1. Introduction
Understanding the changing climate on regional scales is essential for impacts and adaptation research, yet the monthly annual cycle is not fully analyzed in evaluations of climate projections. Changes in climate are most often discussed in terms of mean annual or seasonal changes, and often with the annual cycle removed, which can mask changes in seasonality. This research examines recent and future changes in the monthly mean annual cycle, and is part of a larger program to explore global climate projections on a regional scale with the northeast United States as a case study. Companion studies that examine observed, simulated, and projected mechanisms of warm season variability (Thibeault and Seth 2014b, 2015) and cold season variability (C. Lynch et al. 2015, unpublished manuscript) are also in progress.
Analysis of global temperature records has demonstrated changes in amplitude and phase of the annual cycle (Thomson 1995; Mann and Park 1996; Wallace and Osborn 2002), which have been attributed in part to increasing CO 2 emissions (Stine et al. 2009) but more recently have been explained largely by changes in atmospheric circulation [predominantly the northern annular mode (NAM) and Pacific–North American (PNA) pattern] (Stine and Huybers 2012). Globally, the amplitude of the surface temperature annual cycle appears to be decreasing with an earlier phase in the northern midlatitude land areas (Stine et al. 2009).
The northeast United States has already experienced larger than global average observed changes and is expected to see larger than average changes in both temperature and precipitation in the twenty-first century (Horton et al. 2014). The region is densely populated with more than 18% of the U.S. population, and is a major engine of the U.S. economy, accounting for roughly 20% of the U.S. gross domestic product (http://factfinder2.census.gov). Population projections suggest a 35% increase in the region between 2000 and 2015, which would place additional stress on resources and make natural hazards more costly.
The climatological annual cycle in the northeast United States is characterized by a large amplitude in surface air temperature and relatively little month-to-month variation in precipitation. Spatial variations in the mean annual cycle across the region are created by synoptic-scale circulation patterns and geophysical forces. The annual temperature range is modulated by latitude and distance from the coast due to the tempering effects of the North Atlantic Ocean. Topographic effects are also important: elevations in the region range from sea level to 6288 ft (1.9 km) above sea level at Mt. Washington (Zielinski and Keim 2003). The precipitation annual cycle exhibits little seasonality due to moisture convergence associated with cyclones and frontal systems. Cyclones forming in the region of the midlatitude jet and storm tracks in winter transport moisture from the North Atlantic, and in summer the North Atlantic subtropical anticyclone guides meridional moisture transport poleward around it western flank into the region. The details of observed spatial patterns are influenced by topography and coastlines. While the intrinsically geographic controls of climate (i.e., latitude, topography, and coastal location) are a steady influence on the regional climate, changes in circulation resulting from natural variability are important (Stine and Huybers 2012) and anthropogenic forcings are expected to accelerate during the twenty-first century.
The northeast United States, variously defined, has been the subject of a number of observational studies that have documented annual and seasonal mean changes in temperature and precipitation (Hodgkins et al. 2002, 2003; Huntington et al. 2004; Trombulak and Wolfson 2004; Henderson and Shields 2006; Hayhoe et al. 2007; Burakowski et al. 2008; Huntington et al. 2009; Brown et al. 2010). These analyses have been updated with the addition of recent data by Kunkel et al. (2013). Since the early 1900s there have been observed trends (Trombulak and Wolfson 2004; Brown and DeGaetano 2011), which now show statistically significant temperature increases in all seasons, as well as in the annual mean value, and in fall [September–November (SON)] for mean precipitation (Kunkel et al. 2013). The observed increases in temperature have resulted in more precipitation falling as rain rather than snow (Bradbury et al. 2002; Hayhoe et al. 2007; Burakowski et al. 2008). Studies have also shown an increase in the intensity of winter cyclones with more snow falling during those events (Bradbury et al. 2003; Huntington et al. 2004; Changnon et al. 2008), despite a decrease in average winter snowfall (Burakowski et al. 2008). Temperature changes have led to earlier spring lake ice-out dates and peak streamflow by an average of 1–2 weeks since the late 1800s (Hodgkins et al. 2002, 2003). Increases in temperature and wet precipitation extreme indices have also been documented for the region (Griffiths and Bradley 2007; Thibeault and Seth 2014a).
Climate model projections have been examined for the northeast United States, again with an emphasis on annual and seasonal averages. A globally robust response to greenhouse warming is the increase in precipitation that is required to balance the atmospheric energy budget when longwave radiation increases (Held and Soden 2000; Seager et al. 2010). Analysis of projections for eastern North America has indicated robust increases in winter precipitation [50th percentile values from phases 3 and 5 of the Coupled Model Intercomparison Project (CMIP3and CMIP5) are 11% and 14%, respectively] by the end of the twenty-first century (Christensen et al. 2007; van Oldenborgh et al. 2013). Seasonal analysis of CMIP3 projections suggests the largest future precipitation increases will be in boreal winter [December–February (DJF)] (Meehl et al. 2007; Kunkel et al. 2013). The CMIP3 models also suggest that global circulation and thermodynamic changes result in a westward shift in the winter U.S. East Coast trough such that the convergence of moisture associated with the eastward flank of the trough increases precipitation in the region (Hakkinen 2011). Projections indicate that storm tracks are expected to migrate poleward with the expansion of the tropics under greenhouse warming scenarios; however, the North Atlantic region is more complex with a decrease in cyclone activity projected on the southern flank of the present storm track and an extension eastward over Europe (Chang et al. 2012). Cyclone tracking analyses in the CMIP5 high-resolution models, in contrast, suggest increased frequency and intensification of winter cyclones affecting the northeast United States. (Colle et al. 2013). Wetter winters and warmer temperatures are projected to drive increases in winter runoff with a shifting of the annual peak runoff to earlier in the year (Huntington et al. 2009; Wetherald 2010).
In summer, when the moisture inflow is predominantly guided by the circulation around the North Atlantic subtropical anticyclone there is substantial disagreement among the models regarding the direction of precipitation change (Christensen et al. 2007; van Oldenborgh et al. 2013). In a warming climate the subtropical anticyclones are expected to expand and migrate poleward, and in the North Atlantic the changes in position and intensity, which vary among the models (see Thibeault and Seth 2015), have been shown to be an important influence on summertime precipitation in the eastern United States (Li et al. 2012; Thibeault and Seth 2014b). Despite these uncertainties, the systematic changes in temperature, evaporation, soil moisture, and runoff are likely to yield increases in the magnitude and frequency of excessively high river discharges as well as summertime dryness (Huntington et al. 2009; Anderson et al. 2010).
The large-scale and regional studies to date have begun to fill in pieces of a grand puzzle to illustrate the climate future of the northeast United States under anthropogenic warming scenarios. The emphasis on annual and seasonal means has been justified by the fact that seasonal and annual averages reduce the noise and provide a clearer signal of the mean projected changes. Here the results focus on the mean annual cycle and are part of a larger research program to improve our understanding of climate projections on regional scales using the northeast United States as a case study. More detailed analysis of present and future variability has been performed for warm season (Thibeault and Seth 2015) and cold season (C. Lynch et al. 2015, unpublished manuscript). We examine the monthly annual cycle and perform the following: 1) evaluate the ability of the latest climate models from CMIP5 (Taylor et al. 2011) to simulate the monthly annual cycle and the observed trends in the region; 2) examine the potential for new information to be gained from projections of the monthly annual cycle, including underlying circulation changes, rather than seasonal or annual means; and 3) based upon the information contained in the monthly projections, investigate the reliability of the projected changes on monthly time scales. We first examine the monthly annual cycle and evaluate the ensemble of coupled climate models in the WCRP CMIP5 archive against observational and reanalysis data for the twentieth century. Once a baseline is established the twenty-first-century annual cycle is examined to investigate projected changes to the temporal and spatial annual cycle of the northeast United States. Section 2 describes the data and methods used in this study. Historical and projected model annual cycle results are given in section 3. Discussion, implications of the results, and conclusions are presented in section 4.
2. Data and methods
a. The northeast United States
The region defined for this study is the greater northeast United States (GNE), which covers the area 38°–48°N, 67°–80°W and includes the states of Connecticut, Maine, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, and Vermont, and also includes portions of the Canadian provinces of Ontario and Quebec. This area roughly coincides with the United States Census Bureau’s designation of “the Northeast” and the National Oceanic and Atmospheric Administration’s (NOAA) Northeast Regional Climate Center (NRCC) region.
b. Observations and reanalysis data
The historical twentieth-century annual cycle for the northeast United States is analyzed using a combination of gridded observations, reanalysis products, and data from archived coupled climate model experiments.
Gridded observational data from the University of Delaware (UDEL) and the University of East Anglia Climatic Research Unit (CRU), version 3.1, obtained from the British Atmospheric Data Centre are employed in this study. The UDEL and CRU datasets are created from the Global Historical Climatology Network (GHCN) as well as other reliable observational and real-time climatological databases (Legates and Willmott 1990; Mitchell and Jones 2005). Both of these datasets are land-only, global monthly time series of precipitation and air temperature that include the period 1901–2010 at a 0.5° spatial resolution.
NCEP–NCAR reanalysis winds and geopotential heights are employed to evaluate the models for the 1971–2000 period (Kalnay et al. 1996). Because moisture fields in early reanalyses are poor and show some improvement in the newest generation reanalyses, particularly over land (Trenberth and Fasullo 2013), the global gridded atmospheric European Centre for Medium-Range Weather Forecasts (ECMWF) interim reanalysis dataset (ERA-Interim) is used to evaluate the difference between precipitation and evaporation (P − E) and also surface air temperature for the shorter time period of 1979–2000 (Dee et al. 2011). Global gridded precipitation data from the Global Precipitation Climatology Project (GPCP) (Adler et al. 2003), also for the period 1979–2001, were obtained from NOAA/Earth System Research Laboratory (ESRL; http://www.esrl.noaa.gov).
c. Climate models
We employ two sets of the CMIP5 experiments in this work. The historical experiment, which includes most known climate forcings from the nineteenth and twentieth century, is used to evaluate individual model and ensemble mean results against observational and reanalysis data for the recent period. For future projections, the CMIP5 models include four representative concentration pathways (RCPs; van Vuuren et al. 2011), which are based on a range of projections of future population growth, technological development, and societal responses (Taylor et al. 2011). Because we expect monthly changes to be noisier than seasonal and annual means, this research employs the high-end RCP8.5 scenario where radiative forcing increases to 8.5 W m−2 through the twenty-first century, which should result in a stronger simulated response.
Sixteen climate models are examined for both the historical and RCP8.5 experiments and are given in Table 1. Climate variables analyzed include surface air temperature and precipitation and all data were regridded to T85 spectral resolution (approximately 1.4°) prior to analysis. Uncertainty (model to model variability) in AOGCM output results from individual model parameterizations, random climate variability, and input for model initialization (Tebaldi and Knutti 2007). Therefore, this analysis incorporates the multimodel ensemble mean as well as the uncertainty represented by the model spread.
List of the CMIP5 models and their respective spatial resolution and organization used in this analysis. (Expansions of acronyms are available at http://www.ametsoc.org/PubsAcronymList.)
d. Data analysis
The models are first evaluated against observed estimates for the recent historical period (1971–2000). Taylor diagrams are used to convey information about pattern similarity between observations and models (Taylor 2001). In a Taylor plot the similarity between two datasets is quantified in terms of their correlation (the azimuthal axis), their centered root-mean-square difference (the x axis), and the amplitude of their variations (represented by their standard deviations; the y axis). The climatological mean annual cycle was calculated for each grid cell, and then a temporal correlation analysis was done between the observation and model at each grid cell. At each grid cell, the RMS error between the model and the observation and the standard deviation of models and observation were also calculated. Then the correlation coefficient, RMS error, and standard deviation for each grid cell were averaged for the GNE region. The RMS error does not imply anything about the sign of bias, but simply quantifies the bias; the standard deviation, on the other hand, shows which models exaggerate the amplitude of the annual cycle and which models underestimate the amplitude based on the reference line indicated on the x axis. For this study each model was evaluated against CRU’s gridded observations from 1951 to 2000 for the annual cycle of temperature and precipitation for the GNE region.
Boxplots are also presented to compare model output against observations and reanalyses. Boxplots show the nonparametric spread among the climate models by summarizing the smallest and highest monthly value (whiskers), lower and upper quartile (bottom and top of the box, respectively), and the ensemble median at each month of the annual cycle.
Recent and projected annual cycles are also presented as longitude–month (Hovmöller) diagrams in order to show the land–sea contrast in the climatological monthly evolution of precipitation and several process-level diagnostics. The annual cycle diagnostics are derived from the authors’ analysis of variability during warm season (Thibeault and Seth 2014b, 2015) and cold season (C. Lynch et al. 2015, unpublished manuscript) in the region and are used to evaluate mechanisms underlying future changes. In addition to the annual cycle of precipitation the following are also included: annual cycles of 850-hPa meridional winds, which are associated with moisture transport into the region during warm season (Thibeault and Seth 2015), and the standard deviation of 500-hPa geopotential height, as an estimate of storm track important for cold season (C. Lynch et al. 2015, unpublished manuscript), both averaged for the GNE region (38°–48°N). Also important in the warm season is 500-hPa geopotential height averaged for the mid-Atlantic (30°–40°N) as an indication of the extension of the Atlantic subtropical anticyclone (Thibeault and Seth 2015). In addition the climatological P − E provides insight into the atmospheric divergence of moisture; that is, where precipitation is less than evaporation there is a net flux of moisture out of the region. Note that it is assumed that month-to-month changes in atmospheric moisture storage are small (5% or less over North America; see Trenberth et al. 2007) and expected to become less important as the amplitude of the annual cycle of temperature decreases in the future. The purpose of these diagnostics is to understand the underlying processes in the model ensemble responsible for changes in precipitation.
Annual cycle difference maps are generated for multimodel ensemble temperature and precipitation, which include measures of model agreement and significance, based on Student t tests, for each model. The methodology for construction of plots is given in Tebaldi et al. (2011). It is important to quantify potential changes in climate variables and to identify regions where projections indicate robust changes. As outlined in Tebaldi et al. (2011), a Student’s t test was done to determine the statistical significance of the projected change between the time periods of 2071–2100 and 1971–2000. If less than 50% of the models show a significant change, then the multimodel mean is shown in color. If more than 50% of the models show a significant change, but disagree in the direction of change, then the grid cell is white. And finally, if more than 50% of the models show a significant change and have at least a 70% agreement in direction of change, then model mean is shown in color with stippling.
Based on the results in section 3, the final calculations are performed using two subregions: the Canadian maritime provinces (CM; 43°–51°N, 61°–71°W) and the Northeast (NE; 38°–45°N, 71°–80°W). Annual cycle anomalies are constructed by removing the annual mean from the annual cycle for the 16 model ensemble for the late twentieth (1971–2000) and twenty-first (2071–2100) centuries. Late twenty-first-century changes in the amplitude and phase of the temperature and precipitation annual cycles are plotted for the NE using a Fourier transform. Fourier analysis takes the periodic function of time and resolves it into a summation of sine and cosine waves at increasing frequencies (harmonics). The coefficients are then used to derive the amplitude, phase, and percent variance explained by each harmonic. In climate analysis, the first harmonic is the annual cycle, which explains the largest amount of variance in the data series. The second to the nth harmonics explain less variance and relate to shorter time scale variations or remote forcings such as the Southern Oscillation or Atlantic variability, which are not the focus of this study.
3. Results
a. Twentieth-century annual cycle
This analysis begins with an evaluation of the observed and simulated monthly mean annual cycle for the recent period in the GNE region. Area averaging was done for land only and a land–ocean mask for each model from the CMIP5 archive is used to designate land classification. (A rectangle defining the GNE region is shown in the top-left panel of Figs. 4 and 5.)
Each model is statistically evaluated against CRU’s gridded observations for the 1951–2000 mean annual cycles of temperature and precipitation in Fig. 1. Most models have small RMS errors in surface air temperature, and standard deviations slightly greater than 1, indicating that the amplitude of the modeled temperature annual cycle is slightly exaggerated. The correlation coefficients (r) are also high for temperature, demonstrating good agreement between the simulated and observed mean annual cycle. Figure 1b shows that for precipitation, the model RMS errors and the standard deviations are both larger than observed, indicating that the models have a wet bias and the simulated amplitude is larger than observed. For precipitation, the correlation coefficients suggest larger errors than seen in temperature for the representation of the annual cycle.
Taylor diagram of GNE (land only) (a) temperature and (b) precipitation annual cycle for CMIP5 models compared to CRU observations (1951–2000). Shown are the standard deviation (dashed), root-mean-square error (solid), and correlation coefficient (dotted).
Citation: Journal of Climate 29, 1; 10.1175/JCLI-D-14-00781.1
The observed (markers) and simulated (box plots) climatological mean monthly annual cycles of temperature and precipitation are examined in Fig. 2. The large temperature annual cycle is well represented by the model ensemble, which exhibits larger spread in winter than summer. In contrast, precipitation shows a weak annual cycle (for this land-only analysis) with larger mean values of precipitation in summer (June–September) than in winter (December–February). The models overestimate the amplitude of the annual cycle, as a consequence of greater than observed precipitation in spring [March–May (MAM)] and particularly in summer [June–August (JJA)]. It is notable that the reanalysis (GPCP) also shows more precipitation in spring and summer than the gridded station observations (CRU and UDEL). Thus both CMIP5 and reanalysis simulate a more pronounced amplitude in the annual cycle with a peak in JJA than do the reference observations.
GNE (land only) (top) temperature (°C) and (bottom) precipitation (cm month −1) annual cycle with the CMIP5 models (boxplots) compared to UDEL (circle), CRU (star), and ERA-Interim or GPCP (diamond).
Citation: Journal of Climate 29, 1; 10.1175/JCLI-D-14-00781.1
In Fig. 3 are presented several annual cycle diagnostics that are important for understanding precipitation processes in warm season and cold season in the region. These diagnostics will be used for comparison of future changes and include annual cycles of precipitation (Figs. 3a,b), 850-hPa meridional winds (Figs. 3e,f), and the standard deviation of 500-hPa geopotential height (Figs. 3g,h) averaged for the GNE region (38°–48°N), and 500-hPa geopotential height (Figs. 3c,d) averaged for the mid-Atlantic (30°–40°N). Evaluation of P − E through the annual cycle (Figs. 3i,j) can provide useful estimates of moisture divergence and its changes, given that month-to-month changes in atmospheric storage of moisture are small (Trenberth et al. 2007). The land–sea contrast is evident in precipitation, which is most abundant in the offshore Atlantic storm track region during the cold season and decreases inland with distance from the coast. The CMIP5 models simulated less than observed oceanic precipitation during winter and more than observed continental precipitation during summer, as discussed above.
GNE pairs of observed estimates and CMIP5 ensemble mean (a),(b) precipitation (cm month −1), (c),(d) 500-hPa geopotential height (m), (e),(f) 850-hPa meridional wind (m s−1), (g),(h) standard deviation of 500-hPa geopotential height (m), and (i),(j) P − E (cm month −1) shown as the monthly mean annual cycle averaged for latitudes 38°–48°N, except for (c),(d), which are averaged for latitudes 30°–40°N as explained in the text, with observed estimates from GPCP in (a), NCEP in (c),(e),(g), and ERA-Interim in (i).
Citation: Journal of Climate 29, 1; 10.1175/JCLI-D-14-00781.1
Previous analysis of observed precipitation variability in the Northeast has shown that in summer a positive geopotential height anomaly off the mid-Atlantic coast is associated with southwesterly flow of moisture into the region (Thibeault and Seth 2014b, 2015). The reanalysis mid-Atlantic height field (Fig. 3c) indicates increased heights during summer with a maximum off the coast. The mean meridional low-level winds (Fig. 3e) show signs of the stationary trough that strengthens between December and March with northerly winds over the continent and southerly flow offshore. During summer the northerly winds weaken and southerly shift westward with the expansion of the mid-Atlantic high. Note also the southerly winds in the western edge of the region during early winter. The models’ simulation of the mid-Atlantic geopotential heights are lower than observed, and the trough is shifted westward with stronger southerly winds, thus resulting in larger than observed rainfall amounts.
Winter precipitation in the Northeast results from transient cyclones, which are more readily seen in measures of variability such as the standard deviation of 500-hPa geopotential height (Figs. 3g,h). A detailed analysis of recent and future cold season variability are presented by the authors in C. Lynch et al. (2015, unpublished manuscript). Here the reanalysis describes the climatological storm track strengthening off the Northeast coast in November and extending westward over the continent in January and February. The models appear to develop the storm track later and maintain some strength over the continent later into March.
The P − E annual cycle (Figs. 3i,j) indicates moisture convergence through winter across the GNE with maximum in the period of November–March and divergence in JJA inland (west of 75°W). In the coastal and western Atlantic region the convergence peaks in summer and shows weak divergence in fall. The models capture this seasonality as well as the observed relationship between southerly winds and moisture convergence during summer. The early winter peak in convergence is also associated with southerly winds and the winter maximum with weakened northerly winds which result from transient cyclones.
b. Twentieth- and twenty-first-century trends
Figure 4 shows the annual and seasonal time series of the multimodel average, with the historical (1901–2005) and RCP8.5 (2006–2100) scenarios, of seasonal temperature and precipitation against CRU gridded observations (1901–2010). Computed values of the observed, simulated, and projected trends and differences (future minus present) are provided in Table 2.
Annual and seasonal time series of GNE (land only) (left) temperature (°C) and (right) precipitation (cm month −1) with CRU (purple) for 1901–2000 and CMIP5 model ensemble mean (black) with ±1 standard deviation (gray shading) and min and max (outer thin black lines) for 1901–2100.
Citation: Journal of Climate 29, 1; 10.1175/JCLI-D-14-00781.1
Seasonal and annual (ANN) trends for temperature (°C decade−1) and precipitation (precip; cm month−1) and future minus present changes for temperature and precipitation. Significance at the 99% (boldface and italic), 95% (boldface), and 90% (italic) levels are indicated. CMIP5 results shown in parentheses, for those models with significant changes, the number of models with significant change which agree on the direction of change of the multimodel mean and, for precipitation trend, the number of models with significant change opposite to the multimodel mean. Note that for the temperature results no models show a significant change opposing the model ensemble mean.
The CRU dataset indicates a significant increase in temperature annually and for every season during the twentieth century, with winter showing the largest trend (0.161°C decade−1; Table 2) These observed trends are consistent with those presented in Kunkel et al. (2013) and show increased significance from earlier analyses for the Northeast region (Burakowski et al. 2008; Hayhoe et al. 2008). The model ensemble time series also show increasing trends, with the largest trends in winter and transition seasons (0.127°C decade−1 in winter, 0.097°C decade−1 in spring, and 0.091°C decade−1 in fall) and smallest in summer (0.076°C decade−1).
Given the RCP8.5 scenario, the models suggest that temperature increases will accelerate as the twenty-first century progresses in all seasons with relatively uniform trends in all seasons. This is in contrast with the observed trends that show twice the rate of increase in winter as in summer and fall. The largest projected seasonal trends are in winter (0.712°C decade−1 or 6.3°C increase by end of century) and summer (0.654°C decade−1 or 5.6°C increase by end of century). This double peak (winter and summer) in temperature increases is small when compared to the scale of the temperature changes and is consistent with neither the observed nor the simulated trends for present day.
The excess precipitation simulated by the CMIP5 ensemble in spring and summer is seen clearly in these time series in Fig. 4. Significant increasing precipitation trends are seen annually (0.116 cm month−1 decade−1) with the largest trend observed in the fall season in the CRU data (Table 2). The observed increase in fall is consistent with the analysis of station observations by Kunkel et al. (2013). The model ensemble precipitation trends show more modest increases of which the trends in winter and spring are significant (at the 95th and 90th percentiles, respectively).
In the future, the model projections indicate significant increases in precipitation in all seasons except for summer, with the largest increases in winter and spring, which is consistent with the simulated trends in the recent period but at odds with the observed trends, which show increases only in fall to date. The lack of an ensemble mean projected trend in summer has been shown previously and is examined further in the following section.
c. Twenty-first-century changes in the annual cycle
It has been shown previously that temperature change in the twenty-first century increases with latitude and distance from the coast in eastern North America (Zielinski and Keim 2003). Monthly temperature difference maps (Fig. 5) indicate that the largest changes in the northern part of the region occur during January, as expected, but the largest changes in the southern part of the region occur during August and September. This effect was recognized in McKinnon et al. (2013), suggesting that the annual cycle amplitude and phase lag are largely due to relative land and ocean influences, and the spatial heterogeneity in monthly temperature changes is very likely related to water proximity. This late summer peak in temperature change would be masked by seasonal averages. Note that for temperature every grid cell meets the dual requirements for significant changes (50% of the models) and agreement in the sign of change (70% agreement of models).
Future (2071–2100) minus present (1971–2000) monthly mean temperature (°C) change from CMIP5 models. Rectangle in the top-left panel shows the GNE region that is used for area-averaged calculations.
Citation: Journal of Climate 29, 1; 10.1175/JCLI-D-14-00781.1
Figure 6 shows the spatial precipitation changes for each month. Northeast winter precipitation is expected to increase, with confidence based on model agreement, while changes in summer precipitation are uncertain (Field et al. 2007; Romero-Lankao et al. 2014). As expected from Fig. 4, the largest and most significant precipitation changes are projected to be in winter and spring, particularly in the northern and interior portions of the GNE. The monthly analysis suggests that the largest increases in precipitation would be seen in early spring, with the greatest agreement and largest change in March. This information, also, is masked by seasonal averages.
Future (2071–2100) minus present (1971–2000) monthly mean precipitation (cm) change from CMIP5 models. Rectangle in the top-left panel shows the GNE region that is used for area-averaged calculations. Colors indicate areas where the minimum model agreement on direction of change (70%) is met. Stippling indicates at least 50% of models show significant change at the 0.05 significance level.
Citation: Journal of Climate 29, 1; 10.1175/JCLI-D-14-00781.1
While Figs. 5 and 6 show the spatial patterns of projected monthly changes with a measure of model agreement, the area averaged monthly changes are given in Figs. 7a and 7b to provide individual model results. The projected seasonal temperature trends (Table 2) suggested a weak double peak in temperature change (DJF and JJA). The monthly temperature changes, shown here for each model (Fig. 7a), corroborate this for DJF, but the warm season maximum change is actually shifted by a month to July–September (JAS).
GNE (land only) (a) temperature (°C), (b) precipitation (cm month −1), (c) downwelling shortwave at surface (W m−2), and (d) near-surface relative humidity (%) from each of the CMIP5 models future (2071–2100) minus present (1971–2000) annual cycle of change (bars) with ensemble mean (blue line).
Citation: Journal of Climate 29, 1; 10.1175/JCLI-D-14-00781.1
The individual model projections for monthly precipitation (Fig. 7b) show that the maximum increase is in late winter–early spring [February–April (FMA)], and the maximum uncertainty is in late summer (JAS), which appears to be coincident with peak temperature changes. The monthly annual cycle of precipitation changes can be further understood by examining changes in downwelling shortwave radiation at the surface and near-surface relative humidity (Figs. 7c,d). There is strong model agreement that decreases in surface shortwave radiation are largest in late winter and somewhat less model agreement regarding increases in relative humidity, coincident with the precipitation maximum in February–March. While the models lack agreement in the direction of precipitation change during late summer, they show substantial agreement that shortwave radiation increases and relative humidity decreases, suggesting decreasing cloud cover.
The weak double peak in temperature change is also consistent with surface energy and moisture changes. Because precipitation is shown to increase during the cold season (including fall and spring), it is possible that increased cloud cover associated with precipitation would reduce the solar shortwave received at the surface in spring when sunlight is increasing and in fall when sunlight is decreasing. The increased cloud cover would also increase nighttime longwave heating of the surface. In Fig. 7c, the shortwave flux at the surface is indeed reduced in the spring, but in the fall there is an increase in shortwave flux even as precipitation increases, potentially due to changes in precipitation distribution from fewer but heavier rainfall events.
Figure 8 presents future changes in annual cycle diagnostics shown in Fig. 3 and illustrates processes driving the changes in the precipitation annual cycle. The largest precipitation increases are over the continent in late winter and spring, while during summer changes are modest, with weak drying in late summer (resulting from model disagreement). Mid-Atlantic 500-hPa geopotential height increases year round as expected; however, the maximum increase is near the coast and over land in summer. This induces a weakening of the northerly winds during most of the year over the continent and weakens the southerly winds offshore during summer. Thus, during summer a westward extension of the high weakens the trough over the region. Atmospheric divergence increases through the warm season inland, and in late summer and fall in the coastal and Atlantic region, as evaporation increases outpace precipitation change (which is small during summer) in the CMIP5 projections.
GNE (a) precipitation (cm month−1), (b) 500-hPa geopotential height (m), (c) 850-hPa meridional wind (m s−1), (d) standard deviation of 500-hPa geopotential height (m), and (e) P − E (cm month−1) shown as the future (2071–2100) minus present (1971–2000) monthly mean annual cycle (colors) with simulated present climatology (black contours) averaged for latitudes 38°–48°N except for (b), which are averaged for latitudes 30°–40°N as explained in the text.
Citation: Journal of Climate 29, 1; 10.1175/JCLI-D-14-00781.1
During the cold season moisture convergence increases moderately along with increased southerly winds, which together outweigh the effect of a weakened storm track (Fig. 8e) in fall and early winter. In late winter and spring increased southerly winds (Fig. 8c) and a strengthened storm track (Fig. 8d) combine with increased moisture convergence to yield the maximum increase in precipitation (Fig. 8a). This is interesting because the literature that has examined the future of the North Atlantic storm track suggests increases in the frequency and intensity of cyclones affecting the coast and inland regions of the Northeast (Colle et al. 2013) but a decrease in the storm track using standard variance measures (Chang et al. 2012). Here we find a weakened storm track in early winter and strengthened storm track in late winter. Monthly changes in late winter variability in 500-hPa geopotential heights affirms an increase in March with most models in agreement on the sign of the change (Fig. 9), while changes in February indicate a reduction in variability off the coast, and in April changes are modest and show less agreement.
(left) CMIP5 ensemble mean future (2071–2100) minus present (1971–2000) standard deviation of 500-hPa geopotential height (m) and (right) model agreement on the sign of the change, where colors represent number of models that show positive change (red) and negative change (blue) for (top)–(bottom) February–April.
Citation: Journal of Climate 29, 1; 10.1175/JCLI-D-14-00781.1
Figure 10a shows the spatial change in amplitude of the model ensemble temperature annual cycle. For much of the GNE, the amplitude of the annual cycle decreases due to larger projected increases in winter than in summer temperature. However the zero line crosses the southern portion of the GNE, and to the south of it, the annual cycle amplitude increases due to larger increases in summer temperatures. In Fig. 10b the spatial change in amplitude of the model ensemble precipitation annual cycle shows that along the coast the seasonal precipitation cycle increases in amplitude, but northern inland areas show a decrease in amplitude. This is likely because precipitation over the northern, inland areas increases throughout much of the year, with most of this increase occurring in winter and spring, and less of an increase in the summer and fall months, thereby dampening the amplitude. Along coastal and southern locations, the precipitation annual cycle amplitude is increasing as wet months become wetter (MAM and JJA) and dry months become drier (SON). For precipitation, then, the zero line crosses through the middle of the GNE region as we have defined it, with some variation among individual models (not shown). Thus, the transition from positive to negative amplitude change is lost in the spatial average for the GNE region.
Future (2071–2100) minus present (1971–2000) annual cycle amplitude change plots of (a) temperature (°C) and (b) precipitation (cm month−1) from the CMIP5 model ensemble. Rectangles in (a) show the areal extent of the NE and CM regions used in the calculation for Fig. 11.
Citation: Journal of Climate 29, 1; 10.1175/JCLI-D-14-00781.1
To capture this transition we select two subregions previously defined as CM (43°–51°N, 71°–61°W) and the smaller NE (38°–45°N, 71°–80°W). These regions are shown as rectangles in Fig. 10a. The changes in the annual cycle (amplitude and phase) can be compared by examining the deviation of a climate variable from its annual mean with the trend removed. We examine the amplitude and phase of the annual cycle for the late twentieth and twenty-first centuries. Figure 11 (top) shows the change in the annual cycle of temperature between the late twentieth and twenty-first centuries for these two regions. For temperature there is a small decrease in amplitude in CM and a small increase in NE, as described above, with no phase shift. Precipitation (Fig. 11, bottom), however, shows a statistically significant increase in DJF and MAM precipitation and a decrease in JJA precipitation. In CM, the decrease in summer and increase in winter work to reduce the amplitude of the annual cycle. In NE, the increases are largest in late winter and large decreases in late summer act to increase the amplitude, but also shift the phase to earlier in the calendar year.
NE and CM regional (land only) (top) temperature (°C) and (bottom) precipitation (cm month−1) annual cycle anomalies (with annual mean removed) from CMIP5 models for late twentieth century (1971–2000, solid line) and late twenty-first century (2071–2100, dashed line). Shading represents ±1 standard deviation computed from the recent period.
Citation: Journal of Climate 29, 1; 10.1175/JCLI-D-14-00781.1
4. Discussion and conclusions
Relatively few studies focusing on the northeast United States have examined the monthly mean changes in climate model projections for the twenty-first century. In this research we have evaluated a suite of 16 CMIP5 models, with an emphasis on the annual cycle for the land-only region of the GNE. The results presented here are consistent with previous analyses for the region, and add new insight regarding changes in seasonality.
The CMIP5 model ensemble simulates the amplitude and phase of the temperature annual cycle quite well, with small differences seen between the observational estimates, reanalyses, and model ensemble average. In the Northeast, observed precipitation exhibits a weak annual cycle, with a maximum in summer (June–July). The CMIP5 models and the reanalyses both simulate excess precipitation in spring and especially summer, which contributes to a larger than observed amplitude of the precipitation annual cycle. The models’ simulation of the monthly evolution of circulation diagnostics and P − E show good agreement with observed estimates, while also indicating stronger than observed southwesterly winds that result from a westward shift in the quasi-stationary trough—the proximate cause of summer precipitation bias.
The seasonality of the simulated temperature trends is consistent with observations in the twentieth century, while the seasonality of precipitation trends is not. The models are able to simulate the observed temperature trends and show the largest trends in winter, consistent with the observations. The more rapid winter warming has been associated with observed decreases in snow cover in the region (Burakowski et al. 2008). CMIP5 simulated precipitation trends are significant and positive in winter and spring, while observed trends in the region are significant and positive in fall and annually. Causes for the discrepancy between the observed and simulated trends have not been identified. However, the recent transition from drought (1960s) to the current pluvial in the Northeast has been ascribed to atmospheric internal variability given that models were unable to simulate the transition even when driven by observed sea surface temperatures (Seager et al. 2012).
Projections indicate more uniform changes in temperature through the annual cycle than have been observed in trends thus far, with a weak double peak during winter (DJF, in the high latitudes of the GNE region) and late summer (JAS, in the low latitudes of the GNE region). The models suggest that increases in precipitation would occur much of the year (November–June) with the largest increases in late winter–early spring (FMA). During late summer and early fall (July–October) the models disagree regarding the sign of the precipitation change.
Process-level diagnostics suggest that during late summer the lack of precipitation increase can be understood as a response to westward extension of the North Atlantic subtropical anticyclone, which acts to reduce the amplitude of the quasi-stationary trough over the region and results in weak moisture divergence. During cold season, while moisture convergence increases, the storm track measure indicates reductions during early winter (November–December) and increases during late early spring (March–April).
The model ensemble mean suggests that the amplitude of the annual cycle, for both precipitation and temperature, will decrease in the northern areas and increase in the southern areas, with the transitions (zero lines) crossing New England. Separate evaluations of northern (CM) and southern (NE) areas show that the temperature response is due to larger increases in winter than in summer at higher latitudes, and larger increases in summer than in winter at lower latitudes. The precipitation response results from increases in winter and no change or a decrease in summer and early fall. Precipitation changes also suggest a phase shift, with more of the annual rainfall occurring between November and April and less occurring between May and October.
There are several implications of these results. First, potentially important information exists in the annual cycle for understanding the changes in the physical climate system and for evaluating impacts and adaptation strategies. Second, we find here that while observed temperatures have been increasing most rapidly in winter (twice as fast as summer and fall), projected temperature increases are more uniform through the seasons. A relatively small double peak in temperature change is seen in the projections in midwinter (DJF) and late summer (JAS). Precipitation increases peak in late winter–early spring (FMA) and exhibit greatest model disagreement in late summer (JAS). While the model disagreement regarding summer precipitation has been discussed previously, a strong model agreement in the shift toward late winter–early spring maximum in precipitation increase and late summer–early fall temperature increase has not been shown prior to this analysis. Also new in these results is the indication that in early winter moisture convergence and precipitation increase despite a weakened storm track, which implies that the increase in atmospheric humidity due to higher temperatures is the dominant driver. During early spring the maximum precipitation increase results from the combination of moisture increase and a more active storm track, suggesting that both increased moisture and changes in circulation are important. The seasonal and annual mean analyses conducted previously would have masked such shifts. The seasonality of changes in precipitation is similar in the CMIP3 models (not shown). Third, the Northeast region sits in a transition where the northern areas are expected to be wetter with reduced annual cycle and the southern areas are expected to be drier in summer with an increased annual cycle. The exact location of this transition will be simulated by individual models based on their formulations and equally good models may disagree. Thus, the uncertainty in summertime precipitation may not be substantially reducible.
The causes of seasonality in temperature and precipitation projections globally remain an active topic of research with possible mechanisms including increased thermal inertia in high-latitude oceans due to reductions in sea ice (Dwyer et al. 2012) and poleward expansion of the general circulation (Scheff and Frierson 2012). Here we have conducted a regional analysis of the seasonality of changes and find the maximum in temperature changes to occur in late summer and maximum precipitation changes to occur in late winter. This late winter maximum precipitation change may be related to dynamical controls on the annual cycle (PNA pattern and/or NAM; see Stine and Huybers 2012) and is the subject of ongoing study. A separate analysis has been conducted for late summer precipitation and shows that model disagreement is related to differences in the simulated location and intensity of the Atlantic subtropical anticyclone among equally credible models (Thibeault and Seth 2015). Further analysis is ongoing to evaluate connections to changes in the large-scale circulation and or local processes that might be driving these changes.
There is substantial evidence that climate in the northeast United States is changing, and potential shifts in the annual cycle of temperature and precipitation presented here would have important implications for future water availability, drought, human health, and ecosystem functions. Improved understanding of seasonal transitions has potential to increase confidence in projections, and to provide additional information of use to the impacts and decision-maker communities.
Acknowledgments
The authors thank two anonymous reviewers for insightful and constructive suggestions to improve this research and its presentation. We acknowledge the World Climate Research Programme’s Working Group on Coupled Modelling, which is responsible for CMIP, and we thank the climate modeling groups (listed in Table 2 of this paper) for producing and making available their model output. For CMIP the U.S. Department of Energy’s Program for Climate Model Diagnosis and Intercomparison provides coordinating support and led development of software infrastructure in partnership with the Global Organization for Earth System Science Portals. This research was funded by NSF CAREER Award 1056216.
REFERENCES
Adler, R. F., and Coauthors, 2003: The Version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor., 4, 1147–1167, doi:10.1175/1525-7541(2003)004<1147:TVGPCP>2.0.CO;2.
Anderson, B., K. Hayhoe, and X.-Z. Liang, 2010: Anthropogenic-induced changes in twenty-first century summertime hydroclimatology of the northeastern US. Climatic Change, 99, 403–423, doi:10.1007/s10584-009-9674-3.
Bradbury, J. A., B. D. Keim, and C. P. Wake, 2002: U.S. East Coast trough indices at 500 hPa and New England winter climate variability. J. Climate, 15, 3509–3517, doi:10.1175/1520-0442(2002)015<3509:USECTI>2.0.CO;2.
Bradbury, J. A., B. D. Keim, and C. P. Wake, 2003: The influence of regional storm tracking and teleconnections on winter precipitation in the northeastern United States. Ann. Assoc. Amer. Geogr., 93, 544–556, doi:10.1111/1467-8306.9303002.
Brown, P. J., and A. T. DeGaetano, 2011: A paradox of cooling winter soil surface temperatures in a warming northeastern United States. Agric. For. Meteor., 151, 947–956, doi:10.1016/j.agrformet.2011.02.014.
Brown, P. J., R. Bradley, and F. Keimig, 2010: Changes in extreme climate indices for the northeastern United States, 1870–2005. J. Climate, 23, 6555–6572, doi:10.1175/2010JCLI3363.1.
Burakowski, E., C. Wake, B. Braswell, and D. Brown, 2008: Trends in wintertime climate in the northeastern United States: 1965–2005. J. Geophys. Res., 113, D20114, doi:10.1029/2008JD009870.
Chang, E. K. M., Y. Guo, and X. Xia, 2012: CMIP5 multimodel ensemble projection of storm track change under global warming. J. Geophys. Res. Atmos., 117, D23118, doi:10.1029/2012JD018578.
Changnon, D., C. Merinsky, and M. Lawson, 2008: Climatology of surface cyclone tracks associated with large central and eastern U.S. snowstorms, 1950–2000. Mon. Wea. Rev., 136, 3193–3202, doi:10.1175/2008MWR2324.1.
Christensen, J., and Coauthors, 2007: Regional climate projections. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 847–940.
Colle, B. A., Z. Zhang, K. A. Lombardo, E. Chang, P. Liu, and M. Zhang, 2013: Historical evaluation and future prediction of eastern North American and western Atlantic extratropical cyclones in the CMIP5 models during the cool season. J. Climate, 26, 6882–6903, doi:10.1175/JCLI-D-12-00498.1.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, doi:10.1002/qj.828.
Dwyer, J. G., M. Biasutti, and A. H. Sobel, 2012: Projected changes in the seasonal cycle of surface temperature. J. Climate, 25, 6359–6374, doi:10.1175/JCLI-D-11-00741.1.
Field, C., L. Mortsch, M. Brklacich, D. Forbes, P. Kovacs, J. Patz, S. Running, and M. Scott, 2007: North America. Climate Change 2007: Impacts, Adaptation and Vulnerability, M. Parry et al., Eds., Cambridge University Press, 617–652.
Griffiths, M. L., and R. S. Bradley, 2007: Variations of twentieth-century temperature and precipitation extreme indicators in the northeast United States. J. Climate, 20, 5401–5417, doi:10.1175/2007JCLI1594.1.
Hakkinen, S., 2011: Atmospheric blocking and Atlantic multidecadal ocean variability. Science, 334, 655–659, doi:10.1126/science.1205683.
Hayhoe, K., and Coauthors, 2007: Past and future changes in climate and hydrological indicators in the US Northeast. Climate Dyn., 28, 381–407, doi:10.1007/s00382-006-0187-8.
Hayhoe, K., and Coauthors, 2008: Regional climate change projections for the Northeast USA. Mitig. Adapt. Strategies Global Change, 13, 425–436, doi:10.1007/s11027-007-9133-2.
Held, I., and B. Soden, 2000: Water vapor feedback and global warming. Annu. Rev. Energy Environ., 25, 441–475, doi:10.1146/annurev.energy.25.1.441.
Henderson, K. G., and B. P. Shields, 2006: Characteristics of autumn precipitation trends in the northeastern United States. Prof. Geogr., 58, 184–196, doi:10.1111/j.1467-9272.2006.00525.x.
Hodgkins, G. A., I. C. James, and T. G. Huntington, 2002: Historical changes in lake ice-out dates as indicators of climate change in New England, 1850–2000. Int. J. Climatol., 22, 1819–1827, doi:10.1002/joc.857.
Hodgkins, G. A., R. Dudley, and T. Huntington, 2003: Changes in the timing of high river flows in New England over the 20th century. J. Hydrol., 278, 244–252, doi:10.1016/S0022-1694(03)00155-0.
Horton, R., and Coauthors, 2014: Northeast. Climate Change Impacts in the United States: The Third National Climate Assessment, J. Melillo, T. T. Richmond, and G. W. Yohe, Eds., U.S. Global Change Research Program, 371–395, doi:10.7930/j0sf2t3p.
Huntington, T. G., G. A. Hodgkins, B. D. Keim, and R. W. Dudley, 2004: Changes in the proportion of precipitation occurring as snow in New England (1949–2000). J. Climate, 17, 2626–2636, doi:10.1175/1520-0442(2004)017<2626:CITPOP>2.0.CO;2.
Huntington, T. G., A. D. Richardson, K. J. McGuire, and K. Hayhoe, 2009: Climate and hydrological changes in the northeastern United States: Recent trends and implications for forested and aquatic ecosystems. Can. J. For. Res., 39, 199–212, doi:10.1139/X08-116.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437–471, doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.
Kunkel, K., and Coauthors, 2013: Regional climate trends and scenarios for the U.S. National Climate Assessment. Part 1: Climate of the Northeast. U.S. NOAA NESDIS Tech. Rep. 142-1, 79 pp. [Available online at http://www.nesdis.noaa.gov/technical_reports/142_Climate_Scenarios.html.]
Legates, D. R., and C. J. Willmott, 1990: Mean seasonal and spatial variability in gauge-corrected, global precipitation. Int. J. Climatol., 10, 111–127, doi:10.1002/joc.3370100202.
Li, W., L. Li, M. Ting, and Y. Liu, 2012: Intensification of Northern Hemisphere subtropical highs in a warming climate. Nat. Geosci., 5, 830–834, doi:10.1038/ngeo1590.
Mann, M. E., and J. Park, 1996: Joint spatio-temporal modes of surface temperature and sea level pressure variability in the Northern Hemisphere during the last century. J. Climate, 9, 2137–2162, doi:10.1175/1520-0442(1996)009<2137:JSMOST>2.0.CO;2.
McKinnon, K. A., A. R. Stine, and P. Huybers, 2013: The spatial structure of the annual cycle in surface temperature: Amplitude, phase, and Lagrangian history. J. Climate, 26, 7852–7862, doi:10.1175/JCLI-D-13-00021.1.
Meehl, G., and Coauthors, 2007: Global climate projections. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 747–845.
Mitchell, T. D., and P. D. Jones, 2005: An improved method of constructing a database of monthly climate observations and associated high-resolution grids. Int. J. Climatol., 25, 693–712, doi:10.1002/joc.1181.
Romero-Lankao, P., J. Smith, D. Davidson, N. Diffenbaugh, P. Kinney, P. Kirshen, P. Kovacs, and L. Villers-Ruiz, 2014: North America. Climate Change 2014: Impacts, Adaptation and Vulnerability. V. R. Barros et al., Eds., Cambridge University Press, 1439–1498.
Scheff, J., and D. Frierson, 2012: Twenty-first-century multimodel subtropical precipitation declines are mostly midlatitude shifts. J. Climate, 25, 4330–4347, doi:10.1175/JCLI-D-11-00393.1.
Seager, R., N. Naik, and G. A. Vecchi, 2010: Thermodynamic and dynamic mechanisms for large-scale changes in the hydrological cycle in response to global warming. J. Climate, 23, 4651–4668, doi:10.1175/2010JCLI3655.1.
Seager, R., N. Pederson, Y. Kushnir, J. Nakamura, and S. Jurburg, 2012: The 1960s drought and the subsequent shift to a wetter climate in the Catskill Mountains region of the New York City watershed. J. Climate, 25, 6721–6742, doi:10.1175/JCLI-D-11-00518.1.
Stine, A. R., and P. Huybers, 2012: Changes in the seasonal cycle of temperature and atmospheric circulation. J. Climate, 25, 7362–7380, doi:10.1175/JCLI-D-11-00470.1.
Stine, A. R., P. Huybers, and I. Y. Fung, 2009: Changes in the phase of the annual cycle of surface temperature. Nature, 457, 435–440, doi:10.1038/nature07675.
Taylor, K. E., 2001: Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res., 106, 7183–7192, doi:10.1029/2000JD900719.
Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2011: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485–498, doi:10.1175/BAMS-D-11-00094.1.
Tebaldi, C., and R. Knutti, 2007: The use of the multi-model ensemble in probabilistic climate projections. Philos. Trans. Roy. Soc., 365A, 2053–2075, doi:10.1098/rsta.2007.2076.
Tebaldi, C., J. M. Arblaster, and R. Knutti, 2011: Mapping model agreement on future climate projections. Geophys. Res. Lett., 38, L23701, doi:10.1029/2011GL049863.
Thibeault, J. M., and A. Seth, 2014a: Changing climate extremes in the northeast United States: Observations and projections from CMIP5. Climatic Change, 127, 273–287, doi:10.1007/s10584-014-1257-2.
Thibeault, J. M., and A. Seth, 2014b: A framework for evaluating model credibility for warm-season precipitation in northeastern North America: A case study of CMIP5 simulations and projections. J. Climate, 27, 493–510, doi:10.1175/JCLI-D-12-00846.1.
Thibeault, J. M., and A. Seth, 2015: Toward the credibility of Northeast United States summer precipitation projections in CMIP5 and NARCCAP simulations. J. Geophys. Res. Atmos., 120, 10 050–10 073, doi:10.1002/2015JD023177.
Thomson, D. J., 1995: The seasons, global temperature, and precession. Science, 268, 59–68, doi:10.1126/science.268.5207.59.
Trenberth, K. E., and J. T. Fasullo, 2013: Regional energy and water cycles: Transports from ocean to land. J. Climate, 26, 7837–7851, doi:10.1175/JCLI-D-13-00008.1.
Trenberth, K. E., L. Smith, T. Qian, A. Dai, and J. Fasullo, 2007: Estimates of the global water budget and its annual cycle using observational and model data. J. Hydrometeor., 8, 758–769, doi:10.1175/JHM600.1.
Trombulak, S. C., and R. Wolfson, 2004: Twentieth-century climate change in New England and New York, USA. Geophys. Res. Lett., 31, L19202, doi:10.1029/2004GL020574.
van Oldenborgh, G. J., J. M. Collins, J. Arblaster, J. H. Christensen, J. Marotzke, S. B. Power, M. Rummukainen, and T. Zhou, 2013: Annex I: Atlas of global and regional climate projections. Climate Change 2013: The Physical Science Basis. T. Stocker et al., Eds., Cambridge University Press, 1311–1394.
van Vuuren, D., and Coauthors, 2011: The representative concentration pathways: An overview. Climatic Change, 109, 5–31, doi:10.1007/s10584-011-0148-z.
Wallace, C. J., and T. J. Osborn, 2002: Recent and future modulation of the annual cycle. Climate Res., 22, 1–11, doi:10.3354/cr022001.
Wetherald, R., 2010: Changes of time mean state and variability of hydrology in response to a doubling and quadrupling of CO2. Climatic Change, 102, 651–670, doi:10.1007/s10584-009-9701-4.
Zielinski, G., and B. Keim, 2003: New England Weather, New England Climate.University Press of New England, 300 pp.
