ATLANTIC CYCLOGENESIS

Friday 31 July 2009 by rascol

Motivations

The areas of cyclogenesis are distributed all around the world over the warm oceans, near the tropics. The major part of the about 80 tropical storms which form each years is located in the North-west Pacific, where oceanic waters are the warmest of the world. The remainder is distributed between six other large basins (North-East Pacific, the northern Atlantic, northern Indian Ocean, south-western and south-eastern Indian Ocean and South Pacific). The southern Atlantic is not regarded as an area of cyclogenesis, although the Catarina event occurred off Brazil, in March 2004. When speaking about cyclonic phenomena, it is important first of all to lay down the used terms. When a tropical depression intensifies, it first reaches the tropical storm (if maximum winds exceed 17 m/s) then the tropical cyclone (if winds exceed 34 m/s) stage. One must be wary of the term cyclone, which, in English, refers to a depression, tropical or not, while in French, this term is specific to the cyclonic phenomenon, with winds above the threshold of 34 m/s. According to the basin, different terms can be used to designate a tropical cyclone. For example, in the Atlantic and North-East Pacific, it is referred as hurricane while in the North-west Pacific, one uses the term of typhoon... For cyclogenesis to develop, several factors have to combine. Among these factors, first is the ocean thermal content, i.e. the quantity of heat contained in the layer of oceanic mixed layer, genuine fuel of cyclogenesis. As it is not easily measurable, it is generally approached by the sea surface temperature (SST), easier to measure. Cyclogenesis area over the world coincides coarsely to the part of ocean delimited by the 26°C isotherm. If this threshold has proved to be useful to give an account of the climatology of the past cyclogenesis activity, one should be avoid giving too much physical interpretation of this threshold. It represents only the translation in terms of SST of the vertical temperature profile of the atmopshere. For cyclogenesis to occur environment should be unstable, i.e. an for which atmosphere the decrease of the temperature with altitude is more rapid than that of the standard atmosphere. As it can be seen, this 26°C threshold, usually widespread in literature, synthetises information from the ocean thermal content and the unstability of the atmosphere, what makes its physical interpretation uneasy. A second important factor for cyclogenesis is the lack of vertical wind shear, that is little change in the direction and the intensity of horizontal wind with altitude. These two factors are essential in the formation of hurricanes. Tropical cyclones of other basins tend to be sensitive to other factors, such as the vorticity of the low layer environment or the humidity. Modelling tropical cyclones with GCMs Given the spatial scale of the cyclonic phenomena, the representation hurricanes in the general circulation models (GCM) strongly depends on their resolution. At the climatic scale, the standard resolution is about 300 kms, which is definitely insufficient to represent the cyclones in all their complexity. Consequently, one can wonder whether it is reasonable to study, in these simulations, hurricane-type vortices or if it is preferable to focus on the study of the favourable environments for the cyclogenesis. This question gave rise, in literature to the development of two branches of cyclonic studies in GCMs: a direct approach based on the determination of the tracks (Bengtsson 1982) and an indirect approach based on the study of the environment (Gray 1975). The direct approach consists in objectively choosing what resemble cyclonic phenomena and to follow them in their development. Several criteria can be retained to characterise a hurricane-type vortice relatively to its environment and the list of these factors varies with authors. Nevertheless, most often used are as follows:

minimum of the mean sea level pressure,

temperature anomaly higher in altitude than in low layer,

tangential wind stronger in low layer than in altitude,

temperature anomaly in altitude (between 700 and 300 hPa) > TEMP,

850 hPa vorticity > TRB,

850 hPa wind > WIND,

TEMP, TRB and WIND being thresholds to customise in the tracking procedure. It is obvious that this approach can give good results only if the resolution of the model used is sufficient. Studies were undertaken with resolutions of 300 kms but for a few years, it has clearly refined itself and it seems now reasonable to use resolutions lower than 100 kms to obtain convincing results. Lately, Japanese teams one carried out global simulations with 20 kms resolution (Oouchi er al. 2006) to study hurricanes. But that remains very expensive in computing time and in volume of stored data. In the CNRM, we carried out global experiments with 50 kms of resolution as well as experiments with rotated/stretched grid (see hereafter).

Tracks of the hurricanes modelled in ten-year simulations for the end of the 20th (top) and 21st (bottom) century according to the scenario A2. The portions of tracks in red show that the system is at the hurricane stage, those in yellow the initial tropical (the blue spot represents its origin) or final extra-tropical depression.

The indirect approach makes it possible to assess cyclogenesis through the favourable conditions to the release of hurricanes. The resolution necessary for this kind of approach is much coarser than that required by the direct approach. Taking up again the various factors described above, it is possible to build parameters of cyclogenesis. Gray (1968) was the first to be concerned with this kind of approach. Since, numerous contributions have been tending to refine the criterion developed by Gray, in particular to get rid of this 26°C threshold introduced by the author and incompatible with studies of climate change.

Contribution of the rotated/stretched ARPEGE-Climat configuration

In the CNRM, a recent study made it possible to demonstrate the advantages using a rotated/stretched version of Arpege-Climat in the study of the Atlantic cyclogenesis (Chauvin et al. 2006). Using simulations in T79 truncation, with the pole of the grid at 60°O and 20°N and with a stretching factor of 2.5, a resolution of 50 kms on the Atlantic for the collocation grid is obtained, i.e. the same as that obtained with simulation with uniform grid in T319 truncation. In this study, we showed that the distortion of the grid did not amend the response of cyclogenesis to anthropogenic warming. Thus, the methodology of the rotated/stretched should make it possible, in the future, to carry out longer and more simulations on the topic of the cyclonic activity. The counterpart of the method is that the focus of the study is inevitably located on a basin and cannot therefore be used for general purposes.

And what models say on the impact of the climate change?

All the studies which have already been carried out on the subject, within the scientific community, did not make it possible to achieve a consensual response on the subject. Some simulations suggest an overall increase of the cyclones for the XXIst century and others, a decrease. For the CNRM study, and only regarding the Atlantic, we showed that according to the structure of the SST anomaly prescribed to the model, the number of hurricanes could increase or decrease. On the other hand, whatever the structure of the SST anomaly prescribed to the model, the precipitations associated with the hurricanes tend to increase. For regions such as the Antilles, this point is important since the pouring rain associated with the crossing of a hurricane leaves as much death and damage as wind.

Average distribution of precipitations around the centre of the hurricane (marked as a black cross) in the modelled hurricanes (colours) and precipitation difference between the end of the 21st century and the end of the 20th century (contours). Precipitations are expressed in mm/day.

Bibliographical references

Gray, W.M., 1975: Tropical cyclone genesis, page 121. Dept. of Atmospheric Science Paper, No. 234, Colorado State University, Fort Collins, CO.

Bengtsson, L., H. Bottger, and M. Kanamitsu, 1982: Simulation of hurricane-type vortices in a general circulation model. Tellus, 34, 440-457.

Oouchi, K., J. Yoshimura, H. Yoshimura, R. Mizuta, S. Kusunoki and A. Noda, 2006: Tropical cyclone climatology in a global-warming climate as simulated in a 20 km-mesh global atmospheric model: Frequency and wind intensity analyses. J. Meteorol. Soc. Japan, 84, 259-276.