Research Interests


My main research focus is on investigating mid latitude storms, including how to forecast them better from a few days out to a season, how they may change under global warming, and their immense societal impacts. The distinctive characteristic of my research lies in the fact that I employ a wide range of tools in my research, ranging from analyses of gridded atmospheric analyses and state of the art climate model simulations to learn about the basic characteristics of the phenomena, examination of actual observations to validate what have been learnt from the gridded data, and dynamical studies using a suite of intermediate/mechanistic models to achieve better understanding of these observed phenomena. My major research interest areas are:


  • Extratropical cyclones – forecasting and impacts
  • Weather and climate extremes associated with extratropical cyclones
  • Global warming and climate change
  • Dynamics and life cycle of baroclinic waves and cyclones
  • Dynamics, variability, and trends of storm tracks
  • Wave/mean flow interactions
  • Tropical/extratropical interactions, tropical meteorology


For more details, see my publications



Selected Highlights

The following highlights have been published as Highlights in BAMS, EOS, or in news articles. These serve as examples of my research interests over the years.



To Improve Seasonal Storm Track Forecasts, Look to the Tropical Stratosphere

News article link:

People have become familiar with “bomb cyclones” this winter, as several powerful winter storms brought strong winds and heavy precipitation to the U.S. east coast, knocking out power and causing flooding. With strength that can rival that of hurricanes, bomb cyclones get their name from a process called bombogenesis, which describes the rapid intensification they undergo within 24 hours as they move along the coast. These winter storms tend to form and travel within narrow “atmospheric conveyor belts”, called storm tracks, which can change location over a period of years.

Scientists have extensively studied potential causes behind these year-to-year changes in attempt to better forecast storm tracks and their extreme impacts, but new research from scientists at the Stony Brook University (SBU) School of Marine and Atmospheric Sciences, funded by NOAA Research’s MAPP Program, identifies another crucial controlling force. After analyzing 38 years of model data, the research team found that an alternating pattern of winds high up in the tropical stratosphere, called the Quasi-Biennial Oscillation (QBO), affects significant year-to-year changes in both the North Pacific and North Atlantic storm tracks. … Read more

Journal Article: Wang, J., Kim, H.-M. & Chang, E. K. M. (2018). Interannual Modulation of Northern Hemisphere Winter Storm Tracks by the QBO. Geophysical Research Letters, 45.



Projected Significant Increase in Extreme S.H. Extratropical Cyclone Frequency

BAMS 98, May 2017, p.886-887, Paper of Note


Extratropical cyclones are responsible for much of the high impact weather in the mid-latitudes, including heavy precipitation, high winds, and coastal storm surges. Thus how these cyclones may change in the future is of much general interest. Previous studies have suggested a poleward shift in the location of these cyclones, but how their intensity may change remains uncertain, especially in terms of maximum wind speed. This study shows that under global warming, state-of-the-art climate models systematically project a significant increase in the frequency of extreme cyclones in the Southern Hemisphere (S.H.).

In the study, projected changes in extreme cyclones in the Southern Hemisphere, based on 26 models participating in the Coupled Model Intercomparison Project Phase 5 (CMIP5), have been presented. Multiple definitions of extreme cyclones have been examined, including intensity exceeding constant thresholds of sea level pressure perturbations, 850 hPa vorticity, and 850 hPa winds, as well as variable thresholds corresponding to a top-5 or top-1 cyclone per winter month in these three parameters and the near surface winds. Results show that CMIP5 models project a significant (much more than 10% in most cases) increase in the frequency of extreme cyclones in all four seasons regardless of the definition, with over 88% of the models projecting an increase. Spatial patterns of increase are also consistent, with the largest increase projected between 45°S and 60°S, extending from the South Atlantic across the South Indian Ocean into the South Pacific. Projected increase in cyclone intensity is largely consistent with the projected increase in the temperature gradient in the Southern Hemisphere upper troposphere and lower stratosphere.


(a) Climatology and (b) projected change in the frequency of cyclones with extreme near surface winds in Southern Hemisphere winter. Contours show frequency of cyclone occurrences (in % of time). Shades in panel (b) indicate model agreement, with orange (red) indicating over 2/3 (4/5) of the CMIP5 models project an increase in the number of extreme cyclones. There are no grid boxes at which over 2/3 of the models project a decrease.

Results of this study, together with those from previous studies that suggest cyclone extreme precipitation may increase under global warning, indicate that extratropical cyclones may give rise to more severe impacts in the future. Current research efforts focus on understanding how Northern Hemisphere extreme cyclones may change under the competing influences of the projected increase in the upper level temperature gradient and substantial decrease in the near surface temperature gradient under global warming.—Edmund Chang (Stony Brook University), “Projected Significant Increase in the Number of Extreme Extratropical Cyclones in the Southern Hemisphere,” Journal of Climate, 30, 4915-4935 (2017).



Stony Brook Professors Create Hurricane Prediction Model for New York State

News article link:

A new prediction model tailored to forecast the number of tropical cyclones that could hit New York State in a given season has been developed by three Stony Brook University professors. For the coming Atlantic hurricane season, June 1 through Nov. 30, their forecast is for below-normal activity for tropical storms and depressions, hurricanes and post-tropical storms. Over the past 35 years, the number of storms hitting the state has run from zero to two a year, said Hye-Mi Kim, assistant professor in the School of Marine and Atmospheric Sciences. This year there’s a 19 percent probability, with 43 percent being the average, for one or more storms hitting the state, said Kim, who developed the new forecast approach with professors Edmund K.M. Chang and Minghua Zhang of the School of Marine and Atmospheric Sciences. Their article on the model was posted Thursday on “Weather and Forecasting,” an online journal of the American Meteorological Society. … Read more

Journal Article: Kim, H.-M., E.K.M. Chang, and M. Zhang, 2015: Statistical-dynamical seasonal forecast for tropical cyclones affecting New York State. Wea. Forecasting, 30, 295-307, doi:10.1175/WAF-D-14-00089.1.



Why is the Atlantic Storm Track Stronger Than its Pacific Counterpart?

BAMS 90, October 2009, p. 1438-1440, Conference Notebook


One puzzle confronting atmospheric dynamists is why during the Northern Hemisphere (NH) mid-winter, the Atlantic storm track is stronger than its Pacific counterpart. In this study, Chang and Lin used a suite of mechanistic and GCM experiments to examine this issue, and found that diabatic heating in the mid-latitudes, which is mainly forced by the continent/ocean distribution, appears to be the most important factor leading to this phenomenon.


During the cool season, weather in the mid-latitudes is dominated by passages of cyclones and anticyclones. In the NH, these weather producing systems are organized into two storm tracks, with peaks over the Pacific and Atlantic oceans. Theory suggests that these storms develop by tapping the available potential energy associated with the north-south temperature gradient of the background flow. It is observed that the temperature gradient is much stronger near the entrance of the Pacific storm track, thus one might expect that the Pacific storm track should be stronger than its Atlantic counterpart. However, under various quantitative measures, such as eddy kinetic energy and poleward heat transport by these storms, the Atlantic storm track comes out to be stronger.


In this study, a dry global circulation model is used to examine the contributions made by orographic and diabatic forcings in shaping the zonal asymmetries of the NH winter storm tracks. The model is forced by orography and fixed diabatic heating mimicking observed heating. By design, the model mean flow is forced to bear a close resemblance to the observed mean flow. The model also provides a decent simulation of the storm tracks. In particular, the maxima over the Pacific and Atlantic, and minima over Asia and North America, are fairly well simulated. The model also successfully simulates the observation that the Atlantic storm track is stronger than the Pacific storm track.


Sensitivity experiments are performed by imposing and removing various parts of the total forcings. Results of these experiments suggest that NH extratropical heating is the most important forcing. Zonal asymmetries in NH extratropical heating (basically heating over the oceans, cooling over the continents) act to force the Pacific storm track to shift equatorward, and the Atlantic storm track to shift poleward, attain a southwest-northeast tilt, and intensify. It appears to be the main forcing responsible for explaining why the Atlantic storm track is stronger than the Pacific storm track. Tibet and the Rockies are also important, mainly in suppressing the storm tracks over the continents, forcing a clearer separation between the two storm tracks. In contrast, asymmetries in tropical heating appear to play only a minor role in forcing the model storm track distribution.


Distribution of NH winter storm tracks, in terms of RMS 500 hPa geopotential height perturbations (contour interval 20 m) for synoptic scale waves, as simulated using the mechanistic model by Chang and Lin: with both orographic and diabatic forcings (top panel), with diabatic forcing alone (middle panel), and with orographic forcing alone (bottom). Note that the Atlantic storm track is stronger than its Pacific counterpart in the top two panels, with the orange shading indicative of perturbations of more than 100 m.


To assess whether these dry model results are robust, several experiments have also been conducted using the full physics CAM, comparing the simulated circulations either in the presence or absence of mountains, forced either with fixed SST distributions or coupled with a slab ocean model. The results of these full GCM experiments confirm the mechanistic model results that even in the absence of mountains, the Atlantic storm track would still be stronger than the Pacific storm track, showing that it is diabatic heating forced by continent/ocean distribution that acts to force a stronger Atlantic storm track.


The question remains as to what aspects of the continent/ocean distribution leads to this phenomenon. Preliminary results suggest that the shapes of the continents may be an important factor, but more experiments need to be conducted to clarify this – Edmund K. M. Chang (Stony Brook University), and W. Lin, “Why Is the Atlantic Storm Track Stronger Than its Pacific Counterpart?” presented at the 17th Conference on Atmospheric and Oceanic Fluid Dynamics, 8012 June 2009, Stowe, Vermont.


Related journal article: Chang, E.K.M.: Diabatic and orographic forcing of northern winter stationary waves and storm tracks. J. Climate, 22, 670-688 (2009).



Estimating the number of pre-satellite era North Atlantic tropical cyclones

EOS 88, #32, 7 August 2007, p. 319, AGU Journal Highlights – Mohi Kumar, Staff Writer


Trends in hurricane occurrence over the North Atlantic are difficult to determine because hurricanes likely went undetected before satellite observations were available. To estimate the occurrence of these undetected hurricanes, Chang and Guo examined ship track records before and during the satellite era. They mapped satellite-derived cyclone tracks from 1976 to 2005 against ship tracks from the same time period to determine the probability that ships recorded wind speeds high enough to detect a tropical cyclone. Then, they computed the probability that ships that sailed between 1900 and 1965 made wind observations at similar wind speeds, assuming that tropical cyclones from the satellite era were present at the same rate during earlier years. From this, the authors found which storm tracks were too distant from ships to be detected. They determined that the number of tropical cyclones not making landfall likely was underestimated by 1 or less per year after World War I. Thus, their results suggest that the characteristics of North Atlantic tropical cyclone track statistics might have changed during recent decades. (Geophysical Research Letters, 34, L14801, doi:10.1029/2007GL030169, 2007)



Based on HURDAT, there is an apparent trend in the number of tropical storms in the Atlantic during the 20th century. Several studies have blamed this trend on undetected storms prior to the satellite era due to sparse observations over the ocean. We have used ship observations to estimate whether this could be the case, and our results suggested that the number of underestimated storms is probably insufficient to explain away the trend. For details, see Chang and Guo (2007).



Wave Packets and Pacific Cyclone Development

BAMS 86, 2005, p. 912-913, Paper of Note


In the upper troposphere, medium scale baroclinic waves tend to be organized into wave packets. During the Northern Hemisphere winter, these wave packets propagate across Asia via two pathways – a northern track across Siberia, and a southern track along the subtropical jet across southern Asia. Results of this study show that as these wave packets propagate out of Asia into the Pacific, surface cyclone development is strongly enhanced, with the frequency of occurrence of deep and explosively deepening cyclones becoming significantly above normal. Similar impacts can be found for wave packets from both the northern and southern branches. Moreover, wave packets from the two branches can interfere with each other, either constructively or destructively. In the case when the wave packets from the two branches both have favorable phase configuration, the subsequent frequency of deep cyclones over the Pacific can reach up to 6 times that of average.


It is well known that in the Northern Hemisphere winter, one of the major peaks in cyclogenesis lies over the western Pacific. While it is clear that the location of the peak is related to the presence of strong ocean-continent temperature contrasts, the precise conditions under which explosively deepening cyclones (or ‘bombs’) develop over the region are not well understood. Previous studies have documented some environmental preconditioning that can significantly impact the rate of cyclone deepening. In this study, by stratifying cyclone development statistics based on the configuration of wave packets over Asia prior to the cyclogenesis event, it is demonstrated that these upstream wave packets exert a significant influence on explosive cyclogenesis over the Pacific.



Lagged composites, based on cyclone event over West Pacific on day 0 (panel i), showing development of cyclone over the west Pacific (panels f-j), and evolution of upper tropospheric wave train (panels a-e), propagating from Asia across the Pacific, approaching western US on day +1. Note the two branches of wave activity over Asia on days -3 and -2 (panels a-b). Adapted from Chang (2005).


More detailed examination of the structure of the cyclones that developed under the influence of the upper level wave packets indicates that the dynamics involved in the cyclone development may well depend on whether the wave packet has originated from the northern or the southern branch. Moreover, while cyclone development is found to be clearly enhanced when wave packets propagate from Asia into the Pacific, not all such wave packets are followed by the development of a significant surface cyclone. Studying how wave packets from these two branches interact dynamically with the environmental preconditioning documented by previous studies may provide additional insights on the conditions under which rapid cyclone development can be expected over the Pacific. – Edmund K. M. Chang (Stony Brook University). “The Impact of Wave Packets Propagating across Asia on Pacific Cyclone Development”, in Monthly Weather Review, 133, 1998-2015 (2005).



Jet Variations and Storm Track Variability

BAMS 85, February 2004, p. 164, Paper of Note


A striking observation of the Pacific storm track is the negative correlation between the storm track and jet strengths during midwinter (where storm track strength is measured by transient eddy variance). This relation is not obvious, given that geographically, the storm tracks peak in regions where the jet stream—and correspondingly, the vertical wind shear—are strongest.


Observations show that the midwinter Pacific jet is narrower during years when it is strong. With linear models we are able to reproduce the observed decrease of storm strength with shear, if we take into account the narrowing of the jet. This is the case because a narrower jet can decrease the meridional scale of the perturbations, which, according to baroclinic instability theory, hampers growth. A common suggestion is that perturbations are weaker when the jet is strong because they move faster out of the unstable storm-track region and thus have a shorter period of growth. We find that we need to take into account that the jet narrows when it strengthens; otherwise, the increase in growth rate is strong enough to counteract the effect of increased advection speed.


One question that arises is whether our results can also explain the observation that the midwinter Pacific storm track is weaker than the fall and spring tracks (the midwinter suppression). While a similar jet width-strength relation is observed on a seasonal time scale, linear model calculations suggest that seasonal variations in jet width are not large enough to counter the effects of seasonal changes in baroclinicity to fully explain the midwinter suppression. Seasonal variations in jet width can, however, explain the observation that the midwinter suppression is much more prominent during strong jet years.


In the Atlantic, the storm track and jet strength are positively correlated. A positive correlation is found in our model for weaker jets—hence, it could also explain the Atlantic variations. However, the much more nonlinear characteristics of waves there raise the question of whether our model is relevant. It is left for future research to examine this, and also whether the Atlantic storm track can, under other climate forcings, become more like that in the Pacific.-- Nili Harnik (Lamont-Doherty Earth Observatory) and Edmund K. M. Chang. “The Effects of Variations in Jet Width on the Growth of Baroclinic Waves: Implications for Midwinter Pacific Storm Track Variability,” in the 1 January Journal of the Atmospheric Sciences, 61, 23-40 (2004).



Possible Trends in Northern Hemisphere Storm Tracks

BAMS 83, March 2002, p. 348, Paper of Note


Results of a new analysis suggest that both the Pacific and Atlantic storm tracks intensified by about 30% from the 1960s to the 1990s, and that the month-to-month and interannual variations of the two storm tracks are significantly correlated. Our analysis also shows that much of the storm track variability is not related to previously documented modes of low frequency variability.

Mid-latitude weather and climate during the cool season are closely related to changes in the location and intensity of the storm tracks. For example, precipitation anomalies over the western United States associated with the El Nino-Southern Oscillation cycle are related to storm track variability.

An important issue is that spurious climate perturbations could be introduced into atmospheric analyses due to changes in the observing system. Our preliminary analysis of radiosonde observations along the storm tracks suggests that the upward trend based on radiosonde observations alone appears to be less than that based on the reanalysis data. Unfortunately, since the peaks of the storm tracks lie over the oceans (where there are no radiosonde stations with continuous records), the variations over the storm track peaks cannot be verified. Thus it is crucial that other direct evidence be found to substantiate the magnitude of the storm track variability, especially over the oceans.--Edmund K. M. Chang and Yunfei Fu. “Interdecadal Variations in Northern Hemisphere Winter Storm Track Intensity,” appearing in the 15 March Journal of Climate, 15, 642-658 (2002).



Back to my home page


Back to my vitae