Eurasian Winter Storm Activity at the End of the Century: A CMIP5 Multi-model Ensemble Projection
Abstract
Extratropical cyclone activity over Eurasia has exhibited a weakening trend in the recent decade. Extratropical cyclones bring precipitation and hence supply fresh water for winter crops in the mid- and high-latitude regions of Eurasia. Any changes in extratropical cyclone activity over Eurasia in the future may have a critical impact on winter agriculture and the economies of affected communities. However, potential future changes in regional storm activity over Eurasia have not been studied in detail. Therefore, in this study, we investigate anticipated changes in extratropical storm activity by the end of the century through a detailed examination of the historical and future emission scenarios from six different models from CMIP5. A statistical analysis of different parameters of storm activity using a storm identification and tracking algorithm reveals a decrease in the number of storms over mid-latitude regions. However, intense storms with longer duration are projected over high latitude Eurasia. A further examination of the physical mechanism for these changes reveals that a decrease in the meridional temperature gradient and a weakening of the vertical wind shear over the mid-latitudes are responsible for these changes in storm activity.
Plain Language Summary
Winter storms bring rainfall in winter and thus helps the community by supplying fresh water to the winter crops. However, the climatological storm activity over Eurasia exhibits a weakening trend in the recent years. It is therefore important to understand the future changes in extratropical cyclones over different regions of Eurasia as it can have a critical socioeconomic impact on the community. In this study, we investigated the changes in extratropical storm activity at the end of the century by comparing the historical simulations and future scenarios by six different climate models of CMIP5. The result reveals a decrease in the number of storms over the mid-latitude regions, while more numerous intense storms with longer duration occur over the high-latitude Eurasia. We further linked these changes in extratropical storms to the changes in vertical wind shear and meridional temperature gradient under the high-emission scenarios.
1 Introduction
Extratropical cyclones are an intrinsic part of the daily weather over mid- and high-latitudes, especially during the winter. Extratropical cyclones are often associated with extreme and severe weather events such as blizzard, snowfall, rain, freezing rain, etc., which disrupt daily life and can cause extensive damage in terms of loss of property and life. However, these cyclones are also the primary source of precipitation and thus supply fresh water for agriculture and winter crops in different regions. Extratropical cyclones bring more than half of the winter precipitation in the Northern Hemisphere (Hawcroft et al., 2012). Thus, changes in extratropical cyclone activity in the future can have a significant impact on society and the economy.
Extratropical cyclones are most frequent over certain regions of the mid- and high latitudes. Regions with relatively higher density of extratropical cyclones are known as storm tracks and play a dominant role in the transfer of heat, moisture, and momentum. In recent decades, apart from interannual variability in the number, intensity and duration of storms, extratropical storm tracks have undergone many changes in response to the warming climate. Previous studies show an overall poleward shift in storm tracks under the warming scenarios (McCabe et al., 2001; Yin, 2005; Ulbrich et al., 2009; Chang et al., 2012; Tilinina et al., 2013). However, Zhang et al. (2004) show that, along with an overall poleward shift, the climatological behavior of storm tracks exhibit distinctive variability in terms of numbers, intensity, and duration over different geographical regions of the mid-latitudes. Recent studies indicate a decrease in Eurasian extratropical storm activity over the last decade (Zhang et al., 2004, 2012). The frequency of intense cyclones is projected to increase on the poleward and downstream sides of the North Pacific storm track (Mizuta, 2012). Chang et al. (2012) show a decrease in overall North American storm activity in different seasons. Wuebbles and Hayhoe (2004) reveal a weakening of the western North Atlantic storm track but an intensification of extratropical storms inland over the east coast of North America in the future. Colle et al. (2013) and Michaelis et al. (2017) agree with Wuebbles and Hayhoe (2004) and also show increased storm activity off the U.S. east coast. Additional studies show strengthening and eastward (Woollings et al., 2012) and northeastward extension of the North Atlantic storm track (Della-Marta & Pinto, 2009; Catto et al., 2011; Zappa et al., 2013, 2013; Michaelis et al., 2017). Lionello and Giorgi (2007); Raible et al. (2010) and Zappa et al. (2014) project a decrease in precipitation in association with the decrease in extratropical cyclones over the Mediterranean region. Yettella and Kay (2017) show an increase of cyclone-associated precipitation in winter especially over the high-latitude regions of Eurasia along with a decrease over the Mediterranean. The vast landmass of Eurasia is characterized by the Indian Ocean to the south; Atlantic Ocean to the west; Pacific Ocean to the east and Arctic Ocean to the north; with tall mountains in the south-central region. Therefore, different parts of Eurasia possess their own distinct geographical features. The projected changes in extratropical cyclone activity in the future in response to the warming climate over these different subregions of Eurasia have not been studied.
The only approach to study the future changes in extratropical storm cyclone activities is by examining climate model simulations of present and future climates (Sinclair & Watterson, 1999; Leckebusch & Ulbrich, 2004; Bengtsson et al., 2006, 2009; Greeves et al., 2007; Pinto et al., 2007; Catto et al., 2011). The Coupled Model Intercomparison Project (CMIP5) has historical model runs for the period 1950–2005 and future scenario model runs (RCP4.5 and 8.5) for 2006–2100 with high temporal (6-hourly) resolutions. The high frequency of the model data is essential for implementing a Lagrangian approach of studying changes in extratropical cyclone activity.
Therefore, in this study we investigate the future changes in storm activity over different regions of Eurasia in winter using a sea level pressure (SLP)-based storm identification and tracking algorithm (Zhang et al., 2004) with 6-hourly output from the historical, RCP 4.5 and RCP 8.5 scenarios from different models of CMIP5 (Taylor et al., 2012).
2 Methodology
This study considers extratropical storm activity over the last 30 years of the historical period (1975–2005) in the historical model runs (HIST) and the last 30 years of the end of the century, that is, 2070–2100 in two future emission scenarios (Moss et al., 2010), RCP4.5 and RCP8.5.
Considering the distinct geographical features of climatological storm activity, we divided Eurasia into six subregions for this study (Figure 1)—Mediterranean (MS), Central Eurasia (CE), East Asia (EA), Sea of Okhotsk (SO), Arctic Eurasia (AE), and Scandinavia (SC).
In this study, 6-hourly output data from the following six climate models of the CMIP5 experiments are used: CCSM4, CNRM-CM5, GFDL-CM3, MIROC5, MPI-ESM-LR, and NorESM1-M. Based on the availability of 6-hourly output six models from different institutes and different countries are chosen from the list of top performing models (Walsh et al., 2008). Table 1 shows the details and spatial resolutions of the models. The number of ensemble members used for each model from each scenario is determined based on the availability. Ensemble member denoted r1i1p1 is used for CNRM-CM5, GFDL-CM3, NorESM1-M and the ensemble member r6i1p1 is used for CCSM4. Three ensemble members (r1i1p1, r2i1p1, and r3i1p1) are used for each of MIROC5 and MPI-ESM-LR.
| Model | Model full name | Institution | Horizontal resolution (degree longitude × degree latitude) |
|---|---|---|---|
| CCSM 4.0 |
Community Climate System Model (CCSM), version 4 |
National Center for Atmospheric Research (NCAR), United States |
1.3 × 0.9 |
| CNRM-CM5 |
Centre National de Recherches Meteorologiques (CNRM) Coupled Global Climate Model, version 5 |
CNRM–Centre Europeen de Recherche et de Formation Avancee en Calcul Scientifique (CERFACS), France |
1.4 × 1.4 |
| MIROC5 |
Model for Interdisciplinary Research on Climate (MIROC), version 5 |
Atmosphere and Ocean Research Institute, University of Tokyo, Japan | 1.4 × 1.4 |
| GFDL-CM3 |
Geophysical Fluid Dynamics Laboratory (GFDL) climate Model version 3 |
Geophysical Fluid Dynamics Laboratory, United States | 2.5 × 2.0 |
| NorESM1-M |
Norwegian Earth System Model |
Norwegian Climate Centre (NCC), Norway |
2.5 × 1.9 |
| MPI-ESM-LR | Max Planck Institute (MPI) Earth System Model, low resolution | Max Planck Institute, Germany | 1.9 × 1.9 |
Six-hourly outputs of SLP from the abovementioned models are used for implementing a storm identification and tracking algorithm (Zhang et al., 2004). The algorithm first identifies storm centers that have at least a minimum SLP gradient of 0.15 hPa per 100 km with the surrounding grid points and survives for more than 12 h. The algorithm provides various parameters of storm activity, such as duration or lifetime, central location, and central SLP for each individual storm occurring over the study area. By using the data set of these parameters from each ensemble member of our experiment, the number of storm trajectories, mean storm duration, and mean storm intensity for winter (December–February) are obtained for each Eurasian subregion. The storm trajectory is defined from the time of storm generation until dissipation within the region or from the time when the storm entered the region until the time it left the region. The mean duration for each region is the averaged duration of the total number of storm trajectories. The storm intensity for each individual storm is the difference between the monthly mean climatological SLP and the storm's central SLP at the corresponding location. Storm intensities are averaged throughout their duration and over the number of storm trajectories within each area to get the regional mean storm intensity.
3 Results
3.1 Changes in Storm Activity
A probability density function (PDF) analysis is performed on number of storm trajectories, mean intensity and mean duration in the Historical, RCP4.5 and RCP8.5 experiments (Figure 2) for the subregions of Eurasia mentioned in Section 2. The PDF distributions show the frequency of each value in the dataset. Therefore, when PDF distributions of HIST and future scenarios are compared the differences in the PDF distribution curves have two aspects—(1) the peak of the distribution shifts right or left which indicates an increase or decrease in peak frequency or the mean and (2) the peak of the distribution moves up or down which denotes changes in the percentage of occurrences of the peak frequency. The significance of the changes in the mean in the PDF distribution is tested with a Student's t-test.
A comparison of the PDFs of the number of storm trajectories, mean intensity, and mean duration for each of the subregions reveals many changes in the climatological features of storm activity by the end of the century compared to recent decades (Figure 2). PDFs of the number of storm trajectories in RCP4.5 and RCP8.5 exhibit a decrease over all subregions of Eurasia except SO. In the future scenarios the peak frequency shifts toward fewer numbers of storm trajectories than HIST over MS, EA, AE, and SC. The mean number of trajectories decreases from 65 ± 1 to 58 ± 0.5 (61 ± 1) in RCP8.5 (RCP4.5) over MS. Over EA, there are fewer storms as the frequency of years with fewer storm trajectories increases from 45% in HIST to 55% (60%) in RCP8.5 (RCP4.5) and the mean number of trajectories decreases from 40 ± 1 to 36 ± 0.6 (38 ± 0.5) in RCP8.5 (RCP4.5). However, the peak frequency shifts toward a higher number of trajectories over SO in RCP4.5 and RCP8.5. The mean increases from 15 ± 1 in HIST to 16 ± 1 (16 ± 0.5) in RCP8.5 (RCP4.5). The number of storm trajectories decreases over SC as the mean shifts from 76 ± 1 to 73 ± 1.5 (75 ± 1) in the future scenarios.
Comparison of the HIST and future scenarios reveals noticeable changes in the mean intensity of the storms by the end of the century. Climatologically, the storms are most intense than the HIST period over the higher latitudes, that is, in SO, AE, and SC. The storms are weaker over MS as the frequency of storms with intensity lower than the historical mean intensity increases from 24% to 42%. However, storms intensify over EA, SO, and AE. Storms are more intense over SO as the PDF distribution indicates a higher frequency of more intense storms in the future scenarios where the percentage of occurrences of peak frequency increases from 22% in HIST to 38% (34%) in RCP8.5 (RCP4.5) over SO. Over AE, the peak frequency shifts toward higher intensity as the mean intensity increases from 16.5 ± 1 hPa to 19.1 ± 1.3 (17.1 ± 1.2) hPa in RCP8.5 (RCP4.5).
The PDFs of mean storm duration exhibit a general increase in mean storm duration over the high latitudes. The mean duration increases noticeably over AE, as the PDF distribution indicates an increase in the frequency of storms with longer durations in the future scenarios. The percentage of occurrence of peak frequency increases from 20% in HIST to 40% in RCP8.5. The mean duration over AE increases from 40.2 ± 0.7 h to 41.8 ± 0.3 h by the end of the century in RCP8.5. The mean duration of storms over SO and SC increases in the future scenarios. However, the mean duration over MS decreases from 36.3 ± 0.5 h in HIST to 34 ± 0.7 (34.5 ± 0.7) h in RCP8.5 (RCP4.5). Over EA, the lifetime of the storms gets shorter since the frequency of storms with shorter durations than the mean duration in HIST increases. The percentage of occurrence of peak frequency decreases from 34% to 24% (26%) along with a slight decrease in the peak frequency at the end of the century in RCP8.5 (RCP4.5).
The PDF analysis shows a systematic change in different storm parameters over different geographical regions of Eurasia at the end of the century in the future emission scenarios. There are fewer and less intense storms with a shorter lifespan over MS and CE. Over EA, storms are fewer in number but undergo intensification with shortened duration. Over AE, the storms are fewer in number but more intense with a longer duration. However, storms are more numerous and intense with longer durations over SO at the end of the century in the high future emission scenarios. Over SC, the storms are fewer in number with a slight increase in duration. For a better understanding of the changes in storms in the future emission scenarios by the end of the century, we will further examine the synoptic-scale storm activities in the context of the large-scale atmospheric general circulation.
3.2 Changes in Large-scale Circulation Processes
In the previous section, a Lagrangian approach identifies and tracks individual storms across different subregions of Eurasia. However, in this section, we further investigate the spatial changes in storm activity at each grid point. The results shown in this section are multi-model averages of all of the models used for this study. The multi-model average is conducted by interpolating all model outputs to a common resolution of 1.4° × 1.4°.
We examine cyclone density (Figure 3) at each grid point, which is the total number of cyclone centers that crosses the grid box. The general climatology shows a higher density of cyclones over the Mediterranean region with one branch extending toward India on the southern side of the Himalayas while another branch moves northward into continental mid-latitude Eurasia, which is known as the Eurasian storm track. The Scandinavian region also exhibits a high density of cyclones along the coast of Norway. The east coast of Asia along with the Sea of Okhotsk region also shows a high cyclone density. A comparison between HIST and the future scenarios shows a prominent decrease in cyclone density over the Mediterranean (MS), East Asia (AE), and Scandinavian (SC) regions by the end of the century. However, in the future scenarios cyclone density increases over the Arctic especially in the Barents and Kara Sea regions and Sea of Okhotsk region (SO).
Extratropical cyclones are primarily driven by vertical wind shear where the jet stream provides the necessary upper level divergence for the genesis and deepening of the surface low pressure systems. Therefore, in this section, we investigate the vertical structure of the zonal wind averaged between 0°E and 150°E longitude (Figure 4). In future scenarios the zonal wind shows a prominent weakening between 500 and 250 hPa over the mid-latitudes. However, the zonal wind strengthens over high-latitudes between 600 and 400 hPa pressure levels. In the future scenarios, the vertical wind shear weakens over the mid-latitude and intensifies over the high-latitude regions between 80°N and 90°N and to some extent over the subtropics. What causes these changes in the vertical wind shear over the mid- and high- latitude regions?
Vertical wind shear primarily depends on the meridional temperature gradient (MTG) in the troposphere. Anomalous warming at the surface in the future scenarios may result in a redistribution of the thermal gradient at the lower troposphere (Figure 5). Hence in this section, we examine the spatial distribution of lower troposphere MTG (dT/dy) at 850 hPa, which is the lowest pressure level of data obtained from the CMIP5 models. In HIST, a stronger MTG is present over the south-central Asia, East Asia, Sea of Okhotsk, and Scandinavian regions. The regions with a strong MTG correspond well with the regions having prominent cyclone density in HIST. However, the changes in MTG at the end of the century in RCP4.5 and RCP8.5 indicate a noticeable weakening of the MTG over the region between 40°N and 75°N. MTG also intensifies over the Sea of Okhotsk region and Arctic region. The regions with a prominent increase (decrease) in storm activity and intensified upper level jet are consistent with the regions with intensified (weakened) MTG.
A weakening of MTG causes a decrease in the vertical wind shear over the mid-latitudes in RCP4.5 and RCP8.5. This reduced MTG and vertical wind shear results in a decrease in the frequency of extratropical storms over the mid-latitudes. However, the increase in more intense storms with a longer lifespan over the subregion AE in this study may result from an increase in MTG and vertical wind shear over the Arctic region.
4 Summary and Conclusion
Extratropical cyclones are the primary source of precipitation in the winter for the mid- and high-latitude regions of Eurasia that provides fresh water for winter crops. Any changes in extratropical cyclone activity can have a critical economic impact on the communities. In this study, we examine the projected changes in Eurasian storm activity by the end of the current century using future emission scenarios (RCP4.5 and RCP8.5) from the historical and future emission scenario simulations of six CMIP5 models. An investigation of the regional storm activities over different geographical sectors of Eurasia using a storm identification and tracking algorithm (Zhang et al., 2004) reveals a decrease in the number of storms over the mid-latitudes. Over the Mediterranean region storms are reduced in number and intensity in RCP4.5 and RCP8.5 relative to the historical simulation. There are less numerous but more intense storms over the East Asian region. However, the Sea of Okhotsk region experiences more numerous and more intense storms. Over the high-latitude region, that is, the Arctic Eurasia region in this study, the storms are more intense with longer duration at the end of the century.
To investigate further, we link the statistical findings from the Lagrangian approach to the large-scale circulation. A spatial representation of the changes in storm activity by expressing it in terms of cyclone density reveals a prominent decrease in cyclone density over the Mediterranean Sea, Yellow Sea, Sea of Japan, Japan, and along the coast of northeast China. Over the Sea of Okhotsk and Kara and Barents Seas cyclone density increases noticeably in the future emission scenarios. Changes in cyclone density over Eurasia and the Arctic are consistent with the findings from the storm tracking algorithm. Further investigation reveals a weakening of the vertical wind shear in the future scenarios over the Eurasian storm track region between 40°N and 70°N with an intensification of the westerly jet stream at the upper levels of the troposphere over the region located north and south of the climatological storm track. Studies project the recent warming trend over the Arctic to persist in all the future scenarios (Zhang et al., 2008; Overland et al., 2013). In response to the anomalous warming the lower troposphere MTG decreases considerably over the region between 40°N and 70°N which includes MS, EA, and SC and intensifies at high latitudes especially over the Barents and Kara Seas and the Sea of Okhotsk region. In response to a weaker MTG, the jet stream weakens over the mid-latitude regions, which reduces the vertical wind shear and makes conditions unfavorable for extratropical cyclones in the future emission scenarios. However, intensified MTG and strengthened vertical wind shear provides favorable conditions for the propagation of extratropical cyclones over the high-latitude regions in RCP4.5 and RCP8.5.
Decreased storm activities over mid-latitudes and increased storm activities over high-latitude has significant implications for society and the environment. These extratropical cyclones are the primary source of water for agriculture of winter crops in many regions so a reduction in storm activity could be harmful to agriculture. In contrast, an increase in storm activity could lead to heavy precipitation in winter that may cause loss of property and life. Thus, a better understanding of the future changes in storm activities over different regions of Eurasia will empower us with more information for efficient planning and better decision making.
Acknowledgments
We acknowledge the World Climate Research Programme's Working Group on Coupled Modeling, which is responsible for CMIP, and we thank the climate modeling groups (listed in Table 1 of this paper) for making the datasets available. For CMIP the U.S. Department of Energy's Program for Climate Model Diagnosis and Intercomparison provides coordinating support and led development of software infrastructure in partnership with the Global Organization for Earth System Science Portals. The CMIP5 model data used in this study can be accessed at the ESGF portal (https://esgf-node.llnl.gov/projects/esgf-llnl/). We thank the Research Computing System for providing computational resources. We thank Dr. Peter Bieniek for language editing the paper. S. Basu and Z. Wang were supported for participating in this study by the Global Change Research Program of China (2015CB953904). Z.W. was supported by the Fundamental Research Funds for the Central Universities (2017B04814; 2017B20714).
