Large-scale synoptic circulations have a profound effect on western U.S. summer weather and climate. Heat waves, water availability, the distribution of monsoonal moisture, fire-weather conditions, and other phenomena are impacted by the position and amplitude of large-scale synoptic circulations. Furthermore, regional weather is modulated by the interactions of the large-scale flow with terrain and land–water contrasts. It is therefore crucial to understand projected changes in large-scale circulations and their variability under anthropogenic global warming.
Although recent research has examined changes in the jet stream, storm tracks, and synoptic disturbances over the Northern Hemisphere under global warming, most papers have focused on the cold season. In contrast, this work analyzes the projected trends in the spatial distribution and amplitude of large-scale synoptic disturbances over the western United States and eastern Pacific during July and August. It is shown that CMIP5 models project weaker mean midtropospheric gradients in geopotential height as well as attenuated temporal variability in geopotential height, temperature, vorticity, vertical motion, and sea level pressure over this region. Most models suggest reduced frequency of troughs and increased frequency of ridges over the western United States. These changes in the variability of synoptic disturbances have substantial implications for future regional weather and climate.
The summer weather and climate of the western United States is profoundly influenced by the variability of large-scale circulations over the region. Figure 1 shows mean 500-hPa geopotential heights for July and August 1970–99, using data from the NCEP reanalysis (Kalnay et al. 1996). In the mean, there is a trough immediately offshore of western North America, with a ridge just to the east of the crest of the Rocky Mountains. This ridge–trough configuration is modulated during the warm season on a variety of time scales, resulting in substantial variability in lower-tropospheric conditions.
The modulation of synoptic ridging in the lower to middle troposphere during summer has a substantial impact on lower-tropospheric temperatures and precipitation over the western United States (Gilliland 1980; McKendry 1994; Mass and Bond 1996; Chien et al. 1997). Ridging is associated with above-normal temperatures due to subsidence warming and warm advection on the western side of a ridge (Screen and Simmonds 2014; Meehl and Tebaldi 2004). Clear or near-clear skies accompanying ridging result in enhanced solar heating at the surface. The position and amplitude of warm-season ridging over the United States impact the advection of monsoonal moisture into the southwest United States, with large impacts on precipitation (Diem et al. 2013; Castro et al. 2001; Cavazos et al. 2002). Finally, the modulation of ridging substantially alters western U.S. onshore–offshore flow and interactions with terrain, with major impacts on surface temperature and precipitation (Mass et al. 1986; Brewer et al. 2013; Bumbaco et al. 2013; Brewer et al. 2012).
Variations in land surface characteristics and terrain can modulate the impacts of synoptic-scale forcing over the western United States. For example, the interaction between the synoptic-scale flow and major western U.S. topographic features such as the Rockies, Cascades, and Sierra Nevada mountains can result in downslope flow and compressional warming, producing mesoscale warming and drying over and downwind of lee slopes (Mass et al. 1986; Brewer et al. 2013; Bumbaco et al. 2013; Brewer et al. 2012). In addition, the existence of cold Pacific waters adjacent to warmer land during the warm season results in large temperature variations as the synoptic flow alternates between onshore and offshore directions. Similarly, large low-level moisture gradients exist between the Pacific coastal zone and the arid interior, with changes from onshore to offshore flow greatly reducing low-level moisture availability over the coastal zone. Thus terrain, land–water contrasts, and large regional moisture contrasts can greatly amplify the meteorological signal produced by variability in summer synoptic circulations over the western United States. It is therefore critical to understand how synoptic variability will change over western North America under anthropogenic global warming.
Recently, several papers have described future changes in midlatitude circulation, synoptic variability, and storm tracks under anthropogenic global warming. Some have argued that Arctic amplification, associated with enhanced warming over the Arctic compared to midlatitudes (Holland and Bitz 2003), leads to a more amplified and wavy jet (Francis and Vavrus 2012; Liu et al. 2012). Tang et al. (2014) suggested that a more amplified circulation pattern forced by Arctic amplification would result in more extreme summer weather such as heat waves and drought. Others have questioned the theories of more amplified circulations under global warming (Screen and Simmonds 2013; Barnes 2013; Hassanzadeh et al. 2014). Screen and Simmonds (2013) found that seasonal trends (1979–2011) in 500-hPa variability are not statistically significant. Others have suggested that it is difficult to determine whether Arctic amplification has impacted midlatitude circulations because of large natural variability and short observational records of midlatitude circulation (Screen and Simmonds 2013; Walsh 2014; Barnes 2013).
Most studies of synoptic variability under global warming are for the winter season, with only a few examining warm-season trends. Chang et al. (2012), analyzing trends in the variance of bandpass filtered sea level pressure and meridional wind in CMIP5 model simulations, found a weaker storm track over the Northern Hemisphere during winter, with modest summer (JJA) attenuation as well. Barnes and Polvani (2015) recently analyzed CMIP5 model projections and found robust declines in the meridional extent of large-scale Rossby waves over the North Atlantic and North America during spring and summer at the end of the twenty-first century. Coumou et al. (2015) used vertically averaged band-passed (2.5–6 days) eddy kinetic energy to show that synoptic activity declined in summer over the Northern Hemisphere during the 1979–2013 period, with substantial regional variations. Lehmann et al. (2014) noted a summer weakening of eddy kinetic energy in CMIP5 model projections at the end of the twenty-first century over the Northern Hemisphere.
Screen (2014) found that CMIP5 models under the representative concentration pathway (RCP) 8.5 scenario project declines in surface air temperature variance over land in the middle and high latitudes for every season except summer, during which variance increases. Regarding increased summer variability, that paper, as well as others (Schär et al. 2004; Fischer et al. 2012), suggests that soil moisture–temperature feedbacks may be overcoming the effects of weaker meridional gradients. After bandpass filtering (3–15 days) potential temperature, Schneider et al. (2015) showed that CMIP5 models project declines in 850-hPa temperature variability over the midlatitude Northern Hemisphere oceans for summer (JJA), with lesser declines inland.
Considering the importance of summer synoptic circulations over the western United States, particularly in their modulation of heat waves and drought, it is important to determine the temporal trends in regional synoptic variability under global warming during the warm season. To that end, this work focuses on future changes in summer (July and August) synoptic circulation and variability over western North America and the eastern Pacific as projected by CMIP5 climate model simulations. Temporal changes in the variability of vorticity, vertical motion, and other variables are analyzed, as well as trends in the amplitude and position of ridges and troughs over western North America and the adjacent Pacific. This work is the first of a two-paper sequence, with the second analyzing the implications of projected changes in synoptic variability for the conditions that drive heat extremes over the northwest United States.
2. Data and methods
a. Model data and selection
This research used daily averaged grids from phase 5 of the Coupled Model Intercomparison Project (CMIP5; Taylor et al. 2012), examining simulations based on the high-emission scenario RCP8.5. This high-emission scenario is used for its relatively higher signal-to-noise ratio compared to RCP4.5, and thus provides a clearer view for how large-scale circulations may change. Furthermore, during the past decades emissions have followed RCP8.5 more closely that more moderate scenarios (Peters et al. 2013). Historical (1970–99) and future (2071–2100) July–August periods were analyzed using only one ensemble member from each model. Model grids were regridded to a common grid of 2° longitude by 1.5° latitude.
To reduce excessive impacts of modeling groups with large numbers of simulations, the lower-resolution models in such groups were removed (e.g., the lower-resolution MPI-ESM-LR was removed while the MPI-ESM-MR was retained). In addition, there are a few modeling groups that made available more than two models with the same resolution. To that end, daily averaged 500-hPa geopotential heights and temperature, 2-m temperature, and sea level pressure from all available CMIP5 models were compared to the NCEP reanalysis (Kalnay et al. 1996) over the domain of 20°–70°N, 170°–90°W for the historical period. The two models with the smallest mean absolute differences with the reanalysis were retained from each modeling group. For example, the GFDL-ESM2G was found to have higher mean absolute differences of the fields mentioned than the GFDL-ESM2M and GFDL-CM3, so the GFDL-ESM2G model was not used while the other two remained. After these selections were made, 17 GCMs remained and are listed in Table 1.
b. Analysis of synoptic variability
Two approaches for analyzing changes in synoptic-scale variability over western North America are used in this study, one based on spatial variability and the other on temporal variability. In the first, an algorithm was developed to identify ridge and trough axes in daily 500-hPa geopotential height fields from the CMIP5 models. This algorithm, illustrated in Fig. 2, identifies ridge axes when the geopotential height at a grid point g(i,j) is higher than the geopotential heights at the grid points on either side along the same latitude line, out to g(i,j−x) and g(i,j+x). The x term in g(i,j−x) and g(i,j+x) is the number of grid points needed to get nearest to 750 km away from g(i,j), thus making the red line in Fig. 2 (dist) approximately 1500 km in length.1 Other distances ranging from 500 to 1000 km were tested and the results were not sensitive to the distance used. This method was applied in an analogous manner for identifying trough axes. The locations of trough and ridge axes were collected for historical and future periods; with this information, changes in the frequency and amplitude of ridge and trough occurrence were analyzed for 170°–90°W (eastern Pacific and western North America). The amplitudes of ridges and troughs were quantified by
where z1 is the geopotential height difference between g(i,j) and g(i,j−x) and z2 is the geopotential height difference between g(i,j) and g(i,j+x), and dist is the distance (roughly 1500 km) shown by the red line in Fig. 2.
While the above method looks at ridge and trough frequency and amplitude based on structure, a second approach considers synoptic temporal variability. Specifically, monthly means were removed from daily fields of 500-hPa u- and υ-wind components, 500-hPa omega, 500-hPa geopotential height, 500- and 850-hPa temperature, and sea level pressure for each individual model. A high-pass filter (ninth-order Butterworth filter) was then applied to isolate periods less than 6 days. A similar high-pass filter with a cutoff between 6 and 10 days has been used in a variety of papers quantifying synoptic activity (Coumou et al. 2015; Lehmann et al. 2014; Blackmon 1976). Vorticity was calculated from the high-pass u- and υ-wind components. To analyze changes in variability, the fields were squared, and the mean differences were computed between the historical (July–August 1970–99) and future (July–August 2071–2100) periods for each model and the ensemble mean. The variability in unfiltered 500-hPa geopotential height and 850-hPa temperature was also computed.
a. Changes in the mean state
Before examining the projected changes in synoptic variability over the next century, it is useful to understand how the mean state will change as the atmosphere evolves. The ensemble mean historical, future, and difference (future minus historical) fields for four variables are shown in Fig. 3. Although the 500-hPa geopotential height patterns for the historical and future periods look superficially similar, there are important differences. The most obvious is that heights rise everywhere, consistent with general warming. However, the height rises are not spatially uniform, with the greatest increases in the midlatitudes between 45°–55°N over North America and the Gulf of Alaska. A maximum in height increases is found over the northwest United States and southwestern Canada, with nearly equal rises over the North Pacific.
Although heights rise everywhere at 850 hPa, there are significant regional variations. For example, a relative minimum in 850-hPa height increases occurs in the center of the east Pacific anticyclone, while a local maximum is found over the Gulf of Alaska, suggesting a northward shift of the Pacific high. Over land, changes in the meridional height gradients are weak, unlike at 500 hPa.
Turning to temperature, the greatest warming at 700 hPa and 2 m is over the northwest U.S. interior and southwest Canada, proximate to the largest height rises at 500 hPa. Enhanced surface warming over land that is adjacent to a more slowly warming ocean results in increases in land–sea temperature gradients along the Oregon, Washington, and British Columbia coasts.
The majority of the models evince temporal changes similar to the ensemble mean in Fig. 3, but there are outliers with different evolutions. To illustrate, Fig. 4 shows changes in 700-hPa temperature between the historical and future periods for 16 CMIP5 models. The MRI-ESM1 was not included since its pattern is almost identical to MRI-CGCM3. The majority of the models have warming maxima over the northwest United States and southwest Canada, with a few having relative maxima elsewhere. For example, MPI-ESM-MR has a maximum over the Great Lakes. Models such as IPSL-CM5A-MR and MRI-CGCM3 have maxima over the western United States but with lesser magnitudes compared to the others.
To examine the variability among the models regarding changes in meridional height gradients, Fig. 5 presents the latitudinal variation of the absolute value of the meridional gradients in 500-hPa geopotential height for 170°–90°W for the historical and future periods, as well as the difference between them. For both periods, the individual model (blue) gradients generally agree with the ensemble mean (black), particularly south of 45°N. The ensemble mean has a northward shift of the maximum mean gradient by roughly 2° (49° to 51°N), weakening south of approximately 57°N and strengthening to the north. The maximum in meridional geopotential height gradients at 500 hPa is projected to weaken by roughly 10%–15%.
The changes in the meridional geopotential height gradients are reflected in the midlatitude summer circulation. Figure 6 shows 500-hPa u-wind for the historical period, the difference between the future and historical periods, and the agreement of the models regarding the sign of the changes. The u-wind (westerlies) weakens over the midlatitudes (by approximately 2 m s−1), with the greatest decline over the southwest United States. In contrast, winds increase over Alaska and Canada, with the greatest enhancement over northeastern Canada. The large weakening of the u-wind over the southwest United States is likely associated with the maximum in lower-tropospheric warming over the Pacific Northwest and southwest Canada (Fig. 3). Weaker westerly winds reduce the marine influence over the western coastal zone. Model agreement is high over the areas of large changes, particularly in the regions of midlatitude weakening.
b. Changes in the longitudinal distribution, and amplitude of troughs and ridges
To illustrate changes in the longitudinal distribution of ridge and trough occurrences over the western United States in the different CMIP5 models, Fig. 7 shows the number of historical (July–August 1970–99) and future (July–August 2071–2100) occurrences of ridge axes within specified longitude bins along 45°N at 500 hPa. Ridges were identified using the method described in section 2 and Fig. 2. For each box, the historical number of ridge occurrences is on the left and the number of future occurrences is on the right.
Examining the frequency of ridges and using the ensemble mean statistics at the bottom of Fig. 7, the bin with the maximum historical frequency of ridges (bold value) is 111°–105°W (447 occurrences), with 11 of 17 models in agreement. The mode of ridge frequency remains in this bin under global warming, with a modest increase in ridge frequency west of this bin and a decline to its east. Although the majority of models agree with the ensemble mean, models such as the CMCC-CM, MRI-CGCM3, MRI-ESM1, and MPI-ESM-MR have declines in ridge frequency over nearly the entire domain. Other models such as IPSL-CM5A-MR and MIROC5 suggest more ridging where ridge occurrence is greatest, but no shift in position.
Figure 7 also shows that there is unanimous model agreement for a decline in future ridge frequency over the eastern Pacific (135°–129°), with the ensemble mean frequency declining by 27%. Most models also suggest declines in ridging in the 129°–123° bin, which straddles the coastline, with the ensemble mean declining by 20%. Such declines could have important implications for the weather and climate of western Oregon and Washington, given the profound effects of offshore high pressure in controlling the frequency and strength of onshore and offshore flow in this region (Brewer et al. 2012). Also, an increase in ridging east of the coast, as shown in Fig. 7, would imply an enhanced frequency of western U.S. near-surface warming events.
Figure 8 is similar to Fig. 7, but describes trough occurrence. The ensemble mean for the historical period has maximum trough frequency in the 129°–123°W bin, just offshore of the West Coast. Almost all models suggest a decline in troughing over the western United States, namely in bins 123°–117°W and 117°–111°W. The models with the greatest declines in western U.S. trough occurrence show increased ridging over the West Coast (Fig. 7) and largest warming over the northwest United States and southwest Canada (Fig. 3).
The projected change in the amplitude of ridges and troughs in CMIP5 models over the eastern Pacific and the western United States as a function of latitude is shown in Fig. 9. Averaged over 170°–90°W, these statistics are based on the algorithm [Eq. (1)] described above. The amplitudes of troughs and ridges weaken over most of the midlatitudes (approximately 35°–65°N). The ridge weakening reaches 11% near 55°N for the ensemble mean, and all models manifest roughly the same latitudinal modulation. The outlier is the GFDL-CM3 model, with weakening of over 30% at 50°N. Virtually all CMIP5 models project weakened midlatitude troughs, with the largest ensemble mean weakening (15%) near 41°N. This same exercise was done for many subdomains within the larger domain, and similar results were found.
As an alternative approach to analyzing changes in the amplitude of synoptic disturbances under global warming, we next evaluate the trends in the temporal variability of geopotential height, relative vorticity, and vertical motion variability at 500 hPa using daily averaged CMIP5 model output. Figure 10 shows unfiltered 500-hPa geopotential height variability for the historical period (top left), computed for each model by removing the monthly means from the daily data, squaring the differences, and then computing the means over the historical period. The difference between the historical (1970–99) and future (2071–2100) July–August variability is also shown, as is the number of models that agree with the sign in the difference map. The bottom row shows the same quantities but after high-pass filtering the 500-hPa geopotential height differences from climatology.
For the historical and future periods, the pattern of the unfiltered 500-hPa geopotential height variability indicates large variability over the Gulf of Alaska, where troughs are found in the mean (Fig. 3, top row). Low variability is found over the southern latitudes and immediately downstream of the Rocky Mountains, where ridging is persistent. The unfiltered variability shows declines nearly everywhere between the historical and future periods, with the largest reductions over the Gulf of Alaska and the northeast United States, within and on the southern periphery of areas of low heights in the mean. For virtually all of the major features, there is substantial agreement among the CMIP5 models.
Ensemble mean high-pass filtered 500-hPa geopotential height variability is shown on the bottom of Fig. 10. The change in high-pass filtered variability has a similar distribution as the unfiltered data, with the exception of declines in the high-passed variability over southern Canada and the northern Rockies/Great Plains. Model agreement is high across most of the midlatitudes, where synoptic variability is declining. Clearly, variability associated with mobile synoptic systems is projected to decline over the midlatitudes during summer by the end of the century.
Figure 11 shows changes in the variability of 500-hPa omega and relative vorticity between the historical and future periods after high-pass filtering the daily data of each individual model, squaring these fields to provide changes in the amplitude, and then computing the individual model means and ensemble mean for each period. Both omega and vorticity variability weaken over the middle to high latitudes, with the models nearly unanimous in the projected trends. There are no regions where variability is strengthening. The weakening of omega variability is larger over land than over water, perhaps reflecting the importance of convection in enhancing vertical motion over land during summer, with convection amplifying the synoptic signal. It should also be noted that unfiltered omega and vorticity possess declines in the same locations as the filtered maps.
High-pass filtered sea level pressure variability is shown in Fig. 12, computed in the same way as the other variables. Sea level pressure variability is often used as a proxy for storm track activity (Chang et al. 2012). During the historical period, sea level pressure variability is low over southern latitudes and along the West Coast, and is largest near the Aleutians and over southeastern Canada. The future projection is for pressure variability to weaken from approximately 35° to 60°N, with the largest declines over the northern Rockies and upper Plains, with nearly all models in agreement. Similar results are shown in Harvey et al. (2014) and Chang et al. (2012). Reduced sea level pressure variability over the interior of the western United States would lessen the variability of onshore/offshore flow and downslope warming over the western United States.
Given the weakening variability in large-scale synoptic disturbances, as well as weakening midlatitude geopotential height gradients, one might expect mid- and lower-tropospheric temperature variability to undergo similar changes. Figure 13 shows both unfiltered and high-pass filtered 500- and 850-hPa temperature variability. The pattern of historical 500-hPa unfiltered temperature variance (top row) is very similar to the variance in unfiltered 500-hPa geopotential height shown in Fig. 10. This field shows robust weakening across the midlatitudes, with most models in agreement. Filtered 500-hPa temperature variance is similar, except model agreement regarding weakening is even more robust.
Temperature variability in the lower troposphere reveals a different picture than at 500 hPa, since terrain and land–water contrasts modulate the variability at the lower levels. Unfiltered 850-hPa temperature variability in Fig. 13 (third row) shows weakening over the northeast Pacific Ocean, similar to 500 hPa, but with variability increasing over most of North America. Screen (2014) also found increases in unfiltered surface temperature variance over land during summer over the Northern Hemisphere and suggested that soil moisture–temperature feedbacks are likely the cause. However, model agreement at 850-hPa for the weakening is relatively low over most of North America, with the exception of the southeast United States.
Ensemble mean high-pass filtered temperature variability at 850 hPa (Fig. 13, bottom row) shows that the variance is largest over the continent, particularly eastern Canada. Filtered 850-hPa temperature variability declines over most of the midlatitudes in the future, with general model agreement. These results are consistent with Schneider et al. (2015), who attributed reduced low-level temperature variability to attenuated meridional temperature gradients, which interacts with reduced synoptic variability to lessen the high-pass temperature variability.
4. Discussion and conclusions
The strength and variability of synoptic disturbances play a crucial role in the summer weather and climate of North America and the eastern Pacific, with the distribution, persistence, and amplitude of ridges and troughs modulating temperature and precipitation over the region. Given the large impacts, it is important to understand how the large-scale summer circulations and synoptic variability over western North America and the eastern Pacific may change under anthropogenic global warming.
This study used daily data from CMIP5 global climate models for historical (1970–99) and future (2071–2100) summer (July–August) periods. Given the clustering of solutions within modeling groups, criteria were developed to reduce the number of models from the same modeling group. After these criteria were applied, 17 models from the CMIP5 data remained (Table 1).
Using this multimodel ensemble, projected changes in geopotential height and temperature over the region were analyzed. Although heights rise everywhere, consistent with general warming, the largest height increases are between 45° and 55°N over North America and the Gulf of Alaska. A local maximum in 500-hPa height change and lower tropospheric temperature is projected over the northwest United States and southwestern Canada. Coincident with the greatest height rises, the largest increases in low-level temperature were found over the northwest United States and southwestern Canada.
The 500-hPa meridional geopotential height gradient over western North America and the eastern Pacific is projected to weaken by roughly 10%–15% under global warming by the end of this century, with the maximum gradient predicted to shift northward by approximately 2°. As expected, changes in 500-hPa u-wind follow roughly the same pattern of weakening and northward shift.
Using an algorithm that identified the longitudes of ridge and trough axes along 45°N over the eastern Pacific and western North America, it was found that most models project more trough and ridge occurrences west of the climatological trough and ridge axes and fewer occurrences to the east during the next century. Most models project a decline in trough frequency and an increase in ridge frequency over the western United States, with the ensemble mean showing declines of roughly 30%. The models that suggest increases in ridge occurrence over the northwest United States and southwest Canada also project a maximum in lower-tropospheric warming over that region.
The zonal amplitude of summer troughs and ridges over western North America and the eastern Pacific, based on the associated zonal gradients of geopotential height, are projected to weaken between the July–August historical (1970–2000) and future (2071–2100) periods. Specifically, the weakening was roughly 11% for ridges and 15% for troughs, with all models in agreement on the sign but varying in the magnitude.
Long-term changes in the temporal variability of several tropospheric variables and parameters were described. The spatial structure of 500-hPa geopotential height variability is the inverse of the mean height pattern, with the greatest (least) variability associated with the area of climatological troughing (ridging). By the end of the century, geopotential height variability at 500 hPa shows a substantial decline over the eastern Pacific, no change over the central United States, and significant declines over the eastern United States.
The variability in high-pass filtered geopotential height, vertical motion, and vorticity at 500 hPa declines by approximately 15%–35% in the midlatitudes, with models agreeing in most locations. Similar findings were found for high-pass filtered sea level pressure variability. These results are similar to those found in previous papers (Lehmann et al. 2014; Hassanzadeh et al. 2014; Barnes and Polvani 2015). Unfiltered and high-pass filtered 500-hPa temperature variability weakens over the midlatitudes, with excellent model agreement. Unfiltered and high-pass filtered 850-hPa temperature variability weakens over the northeastern Pacific, similar to 500 hPa. However, unfiltered 850-hPa temperature variability increases across most of North America in the ensemble mean, similar to results found by Screen (2014), with the suggestion that soil moisture feedbacks may be important. However, most areas over North America have weak model agreement, suggesting a result that is not robust at 850 hPa.
In summary, during the summer, the CMIP5 GCMs suggest a weakening of synoptic-scale variability over North America and the eastern Pacific. This reduced synoptic variability could have a significant impact on summertime weather and climate over the western United States. For example, it has been shown in Brewer et al. (2012) that heat waves along the West Coast are usually preceded by high pressure building over southwestern Canada and the northwest U.S. interior. Offshore flow subsequently develops, resulting in adiabatic warming on the western slopes of the Cascade Mountains and the subsequent formation of the West Coast thermal trough. Reduced synoptic variability could lessen the amplitude and frequency of large offshore flow events, thus working against heat wave conditions. Synoptic disturbances also have large impacts on transitions between onshore and offshore flow, which have a profound impact on coastal zone temperatures (Mass et al. 1986). Therefore, weakening of synoptic disturbances could have substantial implications for coastal weather. These important effects will be discussed in the second of this two-paper sequence. Although some insights can be gained from CMIP5 models, given the complex terrain of the region, downscaled regional climate models driven by GCMs will be important tools for such future work.
We acknowledge the various CMIP5 modeling groups, the Program for Climate Model Diagnosis and Intercomparison (PCMDI), and the WCRP’s Working Group on Coupled Modelling (WGCM) for their roles in making available the WCRP CMIP multimodel datasets. Support of this dataset is provided by the Office of Science, U.S. Department of Energy. The NCEP reanalysis data is provided by NOAA/OAR/ESRL PSD, Boulder, Colorado, from their website at http://www.esrl.noaa.gov/psd/.
This work was funded by Grant AGS-1349847 from the National Science Foundation.
The distance of 750 km varies by approximately ±100 km, depending on the distance between grid points as a function of latitude.