Climate Variability of Tropical Cyclones around the World

Climate Variability of Tropical Cyclones around the World

Dr. Christopher W. Landsea

Mr. Stan Goldenberg

Dr. Lixion Avila (NOAA/NWS/National Hurricane Center)
Dr. Gerry Bell (NOAA/Climate Prediction Center)
Dr. David Enfield (NOAA/AOML/Physical Oceanography Division)
Mr. Jon Gill (Bureau of Meteorology, Australia)
Prof. William M. Gray, Colorado State University
Mr. Todd Kimberlain (Colorado State University)
Dr. Neville Nicholls (Bureau of Meteorology Research Centre, Australia)
Dr. Roger Pielke, Jr. (National Center for Atmospheric Research)
Dr. Lloyd Shapiro (Colorado State University)

Mr. John Sheaffer (Colorado State University)


To analyze past tropical cyclone variability and its accompanying climate and to quantify how changes have occurred on the timescales of intra-seasonal, interannual, interdecadal and century ranges.


Tropical cyclones are the costliest and deadliest natural disasters around the world, as the approximate 300,000 death toll in the infamous Bangladesh Cyclone of 1970 and the $26.5 billion (U.S.) in damages due to Hurricane Andrew can attest. In the United States over a recent ten year period, hurricane property damages - exceeding that due to earthquakes by a factor of four - account for 40% of all insured losses. Understanding how tropical cyclone activity has varied in the past and will vary in the future is a topic of great interest to meteorologists, public and private decisionmakers and the general public alike. Even of more concern is the possibility that anthropogenic climate change due to increases in "greenhouse" gases may alter the frequency, intensity and area of occurrence of tropical cyclones.


Current work has been focussing upon tropical cyclone variations on four distinct timescales: intraseasonal,interannual, multidecadal and "greenhouse gases" impact. To examine fluctuations on the intraseasonal (month-to-month) scale, monthly and even weekly large-scale fields are analyzed to better understand the inter-relationships between early, peak, and late-season activity. On the interannual variability, understanding is gained on why individual years are very active with multiple, strong tropical cyclones and other seasons can be quite quiet. This is accomplished in part by analyzing anomalies of relevant large-scale fields such as sea surface temperature, tropospheric vertical shear, ENSO phase, lower tropospheric vorticity, deep layer mean flow, etc. Much of the work for interannual variability is enhanced by utilizing the NCAR/NCEP reanalysis effort currently underway. Multidecadal variation research has centered on describing low-period modes of preferred global sea surface temperatures - especially with regards to Atlantic changes - and relating these modes to Atlantic basin hurricane activity through direct thermodynamic modulations of the hurricane environment and by indirect changes of the atmospheric circulation overlaying the ocean. Finally, investigations into possible changes caused by anthropogenic "greenhouse gas" increases have been carried out by reviews and critiques of numerical global model simulations of possible future climates.


  1. One recent project examining causes for fluctuations in intraseasonal variability has been completed. Two additonal projects are ongoing. The completed project (in collaboration with Gerry Bell and Bill Gray) was undertaken as part of the examination of the 1995 hurricane season (see interannual projects below). Though the season as a whole was extremely active, with above normal activity during August and October, the activity was actually below average during September, the month that is climatologically the peak of the season. It was found that this variability was due to near-normal vertical shear in September. In addition, another key factor may have been the enhanced activity in August producing a temporary suppression of the above normal SSTs in the tropical and subtropical Atlantic. Ongoing projects are looking at ways to enchance intraseasonal predicition of Atlantic tropical cyclone activity. Previous research has concluded that early season activity (i.e. June-July) is unrelated to peak season activity (i.e. August-September-October). The first projecct has documented that, in some cases, proper geographical stratification of early season activity is able to resolve a predicitive element for peak season activity. In particular, early season activity in the central and eastern portions of the deep tropics (e.g. 1996) is almost always indicative of average or above average activity for the remainder of the season, while even a high amount of early activity farther north can be followed, in some cases, by very low activity in the following months (e.g. 1997).

    The second ongoing project deals with activity at the end of the peak three months. Results have ellucidated the existence of a necessary (but not sufficient) condition, the presence of very favorable climatological factors related in overall activity, for October major hurricane activity to take place in the Atlantic basin.

  2. Four recent projects are being completed for the interannual timescales. Two additional projects are just being initiated. The first completed project (in collaboration with Gerry Bell and Bill Gray) investigated the physical mechanisms for the hyperactive 1995 Atlantic hurricane season. We found that 1995's extreme activity was due to a juxtaposition of very low vertical wind shear, warm local sea surface temperatures, high amounts of total precipitable water, low sea level pressures, a west phase of the stratospheric quasi-biennial oscillation and a cold phase of the ENSO.

    The second (in collaboration with Neville Nicholls and Jon Gill) analyzed the trends of Australian tropical cyclones and the ENSO. We found a strong downward trend in the cyclones (a substantial portion of which was likely artificial) that can be primarily explained by decreases in the ENSO Index.

    The third (in collaboration with Roger Pielke, Jr.) normalized United States hurricane damages back several decades by population changes, inflation and changes in wealth. The resultingtime series provides realistic measurements of how damaging previous hurricanes would be if they would strike today.

    Top 30 Damaging Hurricanes

    Normalized to 1995 dollars by inflation, personal property increases, and coastal county population changes.
    1.SE Florida/Alabama 1926 4$72,303,000,000
    2.ANDREW (SE FL/LA) 1992 433,094,000,000
    3.N Texas (Galveston)1900 426,619,000,000
    4.N Texas (Galveston)1915 422,602,000,000
    5.SW Florida 1944 316,864,000,000
    6.New England1938 316,629,000,000
    7.SE Florida/Lake Okeechobee 1928 413,795,000,000
    8.BETSY (SE FL/LA) 1965 312,434,000,000
    9. DONNA (FL/Eastern U.S.)1960 412,048,000,000
    10.CAMILLE (MS/LA/VA) 1969 510,965,000,000
    11.AGNES (NW FL, NE U.S.) 1972 110,705,000,000
    12.DIANE (NE U.S.)1955 110,232,000,000
    13.HUGO (SC) 1989 4 9,380,000,000
    14.CAROL (NE U.S.)1954 3 9,066,000,000
    15.SE Florida/Louisiana/Alabama 1947 4 8,308,000,000
    16.CARLA (N & Central TX) 1961 4 7,069,000,000
    17.HAZEL (SC/NC) 1954 4 7,039,000,000
    18.NE U.S.1944 3 6,536,000,000
    19.SE Florida 1945 3 6,313,000,000
    20.FREDERIC (AL/MS) 1979 3 6,293,000,000
    21.SE Florida 1949 3 5,838,000,000
    22.S Texas1919 4 5,368,000,000
    23.ALICIA (N TX) 1983 3 4,056,000,000
    24.CELIA (S TX) 1970 3 3,338,000,000
    25.DORA (NE FL) 1964 2 3,108,000,000
    26.OPAL (NW FL/AL)1995 3 3,000,000,000
    27.CLEO (SE FL) 1964 2 2,435,000,000
    28.JUAN (LA)1985 1 2,399,000,000
    29.AUDREY (LA/N TX) 1957 4 2,396,000,000
    30.KING (SE FL) 1950 3 2,266,000,000
    - from (Pielke and Landsea 1998).

    The fourth project (in collaboration with Lloyd Shapiro) attempted to isolate the physical mechanism reponsible for the long accepted relationship between interannual fluctuations in Atlantic SSTs and fluctuations in Atlantic basin tropical cyclone activity. The method of singular value decomposition (SVD) was used to demonstrate that although large-scale SSTs are of secondary importance to vertical shear in modulating hurricane formation, explaining ~10% of interannual variablity in hurricane frequency over the ~50% explained by vertical shear, the warmest SSTs directly enhace development.

  3. A new line of investigation (in collaboration with Dave Enfield) will analyze the contributions of tropical cyclones toward total rainfall and extreme rain events from a climatological and interannual perspective.

    A second new research topic (in collaboration with Todd Kimberlain) is to thoroughly document the effect of ENSO on tropical cyclones throughout the global tropics.

  4. On the multidecadal timescales, two projects have been completed and one is work in progress. The first (in collaboration with Neville Nicholls, Bill Gray and Lixion Avila) involves analyzing trends of Atlantic hurricane activity. We show that - instead of the popular notion that intense hurricane activity has been on the rise - Atlantic hurricanes actually decreased during the 1970s through the early 1990s.

    The second (in collaboration with Bill Gray and John Sheaffer) provides a hypothesis - oceanic conveyor belt variability - to explain the observed multidecal changes in Atlantic sea surface temperature, Sahel rainfall and Atlantic hurricanes. Lastly (in collaboration with Lloyd Shapiro and Bill Gray), we are attempting to determine whether we have re-entered an era of increased Atlantic intense hurricane activity. The implications of which - if it does turn out to be the case - are staggering for coastal communities in both the United States and the Caribbean.

  5. Most of the years 1970-94, the North Atlantic hurricane basin had experienced a relative lull in overall activity. These years were floowed by the "hyper-active" years 1995 and 1996, each containing more overall activity than any year since the 1960s. The chief issue being addressed in the current study is whether or not the activity of these two years were from short-term variablity or possibly the result of multi-decadal climate shifts. Preliminary results from analyses of Atlantic SSTs, vertical shear, and tropical cyclone activity itself, do seem to indicate the presence of the long-term shift toward favorable conditions.

  6. The question of man-made changes of our climate is also a topic that has been investigated as well in two projects nearing completion. The first of which is a critique of a global modeling run that suggested a doubled carbon world would have fewer tropical cyclones than in pre-industrialized times. However, my analysis indicates that this result is fundamentally flawed because of an incompatibility in the downscaling technique utilized. The second project was in participation in a World Meteorological Organization commissioned panel to evaluate current evidence for how tropical cyclones will change in the future. Our assessment is that we have no conclusive evidence for changes (either upward or downward) in frequency or mean intensity, though the maximum potential intensity may rise by 10%. We also conclude that the genesis regions of tropical cyclones will not change substantially in a doubled carbon dioxide world.