones2015).
ROS events in the middle latitudes have been studied extensively, including how their occurrence relates to geographic position relative to sources of maritime moisture or the number of rain days a location may experience (Kattelmann, Reference Kattelmann1997; McCabe and others, Reference McCabe, Clark and Hay2007; Cohen and others, Reference Cohen2014; Garvelmann and others, Reference Garvelmann, Pohl and Weiler2015). Impacts from flooding are well known, with the combination of heavy rainfall and melting of the underlying snowpack (McCabe and others, Reference McCabe, Clark and Hay2007). An area may be more susceptible to ROS-generated flooding due to several factors: rain over a large catchment area (which leads to a high amount of runoff), the potential for additional snowmelt conditions and changes in snow cover dynamics (snow metamorphism) and elevated rainfall over extended periods (when storm systems should be producing snow) (Kattelmann, Reference Kattelmann1997; Singh and others, Reference Singh, Spitzbart, Hübl and Weinmeister1997; Garvelmann and others, Reference Garvelmann, Pohl and Weiler2015; Guan and others, Reference Guan, Waliser, Ralph, Fetzer and Neiman2016).
ROS events can disrupt ground transportation and aviation operations, and wet-snow (or slush) avalanches resulting from ROS can damage infrastructure (Putkonen and Roe, Reference Putkonen and Roe2003; Hansen and others, Reference Hansen2014). Officials may close roads and airports due to ice formation, isolating Arctic communities (Hansen and others, Reference Hansen2014). As a notable example of infrastructure impacts, a slush avalanche in Longyearbyen (Svalbard) destroyed a pedestrian bridge, and major roads that serviced the community had to be closed for several days (Hansen and others, Reference Hansen2014). The authors add that Arctic locations are susceptible to wet-snow avalanches in a warming climate, as current infrastructure was not originally built with these natural disasters in mind.
Following ROS occurrence, the ice layers that accumulate on, or within, the snowpack act as barriers to foraging, sometimes leading to mass starvation of caribou, reindeer and musk oxen (Rennert and others, Reference Rennert, Roe, Putkonen and Bitz2009; Forbes and others, Reference Forbes2016; Serreze and others, Reference Serreze2021). Ice formation may also force animals to seek other sources of food further away from their regular environments, exacerbating the conditions leading to starvation (Serreze and others, Reference Serreze2021). Examples include an Arctic ROS event that occurred on Banks Island in Canada during October of 2003 that led to the demise of an estimated 20 000 musk oxen, an event in Svalbard in January of 2012 – which produced one of the largest numbers of reindeer carcasses found in the following summer – and an event in the Yamal Peninsula in northern Russia during the autumn of 2013 that starved ~61 000 reindeer (out of a total of 275 000) (Rennert and others, Reference Rennert, Roe, Putkonen and Bitz2009; Hansen and others, Reference Hansen2014; Serreze and others, Reference Serreze, Crawford and Barrett2015; Forbes and others, Reference Forbes2016).
Known meteorology concerning Arctic rain-on-snow events
Generally, near-surface air temperatures in the region of an Arctic ROS event increase dramatically preceding the onset of precipitation, typically over a relatively short time period (Rennert and others, Reference Rennert, Roe, Putkonen and Bitz2009; Hansen and others, Reference Hansen2014; Serreze and others, Reference Serreze2021). This causes rain to fall during part of the event (or throughout the entirety of the event) and may cause additional surface melting. Air temperatures then decrease following the event, often to well below freezing (Serreze and others, Reference Serreze2021). Liquid water freezes and forms a thick glaze of ice along the surface of the snow layer or within the snowpack (Serreze and others, Reference Serreze2021).
This sequence of shifting temperatures frequently involves an advancing extratropical cyclone (which generates the initial precipitation), with a cold front then progressing through the area (Rennert and others, Reference Rennert, Roe, Putkonen and Bitz2009). Working in tandem with the overall precipitation event, increased warm air advection with these systems causes additional melting of the surface snow layer through amplified mixing and turbulent fluxes (Semmens and others, Reference Semmens, Ramage, Bartsch and Liston2013). For example, Rennert and others (Reference Rennert, Roe, Putkonen and Bitz2009) described a strong anticyclonic ridge at the synoptic scale that initially developed over the Banks Island region preceding the October 2003 ROS event. This feature led to strong, southwesterly flow bringing warmer, moister air into the area. Lift (upward motion), triggered by an approaching shortwave trough (extratropical cyclone), initiated precipitation across the region. The precipitation first began as snow then transitioned to rain as air temperatures rose (Rennert and others, Reference Rennert, Roe, Putkonen and Bitz2009).
Some studies noted ROS connections with atmospheric blocking and atmospheric rivers (ARs). Crawford and others (Reference Crawford, Alley, Cooke and Serreze2020) described a link between blocking patterns and ROS conditions in Alaska, in which a strong pressure gradient builds between a ridge of high-pressure and an approaching extratropical storm system. This gradient instigates further warm air advection and transport of positive anomalies in precipitable water, the total atmospheric water vapor contained within an atmospheric column. Serreze and others (Reference Serreze, Crawford and Barrett2015) associated an AR with the January 2012 Svalbard ROS event, which coincided with an extreme event in total precipitation. Studies for the middle latitudes documented links between ROS events in mountainous regions in the inland Western US and landfalling ARs along the US West Coast (Guan and others, Reference Guan, Waliser, Ralph, Fetzer and Neiman2016; Trubilowicz and Moore, Reference Trubilowicz and Moore2017). Nevertheless, it is widely acknowledged that additional research is needed to understand the weather patterns influencing Arctic ROS events and how features like ARs and blocking setups impact their formation (Rennert and others, Reference Rennert, Roe, Putkonen and Bitz2009; Bieniek and others, Reference Bieniek2018).
Research questions posed for this study
The Arctic Rain-on-Snow Study, a team of interdisciplinary researchers and part of the National Science Foundation's Navigating the New Arctic initiative, focuses on better understanding Arctic ROS events and their impacts on Arctic communities. One of the project goals is to assess the meteorological conditions most influential in setting up ROS occurrences. As mentioned, Arctic ROS events occur at times of the year (October through April) when precipitation should usually be falling as snow and during which solar radiation is limited or even absent, depending on the latitude. This implies a key role of warm (and moist) air advection and transport from lower latitudes. As the Arctic continues to warm, one can expect both the seasonality and intensity of ROS events to change. This prompts three questions:
(1) What are the primary meteorological conditions at varying spatial scales linked to Arctic ROS occurrence?
(2) Do synoptic scale blocking patterns and ARs play important roles in Arctic ROS initiation?
(3) Are the strength and presence of these features (and others) influential in the severity of ROS impacts?
Case study selection, data sources and methodology
Selection of case studies
We selected two divergent ROS events for this study. The first event occurred in Western Greenland in mid-April of 2016. A team of researchers confirmed it by investigating wet-snow avalanches near Nuuk, Greenland, using remote-sensing data (Abermann and others, Reference Abermann, Eckerstorfer, Malnes and Ulf Hansen2019). The authors concluded that over 800 wet-snow avalanches initiated during this ROS event and documented that an automated surface observation station was destroyed on 11 April. This coincided with the day of highest air temperatures and precipitation rates.
Abermann and others (Reference Abermann, Eckerstorfer, Malnes and Ulf Hansen2019) briefly examined the associated weather conditions. A high-pressure system built over the region beginning on 9 April 2016. It continued to strengthen, progressing through April 10. As a warm front approached southwestern Greenland, air temperatures increased rapidly, and this feature provided additional moisture advection needed to produce significant precipitation. Abermann and others (Reference Abermann, Eckerstorfer, Malnes and Ulf Hansen2019) documented that the Greenland Ecosystem Monitoring Program's automated meteorological station near Nuuk, Greenland, recorded this rapid warming (an increase of 22.2°C from 9 to 11 April) and a precipitation total of 25 mm during this same 2 d period. An additional station with the Asiaq monitoring network for Nuuk recorded a slightly smaller temperature increase of 14.6°C but a larger precipitation total of 49 mm for the 2 d period, being a location closer to the maritime environment.
The second event occurred on 19 January 2021, in Iqaluit (Nunavut, Canada), confirmed by an eyewitness report on social media and relayed to the Arctic Rain-on-Snow Study team by one of its members. This case presented an opportunity to research a ROS event that had not been studied previously. In the lead up to this event, much of Canada had experienced a swing from seasonally cold to above normal air temperatures during the early part of January 2021. Around the beginning of the new year, atmospheric reanalysis data indicated surface air temperature anomalies of −5 to −20°C across much of the Canadian Arctic. By the second week in January, atmospheric reanalysis showed the flip to +5 to +20°C surface air temperature anomalies over the same area, likely due to the combination of a building ridge of high-pressure over western Canada and a blocking feature over eastern Canada.
According to an article published for the Nunatsiaq News on 18 January 2021, and posted on the Local Environmental Observer Network (a web-based platform built for community sharing of unusual weather events), Iqaluit was expected to continue experiencing unseasonably high air temperatures through the middle of the month. A low-pressure system was projected to move into the area through the coming week, bringing warm air from the south, and allowing air temperatures to remain around the freezing mark (Nunatsiaq News, 2021). This was likely the same system that produced ROS conditions on 19 January. Climate data provided by the Government of Canada reported that the snow depth at Iqaluit was 25 cm (9.84 inches) on 6 January 2021. There was a period of missing data from 7 January through 19 January, but the next observation for Iqaluit on 20 January indicated a snow depth of 25 cm. Therefore, we can safely assume that there was a snowpack at the time rain was reported on 19 January.
Documented impacts related to ROS for the 2021 Iqaluit event were very limited. However, the effects resulting from the increased air temperatures, including on the day of the event, were noted. Iqaluit broke its maximum air temperature record for 19 January, with a new record of 0.5°C (CBC News, 2021). According to the article from CBC News, the new temperature record broke the previous record of −2.2°C, set in 1958. Subsequently, the differences in impacts resulting from ROS during these two events allow us to compare the meteorological features linked to ROS occurrence and the difference in strength of these features that led to these varying impacts.
Data types and acquisition
We use ERA5 reanalysis output from the European Centre for Medium-Range Weather Forecasts (ECMWF) for this study. ERA5, like other reanalysis datasets, is a combination of observational and modeling datasets and performs well in its depiction of meteorological variables in comparison to direct observations, including temperature, specific humidity and wind speed (Graham and others, Reference Graham, Hudson and Maturilli2019). However, relative to its predecessor, ERA-Interim, ERA5 advances the discernment of synoptic- and meso-scale features, which includes cyclones, a key precipitation forcing for these two ROS events. Comparisons of radiosonde and PILOT data (an alternative upper-air balloon observation) prior to data assimilation shows an improved fit for tropospheric temperature, winds and humidity (Hoffmann and others, Reference Hoffmann2019; Hersbach and others, Reference Hersbach2020). Another key finding from the Hoffmann and others (Reference Hoffmann2019) effort demonstrates that ERA5 trajectories (based within Lagrangian transport models) better conserve stratospheric-level potential temperature, leading to smaller data assimilation increments that improve the uniformity of ECMWF's forecast model and observations. ERA5 also produces hourly output at a 31 km horizontal resolution, higher than ERA-Interim's 80 km resolution (Hersbach and others, Reference Hersbach2020).
We access ERA5 data from the 1979 to present dataset. ERA5 data are available hourly, and we selected the 00Z (UTC) and 12Z files for this study. These times coincide with upper-air launch times (radiosondes), and 12 h increments still provide sufficient temporal resolution to examine the synoptic makeup of ROS events. An application program interface allows for different meteorological parameters to be downloaded to data files.
We supplement this study with direct observations, including automated surface observation stations and atmospheric sounding data (radiosondes). The observation network across the Arctic remains sparse, with many of these automated sites restricted to airports. Other limitations include intermittent outages and upper-air data only being provided twice a day. Sounding data utilized with this study presents a unique dataset not analyzed in previous Arctic ROS studies. We obtained sounding data for Aasiaat, Greenland (north of Nuuk), and for Iqaluit from the University of Wyoming sounding archive using python code provided by MetPy, a program developed by the University Corporation for Atmospheric Research. Automated station data for Godthaab (Nuuk) and Iqaluit were available from the Iowa Environmental Mesonet surface observation archive from Iowa State University (https://mesonet.agron.iastate.edu/ASOS/).
Data visualization methodology
We created data visualizations of ERA5 reanalysis using select python programming modules. Visualizations were divided into three atmospheric heights: upper levels, middle levels and lower levels. Weather variables were chosen at these levels to best characterize various meteorological processes. Figure 1 provides an overview of the meteorological variables used in each analysis and at each atmospheric level. Additionally, MetPy code (noted above) was deployed in building skew-ts, commonly used to visualize sounding data. A separate python-based program was written to provide time series data visualizations of the surface station observations. We created these time series graphs to show the progression of temperature and dewpoint temperature changes and precipitation type transitions, including during each ROS event.
Weather variables examined with the reanalysis data.
Results
April 2016 Western Greenland rain-on-snow event
Atmospheric reanalysis
As introduced earlier, the April 2016 Western Greenland ROS event generated numerous wet-snow avalanches near Nuuk (Abermann and others, Reference Abermann, Eckerstorfer, Malnes and Ulf Hansen2019). Several meteorological features stand out in the ERA5 data. In the upper levels (Fig. 2), an Omega Block is prominent at both the jet stream level (250 mb level) and the 500 mb level. An Omega Block resembles a capital Greek letter Omega through the shape of the tropospheric, synoptic scale waveform. In this case, the ridge extending across Greenland is sandwiched between a broad trough across eastern Canada and a cutoff low over western Europe. It represents a block because of the predominant meridional flow and the persistent disruption to the general west-to-east progression of weather systems.
The 2016 Western Greenland case upper atmospheric levels. The 250 mb geopotential heights are plotted on the left, and the 500 mb heights and winds are plotted on the right. The 500 mb panel also includes isotachs (lines of constant wind speed) in m s−1 in filled contours, in addition to wind barbs that indicate both direction and speed.
Strong wind speeds (above 50 m s−1 over Baffin Bay) are present at the 500 mb level along the gradient between the ridge over southern Greenland and the broad trough over eastern Canada. This area of higher wind speeds represents a jet streak. The location of interest – in this case Nuuk (Godthaab), Greenland – falls within this area that would likely be experiencing greater precipitation rates due to the additional jet dynamics. In addition, these winds are predominantly out of the south on the left (western) side of the ridge, as seen in the flag direction of the wind barbs in the 500 mb analysis. This implies a warmer air mass moving into the region. An opposing northerly flow follows on the right (eastern) side of the ridge, which likely assisted in maintaining the block.
Precipitation associated with this event can be linked to the approaching shortwave trough extending over much of the province of Quebec (Fig. 2). ERA5 reanalysis fields show dynamically induced rising motion on 11 April 2016, coinciding with precipitation over much of the area near Nuuk. Nuuk lies within a fjord, with surrounding terrain approaching 1000 m. While suggesting that orographic lifting contributed to the elevated precipitation amounts, our interpretation is that dynamic lift was the dominant precipitation forcing.
In the atmospheric mid- to low-levels, strong moisture transport and warm air advection accompany a cyclone-induced low-level jet (Fig. 3). These features are usually associated with the pre-cold frontal sector of an extratropical cyclone (Ralph and others, Reference Ralph, Neiman and Rotunno2005). They can be important factors in determining what locations experience precipitation and how much of it. The cited study also notes that ‘the low-level jet is an integral part of extratropical cyclones and is characterized by warm temperatures, weak stratification, large water vapor content, and strong low-altitude winds’ [Browning and Pardoe (1973), as referenced in Ralph and others (Reference Ralph, Neiman and Rotunno2005)]. Most studies agree that low-level jets may be found at an altitude of ~1 km, but wind speeds may range from 23 to 35 m s−1 (Lackmann, Reference Lackmann2002; Ralph and others, Reference Ralph, Neiman and Rotunno2005). However, in Arctic locations, weaker low-level jets may be just as impactful. For example, an event near Barrow, Alaska, associated with a low-level jet of only 16 m s−1 sufficiently warmed and moistened the boundary layer (Intrieri and others, Reference Intrieri2014).
The 2016 Western Greenland case middle to lower atmospheric levels. The 850 mb geopotential heights are plotted on the left (which also includes mixing ratio values, winds and air temperatures above 0°C). The 925 mb heights, winds, air temperatures and relative humidity above 85% (filled green contour) are plotted on the right.
While the 850 and 925 mb analyses in Figure 3 show the zones of higher air temperatures and moisture content, they also depict areas of warm air advection, moisture transport and the important low-level jet. These analyses reveal how the low-level jet links to these narrow corridors of enhanced water vapor transport and unusually high air temperatures, with higher wind speeds in line with these zones. The warm air and moisture advection is clearly linked to the largely southerly flow (winds blowing from south to north) across much of southern Greenland. The height equivalent of the 850 mb level ranges from 1000 to 1500 m and from 400 to 800 m at the 925 mb level. Incorporating these levels with these case studies, one should be able to discern a low-level jet from the reanalysis.
The 2016 Greenland ROS case provides an impressive example of a corridor of high air temperatures and a narrow area of elevated moisture associated with a low-level jet. ERA5 fields for 11 April show air temperatures >5°C at 850 mb and as high as 10–15°C across southern Greenland nearer the surface at 925 mb. Figure 4 also shows precipitable water values of 12–16 mm over much of the southwest coast of Greenland, peaking in some locations at 20–24 mm on 11 April. Climatological values of precipitable water in this area on 11 April average between 2 and 7 mm from the southwest coastline and extending inland. Wind speeds reach an impressive 40 m s−1 at some locations along the southwest coastline in the low- to mid-levels of the atmosphere. The strong low-level jet, the position of moisture sources (North Atlantic) and the overall blocking setup appear to have worked in combination to produce this ROS event.
The 2016 Western Greenland case moisture variables. Integrated water vapor transport, with mean sea level pressure as black contours, is plotted on the left, and precipitable water, with similar mean sea level pressure contours, is plotted on the right. The vapor transport visualization includes magnitudes as filled contours, and vector arrows provide the direction.
An extended area of elevated values of vertically integrated water vapor transport, or IVT, is also captured in Figure 4, with maximum values between 800 and 1000 kg m−1 s−1 just off the southwest coast of Greenland. Based on previous AR studies, the presence of an IVT contour of 250 kg m−1 s−1 stretching over 2000 km – usually from a subtropical source south of (or near) 30°N latitude – typically meets the AR classification (Rutz and others, Reference Rutz, Steenburgh and Ralph2014; Ralph and others, Reference Ralph2017; Zhou and others, Reference Zhou2021). Additionally, the Ralph and others (Reference Ralph, Neiman and Rotunno2005) study shows that the combination of strong winds associated with a low-level jet (as evidenced in Fig. 3), as well as the high water vapor content and transport, generally creates an environment suitable for AR initiation. Subsequently, this represents a case where an AR made direct landfall at the location of interest. The AR likely contributed to the increased warm air advection and moisture transport necessary for this ROS event and played a role in the extreme precipitation conditions experienced in southwest Greenland. Comparing Figure 4 to Figure 2, the blocking pattern likely had an influence on this AR setup, allowing for a large enough gradient to form between the ridge and westernmost low-pressure system.
Atmospheric soundings
The sounding for 11 April at 12Z (top panel of Fig. 5) captures the high moisture content and high air temperatures in the lower atmospheric levels with the 2016 Greenland ROS event. Recall that 11 April had the highest recorded wet-snow avalanche activity for this ROS event. Note the strength of the ‘warm nose’ with this sounding. An air temperature inversion extends from the surface to about the 950 mb level, with air temperatures reaching just above 5°C between the surface and the 900 mb level. This low-level inversion is what gives it the name ‘warm nose’. The low-level jet, with winds above 26 m s−1 in the middle to low levels of the atmosphere, is one of the more pronounced meteorological features in this case. The strength of the low-level jet (with wind directions largely out of the south, southwest) likely assisted in enhanced warm air advection and moisture transport. In addition, the precipitable water value of 19.58 mm calculated from this sounding is well above average for this time of year.
Aasiaat, Greenland (north of Nuuk), during and post rain-on-snow event soundings. The top sounding shows the atmospheric conditions on 11 April 2016. The bottom sounding shows the atmospheric conditions 5 d later. The red and green lines represent air temperature and dewpoint temperature plotted with height, respectively, and winds are plotted as both barbs on the sounding's right and as a hodograph on the right-hand side of the figure.
The lower sounding panel of Figure 5 reveals how the atmospheric profile evolved in the days following the ROS event. This sounding from 16 April (5 d later) is comparatively drier, with a precipitable water of 3.99 mm. A different air mass moved into the region following the passage of a likely cold front and brought much lower temperatures and the drier conditions. Air temperatures are below freezing through the entire atmospheric profile. Also noticeable are the changes in both the wind directions and speeds. Where the low-level jet was prominent in the sounding taken on 11 April, with largely southwesterly winds, the sounding on 16 April shows light wind speeds around 10–15 m s−1 throughout the entire column and a northwesterly direction above the mid-levels.
Automatic surface observing stations
Surface observations from the Godthaab station (Fig. 6) revealed a pattern of rising air temperatures prior to the ROS event with a corresponding change to liquid precipitation. Like the Abermann and others (Reference Abermann, Eckerstorfer, Malnes and Ulf Hansen2019) analysis presented earlier, the surface air temperature increased by almost 20°C in a 2 d period from 9 to 11 April, coinciding with a liquid precipitation event. The surface station data showed the temperature rising from −2°C at 1050Z on 9 April to its highest temperature recorded during the period (17°C) at the same time on 11 April, a highly anomalous value for Greenland in April.
Godthaab (Nuuk), Greenland, surface station observations (9–17 April 2016). Temperature and dewpoint temperature are plotted in the upper panel in °C, and the corresponding precipitation types are plotted in the lower panel.
Temperatures then decreased in the following days, with air temperatures dropping below freezing beginning on 13 April. Precipitation generally transitioned to solid categories at this point, with one brief period of liquid on 14 April and intervals of mixed precipitation scattered throughout. Air temperatures then remained mostly below freezing following on 15 April. This would have allowed the previous liquid that fell with the initial storm system to freeze on or within the snowpack.
January 2021 Iqaluit, Canada, rain-on-snow event
Atmospheric reanalysis
The Arctic Rain-on-Snow Study team was made aware of ROS conditions occurring in Iqaluit, Canada, on 19 January 2021, through an eyewitness. The weaker Iqaluit ROS case exhibited some similarities with the 2016 Greenland case, as well as notable differences. As seen in the left-hand panel of Figure 7, an upper-level block was present, but it was more representative of a Rex Block or Dipole Block, as opposed to the Omega Block seen in the 2016 Greenland case. Rex Blocks form when a trough undercuts a ridge, so the synoptic pattern appears as a ridge positioned poleward over a trough.
The 2021 Iqaluit case upper atmospheric levels. The 250 mb geopotential heights are plotted on the left, and the 500 mb heights and winds are plotted on the right. The 500 mb panel also includes isotachs (lines of constant wind speed) in m s−1 in filled contours, in addition to wind barbs that indicate both direction and speed.
The right-hand panel of Figure 7 shows a strong jet streak at 500 mb (with wind speeds exceeding 55 m s−1) on the equatorward side of a deep trough extending across eastern Canada – likely assisting in maintaining the overall block by undercutting the ridge. Another weaker jet streak (with winds between 30 and 35 m s−1) is located on the western, or left, side of the ridge of high-pressure positioned over southern Greenland. This implies additional dynamics influencing precipitation over the southern tip of Baffin Island. Even though the winds in this jet streak were largely from the south, like the Greenland case, speeds were not as strong. However, it did represent another instance of the location of interest positioned directly beneath the southerly flow aloft.
Precipitation produced during the 2021 Iqaluit event, like the 2016 Greenland event, is largely dynamically driven. As with the Greenland event, a shortwave trough (Fig. 7) progressing north toward Baffin Island on 19 January provides the rising motion necessary for precipitation generation. Iqaluit lies in an inlet linked to Frobisher Bay, and the surrounding terrain lies between 500 and 1000 m. Hence, orographic lifting may have also been a contributing factor, as we also surmise for the 2016 Greenland ROS event.
Like the 2016 Greenland ROS case, the 2021 Iqaluit event is linked to narrow corridors of strong moisture transport and warm air advection. However, air temperatures with this case remained just below freezing in the middle to lower levels, and precipitable water values were lower. Figure 8 shows air temperatures at the 925 mb level (right-hand panel) in the 0 to −8°C range over Iqaluit. The only area with air temperatures above freezing at this level is over southern Greenland, which also extends to the 850 mb level (left-hand panel). Mixing ratios are also lower compared to the Greenland case. However, precipitable water values (while modest) are above average, ranging from 8 to 12 mm across the southern tip of Baffin Island (right-hand panel of Fig. 9). Climatological precipitable water values for 19 January span from 1 to 4 mm across southern Baffin Island.
The 2021 Iqaluit case middle to lower atmospheric levels. The 850 mb geopotential heights are plotted on the left (which also includes mixing ratio values, winds and air temperatures above 0°C). 925 mb heights, winds, air temperatures and relative humidity above 85% (filled green contour) are plotted on the right.
The 2021 Iqaluit case moisture variables. Integrated water vapor transport, with mean sea level pressure as black contours, is plotted on the left, and precipitable water, with similar mean sea level pressure contours, is plotted on the right. The vapor transport visualization includes magnitudes as filled contours, and vector arrows provide the direction.
Southerly winds persist throughout the troposphere over Iqaluit, like the Greenland case. The 2021 Iqaluit event also exhibits a cyclone-induced low-level jet but with weaker speeds compared to the 2016 Greenland case. Wind speeds are around 20 m s−1 in the 850 mb analysis and around 30 m s−1 at 925 mb (Fig. 8). These lower-level winds are associated with the warm sector of a cyclone centered near the northern tip of Quebec.
A comparatively large difference with this case is the AR influence. Where the direct impact of a landfalling AR was associated with the ROS event in Greenland, the Iqaluit ROS event appears to have been indirectly influenced by an AR. The same cyclone that produces precipitation in the Iqaluit region likely stripped moisture from the AR present in the North Atlantic as it rounded the broad low-pressure area over eastern Canada (left-hand panel of Fig. 9). The right-hand panel of Figure 9 shows a similar pattern in precipitable water, with higher values (up to 42 mm) reflecting the position of the AR and a narrow corridor of 12–18 mm flowing north toward Baffin Island.
Atmospheric soundings
Atmospheric soundings for the 2021 Iqaluit case also highlight significant differences compared to the Greenland case. As seen in the top panel of Figure 10, the warm layer is very much limited to the surface on 19 January with no prominent ‘warm nose’. We also saw this in the earlier 925 mb air temperatures in the reanalysis, with no areas exhibiting above freezing conditions. In an extensive study of relationships between sounding profiles and precipitation types, Rauber and others (Reference Rauber, Olthoff, Ramamurthy and Kunkel2000) found that for a majority of soundings with no warm layers – such as this example from the Iqaluit case – freezing drizzle was reported.
Iqaluit, Nunavut, during and post rain-on-snow event soundings. The top sounding shows the atmospheric conditions on 19 January 2021. The bottom sounding shows the atmospheric conditions 6 d later. The red and green lines represent air temperature and dewpoint temperature plotted with height, respectively, and winds are plotted as both barbs on the sounding's right and as a hodograph on the right-hand side of figure.
The Iqaluit sounding from 19 January yields a precipitable water value of 9.19 mm, high for the region and time of year, but modest compared to the 2016 Greenland case. The 19 January sounding also confirms a low-level jet with this case, with wind speeds of 21–26 m s−1 between 1 and 2 km above ground level. There is evidence of veering (turning clockwise with height) winds, indicating warm air advection. Direct onshore flow, with winds out of the southeast, is similar to the 2016 Greenland case and provides the additional moisture transport for precipitation.
The sounding for 25 January, 6 d after the ROS event, shows how the atmospheric profile changed when a new air mass moved in and geopotential heights lowered aloft (bottom panel of Fig. 10). Temperatures fell, especially after the passage of the cold front. The moisture profile also became much drier. This sounding's precipitable water fell to 4.28 mm (closer to climatological values) from the 9.19 mm computed from the 19 January sounding. Wind behavior also changed drastically. Wind speeds slackened and directions appeared to back to the northeast, meaning cold air advection.
Automatic surface observing stations
Another interesting aspect of this case is the lack of liquid or even mixed precipitation types recorded in the automated weather station observations at the time of ROS occurrence (Fig. 11). An eyewitness confirmed this ROS event, so this exemplifies a situation where the automated station data (usually collocated with an airport) was not representative of all regional conditions. Rauber and others (Reference Rauber, Olthoff, Ramamurthy and Kunkel2000) noted that soundings yielding deep cloud-top altitudes and no warm layer throughout the atmospheric column (top panel of Fig. 10) may produce a mix of precipitation types, like light snow, ice pellets or freezing rain. Based on data provided by the Moderate Imaging Spectroradiometer instrument aboard the National Aeronautics and Space Administration's Aqua satellite, cloud-top heights in the area were in the range of 4000 to 7000 m during a daytime overpass on 19 January 2021. Rauber and others (Reference Rauber, Olthoff, Ramamurthy and Kunkel2000) noted that some of the soundings in the category exhibiting no warm layers were associated with cloud tops above 5000 m. The authors of that paper also cautioned that surface observations and sounding data are not usually concurrent in place and time. Weather balloons tend to drift downwind from the launch site, especially in strong winds, as was likely the case here.
Surface station observations from Iqaluit (17–26 January 2021). Temperature and dewpoint temperature are plotted in the upper panel in °C, and the corresponding precipitation types are plotted in the lower panel.
This was a dynamic weather event, with a relatively strong cyclone, enhanced moisture transport and brisk winds, so it is conceivable that observations may not match actual conditions witnessed in Iqaluit. Nevertheless, the station records demonstrate the same pattern seen with the 2016 Greenland event. Surface air temperatures increased – in this case to just the freezing point on 19 January – with solid precipitation (snow) continuing. Falling air temperatures in successive days would have allowed ice to form from any liquid precipitation that accumulated on the existing snowpack. Surface air temperatures rose to the freezing point again for only a brief period on 22 January but then dropped well below freezing and remained so in the days following.
Summary and concluding thoughts
For the two ROS cases examined here, atmospheric blocking acted as a primary causal mechanism. This reaffirms the importance of blocks, as examined in other ROS cases (Voveris, Reference Voveris2022). With a block, the normal west-to-east geostrophic flow becomes disrupted – replaced by meridional flow – and provides the time for warm air masses to move in and for gradients to build for the additional moisture transport. Blocks may come in a variety of forms, and the differences between the two cases (an Omega Block for the 2016 Greenland event and a Rex Block for the 2021 Iqaluit event) likely led to the difference in strength of ROS conditions. The two sites that experienced ROS were similarly situated directly beneath the strongest southerly flow aloft, between the overall ridge of high-pressure and the westernmost trough of the block. This positioned Iqaluit and Nuuk under an upper-level jet streak, which generated additional jet dynamics for precipitation. Blocking patterns in these cases, and other cases examined by Voveris (Reference Voveris2022), developed a few days prior to the day (or days) of ROS conditions and required a few days after the event to weaken and for the geostrophic flow to return to somewhat ‘normal’ conditions.
Another key component to Arctic ROS formation was the presence of an AR and its direct or indirect influences, which reiterates findings from Voveris (Reference Voveris2022). ARs represent a significant source of water vapor transport (and associated high air temperatures) outside of the tropics. When ARs breach the higher latitudes during the winter months, they allow warm, moist conditions to overcome the typical cold, dry Arctic environment and limited solar radiation to produce ROS. We found that the 2016 Greenland ROS event resulted from a direct AR influence (with the AR making landfall), while the 2021 Iqaluit ROS event was influenced indirectly by a cyclone stripping moisture from an AR and carrying this moisture north. The direct AR landfall at Nuuk, Greenland, led to a more pronounced ROS event, and the lack of a direct AR landfall at Iqaluit contributed to lesser ROS-related impacts.
We also revealed the importance of smaller scale atmospheric features, like cyclone-induced low-level jets and ‘warm noses’ associated with increased moisture transport. Soundings showed how the 2016 Greenland case – directly influenced by an AR – exhibited a strong low-level jet, a deep ‘warm nose’ layer, and a mostly saturated atmospheric column. Comparatively, the 2021 Iqaluit case was impacted by the same features but to a lesser extent. Moisture and air temperatures were lower, no ‘warm nose’ was present, and the low-level jet was weaker, a possible consequence of only the indirect influence of an AR. These combined features were enough to adequately warm and moisten the boundary layer, leading to the report of ROS in Iqaluit, but were not enough to produce a strong ROS event as seen at Nuuk, Greenland.
To our knowledge, our study is the first to show how atmospheric blocking, ARs and other mesoscale features, such as low-level jets and subsequent ‘warm noses’, work in tandem to produce Arctic ROS events of differing magnitudes. In a recent study by Serreze and others (Reference Serreze2022), the key meteorological features discussed here, and noted by Voveris (Reference Voveris2022) for other ROS events, can also be associated with Arctic precipitation events of extreme magnitude (either snow- or rain-driven). The interplay between these meteorological drivers is complex and will vary on a case-by-case basis, but some connections have been established between these features themselves and the influence they have on precipitation generation. For example, Ralph and others (Reference Ralph, Neiman and Rotunno2005) note that low-level jets may coincide with AR development when combined with high water vapor content and transport. A study from Benedict and others (Reference Benedict, Clement and Medeiros2019) demonstrates how atmospheric blocking slows the normal progression of shortwave systems, shifting the storm track equatorward, while the high-pressure ridge (resulting from the blocking and developing 7–10 days prior to an AR) directs more systems toward the study area. This leads to higher chances of both AR incidence and extreme precipitation events (Benedict and others, Reference Benedict, Clement and Medeiros2019).
Having established these meteorological links to Arctic ROS events, we can infer how future global warming might influence ROS occurrence and intensity. Uncertainties remain regarding how climate change will affect atmospheric blocking. Some studies argue that blocking patterns might decrease in frequency in the middle latitudes as the climate warms or that areas that experience a climatologically high number of blocking episodes may see a shift in those areas (Woollings and others, Reference Woollings2018). However, Woollings and others (Reference Woollings2018) caution that the ability of climate models to handle blocks remains unclear and that natural variability is likely to have a strong influence on blocking patterns in coming decades. They also add that the effect of wintertime blocking on air temperatures is dependent upon thermal advection – a process expected to weaken in a warming world – but the effect of summertime blocking on air temperatures may strengthen from feedbacks resulting from changes in soil moisture.
Results from climate model studies are in general agreement that future warming will coincide with increased poleward moisture transport and elevated Arctic precipitation with a transition to a more rain-dominated climate (Lenaerts and others, Reference Lenaerts, Camron, Wyburn-Powell and Kay2020; McCrystall and others, Reference McCrystall, Stroeve, Serreze, Forbes and Screen2021; Niwano and others, Reference Niwano2021). More rainfall implies more ROS events. However, a shorter snow cover season may lead to fewer ROS events, especially during the autumn or spring months, or to more incidents of rain falling on frozen ground instead. In addition, Espinoza and others (Reference Espinoza, Waliser, Guan, Lavers and Ralph2018) found that climate models project a 10% decrease in the number of ARs, based on the ‘worst-case’ global emissions scenario from the Intergovernmental Panel on Climate Change. However, they note that models also project ARs to be 25% longer, 25% wider and have larger IVT values, likely due to the increasing moisture available in a warming atmosphere. This implies that more ARs may potentially reach high Arctic regions, and stronger IVT may lead to more extreme precipitation events coinciding with ROS conditions.
Acknowledgements
This study was supported by the National Science Foundation Grant ICER 1928230 ‘NNA Track 1: A Systematic Pan-Arctic Analysis of Rain on Snow and Extreme Precipitation Events and their Impacts on Human-Environment Systems’. We thank Mistia Zuckerman and Tori McLeod for their time and effort in the final editing and submission of this manuscript. We also appreciate the ECMWF and UCAR for their data and MetPy programming codes (respectively) used in this research. We accessed ERA5 data from the Copernicus Climate Data Store API [Application Program Interface], maintained by the ECMWF.
Open access
