1. Introduction
Natural forces have always shaped the earth’s climate, but the Intergovernmental Panel on Climate Change (Stocker et al. 2013) has noted a positive correlation between global air temperatures and carbon dioxide from anthropogenic sources. This finding has considerable implications for the regional climate of an area, especially in the snow-dominated latitudes of the United States. Snowfall is dependent on air temperatures that are at or below freezing. The greatest observed warming in the last century has been in the northern high latitudes (Dai et al. 2001), where Kharin and Zwiers (2005) found an accelerated warming of cold temperature extremes as compared with warm extremes. The increased air temperatures raise mean winter temperatures closer to the freezing threshold, causing more precipitation to fall as rain rather than snow (Knowles et al. 2006); this is especially apparent during the spring (Groisman et al. 2004).
Central New York is one region that would be negatively affected by a transition from snow to rain. The region has grown accustomed to seasonal snowfall totals regularly exceeding 250 cm, which has created a societal dependency on snowfall for winter recreation, spring snowmelt, and insulation of plant biomass from freezing temperatures (Schmidlin 1993; Kunkel et al. 2002). Central New York is located at the intersection of the northeastern United States and the Great Lakes basin. As a result, snowfall in central New York regularly occurs from both lake-effect and non-lake-effect snowstorms. Lake-effect snowstorms are those that result from the advection of a cold air mass (generally a polar or Arctic air mass) over a relatively warm lake (Peace and Sykes 1966; Kunkel et al. 2000). The moisture and instability that generate the snowfall are thus produced solely by the advection of cold air over warmer water. Non-lake-effect snowstorms are those that result from all other mechanisms responsible for organizing moisture, lift, and instability into snowfall. These storms are generally associated with a transient low pressure system (i.e., a midlatitude cyclone or a northeaster). Yet, several studies have found a large contrast between non-lake-effect snowfall trends and lake-effect snowfall trends, as follows.
Braham and Dungey (1984), Norton and Bolsenga (1993), Burnett et al. (2003), Ellis and Johnson (2004), and Kunkel et al. (2009a) found a significant increase in snowfall for stations that experience lake-effect snowfall (those within the extent of the Great Lakes basin) since the early twentieth century. Using observations from over 1200 weather stations from 1951 to 1981, Norton and Bolsenga (1993) found that lake-effect snow in the Lake Ontario snow basin increased by approximately 6.7 cm yr−1 while there was little to no appreciable increase in snowfall outside the basin. Burnett et al. (2003) corroborated an increase in lake-effect snow, noting an approximately 1.5 cm yr−1 increase from 1931 to 2001. Kunkel et al. (2009a) noted that, as a result of inhomogeneities within the dataset of Burnett et al. (2003), the snowfall increase was overestimated by approximately 0.9 cm yr−1, however. Kunkel et al. (2009a) suggest that this difference is attributed to their use of filtered Cooperative Observer Program (COOP) data, which removes biases in the data and reduces an overestimation of snowfall trends by Burnett et al. (2003).
The previously discussed studies all framed snowfall trends as a long-term increase. In contrast, a subsequent study by Bard and Kristovich (2012) proposed the presence of a trend reversal. The authors noted that the long-term snowfall trend actually experienced a reversal in the late 1970s, suggesting that snowfall has decreased for the Lake Michigan basin since the 1970s. Although the authors focused on the Lake Michigan basin, seasonal snowfall totals in the Lake Michigan basin behave similarly to those of the Lake Ontario basin because of the influence of lake-effect snow (Liu and Moore 2004). Notable differences between the two basins do exist, such as the orientation of the two lakes, the influence of coastal low pressure systems (i.e., northeasters), and the water properties (i.e., depth, temperature, surface area, and ice onset and breakup) of the two lakes (Wang et al. 2012). Therefore, it is important to expand upon the findings of Bard and Kristovich (2012) to determine whether a trend reversal is also present for the snowfall in a second Great Lakes basin and, more important, to quantify how significant it might be.
Previous studies have observed snowfall trends for the entire Great Lakes basin. It is also important to understand the spatial and interseasonal snowfall changes at a local level, however. Therefore, an objective of this study is to determine whether snowfall trends in central New York, located in the Lake Ontario basin, behave in a similar manner to trends outlined by Norton and Bolsenga (1993), Burnett et al. (2003), Kunkel et al. (2009a), and Bard and Kristovich (2012). A major finding of previous studies was that snowfall significantly increased throughout the twentieth century for lake-effect stations in contrast with stations distant from the lakes. The spatial variability of snowfall trends for the Great Lakes basin has not been examined, however. Therefore, this study also determines whether changes in seasonal snowfall totals are spatially homogenous throughout central New York. The influence of air temperature and precipitation on seasonal snowfall totals within central New York is also examined.
To analyze the spatiotemporal snowfall patterns in central New York, this paper is divided into three additional sections. Section 2 describes the datasets and the procedures used to analyze the data. Section 3 discusses long-term (1931/32–2011/12) and shorter-term snowfall trends both temporally and spatially and describes potential influences on snowfall patterns in central New York. Section 4 provides a concise synopsis of the results.
2. Methods
a. Study area
With a population of over 1 million people, central New York is defined by using 12 counties (Fig. 1). Because of its location, snow associated with transient low pressure systems and lake-effect snow regularly impact the region, making the city of Syracuse the snowiest metropolitan area in the United States (Kunkel et al. 2000; NOAA 2014). Lake Ontario is the primary source of moisture and warmth for lake-effect precipitation development. Even though it is the smallest of the Great Lakes in surface area, it rarely freezes over during the winter (Niziol 1987; Wang et al. 2012). Because of its east–west orientation and leeward position relative to Lake Ontario, macroscale wind patterns favor the development of lake-effect precipitation over central New York (Niziol et al. 1995).
Study area. Labeled are the 12 counties that make up central New York. Points symbolized with circles and triangles represent the 93 COOP stations in the study area. The 33 stations represented by a triangle report snowfall for the duration of the study period (1931/32–2011/12), and the 60 stations represented by a circle are not included in the long-term (1931/32–2011/12) snowfall calculations.
Citation: Journal of Applied Meteorology and Climatology 53, 12; 10.1175/JAMC-D-14-0084.1
There are five main topographic features in central New York (Fig. 2) that provide a setting that is conducive to the development of lake-effect snow. The “Southern Hills” and the Tug Hill Plateau are two of the most prominent snowfall regions in central New York. Elevations in the Southern Hills quickly rise 500 m within 32 km of the lakeshore, with irregular hills and valleys providing additional topographic features for low-level air convergence and orographic lifting (Clowes 1919). Tug Hill, the region most known for snowfall in New York State, is also conducive to high snowfall totals because of its eastward position relative to Lake Ontario and its steep elevation gradients.
Topographic features of central New York.
Citation: Journal of Applied Meteorology and Climatology 53, 12; 10.1175/JAMC-D-14-0084.1
b. Data
Seasonal snowfall totals, defined here as the total snowfall from 1 July to 30 June, were examined from 1931 to 2012 for 93 National Weather Service COOP stations located in the 12 central New York counties (Table 1). Data were accessed through the National Climatic Data Center (NCDC) online Monthly Summary Observations (digital database “TD3220”) and included precipitation totals, average monthly air temperatures, and monthly snowfall.
COOP stations by central New York county.
Examination of data from COOP stations began with monthly data. If monthly snowfall data were missing for a station, daily COOP snowfall records were acquired for the missing month(s) from NCDC’s Global Historical Climatology Network. If at least 85% of the days within an unreported month were observed at a given station, then the daily data were summed to create a monthly snowfall total for that station. Therefore, if daily snowfall totals are recorded for at least 85% of a month at a given station, it is assumed that the summation of these daily totals approximates the true monthly snowfall totals for that month. If a missing monthly snowfall total was not recovered from the daily data for a given station, and only one winter month (November–April) was missing, a four-point weighted spatial bilinear interpolation using nearby stations was used to estimate the missing monthly snowfall total (Accadia et al. 2003). Once interpolated, monthly snowfall totals were summed from November to April and reported as a seasonal snowfall total at each station. If observations were missing for two or more winter months, however, then the annual snowfall total for that station was not retained.
The reporting consistency for each COOP site varied; therefore, stations without observations for at least five consecutive years at some point during 1931–2012 were not retained. To qualify as having at least five consecutive years of observations, at least 85% of all winter months during the five consecutive years must have values. To increase usable snowfall observations, a slightly lower threshold (85%) was used in this study than that of Kunkel et al. (2009a), who used a 90% reporting frequency. This procedure provided seasonal snowfall data for 93 stations, of a possible 122. Of these 93 stations, only 33 provided seasonal snowfall totals for at least 85% of the 81 years that composed the full length of the study period (1931/32–2011/12). Thus, all trends in seasonal snowfall presented in this paper are from these 33 stations (Fig. 1).
Since lake-effect snowfall is positively correlated with elevation (Clowes 1919), each station was scrutinized for inhomogeneities as outlined in Kunkel et al. (2007), with a particular emphasis on elevation changes greater than 10 m, and changes in latitude/longitude greater than 0.15°. Many reported relocation changes were actually updated geographic coordinates. Therefore, beyond these thresholds, it was judged that a change was likely due to a relocation instead of updated coordinates. A station was also deemed inhomogeneous if the annual reporting frequency during a time series for the station was less than 85%, as described earlier. The determination of homogeneity for a station was used to compare snowfall trends for stations deemed homogenous and those deemed inhomogeneous.
Air temperature and precipitation data were retained for typical snowfall months (November–April). Air temperature was only available for 27 of the 93 COOP stations in Table 1 and was averaged together to calculate the mean monthly temperature for all of central New York. Note that, even though air temperatures were averaged, the temperature variation across central New York is minimal (1.58°C standard deviation). For consistency, monthly precipitation was obtained for the same 27 stations that reported air temperature.
3. Results
Multiple quantitative analyses were employed to assess the temporal and spatial patterns of snowfall in central New York. Analyses included simple linear regressions, Pearson correlations, principal component analyses, and GIS mapping. Results of these analyses were divided into three categories: temporal snowfall patterns, intraregional snowfall variability, and potential factors responsible for snowfall variability.
a. Temporal snowfall patterns
Temporal snowfall trends were calculated using simple linear regressions (Fig. 3). Autocorrelation tests were performed on each time series to investigate periodicities in the dataset, yet no significant (>0.3) correlations existed within the dataset. Snowfall trends were calculated for the entire region by averaging snowfall totals for the 33 stations that reported for the total duration (1931/32–2011/12) of the study period (Fig. 1). To detect nonlinearity in the trendline, trends were calculated for the entire time series (1931–2012), at two 41-yr increments (1931–72 and 1971–2012), and by computing a 21-yr trend with a 1-yr moving window.
Initial and homogenous snowfall trends (significant at the 5% level) for multiple time intervals: (a) long-term trends and (b) two 41-yr trends using initial COOP stations.
Citation: Journal of Applied Meteorology and Climatology 53, 12; 10.1175/JAMC-D-14-0084.1
The long-term (1931/32–2011/12) statistically significant (ρ < 0.05) seasonal snowfall trend (Fig. 3a) for central New York was 1.16 ± 0.31 cm yr−1, comparable to the 1.5 cm yr−1 lake-effect snowfall increase reported by Burnett et al. (2003). After filtering for inhomogeneities within the dataset, the recalculated statistically significant (ρ < 0.05) long-term snowfall trend was reduced to approximately 0.59 ± 0.30 cm yr−1. This supports Kunkel et al. (2009a), who noted the snowfall increase found by Burnett et al. (2003) was reduced from approximately 1.5 to 0.6 cm yr−1 after filtering for inhomogeneities. Note, however, that for the long-term record only nine stations were deemed homogenous in central New York, with nearly one-half of them (four stations) located in Onondaga County. The remaining five stations were randomly dispersed (ρ < 0.01) throughout four other counties (Oswego, Tompkins, Chenango, and Jefferson). Therefore, the decrease from 1.16 to 0.59 cm yr−1 may be a result of regional clustering in Onondaga County instead of inaccuracies in nonhomogeneous data.
If the trend is considered as two distinct time periods (Fig. 3b), the first half of the record exhibits a strong statistically significant (ρ < 0.05) increase in snowfall, 3.27 ± 0.67 cm yr−1, whereas the second half demonstrates a significant (ρ < 0.05) decrease in snowfall. Long-term snowfall trends calculated by previous studies (Norton and Bolsenga 1993; Burnett et al. 2003; Kunkel et al. 2009a) may not best represent changes in the data up to the present time, as also suggested by Bard and Kristovich (2012) for the Lake Michigan basin. This may hold particularly true for the trends in the later studies done by Burnett et al. (2003) and Kunkel et al. (2009a), as both studies’ time series included the lower snowfall totals in the early twenty-first century, in contrast to the higher snowfall totals in the latter twentieth century (1970s and 1990s). Thus, similar to the findings of Bard and Kristovich (2012), snowfall patterns in the Great Lakes may not be best portrayed using a long-term trend, because there is a noticeable discontinuity within the trend. As Moses et al. (1987) describe, the presence of a discontinuity or trend reversal may suggest a multidecadal mode of variability that could be driven by teleconnection patterns and is further discussed later in this study.
Variations in snowfall trends are best demonstrated by the 21-yr moving snowfall trends calculated each winter from 1942/43 to 2000/01 (Fig. 4). The 21-yr snowfall trends during this time period were highly variable, ranging from approximately −7.6 to 7.6 cm yr−1. The greatest change in snowfall trends occurred from 1966 to 1980, during which time the snowfall trends rapidly decreased (
The 21-yr central New York snowfall trends at time-step intervals of 1-yr. The solid line represents the 21-yr snowfall trend for each year from 1942 to 2000; the dashed lines are the associated uncertainty of the trends and are significant at the 5% level.
Citation: Journal of Applied Meteorology and Climatology 53, 12; 10.1175/JAMC-D-14-0084.1
b. Intraregional snowfall variability
To group stations by region within central New York, principal component analyses (PCA) were used to extract hidden temporal and spatial correlations in seasonal snowfall totals. Two PCAs were conducted, the first when missing seasonal snowfall values were excluded (PCA-a) and a second when missing values were replaced with the mean (PCA-b). Unrotated PCAs were first run, with the eigenvalues and scree plots examined, to determine the number of components to retain. Five distinct modes, here referred to as regions, were extracted from the unrotated PCAs for central New York and accounted for 94% of the original variance. After the five regions were determined, PCAs were conducted with a varimax rotation to group stations into a region using their seasonal snowfall totals. Each station was grouped into a single region on the basis of the highest factor-loading score (≥|0.30|) for that station. The five regions were then mapped using the stations grouped within each region and were plotted using the polygon feature of the ArcGIS software (Fig. 5). It is important to note that the borders between the five regions are loose, and crossing regions does not suggest a vast difference between seasonal snowfall totals. Instead, snowfall patterns for stations within the same region behave more similarly than for stations between regions. Note also that all of the regions are generally concentrated to a subsection of central New York, except for region 4, which extends over a narrow and elongated area.
Regional snowfall trends classified by PCA-a and PCA-b. Trends and uncertainties are reported in centimeters per year and are significant at the 5% level. The trends are ordered from top to bottom for each region: first trend is the long-term trend (1931/32–2011/12), the second trend is from 1931/32 to 1971/72, and the third trend is from 1971/72 to 2011/12.
Citation: Journal of Applied Meteorology and Climatology 53, 12; 10.1175/JAMC-D-14-0084.1
Seasonal snowfall totals for stations with a reporting frequency of at least 85% for a given period (as discussed in section 2b) were averaged together for each region, and autocorrelation tests (≥0.30) were used to test for periodicities in the dataset. Snowfall trends for the entire record and two 41-yr records were then calculated using simple linear regressions for each individual region (Fig. 5). Similar to the findings for the entire central New York basin, snowfall trends exhibited a strong trend reversal, as snowfall trends for all regions were considerably higher for the first half of the record (1931/32–1971/72) than for the latter half (1971/72–2011/12). There were also strong regional variations in the snowfall trends. During the long term and first half of the record, regions 4 and 5 experienced the largest positive snowfall trends. Both regions are in areas conducive to lake-effect snow development because of their position to the east of Lake Ontario and because of their inclusion of the Tug Hill Plateau and Adirondack Mountains. Therefore, a larger increase in snowfall totals in these regions suggests an increase in lake-effect snow rather than non-lake-effect snow. A diminished snowfall increase for locations farther from the lake is supported by snowfall changes in region 2. Relative to the other five regions, region 2 experienced the smallest snowfall increase, likely because region 2 is farthest from Lake Ontario and is southeast of the lake. Thus, it is assumed that lake-effect snowfall is less common for this region, with the majority of the region’s snowfall coming from non-lake-effect snowstorms. A lower snowfall increase over the period of record for nontypical lake-effect regions in central New York is consistent with previous studies (Norton and Bolsenga 1993; Burnett et al. 2003; Kunkel et al. 2009a) that noted that stations distant from the lake experienced little to no appreciable increase in snowfall throughout the twentieth and early part of the twenty-first centuries.
The 1971/72–2011/12 snowfall trend for region 3 should be noted, as it is the only region to experience a minimal decrease in the snowfall trend from the first half to the second half of the 1931/32–2011/12 period. Among the regions, region 3 experienced the largest positive snowfall trend over 1971/72–2011/12 (Fig. 5). The cause of this relatively higher positive trend is not readily apparent. Region 3 is located inland and slightly southeast of Lake Ontario, which suggests lake-effect snow is not typical in the region. Increased non-lake-effect snow events (e.g., northeasters) are unlikely to account for the difference in trend between region 3 and the other regions. If non-lake-effect snow events had increased snowfall, it would be expected that all of the regions, especially region 2, would exhibit a trend similar to that of region 3. The relatively higher positive trend in snowfall in region 3 could be related to anomalously low snowfall totals (−43.9 cm) during the 1980s relative to the 1971/72–2011/12 average. Snowfall deviations during the 1980s for the other regions ranged from −20.2 (region 1) to −30.9 (region 5) cm. Thus, the magnitude of the snowfall anomaly in the 1980s for region 3 was greater than that of all other regions. The presence of this comparatively more negative snowfall anomaly early in the 1971/72–2011/12 time series may have contributed to the larger positive snowfall trend observed for region 3 over 1971/72–2011/12. The cause of the anomalously low snowfall in region 3 is beyond the scope of this paper.
Spatial representations of snowfall changes were constructed using ArcGIS, version10.1, and consisted of a 41-yr snowfall-difference map. The 41-yr snowfall-difference map was calculated by subtracting average snowfall totals for the long-term stations of the earlier period (1931/32–1971/72) from average snowfall totals during the latter period (1971/72–2011/12). Recall that 1971/72 represents the break point of the trend reversal. To represent snowfall totals throughout central New York, 10-point nearest-neighbor interpolations were performed using an inverse-distance-weighting function (Chen and Liu 2012).
The snowfall-difference map (Fig. 6) supports the idea of an overall increase in snowfall for central New York, but the increase is not spatially homogenous. The northern reaches of central New York, within the Tug Hill Plateau (see Figs. 1 and 2), experienced the greatest increase (over 75 cm) in average seasonal snowfall totals. An appreciable increase in snowfall totals was also noticed for eastern Oswego County (45–55-cm increase) and for a north–south transect in central New York that extends from northern Lewis County to southern Cortland County (≥15-cm increase). A few areas experienced a decrease in annual snowfall, most notably northwestern Onondaga, southern Chenango, western Oswego, and northern Herkimer Counties. All of these regions, except for northwestern Onondaga and western Oswego Counties, are not commonly associated with lake-effect snow because of their orientation to, and distance from, Lake Ontario. This further supports the notion that snowfall increases in central New York are not evenly distributed and instead are highest in typical lake-effect-snow locations and lowest near the edges or in nontypical lake-effect locations of the Lake Ontario snow basin.
Average annual snowfall for central New York: (top) average annual snowfall (left) from 1931/32 to 1971/72 and (right) from 1971/72 to 2011/12. (bottom) The difference in mean annual snowfall totals between 1971/72–2011/12 and 1931/32–1971/72.
Citation: Journal of Applied Meteorology and Climatology 53, 12; 10.1175/JAMC-D-14-0084.1
c. Potential factors of snowfall variability
Possible factors influencing snowfall changes in central New York were also examined. To compare annual snowfall, winter air temperatures, and winter precipitation, each field was normalized by subtracting the mean long-term value from the seasonal mean for each year and then dividing by the standard deviation of the long-term record. Results were then plotted using a 1.5-yr Gaussian filter, and Pearson correlations were computed on the unfiltered datasets to determine the correlations among the three different variables.
Snowfall totals are highly dependent on air temperatures at or below 0°C and substantial moisture content in the air. Figure 7 shows that winters in central New York have experienced greater variance in snowfall (
Normalized annual snowfall totals, winter air temperature, and winter precipitation data after being run through a 1.5-yr Gaussian filter.
Citation: Journal of Applied Meteorology and Climatology 53, 12; 10.1175/JAMC-D-14-0084.1
The majority of the long-term time series experienced mean winter air temperatures below 0°C, with the lowest average temperatures (under −2.8°C) occurring during the 1930s and 1990s (Fig. 8). The frequency of mean winter air temperatures above the freezing threshold increased, however—especially after the 1983/84 season. The 2011/12 season was the warmest for the time series, with mean air temperatures exceeding 1.6°C. Average winter air temperatures above the freezing threshold favors rain over snow, decreasing seasonal snowfall totals (Knowles et al. 2006). The IPCC report (Stocker et al. 2013) suggest that anthropogenic forcing is the underlying cause of air temperature change, but teleconnections also have a significant impact (Serreze et al. 1998; Livezey and Smith 1999; Kocin and Uccellini 2004, 3–39; Kunkel et al. 2009b). In Fig. 8, there appears to be a slight periodic variation that exists within winter air temperatures, as mean temperatures were at a minimum during the early 1930s, again during the latter 1960s, and finally during the early 1990s. Therefore, the possible 30-yr periodic variation in air temperatures may be linked to an established climate-scale oscillation.
Deviation of average winter air temperatures from the freezing threshold (0°C).
Citation: Journal of Applied Meteorology and Climatology 53, 12; 10.1175/JAMC-D-14-0084.1
Previous studies have examined the possible linkages between the phases of various teleconnection patterns and weather patterns in the northeastern United States and Great Lakes basin. For example, Grimaldi (2008) discovered that during the El Niño (warm) phase of the El Niño–Southern Oscillation (ENSO), early winter months in Syracuse are generally warmer, with anomalously low snowfall totals. This pattern then reverses during later winter months. Also, during El Niño years major snow events are 5 times as likely to occur in Syracuse. Along with ENSO, Kocin and Uccellini (2004, 3–39) found that the North Atlantic Oscillation (NAO) is negatively correlated (−0.64) with increased seasonal snowfall in the eastern United States. Although not as widely studied, the Atlantic multidecadal oscillation (AMO), which is a 65–70-yr variation in Atlantic Ocean sea surface temperatures, has been linked to variations in the North American climate. For example, Zhao et al. (2010) suggested that changes in the AMO result in shifts in the magnitude and location of the jet stream. Also, Fortin and Lamoureux (2009) linked changes in the AMO with snowfall in the northeastern United States, as warmer-than-average North Atlantic sea surface temperatures favor moister conditions and increased snowfall in boreal and Arctic regions. Therefore, previous studies have linked various teleconnections to changes in Northeast snowfall, which may account for the variations in snowfall trends (Figs. 3 and 4) for central New York. Further analyses should be conducted to investigate the influence of teleconnections such as the ENSO, NAO, and AMO on snowfall in central New York.
4. Conclusions
Several studies (Norton and Bolsenga 1993; Burnett et al. 2003; Kunkel et al. 2009a) have noted an increase in snowfall for lake-effect stations in the Laurentian Great Lakes since the early twentieth century. Bard and Kristovich (2012) were the first to recognize a trend reversal within the time series for snowfall in the Laurentian Great Lakes, focusing on Lake Michigan. Because of numerous physical and climatic differences between Lake Michigan and Lake Ontario, this study examined snowfall trends for central New York—located in the heart of the Lake Ontario snow basin.
There are three key findings of this study. The first finding was the recognition that central New York snowfall trends are not linear and instead experienced a trend reversal during the 1960s–1970s. Norton and Bolsenga (1993), Burnett et al. (2003), and Kunkel et al. (2009a) all noted a long-term increase in snowfall for lake-effect stations as compared with non-lake-effect stations. This finding was corroborated by the long-term snowfall trend (1.16 ± 0.31 cm yr−1) for central New York, but further analysis demonstrated a reversal in the trend. It was found, similar to Bard and Kristovich (2012), that above-average snowfall during the 1970s and 1990s likely increased long-term snowfall trends, especially during the latter half of the time series (1971–2012). Annual 21-yr snowfall trends continually decreased from the mid-1970s through the mid-1980s, further suggesting that central New York snowfall trends behave nonlinearly over the period of record.
The second finding is that snowfall patterns in central New York are not spatially homogenous. It is common for a study to overgeneralize a region, for example the Lake Erie basin, the Lake Ontario basin, or even the Great Lakes basin. There are important microclimatic variations that should be accounted for, however. Therefore, one goal of this study was to spatially represent snowfall trends in central New York. It was found that central New York can be divided into five distinct regions, with the regions closer to and leeward of Lake Ontario experiencing higher snowfall trends than do regions farther away from and not leeward of the lake. Since 1931, Tug Hill experienced the largest increase in snowfall for central New York. Tug Hill is commonly known for the vast amounts of lake-effect snow that it receives over a snowfall season. Therefore, it is suggested that, even though central New York is a region that is dominated by lake effect snow, there are spatial differences in which areas typically associated with lake-effect snow (Tug Hill, the Southern Hills, and Oswego County) experienced a higher snowfall trend than did areas that are not characteristically associated with lake-effect snow (northeastern and southern central New York).
The third finding is that winter air temperatures have the greatest correlation with seasonal snowfall totals in central New York. Winter air temperatures are significantly (ρ < 0.05) inversely correlated with seasonal snowfall totals (−0.56), suggesting that colder (warmer) winter temperatures promote increased (decreased) seasonal snowfall totals. This relationship was notably stronger than that between snowfall and precipitation, suggesting that winter air temperatures have a greater influence on seasonal snowfall totals than do winter precipitation totals.
Because of the relative abundance of snowfall throughout the winter season, snowfall is an important resource in central New York. Stable seasonal snowfall totals are vital for the region’s habitats and economy and agriculture, water resources, winter recreation, wildlife, and the activities of the N.Y.S. Department of Transportation depend on stable snowfall totals. The results from this study will help us to understand central New York snowfall trends and will aid in forecasting seasonal snowfall totals. Enhanced forecast predictions will allow better preparation for and adaption to future seasonal snowfall totals.
The most significant limitation during this study was the consistency of the COOP observations. Records for each of the 93 stations were not consistent, with variations in time series, reporting frequencies, missing data, and spatial homogeneity. This consistency is a significant limitation of COOP data and possibly biased the trend calculations, but we emphasize that COOP data are the only data available to allow this type of study to be done. A second limitation was presented during the analysis of temperature and precipitation data. Because of the limited availability of air temperature data, temperature records were averaged throughout central New York with the assumption that the average air temperature is representative of the whole region.
Future work should expand the analysis of driving factors of snowfall patterns in central New York. For example, lake-effect snow is largely dependent on the air–lake temperature contrast; therefore, it would be beneficial to examine yearly air and lake surface temperatures to determine their influence on seasonal lake-effect snow totals. Further analysis should also be done to explore the influence of individual teleconnection patterns (e.g., ENSO, NAO, and AMO) on seasonal snowfall totals in central New York, specifically examining their influence on average winter air temperatures, winter precipitation totals, and factors such as changes to lake ice (Wang et al. 2012), the lake–air temperature difference (Peace and Sykes 1966), and the tracks of low pressure systems (Liu and Moore 2004).
REFERENCES
Accadia, C., S. Mariani, M. Casaioli, A. Lavagnini, and A. Speranza, 2003: Sensitivity of precipitation forecast skill scores to bilinear interpolation and a simple nearest-neighbor average method on high-resolution verification grids. Wea. Forecasting, 18, 918–932, doi:10.1175/1520-0434(2003)018<0918:SOPFSS>2.0.CO;2.
Bard, L., and D. A. R. Kristovich, 2012: Trend reversal in Lake Michigan contribution to snowfall. J. App. Meteor. Climatol., 51, 2038–2046, doi:10.1175/JAMC-D-12-064.1.
Baxter, M. A., C. E. Graves, and J. T. Moore, 2005: A climatology of snow-to-liquid ratio for the contiguous United States. Wea. Forecasting, 20, 729–744, doi:10.1175/WAF856.1.
Braham, R. R., and M. J. Dungey, 1984: Quantitative estimates of the effect of Lake Michigan on snowfall. J. Appl. Meteor., 23, 940–949, doi:10.1175/1520-0450(1984)023<0940:QEOTEO>2.0.CO;2.
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, doi:10.1175/1520-0442(2003)016<3535:IGLSDT>2.0.CO;2.
Chen, F. W., and C. W. Liu, 2012: Estimation of the spatial rainfall distribution using inverse distance weighting (IDW) in the middle of Taiwan. Paddy Water Environ., 10, 209–222, doi:10.1007/s10333-012-0319-1.
Clowes, E. S., 1919: Mountain and valley winds at Syracuse, N.Y. Mon. Wea. Rev., 47, 464–464, doi:10.1175/1520-0493(1919)47<464:MAVWAS>2.0.CO;2.
Dai, A., T. M. L. Wigley, B. A. Boville, J. T. Kiehl, and L. E. Buja, 2001: Climates of the twentieth and twenty-first centuries simulated by the NCAR Climate System Model. J. Climate, 14, 485–519, doi:10.1175/1520-0442(2001)014<0485:COTTAT>2.0.CO;2.
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, doi:10.1175/1525-7541(2004)005<0471:HAOSTA>2.0.CO;2.
Fortin, D., and S. F. Lamoureux, 2009: Multidecadal hydroclimatic variability in northeastern North America since 1550 AD. Climate Dyn., 33, 427–432, doi:10.1007/s00382-008-0422-6.
Grimaldi, R., 2008: Climate teleconnections related to El Niño winters in a lake-effect region of west-central New York. Atmos. Sci. Lett., 9, 18–25, doi:10.1002/asl.166.
Groisman, P. Ya., R. W. Knight, T. R. Karl, D. R. Easterling, B. Sun, and J. H. Lawrimore, 2004: Contemporary changes of the hydrological cycle over the contiguous United States: Trends derived from in situ observations. J. Hydrometeor., 5, 64–85, doi:10.1175/1525-7541(2004)005<0064:CCOTHC>2.0.CO;2.
Kharin, V. V., and F. W. Zwiers, 2005: Estimating extremes in transient climate change simulations. J. Climate, 18, 1156–1173, doi:10.1175/JCLI3320.1.
Knowles, N., M. D. Dettinger, and D. R. Cayan, 2006: Trends in snowfall versus rainfall in the western United States. J. Climate, 19, 4545–4559, doi:10.1175/JCLI3850.1.
Kocin, P. J., and L. W. Uccellini, 2004: Northeast Snowstorms. Vols 1 and 2, Meteor. Monogr., No. 54, Amer. Meteor. Soc., 818 pp.
Kunkel, K. E., N. E. Westcott, and D. A. R. Kristovich, 2000: Climate change and lake-effect snow. Preparing for a Changing Climate: The Potential Consequences of Climate Variability and Change, P. J. Sousounis and J. M. Bisanz, Eds., Office of Research and Development Global Change Research Program, U.S. EPA, 25–28.
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, doi:10.1016/S0380-1330(02)70603-5.
Kunkel, K. E., M. A. Palecki, K. G. Hubbard, D. A. Robinson, K. T. Redmond, and D. R. Easterling, 2007: Trend identification in twentieth-century U.S. snowfall: The challenges. J. Atmos. Oceanic Technol., 24, 64–73, doi:10.1175/JTECH2017.1.
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, doi:10.1016/j.jglr.2008.11.003.
Kunkel, K. E., M. A. Palecki, L. Ensor, D. Easterling, K. G. Hubbard, D. Robinson, and K. Redmond, 2009b: Trends in twentieth-century U.S. extreme snowfall seasons. J. Climate, 22, 6204–6216, doi:10.1175/2009JCLI2631.1.
Liu, A. Q., and G. W. K. Moore, 2004: Lake-effect snowstorms over southern Ontario, Canada, and their associated synoptic-scale environment. Mon. Wea. Rev., 132, 2595–2609, doi:10.1175/MWR2796.1.
Livezey, R. E., and T. M. Smith, 1999: Covariability of aspects of North American climate with global sea surface temperatures on interannual to interdecadal timescales. J. Climate, 12, 289–302, doi:10.1175/1520-0442-12.1.289.
Moses, T., G. N. Kiladis, H. F. Diaz, and R. G. Barry, 1987: Characteristics and frequency of reversals in mean sea level pressure in the North Atlantic sector and their relationship to long-term temperature trends. J. Climatol., 7, 13–30, doi:10.1002/joc.3370070104.
Niziol, T. A., 1987: Operational forecasting of lake effect snowfall in western and central New York. Wea. Forecasting, 2, 310–322, doi:10.1175/1520-0434(1987)002<0310:OFOLES>2.0.CO;2.
Niziol, T. A., W. R. Snyder, and J. S. Waldstreicher, 1995: Winter weather forecasting through the eastern United States. Part IV: Lake effect snow. Wea. Forecasting, 10, 61–77, doi:10.1175/1520-0434(1995)010<0061:WWFTTE>2.0.CO;2.
NOAA, cited2014: Syracuse, New York annual temperature, precipitation, and snowfall data. National Weather Service, Binghamton, NY. [Available online at http://www.weather.gov/bgm/climateSYRAnnualTotals.]
Norton, D. C., and S. J. Bolsenga, 1993: Spatiotemporal trends in lake effect and continental snowfall in the Laurentian Great Lakes, 1951–1980. J. Climate, 6, 1943–1956, doi:10.1175/1520-0442(1993)006<1943:STILEA>2.0.CO;2.
Peace, R. L., Jr., and R. B. Sykes Jr., 1966: Mesoscale study of a lake effect snow storm. Mon. Wea. Rev., 94, 495–507, doi:10.1175/1520-0493(1966)094<0495:MSOALE>2.3.CO;2.
Schmidlin, T. W., 1993: Impacts of severe winter weather during December 1989 in the Lake Erie snowbelt. J. Climate, 6, 759–767, doi:10.1175/1520-0442(1993)006<0759:IOSWWD>2.0.CO;2.
Serreze, M. C., M. P. Clark, D. L. McGinnis, and D. A. Robinson, 1998: Characteristics of snowfall over the eastern half of the United States and relationships with principal modes of low-frequency atmospheric variability. J. Climate, 11, 234–250, doi:10.1175/1520-0442(1998)011<0234:COSOTE>2.0.CO;2.
Stocker, T. F., and Coauthors, 2013: Climate Change 2013: The Physical Science Basis. Cambridge University Press, 1535 pp. [Available online at http://www.climatechange2013.org/images/report/WG1AR5_ALL_FINAL.pdf.]
Wang, J., K. Bai, H. Hu, A. Clites, M. Colton, and B. Lofgren, 2012: Temporal and spatial variability of Great Lakes ice cover, 1973–2010. J. Climate, 25, 1318–1329, doi:10.1175/2011JCLI4066.1.
Zhao, C., Z. Yu, L. Li, and G. Bebout, 2010: Major shifts in multidecadal moisture variability in the mid-Atlantic region during the last 240 years. Geophys. Res. Lett., 37, L09702, doi:10.1029/2010GL043133.
