1. Introduction
The Laurentian Great Lakes are major contributors to regional snowfall through surface heat and moisture fluxes leading to the development of snow within convective liquid and ice clouds (Chang and Braham 1991; Niziol et al. 1995; Kristovich et al. 2003; Barthold and Kristovich 2011). Lake-effect processes can contribute to more than a doubling of snowfall in locations near and downwind of the lakes, relative to regional locations not influenced by the lakes (Braham and Dungey 1984; Kelly 1986; Norton and Bolsenga 1993). The increased snowfall elevates snow-removal costs, increases traffic accidents, reduces retail sales, inflicts severe property damage to homes, disrupts air travel, and raises the rates of injuries and fatalities (Schmidlin 1993; Kunkel et al. 2002). The additional snowfall can benefit sectors of the local economy such as local ski resorts, private snow-removal businesses, and winter-related product sales (Schmidlin 1993; Kunkel et al. 2002), however.
Several studies (Braham and Dungey 1984; Norton and Bolsenga 1993; Burnett et al. 2003; Ellis and Johnson 2004; Kunkel et al. 2009a) found that lake-effect snowfall increased in various locations in the Great Lakes region during the twentieth century. Table 1 qualitatively describes the direction of lake-effect trends determined by previous studies. Braham and Dungey (1984) found wintertime snowfall increased from the 1930s to the late 1970s within the Lake Michigan lake-effect snowbelt but remained nearly constant to the west of Lake Michigan and east of the snowbelt. Similarly, Norton and Bolsenga (1993) found that areas west of Lake Michigan experienced little trend between 1951 and 1980, but increases were noted for the central Great Lakes basin and the lake-effect snow region east of Lake Ontario. Burnett et al. (2003) noted similar patterns near Lake Ontario and pointed out the potential role of whole-lake thermal characteristics, including warmer surface water temperatures and decreased ice cover. In contrast, Grover and Sousounis (2002) found no change in lake-effect snowfall near the western Great Lakes and a possible decreasing trend farther east. Their study may not fully represent lake-contribution snowfall, however, because their analysis focused on large-scale weather systems during autumn months, prior to the peak lake-effect snowfall season. Kunkel et al. (2009a) found an overall increase in snowfall within the Lake Michigan–Huron snowbelt that is approximately one-half of the rate determined by previous studies. They evaluated trends using a dataset of expert-assessed, quality-controlled, temporally homogeneous daily snowfall observations (Kunkel et al. 2009b)—the best-quality data available for their study.
Summary of the direction of twentieth-century lake-effect snowfall trends in the Laurentian Great Lakes region found by previous studies. Braham and Dungey (1984) exclusively focused on total snowfall trends near Lake Michigan, whereas Burnett et al. (2003) examined lake-effect snowfall near Lakes Erie and Ontario. Norton and Bolsenga (1993) and Kunkel et al. (2009a) looked at total snowfall trends near all Great Lakes separately, including Lake Michigan. Ellis and Johnson (2004) examined snowfall trends at sites near the western lakes but placed emphasis on eastern lakes. Grover and Sousounis (2002) took a synoptic approach to look at lake-effect trends for the Great Lakes region but used a single site near Lake Michigan (Grand Rapids, MI).
This study investigates trends in total lake contribution to local snowfall, herein referred to as lake-contribution snowfall (LCS). This nomenclature emphasizes the inclusion of both lake-effect snow and lake enhancement of snowfall accompanying synoptic-scale events, such as midlatitude cyclones. This study updates findings from previous studies using sites with quality-controlled, temporally homogeneous daily snowfall observations identified by Kunkel et al. (2009a,b). The goal of this study is to investigate trends and variability in Lake Michigan’s contribution to snowfall.
2. Method
It is important to screen snowfall time series for inconsistencies and gaps that can affect long-term trends. Station moves, observer changes, measurement practice inconsistencies, and exposure changes are among many sources that affect the long-term continuity of a snowfall record (Kunkel et al. 2007). Kunkel et al. (2009b) assessed the temporal homogeneity of daily snowfall observations from stations participating in the National Weather Service’s Cooperative Observer Program (COOP) in the contiguous United States. Using statistical analyses and expert assessment, they developed a dataset of daily snowfall observations that is appropriate for long-term trend analysis.
We sought to estimate LCS for three locations in the Lake Michigan snowbelt in western lower Michigan (Fig. 1). For a given location, this estimate is made by comparing the snowfall observed at that site with snowfall observed outside of the snowbelt, along an approximately east–west transect. Latitude variation was minimized in selecting sites along each transect to curtail the influences of latitudinal variability in non-lake-effect wintertime precipitation (Groisman and Easterling 1994; Changnon et al. 2006). Two estimates of lake contribution were determined for each of the three snowbelt sites. First, LCSu was determined by subtracting 5-yr snowfall totals at upwind sites from corresponding snowbelt sites along the transects. Likewise, LCSd was determined as the difference between total 5-yr snowfall amounts at snowbelt sites and corresponding downwind sites along the transects.
This figure is adapted from Scott and Huff (1996); the thin contours indicate the amount of snowfall in millimeters added by the presence of the Laurentian Great Lakes. The thick contour is the 80-km boundary used by Scott and Huff to represent the extent of the lakes’ meteorological influence on the region. Superimposed on this map are the snowfall sites chosen for the study presented here, indicated by shaded circles. Within the circles, numbers represent each transect, labeled 1–3 from north to south, and letters indicate whether the site is upwind (U), near the lake (L), or far downwind (D). Sites used in this study are Marshfield, Wisconsin (1U), Wellston, Michigan (1L), West Branch, Michigan (1D), Richland Center, Wisconsin (2U), South Haven, Michigan (2L), Adrian, Michigan (2D), Morrison, Illinois (3U), Valparaiso, Indiana (3L), and Hoytville, Ohio (3D). Temperature data for Wellston were replaced by those from Manistee, Michigan (star).
Citation: Journal of Applied Meteorology and Climatology 51, 11; 10.1175/JAMC-D-12-064.1
Figure 1 gives the transects and sites used in estimating LCS. Each transect includes a site west of Lake Michigan, which is typically upwind of Lake Michigan during lake-effect events (labeled U), one site in the lake-effect snowbelt near the eastern or southern coastline of the lake (L), and one site farther east downwind of the lake (D). Both the upwind and downwind sites shared the attribute of being located at least 80 km from the lakeshore, outside of the lake-effect snowbelts as defined by Scott and Huff (1996). Observations from eight of the sites were identified by Kunkel et al. (2009b) as being temporally homogeneous, the best-quality data for this study. A ninth site, Valparaiso, Indiana (site 3L), was assessed by Kunkel et al. (2009b) as having questionable temporal homogeneity. Snowfall in that region is very sensitive to variations in the occurrence and intensity of north–south-oriented convective bands (e.g., Fig. 21.3 in Kristovich 2009). The apparent lack of temporal homogeneity at the Valparaiso site may be a reflection of such variability (K. Kunkel 2010, personal communication).
Note that for short time scales synoptic-scale snowbands affecting a COOP site could make the calculated LCS incorrect. There is no evidence in the literature that any of the nine COOP sites used in this study would be affected by such bands more often than the other sites, however. Therefore, cyclone-related snowbands are thought to have little influence on the climatological analysis provided here.
Accounting for invalid and missing observations is crucial to the analysis of accumulated snowfall. Daily snowfall observations identified by Kunkel et al. (2009b) as being invalid or missing were removed from the analysis for that site. Remaining daily data were summed into monthly totals, and any months missing more than 3 days of data (~10%) were removed from the analysis dataset. For analysis of lake contributions, it is essential to compare snowfall observations taken coincidentally at sites within each transect. Therefore, if monthly data were missing or removed at any of the three sites in each transect, data from all three sites were removed. Five-year total snowfall was determined by summing monthly snow totals over five snow seasons, October–May, during 1920–2005.
Removal of monthly snowfall observations could have large influences on calculated trends and variations in LCS, particularly if the removed months are during the peak of the lake-effect snow season. Our study utilizes snowfall observations from October through May, the months during which nearly all lake-effect clouds are observed (Kristovich and Steve 1995; Rodriguez et al. 2007). In the Lake Michigan region, lake-effect and lake-enhanced cloud and snow events are most common during December–February (Norton and Bolsenga 1993; Kristovich and Steve 1995; Changnon et al. 2006; Rodriguez et al. 2007), referred to in this study as peak months. To assess the potential impacts of the removal of peak months, the number of missing peak months was determined for each 5-yr period (Table 2). The potential relationship between LCS and removed peak months was examined for each transect (not shown). As expected, years with large numbers of missing peak months tended to have lower LCS values than those with few missing peak months. For years with fewer than five missing peak months, no relationship with LCS was found, however. This finding is taken into account in the following discussions.
The number of missing peak lake-contribution months (December–February) during each 5-yr period for transects 1–3. The maximum possible number of missing peak months is 15 per period. The boldface font indicates ≥5 missing peak months from each period. Periods with zero missing peak months are blank. No analyses were performed during the 1920s for transect 3 because of a lack of data at Hoytville.
3. Results
Lake Michigan’s contribution to snowfall was estimated by using quality-controlled snowfall observations from nine COOP sites (Kunkel et al. 2009b) near the lake. The lake’s contribution is defined as the surplus of snowfall at near-lake sites (Fig. 1: 1L, 2L, 3L) relative to sites (1U, 1D; 2U, 2D; 3U, 3D) outside of Lake Michigan’s snowbelt (Scott and Huff 1996). Figure 2 shows time series of 5-yr total LCSu for each of the three transects. The time series for each transect exhibits substantial temporal variability as well as multidecadal trends. The qualitative linear trends for the entire time series for transects 1 and 2 using 5-yr periods with less than five missing peak months (1930–2005 and 1920–80, respectively; long-term trend lines are not shown) show increases in LCSu, in agreement with the upward trend reported in previous studies. LCSu for transect 3 shows no obvious increasing or decreasing linear trend over the entire time series (1930–2005), however. Similarly, long-term LCSd trends (Fig. 3; long-term trend lines are not shown) for transects 1 and 2 exhibit increases through 1930–2005 and 1920–80, respectively. Transect 3 shows no obvious increasing or decreasing LCSd trend during 1930–2005.
Time series bar graphs of the estimates of 5-yr lake-contribution snowfall (LCSu) determined by subtracting the total snowfall at each upwind site from the total snowfall at their corresponding lake-effect site for (a) transect 1, (b) transect 2, and (c) transect 3 across Lake Michigan. The graphs are aligned north to south. Columns with solid outlines and shading indicate LCSu estimates with <5 missing peak lake-effect months, and columns with dotted outlines without shading indicate LCS estimates with ≥5 missing peak months. See Table 2 for the number of missing peak lake-effect months for each 5-yr period. Dashed lines indicate LCS trends via linear regression. Only the periods with <5 missing peak months were included in the trend calculations.
Citation: Journal of Applied Meteorology and Climatology 51, 11; 10.1175/JAMC-D-12-064.1
As in Fig. 2, but showing estimates of 5-yr total lake-contribution snowfall (LCSd) determined by subtracting the total snowfall at each downwind site from the total snowfall at their corresponding lake-effect site for all three transects across Lake Michigan.
Citation: Journal of Applied Meteorology and Climatology 51, 11; 10.1175/JAMC-D-12-064.1
A closer inspection of the time series in Fig. 2 reveals a distinct trend reversal in LCSu during the latter half of the twentieth century. Lake contribution increased from the early 1900s and reached maximum values between 1950 and 1980. Decreasing LCSu trends were observed after that period. Of interest is that the 5-yr period of maximum values was earliest in southern regions and latest in northern regions. Transect 1 experienced maximum LCSu during 1965–80. Transect 2, the middle transect, experienced a maximum during 1955–70. Transect 3, the southernmost transect, experienced a maximum during 1960–70, although it may have started as early as 1950 (but was not detected because of missing data). Thereafter, LCSu decreased, signifying a distinct trend reversal. In short, LCSu trend-reversal periods and maximum LCSu quantities near Lake Michigan found in this study varied with latitude.
Trend magnitudes for each transect before and after the reversal also vary with latitude. Excluding periods with five or more missing peak months, linear trends starting at the beginning of each time series through the 5-yr period of maximum LCSu show increases of approximately 12, 5, and 3 cm yr−1 for transects 1, 2, and 3, respectively. Linear trends starting at the period of maximum LCSu through the end of each time series show decreases of approximately −21 and −3 cm yr−1 for transects 1 and 3, respectively. No postreversal trends were determined for transect 2 because of the large number of missing peak months for several 5-yr periods at the end of the time series. While the actual trend values are dependent on the specific chosen period of maximum LCS, it is clear that locations farther north experienced greater long-term increases and decreases of local lake contribution and are perhaps more sensitive to changes in regional climate variables such as air temperature.
Linear trend values were also determined for LCSd (Fig. 3) using the same peak-month criterion as was used for LCSu. Although not all 5-yr periods of maximum LCSd agree with those of LCSu, the overall 1950–80 trend-reversal period is consistent. The LCSd pre- and postreversal trend magnitudes are substantially lower than the LCSu trends for transect 1. For transects 2 and 3, prereversal LCSd trend magnitudes are slightly higher than LCSu trends and postreversal trends are comparable. Trends from the beginning of the record to LCSd maxima for transects 1, 2, and 3 show increases of approximately 9, 7, and 5 cm yr−1, respectively. Trends from maxima to end for transects 1 and 3 show decreases of approximately −9 and −3 cm yr−1, respectively. Again, no postreversal trends were calculated for transect 2 because of missing data. The LCSd trend patterns are comparable to those of LCSu, an indication that our method isolates Lake Michigan’s contribution to local snowfall.
4. Discussion
A distinct trend reversal in Lake Michigan’s contribution to snowfall is evident in the latter half of the twentieth century. Because some previous studies (Braham and Dungey 1984; Norton and Bolsenga 1993) only included data up to the 1980s or 1990s, decreasing lake contribution trends would have been less evident; the reversal found here would have occurred late in their study periods. Comparatively, Kunkel et al. (2009a) used a longer period of analysis and noted a smaller increasing trend in lake-effect snowfall on the leeside of Lake Michigan. It is also possible that trends in other regions of the Great Lakes are different than those near Lake Michigan.
A full understanding of the reasons for the observed trend reversal is beyond the scope of this study. It is useful to seek confirmation of the long-term trends from an independent data source. It is well known that individual lake-effect events occur during cold-air outbreaks (Niziol 1987). Braham and Dungey (1984) found that for Lake Michigan seasonal average temperatures climatologically upwind of the lake (west) are inversely related to season-total snowfall downwind of the lake (east). More-complicated relationships between air temperature and snowfall were found for the eastern Great Lakes (Assel et al. 2003; Burnett et al. 2003). Therefore, temperature variations near Lake Michigan were examined to determine if they support the observed trends in LCS. Five-year-average daily maximum, minimum, and average temperatures were determined using all available data during peak lake-effect months (December–February). We used temperature data for the same COOP sites as in the snowfall analysis, with one exception. Because temperatures were not observed at Wellston, Michigan, (site 1L in Fig. 1) during the study period, those observed at Manistee, Michigan, (approximately 30 km west, denoted with a star in Fig. 1) were substituted. Manistee was chosen for its close proximity to both Wellston and Lake Michigan.
Recent increases in minimum air temperature across the region (Fig. 4) appear to support the decreasing trend in LCS in the most recent decades. The early 1930s, early 1950s, and the late 1990s stand out as being the warmest periods for most sites, whereas the late 1970s was the coldest period for all sites. The overall pattern exhibits a decreasing trend in average minimum temperatures from the early 1920s through the late 1970s. These temperature variations are consistent with those found by Braham and Dungey (1984) and Kunkel et al. (2009a). Our analyses show an increasing trend thereafter.
Time series of the 5-yr average minimum temperatures for December–February for the nine COOP sites provided in Fig. 1 with one exception. No data were available for Wellston (1L) and were replaced with data from Manistee (star). Transects 1, 2, and 3 are represented by solid, dashed, and dotted lines, respectively. Near-lake, upwind, and downwind sites are represented by black, dark-gray, and light-gray lines, respectively.
Citation: Journal of Applied Meteorology and Climatology 51, 11; 10.1175/JAMC-D-12-064.1
Scatterplots in Fig. 5 show that transect-1 upwind average minimum temperatures at Marshfield, Wisconsin, (1U) exhibited statistically significant (at the 99th percentile) negative Pearson’s correlation coefficients with LCSu and LCSd. For transect 2, minimum temperatures at Richland Center, Wisconsin, (2U) were negatively correlated with but not significantly related to LCSu and LCSd because of greater lake-contribution variability and fewer data points than other transects. For transect 3, minimum temperatures at Morrison, Illinois, (3U) were negatively correlated with LCSu and LCSd but not statistically significant, possibly because of the less-frequent wind conditions needed for lake-effect snow in Valparaiso (3L). Because minimum temperatures have been found to be inversely related to LCS near Lake Michigan, the temperature observations appear to independently verify the trend reversal in LCS.
Scatterplots of the relationships between 5-yr December–February average minimum air temperatures at upwind sites (1U, 2U, 3U) and both estimates of lake-contribution snowfall [(left) LCSu and (right) LCSd] for (top) transect 1, (middle) transect 2, and (bottom) transect 3. Best-fit least squares lines, Pearson’s correlation coefficients r, and significance p values are displayed.
Citation: Journal of Applied Meteorology and Climatology 51, 11; 10.1175/JAMC-D-12-064.1
Average daily maximum and average temperatures exhibited little relationship with LCS trends. The lack of a relationship would be expected because LCS generally occurs during the coldest periods, such as cold-air outbreaks. In addition, Kristovich and Spinar (2005) found a diurnal cycle in lake-effect precipitation with a maximum during the morning hours, which is typically when daily minimum temperature are observed.
Continuation of this study is strongly suggested. First, it is critical to determine the spatial variability of these long-term variations in LCS. What is the cause of the latitudinal variation in LCS trends in the Lake Michigan snowbelt? How do LCS trends in other regions of the Great Lakes vary from those observed near Lake Michigan in this study? Second, event-scale analysis using highly quality-controlled snowfall observations is essential to the investigation of lake-effect storm characteristics. Daily temporal resolution is much closer to the temporal scale of individual lake-effect events, allowing for a more detailed assessment of long-term lake-effect precipitation trends. Last, investigating the relationship of teleconnection patterns, such as the Northern Hemisphere annular mode, El Niño–Southern Oscillation, and Pacific decadal oscillation, may provide insight into long-term synoptic variability that may impact lake-effect precipitation trends.
5. Conclusions
Wintertime snowfall is substantially augmented in downwind locations of the Laurentian Great Lakes. The region endures many negative consequences caused by the surplus snow, including substantial property damage and increased numbers of traffic accident–related injuries and fatalities. The consensus among previous studies is that lake-effect snowfall increased by varying degrees during the twentieth century. This study investigates variability in Lake Michigan’s contribution to snowfall embedded within long-term trends and updates findings from previous studies using quality-controlled snowfall observations.
A reversal in Lake Michigan’s LCS is evident during 1920–2005. A general increase in LCS was observed from the early 1920s to the early 1950s to early 1980s. LCS decreased thereafter through the early 2000s, indicating a distinct reversal of trends that was not reported by earlier studies. Because some previous studies only included data up to the 1980s or 1990s, recent decreasing lake contribution trends would have been less evident. LCS trend reversal appears to have occurred near the southern part of Lake Michigan before it occurred farther north. The significance of this south-to-north timing variation is unknown. Trend magnitudes before and after the reversal are greatest farther north and are smallest farther south. These latitudinal dependencies suggest sensitivity of LCS to spatially variable climatic conditions and/or topographic differences.
Average air temperature has been shown by previous studies to be inversely related to total snowfall. The study presented here notes the best correlation between LCS and minimum temperature. LCS trend reversal is consistent with observed increasing minimum temperatures during peak lake-effect months after 1970.
Suggestions for future work are threefold. Spatial variations in LCS throughout the Great Lakes area would provide valuable insights into the responsible physical processes. Event-scale analysis of lake-effect storms is necessary for best determining how the events have changed in intensity and frequency. Large-scale analyses (e.g., tracks of high pressure systems and cyclones associated with teleconnection patterns) or local-scale analyses (e.g., influence of local climate conditions or topography) may offer more insight into the factors that control lake-effect events.
Acknowledgments
The authors thank members of the SWS Mesoscale/Boundary Layer Group at the University of Illinois, past and present, for their help and input on this project. Thanks are given to Dr. Ken Kunkel and Ms. Leslie Stoecker for providing quality-controlled snowfall data. Very helpful reviews were provided by Dr. Jim Angel, Mr. Mike Timlin, and three anonymous reviewers. This work was partially supported by the Illinois State Water Survey, the University of Illinois Research Board, and National Science Foundation Grant ATM 05-12954. Opinions expressed are those of the authors and are not necessarily those of the ISWS, Prairie Research Institute, University of Illinois, or the National Science Foundation.
REFERENCES
Assel, R. A., K. Cronk, and D. Norton, 2003: Recent trends in Laurentian Great Lakes ice cover. Climatic Change, 57, 185–204.
Barthold, F. E., and D. A. R. Kristovich, 2011: Observations of the cross-lake cloud and snow evolution in a lake-effect snow event. Mon. Wea. Rev., 139, 2386–2398.
Braham, R. R., and M. J. Dungey, 1984: Quantitative estimates of the effect of Lake Michigan on snowfall. J. Appl. Meteor., 23, 940–949.
Burnett, A. W., M. E. Kirby, H. T. Mullins, and W. P. Patterson, 2003: Increasing Great Lake–effect snowfall during the twentieth century: A regional response to global warming? J. Climate, 16, 3535–3542.
Chang, S. S., and R. R. Braham, 1991: Observational study of a convective internal boundary layer over Lake Michigan. J. Atmos. Sci., 48, 2265–2279.
Changnon, S. A., D. Changnon, and T. R. Karl, 2006: Temporal and spatial characteristics of snowstorms in the contiguous United States. J. Appl. Meteor. Climatol., 45, 1141–1155.
Ellis, A. W., and J. J. Johnson, 2004: Hydroclimatic analysis of snowfall trends associated with the North American Great Lakes. J. Hydrometeor., 5, 471–486.
Groisman, P. Ya., and D. R. Easterling, 1994: Variability and trends of total precipitation and snowfall over the United States and Canada. J. Climate, 7, 184–205.
Grover, E. K., and P. J. Sousounis, 2002: The influence of large-scale flow on fall precipitation systems in the Great Lakes basin. J. Climate, 15, 1943–1956.
Kelly, R. D., 1986: Mesoscale frequencies and seasonal snowfalls for different types of Lake Michigan snowstorms. J. Climate Appl. Meteor., 25, 308–312.
Kristovich, D. A. R., 2009: Climate sensitivity of Great Lakes–generated weather systems. Climatology, Variability, and Change in the Midwest, S. C. Pryor, Ed., Indiana University Press, 236–250.
Kristovich, D. A. R., and R. A. Steve III, 1995: A satellite study of cloud-band frequencies over the Great Lakes. J. Appl. Meteor., 34, 2083–2090.
Kristovich, D. A. R., and M. L. Spinar, 2005: Diurnal variations in lake-effect precipitation near the western Great Lakes. J. Hydrometeor., 6, 210–218.
Kristovich, D. A. R., N. F. Laird, and M. R. Hjelmfelt, 2003: Convective evolution across Lake Michigan during a widespread lake-effect snow event. Mon. Wea. Rev., 131, 643–655.
Kunkel, K. E., N. E. Westcott, and D. A. R. Kristovich, 2002: Effects of climate change on heavy lake-effect snowstorms near Lake Erie. J. Great Lakes Res., 28, 521–536.
Kunkel, K. E., M. Palecki, K. G. Hubbard, D. Robinson, K. Redmond, and D. Easterling, 2007: Trend identification in twentieth-century U.S. snowfall: The challenges. J. Atmos. Oceanic Technol., 24, 64–73.
Kunkel, K. E., L. Ensor, M. Palecki, D. Easterling, D. Robinson, K. G. Hubbard, and K. Redmond, 2009a: A new look at lake-effect snowfall trends in the Laurentian Great Lakes using a temporally homogenous data set. J. Great Lakes Res., 35, 23–29.
Kunkel, K. E., L. Ensor, M. Palecki, D. Easterling, D. Robinson, K. G. Hubbard, and K. Redmond, 2009b: Trends in twentieth-century U.S. snowfall using a quality-controlled data set. J. Atmos. Oceanic Technol., 26, 33–44.
Niziol, T. A., 1987: Operational forecasting of lake effect snowfall in western and central New York. Wea. Forecasting, 2, 310–321.
Niziol, T. A., W. R. Snyder, and J. S. Waldstreicher, 1995: Winter weather forecasting throughout the eastern United States. Part IV: Lake effect snow. Wea. Forecasting, 10, 61–77.
Norton, D., and S. Bolsenga, 1993: Spatiotemporal trends in lake effect and continental snowfall in the Laurentian Great Lakes, 1951–80. J. Climate, 6, 1943–1956.
Rodriguez, Y., D. A. R. Kristovich, and M. R. Hjelmfelt, 2007: Lake-to-lake cloud bands: Frequencies and locations. Mon. Wea. Rev., 135, 4202–4213.
Schmidlin, T. W., 1993: Impacts of severe winter weather during December 1989 in the Lake Erie snowbelt. J. Climate, 6, 759–767.
Scott, R. W., and F. A. Huff, 1996: Impacts of the Great Lakes on regional climate conditions. J. Great Lakes Res., 22, 845–863.
