Abstract

The synoptic behavior of present-day heat waves (HW) over Europe is studied using the GFDL high-resolution atmospheric model (HiRAM) with 50-km grid spacing. Three regions of enhanced and coherent temperature variability are identified over western Russia, eastern Europe, and western Europe. The simulated HW characteristics are compared with those derived from Climate Forecast System Reanalysis products. Composite charts for outstanding HW episodes resemble well-known recurrent circulation types. The HW region is overlain by a prominent upper-level anticyclone, which blocks the passage of synoptic-scale transients. The altered eddy vorticity transports in turn reinforce the anticyclone. The anticyclone is part of a planetary-scale wave train. The successive downstream development of this wave train is indicative of Rossby wave dispersion.

Additional runs of HiRAM are conducted for the “time slices” of 2026–35 and 2086–95 in the climate scenario corresponding to representative concentration pathway 4.5 (RCP4.5). By the end of the twenty-first century, the average duration and frequency of HW in the three European sites are projected to increase by a factor of 1.4–2.0 and 2.2–4.5, respectively, from the present-day values. These changes can be reproduced by adding the mean shift between the present and future climatological temperatures to the daily fluctuations in the present-day simulation. The output from a continuous integration of a coupled general circulation model through the 1901–2100 period indicates a monotonic increase in severity, duration, and HW days during the twenty-first century.

1. Introduction

In many regions of the world, the summer season is marked by the occurrence of extremely hot episodes that persist for more than several days. Such “heat waves” (HW) pose a severe threat to public health and various economic activities and are directly linked to substantial human casualties as well as financial losses (Field et al. 2012). These outstanding events are often accompanied by prominent changes in the ambient atmospheric circulation and precipitation fields, as well as in the conditions of the nearby land and ocean surfaces. Improved understanding of the physical causes of the HW, as well as the impacts of climate change on the future behavior of this phenomenon, are hence of both practical and academic interest to the meteorological community.

Much effort has been devoted to the study of HW in North America. In particular, individual episodes in the Great Plains region have been examined in detail by Klein (1952), Namias (1982), and Kunkel et al. (1996), among others. The long-term trends in the HW characteristics observed over the United States have also been examined by Gaffen and Ross (1998), Kunkel et al. (1999), Easterling et al. (2000a,b), and DeGaetano and Allen (2002), among others. Experimentation with a global climate model by Meehl and Tebaldi (2004), and with a regional model by Diffenbaugh et al. (2005), indicates that more frequent and longer lasting HW will occur over North America in the warmer climate of the twenty-first century.

More recently, Lau and Nath (2012, hereafter LN) examined the synoptic characteristics of HW simulated in various regions of North America by a global atmospheric general circulation model (GCM) with a grid resolution of about 50 km. The model findings for present-day climate conditions were compared with results based on regional reanalysis data with comparable resolution. LN have also noted the projected increase in the severity, duration, and frequency of North American HW from the twentieth to the twenty-first century.

Heat waves occur not only in North America, but are also observed in many other parts of the world. For instance, much attention has been devoted to the behavior of this phenomenon in the European sector. The incidence of the severe HW in western Europe in 2003 (Fink et al. 2004; Black et al. 2004), and in western Russia in 2010 (Dole et al. 2011), is particularly noteworthy. The association of European HW with regional and large-scale circulation patterns, and with boundary forcings (such as those related to sea surface temperature anomalies and land hydrological conditions), has been noted by Della-Marta et al. (2007b), Carril et al. (2008), and Stefanon et al. (2012), among others. The impacts of climate change and decadal-scale variability on HW occurrences in Europe (Schär et al. 2004; Beniston 2004; Della-Marta et al. 2007a; Kyselý 2010) are of strong interest to the local populations.

In the present study, the diagnostic procedure employed in LN is applied to the output from a suite of model runs contributing to phase 5 of the Coupled Model Intercomparison Project (CMIP5), with the focus being shifted to HW in the European sector. These model results are compared with observational findings based on a new high-resolution reanalysis product with global coverage. Although the methodology and model tools used in the current work are similar to those in LN, the present investigation offers new evidence on the model capability to replicate essential characteristics of the circulation patterns that are specific to HW in Europe. Furthermore, a considerable portion of the present paper is devoted to the dynamical interactions between synoptic-scale eddies and the low-frequency component of the flow field during prominent HW episodes in Europe. This mechanism has not been addressed in any detail in LN.

The model and observational data tools used in this study are described in section 2. The typical precipitation and near-surface flow patterns accompanying HW in various European sites are documented in section 3 using both model output and reanalysis data for the present-day climate. Section 4 provides a full account of the temporal evolution of the slowly varying component of the motion field in the upper troposphere during the HW events, the modulation of synoptic-scale activity by the low-frequency circulation changes, and the dynamical feedbacks of the synoptic eddies on the ambient flow pattern. Results on the projected changes in various HW characteristics in Europe during the twenty-first century are presented in section 5. The principal findings are summarized and discussed in section 6.

2. Model specifications and observational dataset

a. GFDL Climate Model version 3

This is a global, fully coupled model developed at the Geophysical Fluid Dynamics Laboratory (GFDL), with new parameterizations of aerosol physics and atmospheric chemistry. The horizontal resolution of the atmospheric component in the GFDL Climate Model, version 3 (CM3), is approximately 200 km. There are 48 vertical atmospheric layers, with the top level being located at about 86.5 km. Further details of the model atmosphere, as well as the treatment of various processes related to ground hydrology and sea ice, are provided by Donner et al. (2011). The oceanic component corresponds to version 4.1 of the Modular Ocean Model (MOM4p1) (see Griffies 2009). Poleward of 30° the MOM has a horizontal resolution of 1° in latitude and longitude. Enhanced meridional resolution is used in the tropics, with grid spacing being reduced to 1/3° at the equator. There are 50 layers in the vertical direction.

The CM3 model is integrated through the 1860–2100 period by subjecting it to the sequence of radiative forcings due to greenhouse gases, ozone, and various types of aerosols. In the historical era (1860–2005), these forcings are computed using observational estimates of the emission rates or concentrations of the trace constituents. The radiative forcings in the 2006–2100 period correspond to a representative concentration pathway (RCP) through which the total forcing attains the 4.5 W m−2 level at the end of that period. This scenario (hereafter referred as RCP4.5) has been developed by the Intergovernmental Panel for Climate Change (IPCC) and is described in more detail by van Vuuren et al. (2011). Altogether five and three parallel runs (i.e., ensemble members) initiated from independent atmospheric conditions have been completed for the 1860–2005 and 2006–2100 epochs, respectively.

b. High-resolution atmospheric model

The C180 version of this atmosphere-only GCM, which is also developed at GFDL, has 32 vertical levels and a horizontal resolution of approximately 50 km. The altitude of the top model layer is about 46 km. Various aspects of the model numerics and physics are documented in detail by Zhao et al. (2009). The radiative forcings due to all trace species are determined from prescribed concentrations derived offline from a given emission scenario. The High-Resolution Atmospheric Model (HiRAM) experiments being analyzed in the present work include the following:

  • Historical (H) run for the 1979–2008 period, during which the model atmosphere is subjected to temporally evolving radiative forcing, as well as sea surface temperature (SST) and sea ice conditions at the lower boundary. All of these forcing fields are obtained from observational estimates. In particular, the intermonthly variations of SST and sea ice are based on the first Hadley Centre Global Sea Ice and Sea Surface Temperature dataset (HADISST1) for the duration of this experiment.

  • Scenario run for the early part of the twenty-first century (S1), in which the radiative forcing corresponding to the RCP4.5 scenario for the 2026–35 period is prescribed. The SST and sea ice conditions are obtained by the following procedure. The difference between the SST and sea ice fields in the 2026–35 and 1979–2008 periods is computed using the output from the CM3 experiment described in section 2a. This seasonally dependent difference is then treated as an anomaly field, and is added to every year of the HADISST1 dataset for the 1998–2008 period. This methodology is intended to remove any systematic biases in the climatology of the CM3 model and to discount any effect of the CM3-simulated climate change on interannual variability of the SST and sea ice fields.

  • Scenario run for the late part of the twenty-first century (S2), which is analogous to the S1 experiment except that the model atmosphere is driven by radiative forcings for the 2086–95 period and that the SST and sea ice “anomaly” field is computed using differences between the CM3 output for the 2086–95 and 1979–2008 periods.

Three parallel runs (i.e., ensemble members) initiated from independent atmospheric conditions have been conducted for each of the H, S1, and S2 experiments with HiRAM.

c. CFSR

Some of the model findings are checked against the corresponding observational results based on the new Climate Forecast System Reanalysis (CFSR) produced by the National Centers for Environmental Prediction (NCEP). The horizontal resolution of this dataset (38 km) is similar to that of the HiRAM model. The global reanalysis fields cover the 1979–2010 period. More details of the data input and data handling procedures used in generating the CFSR product are documented by Saha et al. (2010a). The observational patterns to be presented in this study are obtained from daily grids of maximum near-surface temperature, sea level pressure, precipitation, and near-surface wind, as well as geopotential height and horizontal wind at 250 mb, from the CFSR archives. A comprehensive description of the CFSR archives is provided by Saha et al. (2010b) and Environmental Modeling Center (2010).

3. Near-surface synoptic characteristics of HW occurring in selected regions

We first perform a search for those locations in the European sector where HW preferentially occur, by identifying spatially coherent regions with high-amplitude fluctuations in the surface temperature field. As demonstrated in LN, an appropriate method for achieving this objective is the application of rotated empirical orthogonal function (REOF) analysis (Horel 1981) to the daily maximum surface temperature (Tmax) data in the domain of interest. The results of this analysis are displayed in Fig. 1, which shows the patterns of the regression coefficients of Tmax versus the normalized temporal coefficients of the leading three or four REOF modes. All patterns are based on data for the present-day climate, as simulated in the H experiment (left column) and 1976–2005 segment of the CM3 run (center column), and as portrayed by the CFSR product (right column). The REOF analysis is conducted over the domain of 35°–60°N, 15°W–50°E, and is applied to Tmax variations within the 5-month warm season from 1 May to 30 September (hereafter referred as the MJJAS season). The climatological seasonal cycle, as obtained by computing the multiyear averages for individual calendar days and then performing a 31-day running mean, is removed from the daily time series for various years. The data for all available parallel runs are considered in the analysis of the H and CM3 experiments.

Fig. 1.

Regression charts of daily maximum surface air temperature (Tmax) in the May–September season vs the normalized temporal coefficients of selected first three or four REOF modes of Tmax, as computed using output from (left) the H experiment, (center) the CM3 experiment for the 1976–2005 period, and (right) CFSR data. The values shown in these charts represent the amplitude of local Tmax fluctuations corresponding to one standard deviation of the REOF temporal coefficients. The rank ordering of the REOF mode, and the fraction of variance of the respective dataset explained by that mode, are indicated at the upper left and right corner of each panel, respectively. Each of the rows depicts enhanced temporal variability in a “key region” of Europe. The labels assigned to these regions (western Russia, eastern Europe, and western Europe) are shown on the left side of each row of panels. The spatial extent of the key regions is indicated by a black contour, which corresponds to selected isopleths in the regression patterns.

Fig. 1.

Regression charts of daily maximum surface air temperature (Tmax) in the May–September season vs the normalized temporal coefficients of selected first three or four REOF modes of Tmax, as computed using output from (left) the H experiment, (center) the CM3 experiment for the 1976–2005 period, and (right) CFSR data. The values shown in these charts represent the amplitude of local Tmax fluctuations corresponding to one standard deviation of the REOF temporal coefficients. The rank ordering of the REOF mode, and the fraction of variance of the respective dataset explained by that mode, are indicated at the upper left and right corner of each panel, respectively. Each of the rows depicts enhanced temporal variability in a “key region” of Europe. The labels assigned to these regions (western Russia, eastern Europe, and western Europe) are shown on the left side of each row of panels. The spatial extent of the key regions is indicated by a black contour, which corresponds to selected isopleths in the regression patterns.

It is evident from Fig. 1 that the Tmax field in all model and observational datasets for the present-day climate exhibit pronounced and spatially coherent variability in three specific regions: western Russia centered near 52°N, 50°E (Figs. 1a–c), eastern Europe centered near 45°–50°N, 25°E (Figs. 1d–f), and western Europe centered near 45°–50°N, 10°E (Figs. 1g–i). We shall henceforth focus on the characteristics of HW occurring in these three “key regions.” For a given dataset, each of the key regions is defined as the collection of grid points in the corresponding regression chart (see Fig. 1) with coefficients larger than a certain threshold value, which is selected such that the key region of interest has a spatial extent of approximately 10° in the zonal and meridional directions. The boundary of the key regions as determined by this procedure is indicated in various panels of Fig. 1 using black contours. The locations of the three key regions, as inferred from output of HiRAM (left column of Fig. 1) and CM3 (center column), are in general agreement with observations (right column). There is also a strong spatial correspondence between the western Russian key region (top row of Fig. 1) and the site of the prominent heat wave that occurred in 2010 (see Dole et al. 2011), as well as between the western European key region (bottom row of Fig. 1) and the location of the heat wave in 2003 (Fink et al. 2004). By applying a hierarchical clustering algorithm to a high-resolution gridded observational dataset for the 1950–2009 period, Stefanon et al. (2012) have also identified several characteristic HW patterns in the Euro-Mediterranean sector. Their results for the Russian, eastern Europe, and western Europe clusters bear a considerable resemblance to the regression charts in Fig. 1.

In a given dataset, the search for HW events in a key region is conducted on the basis of the daily time series of the spatial average of Tmax over that key region. Only data at land points are used in computing this average. The 90th and 75th percentile values (hereafter referred to as T1 and T2, respectively) of this population of daily data for the MJJAS season in all available years are noted. A HW event is identified in the key region when the spatial mean of Tmax is higher than T1 for three consecutive days or more, the event-averaged value of Tmax is higher than T1, and Tmax is higher than T2 throughout the event.

The above computation procedure yields altogether 174, 98, and 69 HW events in the 90-yr H simulation (with three parallel runs, each lasting 30 years), and 36, 54, and 44 events in the 32-yr CFSR dataset, for western Russia, eastern Europe, and western Europe, respectively. Hence HiRAM overestimates the number of HW events per year in western Russia and underestimates the number of these episodes in eastern and western Europe. These HW events are then used to construct the composite patterns in Fig. 2 so as to illustrate the typical synoptic patterns associated with this phenomenon. In this analysis, an average is first taken over the duration of each HW event; these means are then averaged (by giving equal weight to each event) to yield the composite values. The distributions of the anomalous sea level pressure (SLP; contours), near-surface horizontal wind vector (arrows), and precipitation (shading) in the H and CFSR datasets are shown in Fig. 2 for HW occurring in each of the three key regions. The anomalous values in these charts and in analogous patterns to be presented in section 4 are obtained by removal of the climatological seasonal cycle, which is computed using the same method for processing the input to the REOF analysis. The boundary of the key region for identifying the HW, as defined on the basis of the regression charts in Fig. 1, is depicted in each panel of Fig. 2 by a bold purple contour.

Fig. 2.

Composite charts of sea level pressure (contours), surface wind vectors (arrows), and precipitation (shading) for heat waves occurring in the key regions located in (top) western Russia, (middle) eastern Europe, and (bottom) western Europe. Results are based on output from (left) the H experiment and (right) CFSR data. The boundaries of the key regions are indicated by a bold purple contour.

Fig. 2.

Composite charts of sea level pressure (contours), surface wind vectors (arrows), and precipitation (shading) for heat waves occurring in the key regions located in (top) western Russia, (middle) eastern Europe, and (bottom) western Europe. Results are based on output from (left) the H experiment and (right) CFSR data. The boundaries of the key regions are indicated by a bold purple contour.

Inspection of Figs. 2a and 2b reveals that both simulated and observed HW in western Russia are accompanied by high SLP anomalies near and to the east of the key region and low SLP anomalies over eastern Europe and Scandinavia. The key region is under the influence of enhanced southerly surface flow from low-latitude zones, or easterly flow with continental origin. Both of these anomalous airstreams are conducive to high temperatures at the key region. The presence of northerly wind anomalies over central/western Europe and the Russian sector east of 70°E is consistent with the abnormally low temperatures in those regions, as is evident from the regression patterns in Figs. 1a and 1c. Dry conditions generally prevail over the key region. Some of these synoptic characteristics are discernible in the Russian HW during the summer of 2010, as well as other outstanding warm episodes observed in that region throughout the twentieth century (Dole et al. 2011).

The synoptic pattern associated with HW in eastern Europe (Figs. 2c,d) is characterized by an anticyclonic anomaly with an axis near 30°–40°E and meridional extension from 40° to 60°N. Southerly or easterly surface wind anomalies prevail over the warm and dry key region in eastern Europe. The association of observed HW in south-central Europe with local anticyclonic circulation and southerlies has been noted by Domonkos et al. (2003). A prominent cyclonic center is evident over the British Isles and Norwegian Sea, thus bringing about wet conditions to some parts of western and northern Europe. By analyzing the weather records at many European stations and the frequency statistics of various large-scale circulation types (Grosswettertypen) classified in the catalogue compiled by Hess and Brezowsky (1952), Kyselý (2008) noted a significantly higher frequency of incidence of the circulation type labeled as SWZ (Southwest C) during HW episodes occurring in eastern European stations. The pattern of this particular circulation type (see Fig. 2 of Kyselý 2008) bears some similarity to the composite charts presented in Figs. 2c and 2d, especially in regards to the cyclonic center in the vicinity of the Greenwich meridian.

The most prominent feature in the composite charts for HW in western Europe (Figs. 2e,f) is the anomalous SLP ridge located at 10°–20°E. Enhanced cyclonic flows are seen over the eastern North Atlantic. The continental ridge drives an offshore circulation over western Europe, thus leading to warm and dry conditions in that region. Kyselý presented evidence on the higher frequency of occurrence of the circulation type “HM” (central European high) during HW episodes in western Europe. The principal features in our composite (Figs. 2e,f) compare well with those of the HM pattern, as documented by Kyselý (2008, see his Fig. 2).

4. Atmospheric dynamical processes in the upper troposphere during HW episodes

To investigate the dynamical interactions between upper tropospheric features of different time scales in the course of the HW events, the following diagnostic tools are utilized:

  • Envelope function (EF) [see more detailed description by Nakamura and Wallace (1990)], which provides a daily measure of the amplitude of synoptic-scale eddies, as computed by taking the square of bandpass (2.5–6 day) filtered geopotential height at 250 mb, and then subjecting the resulting time series to a low-pass (longer than 10 days) filter.

  • Local Eliassen–Palm vectors (E) [see formulation and applications to eddy–mean flow interactions by Hoskins et al. (1983)], with horizontal components given by (, ), where u and υ are the zonal and meridional wind components at 250 mb, the prime indicates bandpass filtered fluctuations, and the overbar represents the low-pass filtering operation. The orientation of E is linked to the direction of the group propagation vector of the synoptic-scale eddies relative to the mean flow. Divergence (convergence) of E implies a tendency for the high-frequency fluctuations to accelerate (decelerate) the mean zonal circulation.

The distributions of the anomalous geopotential height at 250 mb (contours), EF (shading), and E, as obtained by compositing over the HW identified in the three key regions, are presented in Fig. 3 for the H experiment (left panels) and the CFSR dataset (right panels). Comparison of the contour patterns in Figs. 2 and 3 reveals a distinct westward displacement of the geopotential height field at 250 mb relative to the SLP signals at the surface. For a given key region, the center of the positive height anomaly is situated above the site of occurrence of the HW (see boundary of key regions as depicted by bold purple contours in Fig. 3). The prevalence of positive height anomalies in the middle troposphere over the HW sites in Russia, eastern Europe, and western Europe has also been noted in the observational cluster analysis by Stefanon et al. (2012).

Fig. 3.

As in Fig. 2, but for 250-mb geopotential height (contours), envelope function (EF, shading), and E (arrows). The boundaries of the key regions are indicated by a bold purple contour.

Fig. 3.

As in Fig. 2, but for 250-mb geopotential height (contours), envelope function (EF, shading), and E (arrows). The boundaries of the key regions are indicated by a bold purple contour.

The anomalous, quasi-stationary anticyclones in the upper troposphere affect the path of synoptic-scale disturbances (hereafter referred to as storm tracks) in the vicinity of the key region. Eastward penetration of the synoptic-scale eddies to the center and south of the anomalous anticyclones is markedly impeded, as indicated by the negative anomalies of EF (see shading in Fig. 3). The diminished level of transient activity near the key regions is consistent with the dry conditions at these sites (see shading patterns in Fig. 2) and is conducive to the stagnant character of the circulation pattern accompanying the HW. On the other hand, the storm tracks are steered around the northern perimeter of the anticyclones so that positive departures of EF prevail in those locations.

The influences of the anomalous anticyclonic background flow on the group propagation of the bandpass fluctuations are further illustrated by the patterns of E in Fig. 3. The predominantly westward-directed E arrows near and to the south of the positive geopotential height anomaly are indicative of reduced eastward migration (or blocking) of synoptic-scale eddies across those regions. The general alignment of the E arrows with the geopotential height contours in the northern portion of the anticyclones illustrates the steering effect of the background circulation field on the storm tracks. Specifically, the E are diverted poleward as they approach the anticyclonic center from the west, and are seen to turn in a clockwise fashion as they traverse the northern edge of the anticyclone.

The patterns of SLP and geopotential height in Figs. 2 and 3 (contours) indicate the presence of a cyclonic feature to the west of the primary anticyclonic center, thus yielding a wavelike appearance in these composites. The spatial organization of the EF and E fields about the cyclonic center is analogous to that associated with the anticyclone, except for a reversal in the polarity of the anomalies.

The effects of the altered intensity and path of the storm track disturbances on the background flow may be assessed using the eddy-induced geopotential height tendency

 
formula

where f is the Coriolis parameter, g the gravitational acceleration, V the horizontal wind vector, and ζ the relative vorticity; the meanings of the prime and overbar are the same as in the definition of E. The above relationship may be derived by noting that the vorticity tendency induced by transient eddies [i.e., , or ] may be equated to the convergence of eddy vorticity flux, . The data for may be obtained from , and , using the relationships given by Lau and Holopainen (1984). The distributions of anomalous geopotential height (contours) and (shading) at 250 mb, as obtained by composite analysis of the HW identified in the three key regions, are presented in Fig. 4 for the H simulation (left panels) and the CFSR dataset (right panels). For all key regions, a strong spatial correspondence exists between the height and fields in both the HiRAM and observed atmospheres, with the eddy vorticity fluxes acting to impart positive height tendencies to the anomalous anticyclone, and negative tendencies to the cyclone. The eddy–mean flow interaction accompanying the HW is hence characterized by a positive feedback. The slowly varying circulation modulates the strength and trajectory of the storm track disturbances, which in turn reinforce the background flow through eddy vorticity transports.

Fig. 4.

As in Fig. 2, but for 250-mb geopotential height (contours) and eddy-induced height tendency (shading).

Fig. 4.

As in Fig. 2, but for 250-mb geopotential height (contours) and eddy-induced height tendency (shading).

Additional insights into the patterns of in Fig. 4 may be gained by inspecting the distributions of EF and E in Fig. 3. Keeping in mind that storm track axes are sites of divergence of E, thus leading to acceleration of the local zonal wind (Hoskins et al. 1983), poleward displacement of the storm track axis due to the blocking anticyclone accompanying the HW would induce zonal wind acceleration to the north of this anticyclone. Conversely, suppressed synoptic-scale activity at the center and to the south of the anticyclone is coincident with convergence of E and zonal wind deceleration. The divergence/convergence patterns of E, as inferred from the above arguments, are discernible in most of the panels in Fig. 3. The implied patterns of eddy-induced zonal wind tendency (i.e., acceleration north of the anticyclonic center, deceleration south of the center) are consistent with the positive height tendency at the center itself, based on geostrophic considerations.

The wavy character of the composite patterns for SLP and 250-mb height (Figs. 2 and 3) suggests that the HW may be linked to planetary-scale phenomena. The typical spatial extent and temporal evolution of such variations are explored by performing an extended empirical orthogonal function (EEOF) (see Weare and Nasstrom 1982) analysis of the low-pass (10 days and longer) filtered geopotential height field at 250 mb for the MJJAS season in the H and CFSR datasets. This computation is conducted for the domain of 30°–70°N, 60°W–120°E. Four temporal lags, at 0, 3, 6, and 9 days, are used to construct the input correlation matrix for this analysis. The leading pair of modes explains 17% and 16% of the variance in the H and CFSR datasets, respectively.

The distributions of the regression values of the low-pass filtered 250-mb height data versus the normalized temporal coefficients of the leading EEOF mode, as constructed using H (left panels) and CFSR (right panels) data, are presented in Fig. 5. Regression charts are arranged from top to bottom with increasing time lags, from 0 to 9 days at 3-day intervals. These patterns portray the temporal sequence of development of the low-pass height fluctuations associated with the principal mode of variability in the two datasets. An arclike wave chain consisting of alternating positive and negative centers prevails over the North Atlantic–Eurasian sector. The individual centers exhibit almost no spatial displacements throughout the 9-day period considered here. However, the centers in the western portion of the wave chain (i.e., extrema over the North Atlantic and near the British Isles) are seen to attain maximum amplitudes earlier (see charts for 0- and 3-day lags in Fig. 5) than the centers in the eastern portion (western Russia and Siberia, see charts for 6- and 9-day lags). This eastward progression of amplitude growth at geographically fixed anomaly centers is reminiscent of the successive downstream development of dispersive Rossby wave trains with similar time scales, as documented by Blackmon et al. (1984). The essential spatiotemporal characteristics of the regression charts based on CFSR data are reproduced in the H atmosphere, except for some shifts in the location of the individual anomaly centers.

Fig. 5.

Regression charts of low-pass filtered 250-mb height in the May–September season vs the normalized temporal coefficients of the leading EEOF mode of low-pass filtered 250-mb height, as computed using (left) output from the H experiment and (right) CFSR data. The values shown in these charts represent the amplitude of local 250-mb height fluctuations corresponding to one standard deviation of the REOF temporal coefficients. Results are based on temporal lags of (top) 0, (second row) 3, (third row) 6, and (bottom) 9 days.

Fig. 5.

Regression charts of low-pass filtered 250-mb height in the May–September season vs the normalized temporal coefficients of the leading EEOF mode of low-pass filtered 250-mb height, as computed using (left) output from the H experiment and (right) CFSR data. The values shown in these charts represent the amplitude of local 250-mb height fluctuations corresponding to one standard deviation of the REOF temporal coefficients. Results are based on temporal lags of (top) 0, (second row) 3, (third row) 6, and (bottom) 9 days.

The patterns of the leading EEOF mode (Fig. 5) bear a strong resemblance to the composite charts of 250-mb height based on HW identified in western Russia (Figs. 3a,b), thus suggesting that the occurrence of HW in that key region is linked to the downstream influence of Rossby wave trains extending from the North Atlantic to central and southern Asia (see also Lau and Kim 2012). This inference may be tested by a correlation analysis of the time series representing the wave train development (i.e., temporal coefficients of the leading EEOF modes presented in Fig. 5) and temperature variability over western Russia (i.e., areal average of Tmax over this key region). The correlation coefficients between this pair of time series at various time lags are plotted in Fig. 6, for H (solid curve) and CFSR (dashed curve) data. The correlation is seen to peak at a lag of 5–6 days (4–5 days) in the model (observed) atmosphere. The positive height anomaly center over western Russia associated with the EEOF mode also attains maximum amplitude at a lag of 3–6 days (see Figs. 5b,c and 5f,g). These considerations imply that thermal conditions over western Russia are strongly modulated by wave train development over the North Atlantic and Eurasia, with warmest temperatures occurring when the local positive height anomaly reaches maximum strength.

Fig. 6.

Variations with time lag of correlation coefficients between temporal coefficients of leading EEOF of low-pass filtered 250-mb height and areal average of Tmax over the key region of western Russia. Results are based on output from the H experiment (solid curve) and CFSR data (dashed curve). Positive (negative) values on the abscissa correspond to temporal lag (lead) of the Tmax fluctuations relative to the EEOF coefficients. Assuming an independent sample for each 30-day interval and applying a two-tailed Student’s t test, the correlation coefficients are different from zero at the 99% significance level when they exceed 0.12 (0.21) for H (CFSR) data.

Fig. 6.

Variations with time lag of correlation coefficients between temporal coefficients of leading EEOF of low-pass filtered 250-mb height and areal average of Tmax over the key region of western Russia. Results are based on output from the H experiment (solid curve) and CFSR data (dashed curve). Positive (negative) values on the abscissa correspond to temporal lag (lead) of the Tmax fluctuations relative to the EEOF coefficients. Assuming an independent sample for each 30-day interval and applying a two-tailed Student’s t test, the correlation coefficients are different from zero at the 99% significance level when they exceed 0.12 (0.21) for H (CFSR) data.

The results reported in this section are in accord with the findings presented by Schubert et al. (2011), who noted a prominent role for Rossby wave development in extreme climate events during boreal summer, including the 2003 European and 2010 Russian heat waves. By diagnosing the solutions of a stationary wave model, these investigators also demonstrated the dominance of eddy vorticity flux convergence [i.e., ] in the forcing of the Rossby wave trains. In an earlier diagnostic study, Illari and Marshall (1983) have also noted the role of eddy fluxes of potential vorticity in maintaining a blocking episode over western Europe in July 1976, which is associated with a HW episode in that region.

5. Projected changes in HW characteristics in future climate scenarios

In previous sections, we have assessed the fidelity of HiRAM in replicating the synoptic behavior of, and dynamical mechanisms contributing to, the observed HW over Europe in the present-day climate. We now proceed to utilize this model tool to project the changes in various characteristics of the HW in the twenty-first century, by subjecting HiRAM to future climate scenarios (see descriptions of the S1 and S2 experiments in section 2), and compare the output of these runs with that based on the H simulation.

To gain an impression of the projected change in the seasonal mean climate, the difference charts of Tmax in the MJJAS season, as obtained by subtracting the average over the H run from the averages over the (a) S1 and (b) S2 experiments, are displayed in Fig. 7. These results indicate warming in the Eurasian sector of 2°–4°C in 2026–35, and 4°–6°C in 2086–95, relative to the 1979–2008 period. The general pattern in Fig. 7 bears some similarity to its counterpart based on the output from multiple models contributing to phase 3 of the Coupled Model Intercomparison Project (CMIP3), as shown in Meehl et al. (2007, see their Fig. 10.9). The results in Fig. 7 may also be compared with the projected temperature changes deduced from a suite of regional climate models (RCMs) analyzed by Fischer and Schär (2010, see their Fig. 1a and Table 1). There is broad agreement between the projections for the Iberian Peninsula and the Mediterranean region based on HiRAM and the RCMs. However, the temperature increase over central and eastern Europe is higher in the HiRAM experiment than in the RCM runs. The strong meridional contrast in the pattern for temperature change (with larger increases in the south and smaller increases in the north) is much more apparent in the RCM result than in the HiRAM pattern shown in Fig. 7. Diagnoses of RCM output by Kjellström et al. (2007), Boberg and Christensen (2012), and Vautard et al. (2013) indicate that these models share systematic temperature-dependent biases, which might lead to overestimation of the warming in the Mediterranean region and underestimation in high-latitude zones, thus enhancing the north–south contrast in the spatial distribution of temperature change.

Fig. 7.

Distributions of the difference in the climatological mean of Tmax between the (a) S1 and H runs and (b) S2 and H runs for the May–September season.

Fig. 7.

Distributions of the difference in the climatological mean of Tmax between the (a) S1 and H runs and (b) S2 and H runs for the May–September season.

The HW events in the S1 and S2 experiments are identified by following the same procedure as that applied to the H simulation (see section 3), and using the same T1 and T2 criteria as determined based on the H dataset. As in LN, the severity of a given HW is defined as the average of Tmax over the lifetime (referred hereafter as the duration) of that event. For a given key region in a given model or observational dataset, the mean severity and duration are computed by averaging these measures over all identified HW (by giving equal weight to each HW event), the mean frequency is evaluated by dividing the total population of events by the number of available years, and the average number of HW days per year is obtained by multiplying the mean duration and frequency. All of these HW metrics are computed separately for the H, S1, and S2 runs.

As in LN, the impacts of the projected mean shift in Tmax in the 2026–35 and 2086–95 periods (as illustrated in Fig. 7) on the HW metrics are assessed using the following procedure. The differences in the climatological Tmax fields (hereafter referred as ∆Tmax) are computed between the S1 and H experiments. These values (which are analogous to those displayed in Fig. 7a, except that the differences are computed for individual calendar months instead of the MJJAS season), are added to the Tmax data generated in individual days for all years in the H experiment, so as to produce a new dataset, hereafter referred to as S1*. Analogously, the ∆Tmax values based on the S2 and H experiments are added to the daily output from the H experiment so as to produce another new dataset, S2*. HW episodes are identified on the basis of these new datasets using the same scheme previously applied to the S1, S2, and H runs, with the same T1 and T2 thresholds being established from the H experiment. The various statistical measures of HW detected in the three key regions, as computed based on the available datasets, are summarized in Table 1.

Table 1.

(top) Averaged severity (°C) and duration (in days), and (bottom) frequency of occurrence (in episodes per year) and number of HW days per year, for various key regions, and for various datasets (CFSR, as well as output generated by HiRAM in the H, S1, and S2 experiments; see individual columns). The additional datasets labeled S1* and S2* are generated by applying to the daily data in H a constant shift ΔTmax equivalent to the mean summertime warming in the respective key regions as projected in the S1 and S2 runs, respectively. For a given key region, and for duration, frequency, and HW days per year, the statistics based on the S1, S2, S1*, and S2* datasets are displayed in two rows. The first row shows the actual values of the metric in question; the second row shows the ratio between these values and the corresponding estimate based on the H experiment (i.e., S1/H, S2/H, S1*/H, and S2*/H).

(top) Averaged severity (°C) and duration (in days), and (bottom) frequency of occurrence (in episodes per year) and number of HW days per year, for various key regions, and for various datasets (CFSR, as well as output generated by HiRAM in the H, S1, and S2 experiments; see individual columns). The additional datasets labeled S1* and S2* are generated by applying to the daily data in H a constant shift ΔTmax equivalent to the mean summertime warming in the respective key regions as projected in the S1 and S2 runs, respectively. For a given key region, and for duration, frequency, and HW days per year, the statistics based on the S1, S2, S1*, and S2* datasets are displayed in two rows. The first row shows the actual values of the metric in question; the second row shows the ratio between these values and the corresponding estimate based on the H experiment (i.e., S1/H, S2/H, S1*/H, and S2*/H).
(top) Averaged severity (°C) and duration (in days), and (bottom) frequency of occurrence (in episodes per year) and number of HW days per year, for various key regions, and for various datasets (CFSR, as well as output generated by HiRAM in the H, S1, and S2 experiments; see individual columns). The additional datasets labeled S1* and S2* are generated by applying to the daily data in H a constant shift ΔTmax equivalent to the mean summertime warming in the respective key regions as projected in the S1 and S2 runs, respectively. For a given key region, and for duration, frequency, and HW days per year, the statistics based on the S1, S2, S1*, and S2* datasets are displayed in two rows. The first row shows the actual values of the metric in question; the second row shows the ratio between these values and the corresponding estimate based on the H experiment (i.e., S1/H, S2/H, S1*/H, and S2*/H).

The severity of HW in all three regions, as obtained from the H experiment, is in good agreement with the observational estimates (CFSR). The discrepancies between the H and CFSR datasets are larger for frequency and number of HW days per year.

Comparison between the HW metrics for individual regions based on the H, S1, and S2 datasets indicates only a slight elevation of severity level (by less than 1°C) in the S1 and S2 runs. It should be noted here that the change in severity from H to S1 or S2 is critically dependent on the procedure and criteria for identifying HW. Alternative procedures and criteria could yield rather different results. On the other hand, the HiRAM projects a marked increase in duration (by a factor of 1.4–2.0), frequency (by a factor of 2.2–4.5), and number of HW days per year (by a factor of 3–7) through the course of the twenty-first century.

There is an overall good match between the HW measures for the S1 and S2 datasets and those for the S1* and S2* datasets, respectively. The most notable exception is the underestimate by the S1* and S2* datasets of the number of HW days per year over western Europe. As argued in LN, the general agreement between the S1 and S1* results, as well as between the S2 and S2* results, suggests that the changes in various HW metrics from one period to another may largely be attributed to the shift in the mean climate (i.e., ∆Tmax) during these two periods. The correspondence of the HW statistics for western Russia and eastern Europe based on the S1* and S2* datasets with those based on S1 and S2 datasets also implies that the variability of the daily Tmax fluctuations in these regions about the mean climate does not change appreciably from one period to another. The lower estimates of HW days in western Europe in the S1* and S2* datasets as compared to those in S1 and S2 simulations indicates significant changes in both the mean and daily variability of Tmax in future climates for this region.

To examine the detailed spatial pattern of the projected change in HW characteristics, the identification procedure for HW events has been applied to daily time series of Tmax at individual grid points, instead of areal average of Tmax over selected key regions. The severity, duration, and frequency measures for HW occurring at each grid point are then computed using Tmax data for that location. The geographical distribution of the ratio of number of HW days per year in the S2 run versus the same measure in the H run, as obtained from this grid-point-based analysis, is presented in Fig. 8a. The corresponding pattern based on the S2* dataset is displayed in Fig. 8b. The chart in Fig. 8a indicates that the number of HW days is projected to increase throughout the domain of interest in the twenty-first century. The ratio of this metric in 2086–95 to that in the present climate ranges from 2 to 3 in parts of Russia, to as much as 5–6 in many regions in central and western Europe. This large-scale pattern is consistent with the relatively higher S2/H ratios for the key regions in central and western Europe as compared to western Russia (see Table 1). It is worth noting that high ratios (exceeding 5) are projected at many grid points in the Iberian Peninsula, Asia Minor, the British Isles, and Scandinavia, which are not included in the three key regions identified by REOF analysis (see Fig. 1). It would be of interest to further investigate the present-day and future HW characteristics in all of these additional sites. The analysis of a recent suite of RCM projections, as reported by Jacob et al. (2014), also indicates enhanced HW occurrences in many of the locations with high ratios in Fig. 8a. The considerable amount of spatial information in Fig. 8a illustrates the advantages of conducting most of our diagnosis on the output from HiRAM. Such rich regional details are not apparent in the corresponding result based on CM3 data (not shown).

Fig. 8.

Distributions of the ratio of number of heat wave days per year in the (a) S2 and (b) S2* datasets vs the same measure in the H simulation, as computed for heat wave events based on Tmax time series at individual grid points.

Fig. 8.

Distributions of the ratio of number of heat wave days per year in the (a) S2 and (b) S2* datasets vs the same measure in the H simulation, as computed for heat wave events based on Tmax time series at individual grid points.

The pattern for HW days based on the S2* dataset (Fig. 8b) is generally in agreement with that derived from the S2 projection (Fig. 8a). This comparison confirms the key role of mean climate shift ∆Tmax in modifying HW behavior not only for areal averages (see Table 1), but also at individual grid points of a high-resolution dataset. The underestimation in the S2* dataset of HW days over western Europe, as noted in Table 1, is also evident from grid point results in Figs. 8a and 8b.

The “time slice” integrations in the S1 and S2 experiments with HiRAM provide data on projected climate change in only two decades of the twenty-first century. A more detailed depiction of the temporal evolution of the HW characteristics can be made using the output from the CM3 experiment, which is conducted through the entire span of the twentieth and twenty-first centuries (see section 2). The key regions for HW occurrences are chosen on the basis of the REOF analysis of Tmax data from the 1976–2005 period of the CM3 run, as illustrated in Figs. 1b, 1e, and 1h. HW events are chosen according to the areal average of Tmax over the land points in these regions in the MJJAS season and applying the T1, T2 criteria derived from CM3 data for the 1976–2005 period. In each decade of the period from 1901 to 2100, the averaged severity, duration, frequency, and number of HW days are computed for HW identified in that decade. The time series of various decadal HW measures, as well as the decadal mean of the areal average of Tmax, for each of the three key regions are shown in Fig. 9. The thin curves of various colors in this figure depict the results based on each of the five parallel model runs in the historical period (1901–2000) and the three individual runs for the future climate scenario in the 2011–2100 period. The bold black curves represent the ensemble means, as computed using the data from all available model runs for each decade.

Fig. 9.

Time series of the (left) severity, (left center) duration, (center) frequency of occurrence, (right center) number of heat wave days per year, and (right) areal averaged Tmax as computed for the May–September season in individual decades of the twentieth and twenty-first centuries of the CM3 experiment. Results are shown for heat waves occurring in the key regions of (top) western Russia, (middle) eastern Europe, and (bottom) western Europe. Time series for each of the five (three) parallel model runs in the 1901–2000 (2011–2100) periods are displayed using thin colored curves. Ensemble-mean HW measures based on data from all available model runs are depicted using a bold black curve.

Fig. 9.

Time series of the (left) severity, (left center) duration, (center) frequency of occurrence, (right center) number of heat wave days per year, and (right) areal averaged Tmax as computed for the May–September season in individual decades of the twentieth and twenty-first centuries of the CM3 experiment. Results are shown for heat waves occurring in the key regions of (top) western Russia, (middle) eastern Europe, and (bottom) western Europe. Time series for each of the five (three) parallel model runs in the 1901–2000 (2011–2100) periods are displayed using thin colored curves. Ensemble-mean HW measures based on data from all available model runs are depicted using a bold black curve.

Examination of the bold black curves in Fig. 9 reveals that, in the course of the twenty-first century, the ensemble-mean severity increases by up to 1.5°C, and the duration is prolonged by a factor of 2–3. The time series for mean frequency exhibit large multidecadal fluctuations throughout the 200-yr period. A prominent increase in mean frequency (by a factor of more than 3 in some regions) is discernible within the 1980–2040 period. The ensemble-averaged number of HW days exhibits only minor changes during much of the twentieth century, but rises steeply and almost monotonically beyond 2000, with a total increase by a factor of 3–4 above the present-day level. The evolution of the number of HW days (fourth column from left of Fig. 9) bears considerable resemblance to that of the areal mean of Tmax (fifth column), which indicates a warming of 5°–6°C in the twenty-first century. Some of the temporal changes in the mean HW duration and frequency, as shown in Fig. 9, are partially supported by similar estimates reported by Beniston (2004), Della-Marta et al. (2007a), and Kyselý (2010) based on observational data and various model projections.

The HW severity and frequency, as estimated using data for individual model runs (see thin colored curves in first and third columns from left of Fig. 9), exhibit notable excursions from the ensemble-mean values (bold black curves). The scatter of the individual realizations about the mean response is relatively lower in the results for number of HW days per year (fourth column of Fig. 9), thus supporting the arguments presented in LN that the latter metric is a more robust indicator of HW activity. For all four HW measures considered here, the magnitude of projected ensemble-mean changes through the twenty-first century generally exceeds the level of variability among the individual samples.

While interpreting the evolution of various HW metrics based on CM3 output in Fig. 9, it should be borne in mind that the increase in global-mean surface air temperature in the twentieth century, as simulated by this model, is smaller than the observed trend (Golaz et al. 2013). Moreover, the analysis performed by Levy et al. (2013) indicates that CM3 projects more warming for the twenty-first century as compared to a previous version of the GFDL Climate Model (CM2.1), probably due to the treatment of indirect aerosol effects in CM3.

6. Summary and discussion

The capability of a high-resolution GCM developed at GFDL (HiRAM) to replicate the essential characteristics of heat waves observed in different European regions is assessed. This model evaluation is conducted by comparing composite patterns of various atmospheric fields based on simulations with HiRAM under the present-day climate setting with corresponding results derived from a reanalysis product with similar spatial resolution. The typical synoptic flow patterns accompanying prominent HW in the model atmosphere are in agreement with their observed counterparts (see Fig. 2). These model patterns also bear some resemblance to circulation types that are known to be associated with HW in historical weather records.

A ubiquitous synoptic feature in the composite charts for the HW episodes is the upper-level anomalous anticyclone situated above the region affected by the HW (Figs. 3 and 4). This high center steers the incident high-frequency transient disturbances poleward so that the level of transient activity at the HW site itself is reduced considerably. The resulting change in the eddy vorticity forcing leads to reinforcement of the anticyclone. These findings suggest that HW episodes could persist owing to the stagnant flow environment associated with the blocking high, as well as positive dynamical feedbacks between the quasi-stationary and transient components of the circulation. High-resolution models are needed to replicate such interactions between the transient disturbances and the ambient slowly varying flow field, as is illustrated by the diagnostics performed on the output from HiRAM (Figs. 3 and 4).

The diagnostics presented in Figs. 3 and 4 indicate that poleward shifts of the storm track are conducive to anticyclonic development of the quasi-stationary flow, which in turn provides a favorable environment for HW occurrences. Secular trends of poleward migration of the summertime storm tracks in the Eurasian sector have been reported by Zhang et al. (2004) and Bengtsson et al. (2006), among others, based on observational data and model projections, respectively. As pointed out by Kyselý (2008), such trends in storm track behavior could lead to more frequent and persistent HW in Europe.

The anticyclonic anomaly over the HW region is seen to be part of a larger-scale wave train in the North Atlantic–Eurasian sector. The typical spatiotemporal evolution of this phenomenon (Fig. 5) is similar to that associated with Rossby wave dispersion, with successive downstream development of geographically fixed nodes and antinodes. The spatial phase of this wave train is closely related to the location of the HW region, with the center of positive anomaly in 250-mb height lying above the key region of interest (see Figs. 2 and 3). The timing of the downstream development of the wave train also matches well with that of the surface temperature variations within the HW region (see Fig. 6).

After demonstrating the fidelity of HiRAM in reproducing the synoptic aspects and various basic metrics (such as severity, duration, and frequency) of European HW, additional runs are conducted with this model by subjecting it to climate forcings for different segments of the twenty-first century. Analysis of the HW statistics for these scenario runs indicates marked increases in duration, frequency, and HW days per year toward the end of the twenty-first century. The spatial pattern of the change in number of HW days per year is examined in greater detail by computing and mapping this metric based on temperature time series at individual grid points (Fig. 8). The decade-by-decade variations of the HW statistics through the twentieth and twenty-first centuries are presented using output from a coupled model (CM3) experiment with a continuous span over the 1901–2100 period (Fig. 9). These results indicate monotonic increase in duration and HW days per year in various European regions throughout the twenty-first century.

The results presented in this study are generally in support of the key findings reported in LN on the basis of HW in North America. Specifically, the synoptic flow pattern accompanying HW in both North America and Europe is dominated by an upper-level anticyclone that blocks the propagation of transient eddies through the HW region. Moreover, the projected increases in duration and frequency of European HW during the twenty-first century are similar in magnitude to the increases in these measures for the North American HW (cf. Table 1 in this article with Table 1 in LN). It is also demonstrated that, for both Europe and North America, the mean shift in the climatological surface temperature (∆Tmax) is a critical factor in determining the projected change in various HW metrics for the twenty-first century. This inference is based upon the similarity between HW statistics for the S1/S2 datasets and those for the S1*/S2* datasets (see appropriate entries in Table 1 of this study and in LN). In summary, the additional evidence offered here serves as useful independent checks of the validity and robustness of the main conclusions reported in LN.

Although the analysis approach and principal results in this work are similar to those in LN, the following novel aspects of the present study are worth noting. The dynamical interactions between the quasi-stationary component of the circulation and the synoptic-scale transients during European HW episodes are examined here in considerably more detail using E and eddy-induced geopotential height tendencies (Figs. 3 and 4). These diagnostics have not been performed in LN. Furthermore, the successive downstream development of the wavelike pattern in the upper troposphere (as portrayed by EEOF analysis, see Fig. 5), and its temporal link with HW episodes (Fig. 6) are brought into much sharper focus in this study than in LN. The model results in LN were compared with their observational counterparts as computed using the North American Regional Reanalysis (NARR) dataset, which provides coverage over the North American sector only. In the current study, the observational results are based on the more recent Climate Forecast System Reanalysis dataset, which has global coverage. The findings presented here illustrate the utility of the CFSR dataset for studying various regional details of the observed climate system. It would be of interest to make further use of the fine spatial resolution afforded by this new data resource to investigate myriad regional phenomena occurring in different parts of the globe.

As was discussed in LN, many processes could contribute to a HW event in various stages of its development. Such processes include positive temperature advection due to enhanced southerly or offshore flows, adiabatic warming accompanying the subsidence at the anticyclone, interactions between the anomalous circulation and the local orography, effects of cloud cover and precipitation changes on various forms of heat transfer near the land surface, and the remote influences of sea surface temperature anomalies. In particular, Stefanon et al. (2012) have noted that HW episodes in western and eastern Europe are preceded by pronounced droughts in nearby regions, thus suggesting that air–land interactions may contribute substantially to the occurrence and maintenance of such HW events. Improved understanding of the role and relative importance of these individual mechanisms in the evolution of present-day and future HW could be achieved by more in-depth diagnoses of the output from the HiRAM experiments.

Impacts of HW on human society and public health also depend on the diurnal variations of the air temperature and humidity extremes. These aspects of the HW require further study using model and observational datasets. The future changes in various HW metrics are projected here by applying a particular set of criteria for identifying the outstanding events to a particular pair of models (HiRAM and CM3). In view of the differences of the CM3 results for the twentieth and twenty-first centuries from observations and other model projections (see discussion at the end of section 5), the robustness of the findings reported here needs to be evaluated more thoroughly by diagnosing the output from a larger suite of models developed outside of GFDL. The dependence of the results on the criteria for selecting the HW events also needs to be tested. For instance, the HW episodes in future climate scenarios are identified in this study by applying the same thresholds T1 and T2 as determined based on present-day data. Since human society will likely acclimate to a warmer climate to some degree, it is of interest to select HW episodes in future climates using higher T1 and T2 thresholds than those values considered here, and examine the statistics derived therefrom to reassess the implications of climate change on HW occurrences.

Acknowledgments

We thank our colleagues at GFDL for providing the output from various experiments with HiRAM and CM3. We are indebted to Tom Knutson, John Lanzante, and Ming Zhao for their perceptive comments on an earlier version of this manuscript.

REFERENCES

REFERENCES
Bengtsson
,
L.
,
K. I.
Hodges
, and
E.
Roeckner
,
2006
:
Storm tracks and climate change
.
J. Climate
,
19
,
3518
3543
,
doi:10.1175/JCLI3815.1
.
Beniston
,
M.
,
2004
:
The 2003 heat wave in Europe: A shape of things to come? An analysis based on Swiss climatological data and model simulations
.
Geophys. Res. Lett.
,
31
,
L02202
,
doi:10.1029/2003GL018857
.
Black
,
E.
,
M.
Blackburn
,
G.
Harrison
,
B.
Hoskins
, and
J.
Methven
,
2004
:
Factors contributing to the summer 2003 European heatwave
.
Weather
,
59
,
217
223
,
doi:10.1256/wea.74.04
.
Blackmon
,
M. L.
,
Y.-H.
Lee
,
J. M.
Wallace
, and
H.-H.
Hsu
,
1984
:
Time variation of 500 mb height fluctuations with long, intermediate and short time scales as deduced from lag-correlation statistics
.
J. Atmos. Sci.
,
41
,
981
991
,
doi:10.1175/1520-0469(1984)041<0981:TVOMHF>2.0.CO;2
.
Boberg
,
F.
, and
J. H.
Christensen
,
2012
:
Overestimation of Mediterranean summer temperature projections due to model deficiencies
.
Nat. Climate Change
,
2
,
433
436
,
doi:10.1038/nclimate1454
.
Carril
,
A. F.
,
S.
Gualdi
,
A.
Cherchi
, and
A.
Navarra
,
2008
:
Heatwaves in Europe: Areas of homogeneous variability and links with the regional to large-scale atmospheric and SSTs anomalies
.
Climate Dyn.
,
30
,
77
98
,
doi:10.1007/s00382-007-0274-5
.
DeGaetano
,
A. T.
, and
R. J.
Allen
,
2002
:
Trends in twentieth-century temperature extremes across the United States
.
J. Climate
,
15
,
3188
3205
,
doi:10.1175/1520-0442(2002)015<3188:TITCTE>2.0.CO;2
.
Della-Marta
,
P. M.
,
M. R.
Haylock
,
J.
Luterbacher
, and
H.
Wanner
,
2007a
:
Doubled length of western European summer heat waves since 1880
.
J. Geophys. Res.
,
112
, D15103, doi:10.1029/2007JD008510.
Della-Marta
,
P. M.
,
J.
Luterbacher
,
H.
von Weissenfluh
,
E.
Xoplaki
,
M.
Brunet
, and
H.
Wanner
,
2007b
:
Summer heat waves over western Europe 1880–2003, their relationship to large-scale forcings and predictability
.
Climate Dyn.
,
29
,
251
275
,
doi:10.1007/s00382-007-0233-1
.
Diffenbaugh
,
N. S.
,
J. S.
Pal
,
R. J.
Trapp
, and
F.
Giorgi
,
2005
:
Fine-scale processes regulate the response of extreme events to global climate change
.
Proc. Natl. Acad. Sci. USA
,
102
,
15 774
15 778
,
doi:10.1073/pnas.0506042102
.
Dole
,
R.
, and Coauthors
,
2011
:
Was there a basis for anticipating the 2010 Russian heat wave?
Geophys. Res. Lett.
,
38
,
L06702
,
doi:10.1029/2010GL046582
.
Domonkos
,
P.
,
J.
Kyselý
,
K.
Piotrowicz
,
P.
Petrovic
, and
T.
Likso
,
2003
:
Variability of extreme temperature events in south–central Europe during the 20th century and its relationship with large-scale circulation
.
Int. J. Climatol.
,
23
,
987
1010
,
doi:10.1002/joc.929
.
Donner
,
L. J.
, and Coauthors
,
2011
:
The dynamical core, physical parameterizations, and basic simulation characteristics of the atmospheric component AM3 of the GFDL Global Coupled Model CM3
.
J. Climate
,
24
,
3484
3519
,
doi:10.1175/2011JCLI3955.1
.
Easterling
,
D. R.
,
J. L.
Evans
,
P.
Ya. Groisman
,
T. R.
Karl
,
K. E.
Kunkel
, and
P.
Ambenje
,
2000a
:
Observed variability and trends in extreme climate events: A brief review
.
Bull. Amer. Meteor. Soc.
,
81
,
417
425
,
doi:10.1175/1520-0477(2000)081<0417:OVATIE>2.3.CO;2
.
Easterling
,
D. R.
,
G. A.
Meehl
,
C.
Parmesan
,
S. A.
Changnon
,
T. R.
Karl
, and
L. O.
Mearns
,
2000b
:
Climate extremes: Observations, modeling, and impacts
.
Science
,
289
,
2068
2074
,
doi:10.1126/science.289.5487.2068
.
Environmental Modeling Center
,
2010
: NCEP Climate Forecast System Reanalysis (CFSR) Selected Hourly Time-Series Products, January 1979 to December 2010. Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory. [Available online at http://rda.ucar.edu/datasets/ds093.1.]
Field
,
C. B
.,
V.
Barros
, and
T. F.
Stocker
, Eds.,
2012
: Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. Cambridge University Press, 582 pp.
Fink
,
A. H.
,
T.
Brücher
,
A.
Krüger
,
G. C.
Leckebusch
,
J. G.
Pinto
, and
U.
Ulbrich
,
2004
:
The 2003 European summer heatwaves and drought—Synoptic diagnosis and impacts
.
Weather
,
59
,
209
216
,
doi:10.1256/wea.73.04
.
Fischer
,
E.
, and
C.
Schär
,
2010
:
Consistent geographical patterns of changes in high-impact European heatwaves
.
Nat. Geosci.
,
3
,
398
403
,
doi:10.1038/ngeo866
.
Gaffen
,
D. J.
, and
R. J.
Ross
,
1998
:
Increased summertime heat stress in the US
.
Nature
,
396
,
529
530
,
doi:10.1038/25030
.
Golaz
,
J.-C.
,
L. W.
Horowitz
, and
H.
Levy
II
,
2013
:
Cloud tuning in a coupled climate model: Impact on 20th century warming
.
Geophys. Res. Lett.
, 40, 2246–2251, doi:10.1002/grl.50232.
Griffies
,
S. M.
,
2009
: Elements of MOM4p1. GFDL Ocean Group Tech. Rep. 6, 377 pp. [Available online at http://data1.gfdl.noaa.gov/~arl/pubrel/o/old/doc/mom4p1_guide.pdf.]
Hess
,
P.
, and
H.
Brezowsky
,
1952
: Katalog der Grosswetterlagen Europas. Berichte des Deutschen Wetterdienstes in der US-Zone 33, 39 pp.
Horel
,
J. D.
,
1981
:
A rotated principal component analysis of the interannual variability of the Northern Hemisphere 500 mb height field
.
Mon. Wea. Rev.
,
109
,
2080
2092
,
doi:10.1175/1520-0493(1981)109<2080:ARPCAO>2.0.CO;2
.
Hoskins
,
B. J.
,
I. N.
James
, and
G. H.
White
,
1983
:
The shape, propagation and mean-flow interaction of large-scale weather systems
.
J. Atmos. Sci.
,
40
,
1595
1612
,
doi:10.1175/1520-0469(1983)040<1595:TSPAMF>2.0.CO;2
.
Illari
,
L.
, and
J. C.
Marshall
,
1983
:
On the interpretation of eddy fluxes during a blocking episode
.
J. Atmos. Sci.
,
40
,
2232
2242
,
doi:10.1175/1520-0469(1983)040<2232:OTIOEF>2.0.CO;2
.
Jacob
,
D.
, and Coauthors
,
2014
:
EURO-CORDEX: New high-resolution climate change projections for European impact research
.
Reg. Environ. Change
,
14
,
563
578
,
doi:10.1007/s10113-013-0499-2
.
Kjellström
,
E.
, and Coauthors
,
2007
:
Modeling daily temperature extremes: Recent climate and future changes over Europe
.
Climatic Change
,
81
,
249
265
,
doi:10.1007/s10584-006-9220-5
.
Klein
,
W. H.
,
1952
:
The weather and circulation of June 1952: A month with a record heat wave
.
Mon. Wea. Rev.
,
80
,
99
104
,
doi:10.1175/1520-0493(1952)080<0099:TWACOJ>2.0.CO;2
.
Kunkel
,
K. E.
,
S. A.
Changnon
,
B. C.
Reinke
, and
R. W.
Arritt
,
1996
:
The July 1995 heat wave in the Midwest: A climatic perspective and critical weather factors
.
Bull. Amer. Meteor. Soc.
,
77
,
1507
1518
,
doi:10.1175/1520-0477(1996)077<1507:TJHWIT>2.0.CO;2
.
Kunkel
,
K. E.
,
R. A.
Pielke
Jr.
, and
S. A.
Changnon
,
1999
:
Temporal fluctuations in weather and climate extremes that cause economic and human health impacts: A review
.
Bull. Amer. Meteor. Soc.
,
80
,
1077
1098
,
doi:10.1175/1520-0477(1999)080<1077:TFIWAC>2.0.CO;2
.
Kyselý
,
J.
,
2008
:
Influence of the persistence of circulation patterns on warm and cold temperature anomalies in Europe: Analysis over the 20th century
.
Global Planet. Change
,
62
,
147
163
,
doi:10.1016/j.gloplacha.2008.01.003
.
Kyselý
,
J.
,
2010
:
Recent severe heat waves in central Europe: How to view them in a long-term prospect?
Int. J. Climatol.
,
30
,
89
109
, doi:10.1002/joc.1874.
Lau
,
N.-C.
, and
E. O.
Holopainen
,
1984
:
Transient eddy forcing of the time-mean flow as identified by geopotential tendencies
.
J. Atmos. Sci.
,
41
,
313
328
,
doi:10.1175/1520-0469(1984)041<0313:TEFOTT>2.0.CO;2
.
Lau
,
N.-C.
, and
M. J.
Nath
,
2012
:
A model study of heat waves over North America: Meteorological aspects and projections for the twenty-first century
.
J. Climate
,
25
,
4761
4784
,
doi:10.1175/JCLI-D-11-00575.1
.
Lau
,
W. K. M.
, and
K.-M.
Kim
,
2012
:
The 2010 Pakistan flood and Russian heat wave: Teleconnection of hydrometeorological extremes
.
J. Hydrometeor.
,
13
,
392
403
,
doi:10.1175/JHM-D-11-016.1
.
Levy
,
H.
, II
,
L. W.
Horowitz
,
M. D.
Schwarzkopf
,
Y.
Ming
,
J.-C.
Golaz
,
V.
Naik
, and
V.
Ramaswamy
,
2013
:
The roles of aerosol direct and indirect effects in past and future climate change
.
J. Geophys. Res. Atmos.
, 118, 4521–4532, doi:10.1002/jgrd.50192.
Meehl
,
G. A.
, and
C.
Tebaldi
,
2004
:
More intense, more frequent, and longer lasting heat waves in the 21st century
.
Science
,
305
,
994
997
,
doi:10.1126/science.1098704
.
Meehl
,
G. A.
, and Coauthors
,
2007
: Global climate projections. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 747–845.
Nakamura
,
H.
, and
J. M.
Wallace
,
1990
:
Observed changes in baroclinic wave activity during the life cycles of low-frequency circulation anomalies
.
J. Atmos. Sci.
,
47
,
1100
1116
,
doi:10.1175/1520-0469(1990)047<1100:OCIBWA>2.0.CO;2
.
Namias
,
J.
,
1982
:
Anatomy of Great Plains protracted heat waves (especially the 1980 U.S. summer drought)
.
Mon. Wea. Rev.
,
110
,
824
838
,
doi:10.1175/1520-0493(1982)110<0824:AOGPPH>2.0.CO;2
.
Saha
,
S.
, and Coauthors
,
2010a
:
The NCEP Climate Forecast System Reanalysis
.
Bull. Amer. Meteor. Soc.
,
91
,
1015
1057
,
doi:10.1175/2010BAMS3001.1
.
Saha
,
S.
, and Coauthors
,
2010b
: NCEP Climate Forecast System Reanalysis (CFSR) 6-hourly products, January 1979 to December 2010. Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory, Boulder, CO. [Available online at http://rda.ucar.edu/datasets/ds093.0.]
Schär
,
C.
,
P. L.
Vidale
,
D.
Lüthi
,
C.
Frei
,
C.
Häberli
,
M. A.
Liniger
, and
C.
Appenzeller
,
2004
:
The role of increasing temperature variability in European summer heatwaves
.
Nature
,
427
,
332
336
,
doi:10.1038/nature02300
.
Schubert
,
S.
,
H.
Wang
, and
M.
Suarez
,
2011
:
Warm season subseasonal variability and climate extremes in the Northern Hemisphere: The role of stationary Rossby waves
.
J. Climate
,
24
,
4773
4792
,
doi:10.1175/JCLI-D-10-05035.1
.
Stefanon
,
M.
,
F.
D’Andrea
, and
P.
Drobinski
,
2012
:
Heatwave classification over Europe and the Mediterranean region
.
Environ. Res. Lett.
, 7, 014023, doi:10.1088/1748-9326/7/1/014023.
van Vuuren
,
D. P.
, and Coauthors
,
2011
:
The representative concentration pathways: An overview
.
Climatic Change
,
109
,
5
31
,
doi:10.1007/s10584-011-0148-z
.
Vautard
,
R.
, and Coauthors
,
2013
:
The simulation of European heat waves from an ensemble of regional climate models within the EURO-CORDEX project
.
Climate Dyn.
, 41, 2555–2575, doi:10.1007/s00382-013-1714-z.
Weare
,
B. C.
, and
J. S.
Nasstrom
,
1982
:
Examples of extended empirical orthogonal function analyses
.
Mon. Wea. Rev.
,
110
,
481
485
,
doi:10.1175/1520-0493(1982)110<0481:EOEEOF>2.0.CO;2
.
Zhang
,
X.
,
J. E.
Walsh
,
J.
Zhang
,
U. S.
Bhatt
, and
M.
Ikeda
,
2004
:
Climatology and interannual variability of Arctic cyclone activity: 1948–2002
.
J. Climate
,
17
,
2300
2317
,
doi:10.1175/1520-0442(2004)017<2300:CAIVOA>2.0.CO;2
.
Zhao
,
M.
,
I. M.
Held
,
S.-J.
Lin
, and
G. A.
Vecchi
,
2009
:
Simulations of global hurricane climatology, interannual variability, and response to global warming using a 50-km resolution GCM
.
J. Climate
,
22
,
6653
6678
,
doi:10.1175/2009JCLI3049.1
.