Decreasing Snow Cover and Increasing Temperatures Are Accelerating in New England, USA, with Long-Term Implications

Abstract

As the planet warms, not all regions are heating at the same rate. While North America is not one of the fastest warming continents, New England in Northeastern United States is warming faster than most other regions. This research evaluates how fast temperatures are rising and snow cover declining in New England. Three monthly mean air temperature data sets (minimum, average, maximum) from the United States Historical Climatological Network (USHCN) from 1900 to 2025 were used along with two MODIS/Terra satellite data sets: Land Surface Temperature and Emissivity Monthly L3 Global 0.05 Degree (MOD13C3) data and Snow Cover 8-Day L3 Global 0.05 Degree (MOD10C2) data (2000 to 2025). Univariate Differencing and the Mann–Kendall test were used on all three data sets to evaluate change over time at the seasonal and annual levels for New England and each of the six states. A time series regression analysis was undertaken to determine the relationship between snow cover and land surface temperature. Results show six major trends: (1) the existence of three distinct periods of temperature change, with most of the warming occurring since the late-1980s; (2) strong seasonal variations where winter is warming almost twice as fast as any other season; (3) minimum and nighttime temperatures are rising faster than maximum and daytime temperatures, especially since the 1980s; (4) snow cover is decreasing throughout New England, and rapidly in southern New England which has lost over 30–40% of snow cover days between 2000 and 2025; (5) there is a strong inverse relationship between snow cover change and land surface temperature change indicating that snow cover loss is a factor warming New England; and (6) most striking is the acceleration of temperature and snow cover decline in the past 5-year period. This research also discusses six major implications for these temperature and snow cover changes for New England.

1. Introduction

It is well established that the world is warming due to the greenhouse effect, and this warming is leading to a changing climate throughout the world [1]. Human activity, especially the burning of fossil fuels, is primarily driving the current greenhouse effect [2]. Not only is the world warming, but the warming has accelerated in recent decades [3]. Climate change is a global phenomenon with a variety of regional and local impacts, including an increase in severe storm activity, the melting of snow, glaciers and sea ice, warming of the oceans, bleaching of coral reefs, rising sea levels, changes in precipitation patterns, increased drought and flooding, heat waves, increased fire activity, changes in species’ physiology, phenology, species migration and human health impacts to list a few [4,5]. Climate change is already resulting in significant societal and environmental impacts and will induce major socio-economic damage in the future [6]. It is extremely important to understand regional aspects of climate change to create appropriate adaptation planning and to monitor the effects of mitigation strategies [7].

As the planet continues to warm, not all regions are heating at the same rate. The Polar and Arctic regions are warming the fastest followed by Europe and Asia [8,9]. While North America is not one of the fastest warming continents of the world, New England in Northeastern United States is warming faster than continental United States and the northeast’s coastal region is one of the fastest warming regions of the world [10]. Long-term global temperatures are mainly being driven by the increase in greenhouse gases, while regional variations are controlled by factors such as land vs. water (land heats faster than water), changes in wind patterns, land-cover changes and air quality [11,12].

New England, a region of six states in Northeastern United States (Connecticut (CT), Maine (ME), Massachusetts (MA), New Hampshire (NH), Rhode Island (RI), and Vermont (VT)), is warming faster than the global average [13] and is experiencing profound impacts, such as warming temperatures, precipitation changes leading to floods and droughts, rising sea levels, an influx of pests, increasing human health problems, disruptions to urban infrastructure, and others [14]. These changes are reshaping New England’s ecosystems, landscapes, economies, culture, and human activities [15,16,17,18].

New England has a four-season, humid continental climate where most of the region has a Koppen classification of Dfb with southern New England transitioning to a Cfa climate [19]. New England’s humid continental climate is reflected in its ecosystems, agriculture, economies, architecture, tourism, sports and culture [20]. Similarly to a number of regions in the world, this four-season climate is diminishing as New England warms [13]. In particular, minimum temperatures and winter temperatures are warming the fastest, and so, the winter season is disappearing, especially in southern New England, and summer weather is expanding earlier into the spring and later into the fall [16]. In many parts of the world, summers are expanding, and winters are shrinking [21].

With shorter winters, the number of days with snow cover is declining and thus the reflection of incoming solar radiation is declining with more solar radiation being absorbed [22] and leading to warmer land surface temperatures [23,24]. In northeast North America (which includes New England), snow cover is declining and areas with snow cover have a colder land surface temperature than areas without snow cover [25]. Globally, snow cover days are declining, and southern New England has been found to be a major area in the world quickly losing snow cover [26].

There are several ways to measure temperatures on Earth including thermometers in meteorological stations, ocean buoys, and satellites imagining the surface of the Earth, as well as climate proxies such as ice cores. This research uses three separate temperature data sets to study how temperatures are changing in New England, including the United States Historical Climatological Network (USHCN) station data from across New England, Land Surface Temperature (LST) data collected by the MODIS satellite, and snow cover data (temperature proxy) also from the MODIS satellite. These three distinctive data sets were used to analyze temperature and snow cover change in New England.

This research paper is an extension of an earlier publication by the authors [13], where they extensively evaluated the validity of the USHCN temperature data for New England stations and then, using that data, demonstrated that New England was warming faster than the global average. This new research adds five years to the USHCN data along with LST data and snow cover data for New England. The authors explore if past trends are continuing, and more importantly, explore new aspects of climate change in New England, focusing on the rate of change as well as evaluating snow cover change seeing that the region has been identified globally as an important area of declining snow cover. In addition to exploring both temperature and snow cover change, this research investigates the implications of these changes, which were not studied in the earlier publication.

Global warming and climate change are not occurring uniformly across the globe, and so, it is important to understand the regional differences in how warming and climate change occurs. This research paper explores the following:

(1)

How temperatures are changing in each of the New England states annually and seasonally, if the warming in New England is accelerating like other areas in the world, and if there are variations in the warming between different parts of New England.

(2)

The research explores how snow cover is changing in New England annually and seasonally as the temperatures warm, if there are variations in snow cover loss between different parts of New England, and if there is a relationship between declining snow cover and increasing temperatures.

2. Materials and Methods

This research analyzes temperature and snow cover change annually and seasonally for New England and each of the six states.

2.1. Air Temperature Analysis

Air temperature data are from the United States Historical Climatological Network (USHCN) data set (Version 2.5). Three air temperature data sets (monthly mean minimum, monthly mean average, and monthly mean maximum) were downloaded from the National Centers for Environmental Information website https://www.ncei.noaa.gov/products/land-based-station/us-historical-climatology-network (accessed on 6 August 2025). Although some station data start as far back as 1885, 1900 was chosen as the starting point because over 77% of the USHCN stations in New England have data for 1900 and beyond. In New England, annual temperature data can vary from year to year, with some years changing dramatically, thus potentially skewing the analysis. Therefore, in this study, the data were averaged into five-year units starting with 1900 to 1904. The five-year data units were graphed over time as well as differenced. Single-year data were not differenced because of the greater variability in a single year of data. Figure 1 outlines the air temperature methodology used in this research.

The USHCN station data have been extensively analyzed and adjusted over time to take into account the validity of extreme outliers, time of observation bias, changes in instrumentations, random relocations of stations, and urban warming biases [15,27,28]. The quality of the New England USHCN station data used in this study were reviewed using the Mann–Kendall test [29,30,31], and the standard normal homogeneity test [32,33]. Details on the review of the USHCN station data for New England can be found in [13]. There were no major problems discovered with any of the USHCN stations used.

For the temperature analysis of New England, 44 USHCN stations were used: CT: four stations, ME: 12 stations, MA: 12 stations, NH: five stations, RI: three stations, and VT: eight stations (Figure 2). The air temperature data sets (from January 1900 to February 2025) for the six New England states were downloaded and imported into Microsoft Excel. Concerning data preprocessing, if a station had a missing year of data, that year was skipped and not processed (6.1% of years were missing from the entire data set). For some months, there were one to five days of missing observations for the monthly means (9.6% of total months of the entire data set) and we tolerated one to five missing days of observations per month and used these values with the other monthly means. If there were more than five days of missing observations in a month, the data set used “−9999” as a placeholder and we filled these months with averages for that month using the average of four years, two years of that month before and two years after the missing data. If one of the two months before or after were missing, we moved on to the nearest year for that month. Only 2.2% of total months in the entire data set had missing months that were replaced with averages. No state had a disproportionate amount of missing data. All data preprocessing was conducted using the python programming language.

After the data were preprocessed, annual and seasonal (spring, summer, fall, winter) averages and anomalies were created using the following months: annual (January, February, March, April, May, June, July, August, September, October, November, December); spring (March, April, May); summer (June, July, August); fall (September, October, November); and winter (December of previous year, January and February of current year).

Different researchers have created annual and seasonal units using USHCN data and other temperature data sets to reduce noise and provide a clearer signal [34,35,36]. Creating five-year averages further reduces the noise for the annual and seasonal data as others have done [37,38,39]. Anomalies were created using the 30-year base period of 1951–1980 [39], which occurs near the middle of the entire time period and is used by other researchers [40]. Subtracting the 30-year average from each of the yearly annual and seasonal data points created the anomalies. The annual and seasonal anomalies from every station were then averaged into state-wide files (CT, RI, MA, ME, NH, and VT) as well as New England-wide files, and then, five-year (half-decade) units were created and then graphed and analyzed at the state and regional levels. A Univariate Differencing method [41] (subtracting the first five years (1900–1904) from the last five years (2020–2024)) was used to analyze temperature change over time. The differencing was performed for each of the states and for the entire New England region. After the differencing, a t-test (one tailed distribution, sample with equal variance) was run for every result to determine if it was significant at the 95th (p < 0.05) or 99th (p < 0.01) percentile. The five-year averages reduce the annual variability, and the t-tests indicate significance. The standard error and the 95% confidence intervals were also determined for all analysis. Note that because winter data (December–January–February) involves two different years, the winter five-year data sets start with the 1900–1901 winter to 1904–1905 winter, and the time series ends with the 2020–2001 to 2024–2025 period.

The Goddard Institute for Space Studies (GISS) Global Surface Temperature data (in monthly and annual means) were downloaded for the years 1900 to 2025 from the NASA Goddard Institute for Space Science (https://data.giss.nasa.gov/gistemp/ (accessed on 6 August 2025)) [40,42]. This data set used the 1951–1980 base period for the creation of temperature anomalies, the same base period used to process the USHCN data [39]. The GISS data set was used to see how New England’s temperature is changing compared with the global average.

2.2. Land Surface Temperature Analysis

This study used the MODIS/Terra Land Surface Temperature and Emissivity Monthly L3 Global 0.05 Degree Climate Modeling Grid (CMG), Version 6.1 (MOD13C3) data set. The MOD13C3 product provides monthly LST values in a 0.05-degree spatial resolution (5600 m at the equator) latitude/longitude CMG. The monthly LST values are derived from compositing and averaging the values from the corresponding month’s MOD11C1 daily files. The MOD13C3 data set is based exclusively on data collected during clear-sky conditions being filtered by the MODIS Cloud Mask product (MOD35_L2) [43,44]. The MODIS LST product was found to agree well with ground-based LSTs, where differences were comparable, or smaller, than the uncertainties of the ground measurements [45].

LST is a measurement of the “skin” temperature of the Earth as detected by satellites looking through the atmosphere at the surface of the Earth, and LST is not equivalent to air temperature, but they are related [46]. Air temperature measures the heat of the air at a standard height above the ground (generally between 1.5 and 2 m high), while LST measures the heat of the surface of the Earth (lakes, forests, fields, etc.). LST can be significantly higher or lower than air temperature depending on many factors such as time of day, land cover, land vs. water, etc. LST is especially influenced by solar radiation, wind, humidity and surface properties such as color, material (land cover), and moisture content, while air temperature is primarily driven by solar radiation, wind, and humidity [47].

Data were downloaded from NASA’s Land Processes Distributed Active Archive Center (LP DAAC) via the EarthDATA website (https://ladsweb.modaps.eosdis.nasa.gov/archive/allData/61/MOD11C3/ (accessed on 7 November 2025)). Downloaded data were from February 2000 (start of MODIS data collection) to March of 2025. The MOD13C3 data set has both monthly Day and monthly Night images, and these were analyzed separately for day and night. The MODIS sensor for the MOD13C3 product is onboard the Terra satellite which collects data at 10:30 a.m. (daytime) and 12:00 a.m. (nighttime) local time [43]. The MOD13C3 pixel values are in Kelvin. For this research, the pixel values were converted into Celsius (Pixel in Kelvin × 0.02) minus 273.15 [43]. After the pixels were converted into Celsius, the New England region and each of the states were extracted from the global data.

Data were processed into seasonal averages: winter (previous year December plus current year January and February), spring (March, April, May), summer (June, July, August), fall (September, October, November), and annual (January through December). Snow year files (previous year fall + current year winter and spring) were created to use in a time series regression with the snow cover data. For each season, the three monthly LST averages were added together and then divided by three, whereas for the annual data, the 12 months of the year were added together and divided by 12. Like the air temperature data, to reduce the influence of highly anomalous years, this study averaged the data into five-year average units, with the start of the time series being 2000–2004 (2001–2005 for the winter season) and ending with the period 2020–2024 (2021–2025 for winter).

This research used the Mann–Kendall test and Univariate Differencing to determine how LST was changing in New England from 2000 to 2024. The Mann–Kendall test in the Earth Trends Modeler in the TerrSet LiberaGIS software (version 20.04) [48] was used to analyze the MOD13C3 time series data. The Mann–Kendall test produces p- and Z-values for each pixel. The p-value provides the level of significance, and the Z-value is the number of standard deviations in the positive or negative direction of change. Positive Z-values indicate a warming trend, and negative Z-values indicate a cooling trend. To show where significant changes occurred, a significance map (p < 0.05; p < 0.01) was created to show significant changes in LST at the 95th and 99th percentiles for New England. A map of Z-values was created with the following levels: >2.5 standard deviations (SD); 2.5 SD to 1.5 SD; 1.5 SD to 0.5 SD; and +0.5 SD to −0.5 SD. There were no values beyond −0.5 SD. Only nighttime data were mapped because daytime results were highly limited in significance.

The Overlay function in the TerrSet LiberaGIS software was used to perform the Univariate Differencing where, on a pixel-by-pixel basis, the earliest five-year period (2000–2004) was subtracted from the last five-year period (2020–2024). The results indicate how temperature for each pixel in the image changed between the early period to the late period. The MOD13C3 and the MOD11C1 series temperature data have been widely used to measure and analyze LST at the regional and global scales [25,46,49,50,51]. Figure 3 outlines the LST methodology used in this research.

2.3. Snow Cover Analysis

This study used the MODIS/Terra Snow Cover 8-Day L3 Global 0.05 Degree Climate Modelling Grid (CMG), Version 61 (MOD10C2) data to map snow cover change in New England. The MOD10C2 data have a sinusoidal, equal-area projection, and is appropriate for the analysis of snow cover change in New England. The MOD10C2 data set has a spatial resolution of 0.05-degrees and is an aggregation of MOD10A2 products with 500 m spatial resolution [52]. The data were downloaded from NASA’s Earth Data website (https://search.earthdata.nasa.gov/search (accessed on 21 June 2025)). Version 61 data were downloaded and processed for the whole period from 24 February 2000 to 1 March 2025. The MOD10C2 data is an eight-day composite, where the maximum values for each pixel over eight days are recorded and this reduces persistent cloudiness (clouds have lower values than land and snow values). This reduction in cloud cover is particularly important for mid-to-high latitudes [53]. However, MODIS snow products do have some issues concerning cloud cover: the difficulty of detecting snow in forest areas, and topography that may impact the accuracy of the results [54,55,56]. Clouds not only obscure snow cover or lack of snow cover, but there can also be snow-cloud confusion, such as when thin cirrus, ice clouds and cloud shadows may be misclassified as snow, or sometimes snow can be misclassified as clouds because the spectral characteristics overlap each other. This can be particularly evident in forest ecosystems [52,56]. In regions of complex terrain (steep slopes) or under low solar elevation (winter and/or high latitudes), snow reflectance signals are weaker, and snow/cloud discriminations are harder [52,56]. An additional issue with the eight-day composites is that short-lived snowfalls that occur then melt before cloud cover is dissipated can appear as if no snow ever fell. The eight-day composites also cause timing uncertainty because if snow appears or disappears in a new eight-day composite, it is unsure which of the eight days the snow action happened. Despite these issues, the MOD10C2 product is valuable and has been used in numerous studies [25,57,58,59,60].

Concerning our interpretation of these snow cover uncertainties, first, there are very few areas in New England with complex terrain issues, and over the entire time period of the data (2000–2025), there has not been a major transition of land cover in New England; so, for comparability reasons, throughout this period, there has been consistent land cover, and thus, all time periods had similar issues. Concerning the problem of low sun angles, again we are comparing time periods with the same sun angles and similar issues. Concerning the question of not knowing the exact date of snow fall and snow melt, this research does not analyze the exact timing of events but rather analyzes broad trends of snow cover duration. To further reduce the influence of cloud contamination, the MODIS cloud cover layer was added to the MODIS snow cover layer when the eight-day cloud cover pixels plus eight-day snow cover pixels equaled a value of 100% and were surrounded by snow cover pixels with a value of 100%. Here, it was assumed that clouds were covering 100% snow cover and were classified as 100% snow cover. A review of cloud contamination was performed on every eight-day file in New England for the first and last five years of the data set (not including summer data which were not used in the study) and found the average of cloud contamination to be less than 2%, and so, persistent cloud contamination is considered a minor issue here.

MOD10C2 pixel values are the maximum percentage of snow cover (0% to 100%) for the pixel’s area for eight continuous days [61]. Snow cover is created by the Normalized Difference Snow Index (NDSI) which uses MODIS band 4 (green) (0.545 μm to 0.565 μm) and band 6 (near-infrared) (1.628 μm to 1.652 μm). The greatest deficiency of the MOD10C2 data set is cloud cover contamination, while other issues, such as errors of commission, have been found to be very low [62]. Data were processed into seasonal averages: winter (previous year December plus current year January and February), spring (March, April, May), and fall (September, October, November). The summer season was not processed because there is very little snow in New England in the summer and the addition of this season would create more noise than signal. An annual product was created based on the snow year, or hydrologic year, without the summer (previous year fall plus current year winter and spring) [63]. Seasonal data for air temperature, snow, and LST cover the same time periods, but the annual snow cover data are different (hydrologic year) from the air temperature and LST data (calendar year).

This research used the Mann–Kendall test to determine if snow cover values were significantly increasing, decreasing, or staying the same for New England. The Mann–Kendall test is used to statistically assess the upward or downward trend of snow cover over time (2000–2024) [29,30,64,65]. The Earth Trends Modeler in the TerrSet LiberaGIS software was used to analyze the MOD10C2 time series data, including the Mann–Kendall test which has p- and Z-value outputs. The p-value provides the level of significance, and the Z-value is the number of standard deviations in the positive or negative direction of change. Positive Z-values indicate increasing snow cover, and negative Z-values indicate decreasing snow cover.

The Mann–Kendall statistical test has been frequently used to quantify the significance of trends in meteorological time series [31,66,67], and snow cover analyses [59,63,68,69]. To show where significant changes occurred in snow cover, two significant maps (p < 0.05; p < 0.10) were created to show significant changes in snow cover at the 95th and 90th percentiles for New England. When analyzing and mapping the Z-values, a map was created with Z-values clustered into the following levels: <−2.5 standard deviations (SD); −2.5 SD to −1.5 SD; −1.5 SD to −0.5 SD; and −0.5 SD to +0.5 SD. In addition to mapping Z-values, percent of snow cover loss in New England based on the results of the Univariate Differencing of the snow data was also mapped [((2020–2024) minus (2000–2004)) divided by (2000–2004)].

Similarly to the air temperature and LST data, to reduce the influence of highly anomalous years, this study averaged the data into five-year average units, with the start of the time series being 2000–2004 (2001–2005 for the winter season and for the annual snow-year) and ending with the period 2020–2024 (2021–2025 for the winter and snow year). As used in air temperature and LST data analysis, the Univariate Differencing technique was also used to analyze snow cover change. This method is used in a variety of satellite-based change detection studies [70], including MODIS-based studies [25,71,72]. Here, the average of the first five years was subtracted from the average of the last five years. This was performed for each season and for the annual snow year data.

The Univariate Differencing method allows for the use of approximate change in days. For example, the winter season has 90 days, and if a pixel has an average of 100 (100% snow cover), then it had 90 days of snow cover (1.00 × 90 = 90 days). If the pixel had an average value of 50%, then it would have 45 days of snow cover (0.50 × 90 = 45 days). For Univariate Differencing analysis, if the pixel in the last five years had a value of 0.90 and in the first five years it had a value of 1.00, then during the period, snow cover would have declined 9 days (0.9 minus 1.0 = −0.1) (−0.1 × 90 = −9.0 days). These are approximate days because the eight-day maximum value composites create an uncertainty of how many actual days had snow cover. However, because of the averaging of the first and last five years and the use of the same methodology throughout the time series, outliers are not as prominent. The use of days of change creates a way to show the intensity of change and is not an exact measurement of days with and without snow cover.

To investigate if there is a relationship between snow cover loss and land surface temperature rise, a time series regression in the Earth Trends Modeler in the TerrSet LiberaGIS software was run between the snow cover data (MOD10C2) (independent variable) and the LST data (MOD13C3) (dependent variable). The time series used snow year data (previous year fall + current year winter and spring) for the first five years and the last five years of both the snow cover and LST data sets: 2000–2001, 2001–2002, 2002–2003, 2003–2004, 2004–2005, 2020–2021, 2021–2022, 2022–2023, 2023–2024, and 2024–2025. These ten images capture the situation in the early 2000s and the early 2020s, the end periods of the MODIS data set. The outcomes of the regression are values between −1 and +1, where values in the negative direction indicate an inverse relationship and those with a positive value indicate a positive correlation. The closer to −1 and +1, the stronger the negative and positive correlation. The research framework can be seen in Figure 4.

The MOD10C2 snow data (0.05-degrees resolution) were used in this study instead of the MOD10A2 snow data (500 m resolution) because the MOD10C2 data are a global scale data set which captures regional to global trends while the MOD10A2 data captures more local influences, and this research evaluates broader trends and not local trends. Multiple studies have shown that coarse resolution data emphasize large-scale regional patterns, while finer scale data reveal more local patterns [73,74]. This is the same reason why the LST MOD13C3 data were used instead of the MOD11C1 data. No generative artificial intelligence was used for this research.

3. Results

3.1. Air Temperature Change

In New England, minimum, average and maximum air temperatures have been rising since 1900 with minimum temperatures rising the fastest (Figure 5,

Table 1. New England 5-year Temperature Change Values (2020–2024) minus (1900–1904) in degrees Celsius.

State	Annual a	Spring b	Summer c	Fall d	Winter e	Over 1.5 °C f
Connecticut
(4 USHCN stations)
Maximum	1.48 **	1.40 *	0.30	0.89	2.99 **	10/15
Average	2.46 **	1.30 *	1.72 *	1.95 **	4.48 **
Minimum	3.69 **	2.38 **	3.60 **	2.92 **	5.24 **
Maine
(12 USHCN stations)
Maximum	1.73 **	1.10	1.59 *	0.62	3.22 **	11/15
Average	2.63 **	1.48 *	2.02 **	2.18 **	4.38 **
Minimum	3.07 **	1.49 *	2.40 **	2.44 **	5.41 **
Massachusetts
(12 USHCN stations)
Maximum	2.62 **	1.70 *	2.30 *	2.23 **	3.84 **	14/15
Average	2.75 **	1.64 **	2.57 **	2.24 **	4.26 **
Minimum	2.92 **	1.28 *	2.78 **	2.33 **	5.03 **
New Hampshire
(5 USHCN stations)
Maximum	1.88 **	1.17	1.55 *	1.37 *	2.97 **	12/15
Average	2.41 **	1.36 *	2.03 **	1.91 **	3.86 **
Minimum	2.95 **	1.57 *	2.63 **	2.34 **	4.76 **
Rhode Island
(3 USHCN stations)
Maximum	2.73 **	1.95 **	2.00 *	2.51 **	4.41 **	13/15
Average	2.38 **	1.42 *	2.12 **	1.95 **	3.80 **
Minimum	1.89 **	0.89	1.93 **	1.68 *	3.30 **
Vermont
(8 USHCN stations)
Maximum	2.55 **	1.91 *	2.14 **	2.31 **	3.38 **	13/15
Average	2.55 **	1.40	2.07 **	2.21 **	4.01 **
Minimum	2.65 **	0.73	1.89 **	2.47 **	5.53 **
New England
(44 USHCN stations)
Maximum	2.17 **	1.54 *	1.65 *	1.66 **	3.42 **	13/15
Average	2.53 **	1.43 *	2.09 **	2.07 **	4.13 **
Minimum	2.77 **	1.33 **	2.50 **	2.21 **	4.74 **

^a Annual refers to the calendar year (January–December). ^b Spring refers to meteorological spring (March, April and May). ^c Summer refers to meteorological summer (June, July and August). ^d Fall refers to meteorological fall (September, October, and November). ^e Winter refers to meteorological winter (December of prior year plus January and February of current year). ^f This column shows the number of data points over 1.5 °C for the 15 data points per state: 5 (1 annual + 4 seasons) × 3 (Tmax + Tave + Tmin) = 15. * Significant at the 95th percentile (p < 0.05). ** Significant at the 99th percentile (p < 0.01). Numbers with Italic font were at 1.5 °C or above. Numbers with Italic font and Bold were at 2 °C or above.

). Average air temperatures rose a little more than 1 °C from early 1900 to the early 1950s and then cooled a little more than 0.5 °C until the mid-1970s, after which air temperatures have risen about 2 °C, creating a warming of approximately 2.5 °C for New England since the early 1900s (Figure 5, Table 1). The average mean temperature for New England and each of the states has warmed more than 2.38 °C since 1900 (Table 1).

Minimum temperatures have been warming faster than maximum temperatures for New England and all states for every season except for RI and spring in MA and spring and summer in VT (Table 1). Since the 1990–1994 period, New England’s minimum temperatures have been rising faster than maximum temperatures, apart from a brief pause in the 2000–2004 period, with the biggest difference occurring in the past five years as New England’s warming has accelerated (Figure 6). Winter temperatures are warming the fastest, almost twice as fast as in any other season. Winter minimum temperatures for New England and all states have risen more than 4 °C, except RI where winter maximum temperatures have risen faster and above 4 °C. Winter minimum temperatures for CT, MA, ME, and VT have risen more than 5 °C. After winter, summer and fall seasons are warming faster than spring in most cases (Table 1). Spring temperatures at the minimum, average and maximum levels have been highly variable for all states and have the lowest significant levels compared to the other seasons overall. For New England and every state, except for RI and VT summer, all minimum summer temperatures have warmed more than 2 °C. Concerning minimum, average and maximum temperature changes for all seasons and annually for all states and New England, 82% of the data points have warmed above the 1.5 °C threshold that the world is trying to stay below [75] (Table 1). Results at the annual level and the winter season for New England and all states are significant at the 99th percentile, and all change data points for New England are significant at or above the 95th percentile, being the most (12 out of 15 data points) significant at the 99th percentile (Table 1).

Warming in New England has accelerated where temperatures have increased the most in the last five years for every state and New England at the minimum annual level (significant at the 99th percentile for New England and all states, except for Rhode Island, which was significant at the 95th percentile) (

Table 2. New England 5-year Temperature Change Values (2020–2024) minus (2015–2019) in degrees Celsius.

State	Annual a	Spring b	Summer c	Fall d	Winter e	Significance f
Connecticut
(4 USHCN stations)
Maximum	0.60	1.01	0.36	−0.35	1.00	3/15
Average	0.31	0.41	0.15	−0.10	0.66
Minimum	1.04 ***	1.03 **	1.38 ***	0.43	0.91
Maine
(12 USHCN stations)
Maximum	0.55 **	1.14 *	0.33	−0.58	1.24	5/15
Average	1.04 ***	1.35 ***	0.46	0.56	1.72 *
Minimum	1.07 ***	1.18 **	0.60	0.40	2.03 *
Massachusetts
(12 USHCN stations)
Maximum	0.77 ***	1.11 *	0.19	0.59	1.01	7/15
Average	0.89 ***	1.05 **	0.46	0.53	1.30 *
Minimum	1.05 ***	0.71 **	0.78 **	0.65 *	1.83 **
New Hampshire
(5 USHCN stations)
Maximum	0.46	0.85	−0.29	0.10	0.98	5/15
Average	0.94 ***	1.19 **	0.54	0.44	1.38 *
Minimum	1.22 ***	1.38 ***	1.12 ***	0.40	1.70 *
Rhode Island
(3 USHCN stations)
Maximum	0.65	0.87	0.03	0.36	1.29 **	3/15
Average	0.58	0.74	0.41	0.17	1.01
Minimum	0.64 **	0.77 *	0.67 **	−0.15	1.28 *
Vermont
(8 USHCN stations)
Maximum	0.77 **	1.40 **	0.30	0.62	0.72	5/15
Average	0.95 ***	1.28 *	0.63 *	0.57 *	1.20
Minimum	0.88 ***	1.11 **	0.79 *	0.68	0.77
New England
(44 USHCN stations)
Maximum	0.63 *	1.06 *	0.15	0.12	1.04 *	5/15
Average	0.79 ***	1.00 *	0.44 **	0.36	1.21 *
Minimum	0.90 ***	0.97 **	0.81 ***	0.33	1.27 *

^a Annual refers to the calendar year (January–December). ^b Spring refers to meteorological spring (March, April and May). ^c Summer refers to meteorological summer (June, July and August). ^d Fall refers to meteorological fall (September, October, and November). ^e Winter refers to meteorological winter (December of prior year plus January and February of current year). ^f This column shows number of data points at 95 and 99th percentiles for the 15 data points per state: 5 (1 annual + 4 seasons) × 3 (Tmax + Tave + Tmin) = 15. * Significant at the 90th percentile (p < 0.1). ** Significant at the 95th percentile (p < 0.05). *** Significant at the 99th percentile (p < 0.01). Numbers with Bold font were at the 95th and 99th percentiles.

). In the last five years, average temperatures have increased the most for New England and for the states of MA, ME, NH and VT at the 99th percentile (Table 2). Although warming in New England has been happening since at least 1900, most of the warming (>50%) at the minimum, average and maximum levels have happened since the 1985–89 period for New England and all states, except maximum temperatures for CT (Figure 5 and

Table 3. Percent of warming since 1985–1989 compared to warming since 1900–1904 ^a.

Area	Minimum	Average	Maximum
CT	66%	55%	41%
RI	87%	72%	60%
MA	79%	64%	55%
ME	91%	84%	67%
NH	84%	81%	51%
VT	80%	80%	70%
NE	80%	75%	68%

^a [(2020–2024 minus 1985–1989) divided by (2020–2024 minus 1900–1904)].

). For New England, 75% of mean average temperatures since 1900 have warmed since the late 1980s, while 80% of the minimum temperatures have warmed since the late 1980s (Table 3). For the northern New England states of ME, NH and VT, over 80% of the mean annual temperature rise has occurred since the 1985–89 period, with 91% of ME minimum temperature rise occurring since the 1985–89 period (Table 3).

Air temperatures in New England are increasing faster than the global average (Figure 7), and over the last five years, the warming in New England has accelerated, especially for minimum temperatures (Figure 6). For most periods since 1900, New England’s average temperatures have been warming faster than the global average, and for every five-year period since the late 1940s, except for the 1985–89 period, New England has warmed significantly more than the global average with the greatest difference occurring in the last five-year period during which New England’s temperatures accelerated (Figure 7).

3.2. Land Surface Temperature Change

Unlike the USHCN air temperature data which records minimum, average and maximum temperatures, the MODIS LST data records daytime (10:30 a.m.) and nighttime (12:00 a.m.) temperatures. Although not always the case, nighttime temperatures are usually lower than daytime temperatures, and so, there is some correlation between nighttime and minimum temperatures and daytime with maximum temperatures, especially for the mid-latitudes. Another difference, as mentioned before, is that air temperature is measured above the Earth’s surface while LST is a measurement of the Earth’s surface temperature.

The results of the LST data analysis are very similar to the USHCN station data results where the New England region and all states are warming (both the daytime and nighttime data), the winter season is warming faster than any other season for New England and all of the states (both the daytime and nighttime data), and the nighttime (minimum) temperatures are warming faster than the daytime (maximum) temperatures (

Table 4. Land Surface Temperature (2020–2024 average minus 2000–2004 average) Night and Day in degrees Celsius.

Region	Day/Night	Spring a	Summer b	Fall c	Winter d	Annual e	USHCN f
Connecticut	Day	0.370	0.982	1.224	1.600 *	0.457	−0.07
	Night	1.114	1.283 *	1.458 *	1.749 **	1.429 **	1.80 *
Rhode Island	Day	0.218	0.809	0.847	1.270 *	0.248	1.39
	Night	1.238	1.307 *	1.462 *	1.635 **	1.297 **	0.95
Massachusetts	Day	0.621	0.899	1.208	1.536	0.749	0.92 *
	Night	1.367	1.280 *	1.624 *	1.792 **	1.454 **	1.38 *
Maine	Day	0.813	0.917	0.883	0.929	0.586	0.88
	Night	1.595 *	1.483 *	1.445 *	2.129 **	1.694 **	2.29 **
New Hampshire	Day	0.583	0.786	1.188	1.215	0.558	0.81
	Night	1.445	1.103	1.553 **	1.706 **	1.498 **	1.96 *
Vermont	Day	0.971	0.733	1.171	1.638	0.702	1.22 *
	Night	1.679 *	1.061 *	1.615 **	1.659 **	1.529 **	1.57 *
New England	Day	0.738	0.873	1.030	1.199 *	0.600	0.86 *
	Night	1.52 **	1.327 **	1.510 **	1.911 **	1.599 **	1.66 **

^a Spring refers to meteorological spring (March, April and May). ^b Summer refers to meteorological summer (June, July and August). ^c Fall refers to meteorological fall (September, October and November). ^d Winter refers to meteorological winter (December of prior year plus January and February of current year). ^e Annual refers to the calendar year (January–December). ^f USHCN annual data—here for comparison with LST data. USHCN maximum is similar to LST Day and minimum is similar to LST Night. USHCN data is for the same time period for LST data [(2020–2024) minus (2000–2004). * Significant at the 95th percentile (p < 0.05). ** Significant at the 99th percentile (p < 0.01).

). The change in USHCN annual temperatures is similar in magnitude to the change in LST annual temperatures for the differencing between the two five-year periods (2020–2024) and (2000–2004) (Table 4). For the LST data, very few of the daytime results were significant (8%), while most of the nighttime results were significant (85%) (Table 4).

Like the USHCN results, the LST results show that the warming in New England has accelerated (both daytime and nighttime data) over the past five years (Figure 8). Throughout the time period, the nighttime temperatures have risen more than 2.5 times the daytime temperatures (1.599 °C to 0.600 °C) (Table 4). Both the nighttime and daytime data show that the warming of the land surface of New England accelerated in the last five years, similarly to the acceleration of the USHCN air temperature data (Figure 5 and Figure 8). The nighttime LST data shows a significant rise in temperature (Figure 8).

The results of the Mann–Kendall test show widespread warming throughout much of New England and especially in southern New England and eastern coastal New England (Figure 9). Only the nighttime data were mapped because most of the daytime change was not significant. Concerning the nighttime data, more than half of New England saw significant warming at or above the 95th percentile (Figure 9). The results clearly show that the land surface of New England is warming in a similar widespread manner like the warming air temperature. Both the LST and USHCN data show that warming accelerated in the past five years.

3.3. Snow Cover Change

Concerning snow cover analysis for New England, the summer season (June, July, August) was not analyzed because there is almost no snow cover during this season. Every state in New England, and the entire region, has lost snow cover annually (fall, winter, spring) and during every season, between the early 2000s and the early 2020s (

Table 5. Change in number of days with snow cover by season for each region, between five-year average periods: 2000–2004 and 2020–2024.

Region	Spring	Fall	Winter	Annual
CT-2000-04 a	13.48	5.83	62.27	81.59
CT-2020-24 a	7.25	2.00	42.04	51.28
Difference b	6.24	3.83	20.23	30.30
% Change c	−46%	−66%	−33%	−37%
RI-2000-04 a	13.79	7.51	57.71	79.00
RI-2020-24 a	5.40	2.49	39.02	46.92
Difference b	8.39	5.01	18.68	32.09
% Change c	−61%	−67%	−32%	−41%
MA-2000-04 a	19.91	7.76	69.78	97.44
MA-2020-24 a	11.05	3.97	51.34	66.36
Difference b	8.86	3.79	18.44	31.09
% Change c	−45%	−49%	−26%	−32%
ME-2000-04 a	44.38	17.37	85.93	147.68
ME-2020-24 a	35.34	14.41	81.49	131.24
Difference b	9.04	2.96	4.45	16.44
% Change c	−20%	−17%	−5%	−11%
NH-2000-04 a	35.00	11.73	82.21	128.93
NH-2020-24 a	26.06	8.51	75.64	110.21
Difference b	8.94	3.22	6.57	18.73
% Change c	−26%	−28%	−8%	−15%
VT-2000-04 a	39.35	12.34	84.58	136.27
VT-2020-24 a	29.38	9.28	79.16	117.82
Difference b	9.97	3.06	5.42	18.45
% Change c	−25%	−25%	−6%	−14%
NE-2000-04 a	36.74	13.69	81.06	131.49
NE-2020-24 a	27.70	7.57	73.10	108.37
Difference b	9.04	6.12	7.96	23.12
% Change c	−25%	−45%	−10%	−18%

^a Numbers represent the number of days with snow cover average for the entire state. ^b Numbers represent the difference in the number of days with snow cover between five-year average periods: 2000–2004 and 2020–2024. All values are a decrease in number of days with snow on the ground. ^c Per cent change = [(2020–2004 minus 2000–2004) divided by (2000–2004)].

). The number of days with snow on the ground has decreased across the entire region. Southern New England (CT, RI, and MA) has fewer days with snow cover than northern New England (ME, NH, and VT) and has been losing days with snow cover faster than northern New England. In the 2021–2025 annual period, southern New England on average had less than 55 days annually with snow cover (CT + RI + MA/3) and lost more than 36% of snow cover between the 2001–2005 period and 2021–2025, while northern New England had over 119 days of annual snow cover (ME + NH + VT/3) and losses at 15% and lower (Table 5).

For every state and New England, percent of snow cover has declined faster in the spring and fall seasons, especially for southern New England. For example, RI has lost 61% of snow cover in the spring and 67% in the fall while MA and CT lost more than 40% in spring and fall. Northern New England also had the highest percentage of snow cover loss in the spring and fall, but their losses were less than 30% for these seasons (Table 5). In both RI and CT, there is now less than 10 days with snow on the ground for the state average in the spring and fall. RI and CT have been losing snow cover the fastest, while ME has been losing snow cover the slowest, though ME is also losing snow cover, including an 11% annual average decline between the early 2000s and the early 2020s (Table 5). Although spring and fall are the seasons losing the greatest percent of snow cover, for southern New England, the greatest number of days with snow cover loss is the winter season with CT, RI and MA losing more than 18 days of winter snow cover between 2001–2005 and 2021–2025 and northern New England losing less than 7 days of winter snow cover (Table 5).

Similarly to changes in air temperature, the decrease in snow cover has accelerated in the last five-year period. The New England region and every state have been losing snow cover in the last two five-year periods with the greatest loss of any period being the last five years, except for VT which lost more earlier (

Table 6. Percent change from each preceding five-year period for annual change ^a.

	CT	MA	ME	NH	RI	VT	NE
2000–2004	0.00	0.00	0.00	0.00	0.00	0.00	0.00
2005–2009	−1.87	−4.22	−0.82	−2.22	−2.80	−3.48	−1.92
2010–2014	0.48	1.04	2.01	1.54	0.02	1.85	1.66
2015–2019	−5.72	−4.08	−0.21	−2.59	−4.05	−3.02	−1.98
2020–2024 b	−5.75	−5.87	−7.74	−4.86	−6.53	−2.68	−6.13

^a Annual period = Fall (prior year) + Winter (current year) + Spring (current year). ^b All 2020–2024 results significant at the 99th percentile, except VT which is at the 95th percentile. Bold = five-year period of greatest change.

, Figure 10 and Figure 11). ME, which has lost the least amount of annual snow cover days since 2000 (Table 5), has now experienced an acceleration of annual snow cover loss faster than any other state in the last five-year period (Table 6). During the time from 2000 to 2024, only the 2010–2014 period saw a slight increase in snow cover, with the other three periods showing a decline in snow cover (Table 6, Figure 10).

Figure 12 shows the spatial distribution of the percentage of annual snow cover loss in New England and the decreasing Z-values from the Mann–Kendall analysis. This figure clearly shows the greatest loss in snow cover in southern New England, stretching north along coastal ME and stretching north along western VT. The mountainous areas of western MA, VT and NH along with Northern ME are regions of least snow cover decline (Figure 12). Figure 13 shows the areas in New England that experienced a significant decline in snow cover at the 95th and 90th percentiles. The map of the 90th percentile provides an indication of where snow cover decline will potentially continue. The areas of significant LST increase are similar to the areas of decreasing snow cover, indicating a potential relationship between decreasing snow cover and increasing LST (Figure 9 and Figure 12).

To investigate if there is a relationship between snow cover loss and land surface temperature rise, a time series regression was run between the snow cover data (MOD10C2) and the LST data (MOD13C3) where the snow cover data were the independent variables and the LST data were the dependent variables. The time series used snow year data (previous year fall plus current year winter and spring) for the first five years and the last five years of both the snow cover and LST data sets: 2000–2001, 2001–2002, 2002–2003, 2003–2004, 2004–2005, 2020–2021, 2021–2022, 2022–2023, 2023–2024, and 2024–2025. These years were chosen to capture the situation at both ends of the time period.

The results show a strong inverse relationship (values = −0.9 to −1) throughout much of the region (Figure 14). An inverse relationship means that as snow cover declines, land surface temperatures increase. The resulting pattern clearly shows that areas experiencing intense snow cover loss (southern New England and coastal ME) are warming, which would indicate a positive feedback loop between snow cover loss and the warming of areas in New England. Southeastern MA, which has lost snow cover, does not appear significant here probably because although a high percentage of snow cover has been lost, there was not much snow cover here, so the change in albedo was not as great as areas with much greater snow cover. Two areas which show up here more prominently than in the snow cover loss images (Figure 12) are the Champlain Valley in western VT and the Connecticut River Valley along the border between NH and VT. Both Lake Champlain and the Connecticut River have been warming, which would indicate their appearance here [76,77]. There were no positive correlations in the time series regression.

4. Discussion

4.1. Important Observations

Although climate is changing at a global scale, due to geographically specific climate feedback mechanisms, not all regions are experiencing the same rate of change [7,78,79,80], which reflects the fact that our present-day warming is a complex and spatially diverse process. Therefore, it is important to explore how climate change is occurring in diverse regions of the world. Current warming is greater over land than over the oceans, and there are major differences between the world’s land areas that are warming [81]. Areas warming quickly include the following: the Polar regions [82], the Arctic regions [83], Europe, which is the fastest warming continent [8,84], Central Asia [85], the Eastern Mediterranean and West Asia [86], and northeastern North America [87].

This research paper is based on a 125-year (1900–2024) series of annual and seasonal temperature values from 44 USHCN meteorological stations in New England along with 25 years (2000–2024) of LST and snow cover data from the Terra MODIS satellite and has yielded six important observations.

(1)

The existence of three major periods of change in New England are as follows: (1) 1900 to the early 1950s with about a 1 °C temperature rise, (2) the 1950s to the late 1960s with a decline of about 0.5 °C, and (3) the 1970s to 2024 with a rise of about 2 °C, with a sharp rise in temperature since the mid-1980s. New England’s pattern of change roughly parallels the global patterns of change, though New England warms more quickly overall. The early warming period from 1900 to 1950 was driven by both natural factors (higher solar irradiance, volcanic activity, internal variability) and early anthropogenic forcing, from greenhouse gases and land-use changes [88]. The decline in temperatures from the early 1950s to the 1970s is due to air pollution and increased aerosols in the atmosphere [89], which reflected incoming solar radiation out of the Earth system and affected New England more than the globe. The last period of warming from the 1970s to 2025 has been driven largely by human activity, especially the burning of fossil fuels as well as land cover change increasing atmospheric greenhouse gases along with declining atmospheric aerosols [3].

(2)

There are strong seasonal variations in the warming of New England where winter is warming almost twice as fast as any other season, and the winter season shows more consistent significant changes than any of the seasons as indicated in both the air temperature and LST data sets. Europe, across the Atlantic Ocean from New England, which is warming just as fast, saw minimal variations between seasons, though spring and summer are warming faster than the other seasons and the European winter exhibits more variation than in New England [11,84]. Other regions of the world that once experienced prominent winter warming are now warming faster in the spring including Central Asia and northwest China [85,90]. With New England warming the fastest in winter, snow cover has been declining throughout the region, paralleling the warming temperatures. The seasonal changes occurring in New England reflect global and hemispheric models which show spring arriving earlier and autumn arriving later, and projections estimate that summer will last nearly half a year and winter less than 2 months by 2100 [21]. Changes in temperature for different seasons not only alter the boundaries between seasons but also change crop phenology, disrupt migration and animal reproductive periods, expand the range of invasive species, and lead to various risks and disasters, such as increasing heatwaves, wildfires, floods, and early snowmelt among other climate-induced seasonal problems [91].

(3)

In New England, annual minimum air temperatures are rising faster than maximum air temperatures and nighttime LST are rising faster than daytime LST, reducing the diurnal temperature range (DTR), especially since the 1980s for the air temperatures. Minimum air temperatures in every season, except springtime, are rising faster than maximum air temperatures, especially in the winter when they have warmed almost 5 °C. In every state, except RI, annual, fall and winter minimums have been rising faster than maximum temperatures, and in all states except RI and VT summer, minimums are rising fastest. Globally, the opposite is happening where maximum temperatures are rising faster with a major reversal happening around 1988, which affected nearly half of the world’s land areas, particularly in central Eurasia, South America, Western Australia, Northern Europe, and Greenland [92]. Interestingly, in the late 1980s is when the minimum temperatures in New England started to increase much faster than maximum values. With minimum air temperatures in New England rising faster than maximum temperatures, particularly in late autumn, winter, and early spring, results in precipitation arriving as rain instead of snow more often [93,94]. This is reflected in the loss of snow cover seen throughout New England in this study. The LST data showed a similar pattern to the air temperature data with nighttime temperatures (minimum) rising faster, and more significantly, than the daytime (maximum) temperatures. Other LST studies have found nighttime LST warming faster and more significantly than daytime LST [50].

(4)

Snow cover is decreasing throughout New England, especially in southern New England (CT, RI, MA). Although snow cover has been declining quicker in southern New England, snow cover decline is now also accelerating in northern New England (ME, NH, VT). Snow cover has broadly been declining across the globe, especially in the spring season [26,95]. Southern New England is a global hot spot of snow cover decline being in the top 5% of regions losing snow cover between 2000 and 2020, which also includes the Andes of South America, Northeast China, and Southeast Europe [26]. The warming that New England has been experiencing, especially the increasing minimum and nighttime temperatures, has allowed the snow/rain ratio to decline where cold weather snow is now falling as rain more often [96]. Global studies have shown that at the mid- and low-latitudes in the Northern Hemisphere, the snow/rain ratio has a decreasing trend, with latitude and altitude being strong predictors of changes in this ratio [97,98]. The ratio of snow to total precipitation (s/p) is a hydrologic indicator that is sensitive to climate variability, and this ratio has been declining in New England for some decades [93].

(5)

Declining snow cover appears to be a factor in the warming of some areas in New England. A time series regression between snow cover data and LST data showed a widespread strong inverse relationship across southern New England and coastal ME, areas of intense snow cover loss. An inverse relationship means that as snow cover declines, land surface temperatures increase. Southern New England sits at the sweet spot of the mid-latitudes where warming is driving the snow/rain ratio to decline, and thus, snow cover days are declining, which, in turn, might be warming New England with the snow albedo feedback [99]. Snow is highly reflective and sends incoming solar radiation back to space. In the absence of snow, more of the incoming solar radiation gets absorbed and not reflected, thus warming the land [23,99]. The close relationship between areas of snow cover loss and LST warming, as revealed in the time series regression, indicates that the decline of snow cover might be influencing the warming of New England. Not only are snow cover days disappearing in New England, but snowpack is decreasing as well. Steep snowpack reductions, between 10% and 20% per decade, have occurred in New England along with snowpack declines in Southwestern United States, as well as in Central and Eastern Europe [100,101].

(6)

Perhaps the most striking result from this study is the acceleration of air temperature warming, LST warming, and snow cover decline in New England. Global temperatures have been consistently rising since the 1970s [102] and accelerating since the 1980s [3], which we also see in the New England data, but even more striking is the acceleration over the last five years in both air temperature warming and LST warming as well as in snow cover decline. Annual air temperatures for New England and every state have accelerated in the past five years, New England’s nighttime and daytime LST has accelerated in the past five years, and annual snow cover decline for every state, except VT, has accelerated in the past five years while VT’s acceleration has been over the past ten years. Even ME, which has had a relatively consistent high percentage of annual snow cover, saw an acceleration of snow cover decline.

New England continues to warm due to three main issues: (1) The continued increase in greenhouse gases which is driving warming across the globe. Atmospheric concentrations of the three major greenhouse gases (CO₂, CH₄, and N₂O) reached new record observed highs in 2023 [9], while in 2024 CO₂ levels increased by a record amount, more than in any year since global measurements have been made (1958) [103], and real-time observations, such as at Mauna Loa, show continued increases in 2025. (2) A weakening of the Labrador Current, which brings cold water from the Arctic, and a weakening of the Atlantic Meridional Overturning Circulation (AMOC) have led to warmer ocean temperatures in New England because as the Gulf Stream slows down, there is a northward shift and broadening of the Gulf Stream moving more ocean heat into the Gulf of Maine and along New England’s coast, as well as shifts in wind patterns that blow the warmth to the coast [104,105]. The Gulf of Maine is warming faster than 99 percent of marine regions in the world [106] and a persistent positive phase of the North Atlantic Oscillation (NAO) pushes warmer, maritime air over the coastal Northeast, contributing to land-based warming trends [10]. The weakening of the AMOC also leads to increases in sea level along the coasts of New England [3,107]. (3) Like in the Arctic, where the snow-albedo feedback is a key reason for extreme warming [108], New England is beginning to experience warming due to changes in the snow-albedo feedback; as the climate warms, more snow melts and there is less reflection of solar energy and more absorption of that energy leading to late autumn, winter and early spring warming [25,99,109]. The snow-albedo feedback is an important driver of regional climate change over Northern Hemisphere land, where observation-based estimates show a feedback of around a 1% decrease in surface albedo per degree of warming during spring [110] and eliminating snow cover increases local temperature by about 2.8 °C [111]. These three major factors have combined in recent years resulting in an acceleration of warming in New England.

4.2. Implications for Temperature and Snow Cover Change

There are six major implications for New England related to warming temperatures and declining snow cover.

(1)

Warming temperatures and health concerns

Climate change health concerns include temperature extremes which are related to a larger fraction of cardiorespiratory deaths in the Northeastern United States [112], additional increased respiratory illnesses due to higher pollen concentrations, and decrease in air quality due to smoke from climate-induced fires in Canada and other areas [113]. Vector-borne diseases from the expanding ranges of disease-carrying insects, such as ticks, which have expanded across New England recently along with increased water-borne illness, such as cases of harmful algal blooms, are also expanding in New England [112,114]. Climate change is taking a toll on people through an increase in mental health issues [115]. Heat-related health issues will increase throughout New England, and the urban heat island effect (urban areas are warmer than the surrounding countryside) is expected to increase and further complicate the health of urban residents [116].

(2)

Sea-level rise and coastal flooding

New England has a long and broad coastline which is being affected by sea-level rise and increased coastal storm activity [117,118,119]. Projections for 2100 estimate that the Massachusetts coast could see 4 to 10 feet of sea level rise [120]. Sea-level rise and increased coastal storm activity increases erosion which threatens beaches, dunes, and wetlands; threatens drinking water supplies through saltwater intrusion; and damages residences, historic structures, roadways and other infrastructure creating extensive economic losses and potential losses of life [18,121]. Further weakening of the AMOC can lead to even higher sea levels than predicted along the eastern coast of the United States [122].

(3)

Extreme precipitation events, flooding, and drought

Recent trends in precipitation throughout New England has been towards increases in rainfall intensity, exceeding those in other regions of the contiguous United States [91]. Rainfall intensity is expected to increase in the future, with increases in total precipitation expected during the winter and spring but with little change in the summer [91,105]. Intense rainfall tends to increase runoff with less infiltration of the water, leading to more flooding which is beginning to happen in multiple places across New England and will intensify in the future [123]. Summers will be getting hotter and with precipitation not expected to increase much in the summer, there will be more evaporative energy which will lead to increases in summer and fall droughts [124]. Winter precipitation will continue to decrease as snow and increase as rain [96]. Vulnerable populations (low income, the elderly, people without AC, people in flood-prone or coastal zones) in New England will be disproportionately affected [112].

(4)

Impacts on agriculture, fisheries, and ecosystems

Increased heat can induce stress in crops, livestock and fish, decreasing agricultural productivity. Increased spring rains, increased summer drought and weather whiplash (drastic day-to-day changes in weather) will have a negative effect on agriculture in New England [125] and will create additional stress for ecosystems [126]. Increased summer drought may lead to increased forest fires in New England. Warmer ocean waters are already affecting ocean ecosystems, such as kelp forests and species that New Englanders rely on, and this warm water stress will continue in the future. Shellfish and mollusks are not only negatively affected by the warming water but also by ocean acidification which is occurring due to the burning of fossil fuels [127,128]. The northward shift in warmer waters can also encourage the arrival of warm-water marine species while displacing cold-water species from their historical habitats [129].

(5)

Impacts on forests

New England is dominated by forests, covering approximately 81% of the region [130]. These forests sequester carbon, provide habitat for wildlife, create economic benefits and recreational opportunities. The warming of the region is bringing stress on cold-adapted species with new species arriving along with forest pests where warmer winter temperatures allow invasive pests such as Adelges tsugae (Hemlock Woolly Adelgid) and Dendroctonus frontalis (Southern Pine Beetle) to expand their range due to lower winter mortality [96]. The higher elevations remain more resilient to winter warming compared to lower elevations and coastal regions. The altering of seasons will create various mismatches in seasonal events like pollination or migration. Forest ecosystems in New England rely on deep snowpack to insulate the soil [131] and sustained cold winter temperatures reduce the survival of invasive pests [132], and as noted in this research, snow cover days and snowpack are declining in New England [133]. Winter thaw cycles have become deeper and more prevalent in New England, and these warmer winter thaws can damage forest roots, alter ecosystem productivity, and disrupt phenology [134].

(6)

Economic and social costs

Public and private infrastructure will be damaged by various events such as coastal flooding, inland flooding, storm activity, and other climate-induced events [135]. As mentioned above, climate-change induced events in agriculture and fisheries will have large economic costs [128]. With the decline in snow cover and snowpack, winter tourism (including skiing, snowmobiling, snowshoeing, ice fishing, among other activities), which is a major economic engine in rural New England, will diminish with the lack of snow and cold winter weather, as well as a loss of important winter cultural practices [136]. The overall loss of coldness and snow cover has negative consequences for logging, maple sugaring, and forest products [137]. Climate change is raising the costs of insurance [138], raising the costs of municipal bonds [139], and having adverse effects on the stock prices of climate-vulnerable companies [140].

5. Conclusions

This research shows that outside the Polar and Arctic regions, New England is one of the world’s fastest warming areas, warming more than 2.5 °C since 1900, with 75% of this warming occurring since the late 1980s. Winters are warming almost twice as fast as the other seasons, with New England’s winters warming more the 4 °C since 1900, and winters for the states of CT, MA, ME and VT have warmed more than 5 °C. Except for the state of RI, minimum temperatures have been rising faster than maximum temperatures, with New England’s minimum annual temperatures rising 2.77 °C, 80% of this warming occurring since the late 1980s. In addition to the air temperature warming, the temperature of New England’s land surface has been rising as well, especially at night where it has warmed more than 1.5 °C since 2000. Amid this warming, New England is experiencing widespread snow cover loss where the number of days with snow on the ground has decreased across the entire region and substantially in southern New England, which has lost more than 30% of its annual snow cover since the year 2000. Based on regression analysis, throughout southern New England and coastal ME, the loss of snow cover is strongly correlated with rising land surface temperatures, indicating that the loss of snow cover might be a factor addressing why New England’s winters are warming quickly. An important finding in the research is that New England’s rising air temperature has accelerated in the last five years more intensively than in any other five-year period since the start of historical data back to 1900, and the temperature of New England’s land surface and snow cover decline has accelerated faster in the last five years than at any time since the start of the MODIS satellite data in 2000. This acceleration of temperature and snow cover decline means that climate-related changes will be occurring faster than previously expected in this region.

The accelerated warming and snow cover decline in New England are not future possibilities; they are happening now. Addressing these changes requires serious climate actions such as lowering emissions, adapting urban and rural areas for warmer temperatures and increased storm activity along with adapting nature-based economies such as agriculture, forestry, and recreation to these new realities. In many northeastern states, particularly in New England, mitigation and adaptation actions have been codified into state law. States in the region have legally mandated greenhouse gas emissions reductions, consistent with the goals of the Paris Agreement, requiring state agencies to integrate the best available science into land-use planning and zoning, building codes, and regulation development [141,142]. In addition, local communities, such as church groups, towns, cities, counties, and tribes throughout New England, are engaged in efforts to build resilience and adapt to these climate changes [142,143,144]. These efforts need to be conducted carefully and coordinated across the region [145]. Tribal nations, such as the Mi’kmaq Nation in ME, have adopted climate change plans, including the Thirteen Moons Climate Change Adaptation Plan which calls for proactive efforts to address climate change, such as the development of solar energy, community education and outreach on climate change, and forest and wildlife health monitoring [146].

Future research should use finer scale data such as MOD10A2 and MOD11C1 to begin to investigate the more local influences on LST and snow cover loss. Changes in sea surface temperature in the Gulf of Maine, Long Island Sound and neighboring Atlantic Ocean should be investigated to better understand the influence of the coastal waters on the warming of New England. Modeling future climate change with both land and ocean data made up of coarse and finer spatial resolutions should be undertaken to understand potential future changes.

Longer-term acceleration of temperature and snow cover decline in New England ultimately depends on regional events such as the weakening of the AMOC and the changing phases of the NAO, how strongly the climate responds to the various forcing agents (i.e., greenhouse gases, land-cover change, ocean currents, etc.), and our ability to urgently reduce those forcing agents.