The severe 2013/14 winter (December–March) in the Midwest was dominated by a persistent atmospheric circulation pattern anchored to a North Pacific Ocean that was much warmer than average. Strong teleconnection magnitudes of the eastern Pacific oscillation (−EPO), tropical Northern Hemisphere pattern (+TNH), and second-lowest Hudson Bay 500-hPa geopotential height field are indicators that led to severe winter weather across the eastern United States. Unlike in previous cold and snowy midwestern winters, Atlantic Ocean blocking played little to no role in the winter of 2013/14. The primary atmospheric feature of the 2013/14 winter was the 500-hPa high pressure anchored over the North Pacific in response to the extremely warm sea surface temperature anomalies observed of +3.7 standard deviations. Only one other severe midwestern winter (1983/84) since 1950 featured a similar Pacific blocking. The accumulated winter season severity index, which is a metric that combines daily snowfall, snow depth, and temperature data for the winter season, was used to compare the 2013/14 winter with past winters since 1950. Detroit, Michigan, and Duluth, Minnesota, experienced their worst winter of the 55-yr period. Seven climate divisions in northern Illinois, eastern Iowa, and parts of Wisconsin experienced record-cold mean temperatures. These climate conditions were associated with a number of impacts, including a disruption to the U.S. economy, the second-highest ice coverage of the Great Lakes since 1973, and a flight-cancellation rate that had not been seen in 25 years.
The 2013/14 winter in the Midwest, which was characterized by frequent snowstorms and outbreaks of arctic air, brought back memories to long-term residents of winters experienced in the late 1970s. The three consecutive winters from 1976/77 through 1978/79 featured record-breaking weather conditions (Diaz and Quayle 1980) such as severe cold-air outbreaks, record numbers of snowstorms, and a wide range of impacts on society (Changnon and Changnon 1978a,b; Changnon 1979; Changnon et al. 1980). Documented impacts suggest that the 2013/14 winter was severe. Economists reported that the unusually cold and snowy weather conditions experienced in early 2014 negatively impacted the economy by “disrupting production, construction, and shipments, and deterring home and auto sales” (Sharf 2014). Similar to the situation during the winter of 1978/79, more than 90% of the Great Lakes were ice covered by mid-February (Great Lakes Environmental Research Laboratory “Historical Ice Cover”; http://www.glerl.noaa.gov/data/ice/), which increased a typical 3-day commercial shipment through the Great Lakes to 9 days (Guarino 2014). By mid-February of 2014, approximately 5.5% of all U.S. flights had been canceled as a result of conditions of extreme cold or snowfall, which was the highest rate in 25 years (Quirk 2014). By March, the city of Chicago had spent $2.8 million on filling approximately 240 000 potholes. The continuous subfreezing air temperatures experienced in the 2013/14 winter caused soil temperatures to drop below freezing to depths of 1.2–1.5 m in the Midwest (data obtained from http://www.ncdc.noaa.gov/data-access/land-based-station-data/land-based-datasets/cooperative-observer-network-coop). These conditions caused water in underground pipes and water mains to freeze, leaving entire subdivisions without water for periods from days to weeks (see, e.g., La Salle, Illinois: Collins and Abbey 2014; Stella 2014). Communities such as Moline, Illinois, and Des Moines, Iowa, experienced their all-time highest number of broken water mains because of the severe cold.
The goal of this research was to describe and explain the 2013/14 midwestern winter. The two primary objectives were to document the climatological extremes, putting them into a historical perspective, and to identify and discuss the meteorological features and teleconnections that were associated with those climatological conditions.
a. Climatological assessment
Surface climatological data for the period from 1950 through 2014 were obtained from the Midwestern Regional Climate Center’s Application Tools Environment, known as “cli-MATE” (http://mrcc.isws.illinois.edu/CLIMATE/), to characterize the 2013/14 winter and place it in historical perspective. Daily and monthly temperature, precipitation, snowfall, and snow-depth information was obtained from National Weather Service first-order and cooperative stations across the Midwest for the period from December 2013 through March 2014. Station and climate-division (CD) averages and anomalies were determined and mapped using cli-MATE and the “Climate at a Glance” tools from the National Centers for Environmental Information (http://www.ncdc.noaa.gov/cag/).
Determining the appropriate climate conditions that create a severe winter was challenging. A winter at a location may have much-below-normal temperatures but below-normal snowfall. Is this situation more severe than a winter with temperatures that are slightly below normal and snowfall that is much above normal? The severity of a winter is related to the intensity and persistence of cold weather, the frequency and amount of snow, and the amount and persistence of snow on the ground. To evaluate the severity of the 2013/14 winter, the accumulated winter season severity index (AWSSI; Boustead et al. 2015) was used. The AWSSI uses commonly available daily climatic data—maximum and minimum temperature, snowfall, and snow depth. It does not account for wind or freezing rain. The advantages of such an index are that it allows for the creation of a historical database for any location with daily temperature, snow, and snow-depth data and allows for comparisons of season-to-season severity at one location in the context of the climatological mean of that location or between locations. This study utilized available AWSSI calculations during the 1950–2014 period for 12 midwestern locations to assess the severity of the 2013/14 winter.
The AWSSI was used to assign a score to each daily maximum temperature, minimum temperature, snowfall, and snow depth (Table 1). The scores were summed throughout the winter, which resulted in a score that integrated the temperature and snow character of the winter as well as its length (Boustead et al. 2015). At any location, the winter season began when the first of any one of the following instances occurred: first measurable snowfall (≥0.254 cm), maximum temperature at or below 0°C, or 1 December (i.e., latest start to season). The winter season ended at the last occurrence of any of the following: last measurable snowfall (≥0.254 cm), last day with 2.54 cm of snow on the ground, last day with a maximum temperature of 0°C or lower, or on the date 28 or 29 February (i.e., earliest conclusion to season). At a specific location, the final winter score was then compared with other winter scores and ranked.
b. Meteorological conditions examined
This study examined a number of atmospheric variables, sea surface temperature (SST), and teleconnections (TC) in an effort to explain the climate conditions associated with the severe 2013/14 midwestern winter. Surface weather features such as winter cyclone tracks were classified and counted using daily weather maps (obtained at http://www.esrl.noaa.gov/psd/data/composites/day/). Monthly upper-atmospheric height anomalies for 500-hPa pressure fields were examined across the Northern Hemisphere.
Northern Hemispheric modes were related to the North American atmospheric pattern that continuously affected the Midwest. Atmospheric pattern descriptions were defined in terms of TCs and custom bounded geographical boxes of specific weather variables that were calculated with NCEP–NCAR reanalysis data (Kalnay et al. 1996). Modern TC indices (Wallace and Gutzler 1981; Mo and Livezey 1986; Barnston and Livezey 1987; Thompson and Wallace 2000), such as the tropical Northern Hemisphere pattern (TNH) and the North Atlantic oscillation (NAO), were defined by the Climate Prediction Center (CPC) and used to analyze their influence on North American winters. These TC datasets (i.e., NAO and TNH) are numerical time series that were standardized for the most recent climatological 30-yr base period. The polarity and magnitude of TCs are indicative of the type of atmospheric blocking or weather pattern that a certain region is likely to incur. Loading patterns (500 hPa) associated with these specific modes are shown in Fig. 1.
The eastern Pacific oscillation (EPO) index was developed to determine the mean monthly and December–March (DJFM) seasonal time series using two different specific coordinate boxes (20°–35°N, 160°–125°W minus 55°–65°N, 160°–125°W) of the area-weighted mean 500-hPa geopotential height fields. The EPO index was created using data from the NCEP–NCAR reanalysis (Kalnay et al. 1996), was calculated on the basis of an EPO formula from the NOAA/Earth System Research Laboratory (http://www.esrl.noaa.gov/psd/data/timeseries/daily/EPO/), and was standardized for climatological studies (30-yr mean). The “east Pacific–North Pacific” TC, derived from the CPC, was not used because the time series does not include December values.
Four indices were calculated using area-weighted weather variables bounded by a coordinate box (Fig. 2) and averaged spatially into a monthly and seasonal time series for evaluation purposes. The North Pacific SST pool (SST1) dataset was defined by the November mean SST area within the bounds of 35°–50°N and 180°–130°W. The DJFM North Pacific SST pool (SST2) was defined by the same procedure as was used for the SST1 dataset but the seasonal mean of DJFM was accumulated. The next two datasets involved DJFM mean 500-hPa height values across different areas. The North Pacific heights (NPHS) dataset was bounded by 55°–65°N and 160°–125°W, and the Hudson Bay heights (HBAY) dataset was bounded by 55°–65°N and 95°–75°W. NCEP–NCAR reanalysis information was acquired to calculate the EPO and the four bounded indices (i.e., HBAY, NPHS, SST1, and SST2) for the period from January 1950 to March 2014 (Kalnay et al. 1996). These were used to describe how the 2013/14 winter evolved. For the period of 1950–2014, the magnitudes of TC indices were compared with severe winters as defined by the AWSSI scores for the Midwest.
3. 2013/14 climatological conditions
a. Climate summary
Temperature rankings were determined using CD data for the period from December 2013 to March 2014 for the contiguous 48 United States (Fig. 3). The extent of the cold during the winter of 2013/14 was sizable, with much of the central and eastern United States experiencing below-normal and/or record-below-normal temperatures. The core area of the “much below normal” (i.e., CDs experiencing one of their 10 coldest winters) and record-cold region was over the central United States from the Midwest down to the lower Mississippi River valley. At the state level for the DJFM period, both Michigan and Wisconsin experienced their third coldest period ever, Illinois and Indiana had their fourth coldest, Missouri had its fifth coldest, Iowa and Minnesota had their sixth coldest, and Ohio experienced its ninth coldest period ever.
A closer examination of CD temperature anomalies for the Midwest (Fig. 4) shows the magnitude of the winter temperature departures. Areas in eastern Minnesota, much of Wisconsin, eastern Iowa, northern Illinois, and northwestern Indiana were 3.9°–4.5°C below normal for the DJFM period. The rest of the Midwest was 1.6°–3.9°C below normal for the period. Seven CDs in the Midwest experienced their coldest DJFM dating back to 1895.
The people of the Midwest had not experienced such a cold winter in over a generation (Fig. 4). The three winters in the late 1970s were the defining winter experience in the past 40 years. The winter of 2013/14 was similar in respect to temperature. The DJFM midwestern temperature for 1976/77 was −4.9°C (23.2°F), for 1977/78 it was −6.6°C (20.1°F), and for 1978/79 it was −6.3° (20.7°F), as compared with 2013/14 when it was −6.3° (20.7°F).
The other factor that made the 2013/14 winter stand out was the above-normal snowfall accumulation throughout the Midwest. Snowfall for DJFM period (Fig. 5) was 100–150 cm (40–60 in.) above normal in northeastern Illinois, northern Indiana, northwestern Ohio, and western Michigan. Large portions of the rest of the Midwest were 25–100 cm (10–40 in.) above normal for snowfall. This resulted in a wide band of 2–3 times the normal snowfall stretching from Missouri, through Illinois, Indiana, and Ohio as well as northern Kentucky and southern Michigan.
b. Historical climate perspective of the 2013/14 winter using the AWSSI
The coldest weather in the contiguous 48 United States, relative to normal, encompassed an area extending from the northeastern three-quarters of Minnesota through Wisconsin, the Michigan Upper Peninsula, and northern Illinois (Fig. 3). Snowfall departures, in terms of percent of normal, were greatest in the southern half of the Midwest, the Ohio valley, and far northern Minnesota (Fig. 5).
The AWSSI tracks daily climate variables throughout the winter season, in effect accounting for extremes; runs of days with temperatures, snowfall, and snow depth at different values; and the length of the season. The winter score for any selected winter could be compared with the median (i.e., 1950–2014) winter score for a location to determine the relative severity of a winter. Therefore, the higher the AWSSI score is, the colder and snowier are the winter. In a specific year, the raw AWSSI scores for a number of stations can be used to understand winter severity across the region. The median AWSSI (1950–2014) score for Duluth, Minnesota, is 2000.0, whereas the median for Chicago, Illinois, is 543 (Table 2), indicating that the average winter is much colder, snowier, and likely longer in Duluth than in Chicago. To determine the relative severity of the seasonal AWSSI, along with the median, the 25th and 75th AWSSI percentiles were calculated for each location (i.e., they can be used to compare AWSSI values between locations).
The AWSSI 1950–2014 median, 25th-percentile values, and 75th-percentile values and the 2013/14 AWSSI values were determined for 12 selected midwestern locations (Table 2). The rank of the AWSSI (Table 2) with respect to all winters for the period 1950–2014 indicated that 2013/14 was the most severe winter for both Detroit (Michigan) and Duluth (tied with 1995/96 for Duluth). The AWSSI score for Detroit was 1277, which is more than 500 above the 75th-percentile AWSSI score, exceeding the score of 1046 set in the winter of 1977/1978. For all 12 locations, the winter of 2013/14 was one of the 10 most severe since 1950, with eight cities ranked third most severe or higher.
4. Meteorological assessment of winter 2013/14
a. North American 2013/14 winter characteristics
The primary atmospheric feature attributed to the severe winter of 2013/14 involved a prolonged period of atmospheric blocking that was associated with the polar jet stream over Alaska and northwestern Canada. As a result, a persistent trough developed downstream over central North America (Fig. 6). This trough became an anchor for the 500-hPa-level low pressure vortex over Hudson Bay into southeastern Canada. Hudson Bay experienced its second-lowest average winter 500-hPa-level geopotential height (5069 gpm) within the NCEP–NCAR reanalysis dataset (Kalnay et al. 1996). Arctic air masses associated with this trough created the cold conditions experienced in the upper plains, Midwest, and many areas generally east of the Mississippi River (Fig. 3). The air masses were not modified much as they traversed from Siberia, the Arctic, and Canada because of widespread and persistent snow cover over much of Canada and the Midwest in 2013/14.
The frequent incursion of unusually cold air masses into the Midwest was often associated with the occurrence of surface “Alberta clipper” systems. Clipper events occurred more frequently over the Midwest than did any other storm track (Table 3), which was notably different from what typically occurred during the severe winters of the late 1970s (Changnon et al. 1980) and what occurs in an average winter (Changnon 1969). Weak clipper systems rotated cyclonically around the semipermanent Hudson Bay low pressure feature, frequently providing light-to-moderate midwestern snowfall amounts of generally less than 15.2 cm (6 in.). Farther south over the mid-Mississippi River basin, west-to-east-moving extratropical cyclones also brought snow events across eastern portions of the Midwest. Although the Midwest experienced fewer heavy-snow-producing cyclone tracks (i.e., Colorado lows and Gulf lows), many areas still received above-average snowfall (Fig. 5) because of the increased frequency of Alberta clippers.
b. The PNA, North Pacific anomaly, and other teleconnections
A positive temperature anomaly developed over the NPHS SST pool during November of 2013 with an SST1 index value of +2.7 standard deviations σ. A warm SST pool over the North Pacific is a common occurrence when the Pacific decadal oscillation is in a long-term negative phase (Mantua et al. 1997) but is far from common when compared with the other episodes of extremely positive SST anomalies. The SST2 index measured at +3.7σ above the climatological mean for DJFM 2013/14 (Fig. 7). The SST1 domain experienced near-record warmth in November of 2013, and SST2 experienced record warmth in DJFM 2013/14.
How did the warm North Pacific SST anomaly affect the 2013/14 winter over North America? Oceanic SSTs covering the prescribed North Pacific area can force circulation patterns that lead to changes within the height-field distribution pattern over the Northern Hemisphere through bottom-to-top (ocean to atmosphere) processes (Namias 1969, 1970; Kushnir and Lau 1992; Hartmann 2015). North Pacific SSTs are able to modulate the Pacific–North American (PNA) TC pattern and PNA-like pattern variants (Frankignoul and Sennéchael 2007). Surface heat flux was the main driver between SSTs and geopotential height fields, although in some instances atmosphere–ocean feedback (cloud cover, evaporation, radiation, etc.) can occur to balance the heat flux budget (Frankignoul et al. 2004). The SST2 anomaly was likely the primary reason for the persistent blocking ridge over the Bering Sea, Alaska, and northwestern Canada (NPHS anomaly) during the 2013/14 winter. The SST2 anomaly acted as a trigger to displace the polar vortex over central and southern parts of Canada (Hartmann 2015). The mean NPHS (Fig. 2) geopotential height for DJFM 2013/14 was observed at 5355 gpm, the third highest in the NCEP–NCAR reanalysis dataset (Kalnay et al. 1996), exceeding all other DJFM seasons within the dataset record from 1948 to March of 2014 except for two seasons.
Three other teleconnections were examined for their relationship to the winter of 2013/14. The TNH pattern affects central U.S. winter temperature and storm track because its primary 500-hPa anomaly is located slightly south of Hudson Bay, as seen in Fig. 1 (Rodionov and Assel 2000). The physical importance of the TNH on midwestern winter climate conditions is related to its spatial locality, impact on regional weather patterns, and relationship to winter Great Lakes ice cover (Assel and Rodionov 1998; Rodionov and Assel 2000). The NAO and EPO can aid in diagnosing the relationship between Atlantic and Pacific blocking and their influence on midwestern temperature and precipitation patterns (Renken et al. 2014).
The TNH teleconnection was +1.1σ from December through February of 2013/14 (CPC does not calculate the March TNH). Symptomatic of past severe winters that experienced a warm North Pacific state, the positive TNH contributed to a strong Hudson Bay 500-hPa low pressure system. The winter of 2013/14 was no different, recording the second-strongest winter HBAY anomaly in the NCEP–NCAR reanalysis record (Kalnay et al. 1996). A past study has also shown that positive TNH values correlate positively with Great Lakes ice cover (Assel and Rodionov 1998) and regional snow cover over the central United States. This feature was also related to the strong midlevel subsidence from California to the Gulf of Alaska. The positive TNH affected the interior by supporting cold-air outbreaks and securing the persistent western U.S. high pressure feature that helped to establish extreme drought over California (Wang et al. 2014).
The EPO can gauge North Pacific or Alaskan blocking regimes but can also be an indicator relating to the persistence and strength of the Hudson Bay low. Walsh et al. (2001) described the origins of cold air masses that moved into the Midwest and other regions across the United States and Europe. The origins of these cold air masses were delineated very well by the type of blocking pattern imposed upon Alaska and northwestern Canada. Although Walsh and his colleagues did not discuss the EPO, it is evident by the sea level pressure anomaly in Fig. 7 of Walsh et al. (2001) that the EPO spatial component exists. The winter of 2013/14 featured a November EPO at −1.4σ and a December–March mean EPO value measured at −1.4σ (the sixth lowest since 1950).
c. Winter 2013/14 comparisons with past severe winters
The severe midwestern winters in the late 1970s were associated with Greenland-based blocking. The negative NAO during the winters from 1976 to 1979 dominated the planetary circulation; this circulation pattern is associated with midlatitude troughing over the eastern contiguous United States (Table 4). During some winters (e.g., 1983/84 and 2013/14), the primary blocking mechanism was Pacific based and not Atlantic based. The 2013/14 winter was dominated by Pacific-based blocking associated with a negative EPO, as opposed to the late 1970s when high-latitude blocking was closely affiliated with Atlantic-based circulation patterns (Shabbar et al. 2001).
A classification system for blocking that is denoted by the negative EPO, negative NAO, or a combination of both was created to categorize the Midwest’s most severe winters, which were determined by midwestern AWSSI counts (Table 4). AWSSI counts were determined using the top five most severe winters for each of the 12 selected midwestern cities (Table 2). The winters that experienced the greatest frequency (counts of two or more cities) out of these 60 samples (i.e., top five winters for the 12 selected cities) were chosen as the most severe midwestern AWSSI winters (Table 4). Nine winters met the criteria for regional AWSSI status. The three winters in the late 1970s and 2013/14 ranked as the four top severe Midwest winters. Table 4 provides the regional AWSSI counts, rankings, and associated winters (defined by DJFM) with Atlantic, Pacific, or both blocking types. The winter of 2013/14 was interesting because of the degree of extreme cold and snow that occurred with only negative EPO being the dominant blocking. Only two of these severe winters experienced a positive NAO: 1983/84 and 2013/14.
The winter of 2013/14 was the worst in the Midwest since a series of severe winters in the late 1970s, and the resulting impacts affected nearly all weather-related sectors. The climatological and meteorological characteristics of the 2013/14 winter were evaluated, described, and put into historical context.
Average winter (December 2013–March 2014) temperatures ranked at or near the bottom (i.e., coldest) for many areas in the Midwest. The cold temperatures were persistent throughout the winter. This continuous period of below-average temperatures experienced across the center of the United States was related to two anomalous atmospheric features: 1) record warm North Pacific SST forcing near record-high 500-hPa heights over Alaska and 2) the second-lowest midlevel pressure over Hudson Bay.
The 2013/14 winter circulation was primarily driven by North Pacific blocking created by a midlevel pressure anomaly centered over Alaska. Previous extreme winters were associated with North Atlantic blocking or North Atlantic blocking in tandem with North Pacific blocking. The severe winters of 1983/84 and 2013/14 experienced +TNH and −EPO values that occur when only North Pacific blocking dominates. These teleconnections should be monitored in future winters to assist with long-range forecasting and to explain seasonal climate conditions.
Areas in the eastern Great Lakes (e.g., Detroit) experienced record winter snowfall amounts, while other areas in the Midwest experienced above-average snowfall (Fig. 5). Frequent snowfall occurred in association with ever-present Alberta clippers. This type of surface cyclone track was directly linked to the dominant upper-level pattern, a large trough located from Hudson Bay over the central and eastern United States, and produced light-to-moderate snowfalls and an ever-increasing snow depth into the middle of February of 2014.
The combination of severe cold and snowy conditions led to a wide variety of impacts. The U.S. economy was negatively affected because of disruptions in production, construction, and shipments and reductions in home and auto sales. Many workers lost income because of weather-related shutdowns, individuals who own and maintain homes and automobiles suffered losses, and schoolchildren experienced multiple days off because of the dangerously cold conditions. The transportation sector (e.g., travel by ship, train, airplane, or automobile) was severely affected by the severe cold and frequent snowfalls. An important consequence was that the winter provided another indicator that much of our municipal infrastructure (e.g., water supply systems) was rapidly aging and fragile. Although they are infrequent in occurrence, understanding the physical conditions and related impact information associated with severe winters is important as we continue to assess and manage weather-related risks.
The research group appreciates the support provided by the Midwestern Regional Climate Center/Illinois State Water Survey. We specifically thank Zoe Zaloudek for creating maps related to the research, Tyler Smith for compiling station AWSSI data, David Kristovich for providing comments, and Lisa Sheppard for editing the manuscript. We appreciate the comments and suggestions from the anonymous reviewers. The views expressed herein are those of the authors and do not necessarily reflect those of the Illinois State Water Survey.