Storms and Climate Change

[Menydh]

Storms and Climate Change

Real Climate
May 4, 2005


The Atlantic hurricane season will soon be upon us again , and no doubt many people will recall last year's devastating Hurricanes that swept across Florida. There was a great deal of press about these storms, as 3 major hurricanes and 5 tropical storms made landfall in the US. According to HurricaneProtection.com, the last time eight different tropical cyclones impacted the United States coastline in a single season was 1916. There were a total of 15 tropical storms and hurricanes, which means that the total number of storms that year was higher than 95% of the previous years of hurricane observations. There was also a record number of Typhoons over Japan in 2004 (10! The previous record was 6 from 1996) . Typhoons are the same as Hurricanes, but have a different name over the Indian ocean and the western Pacific. They are also known as 'tropical cyclones'. Furthermore, it was the first time that a tropical cyclone had been observed in the south Atlantic (see WMO Climate News, Jan 2005, p. 12)! So, what's going on?

The topic regarding tropical storms have also been a contriversial topic in the scientific community. Has there really been a systematic long-term trend in the storm statistics? Chan and Liu (2004) found no systematic change in typhoons that could be related to changes in the sea surface temperature. Because trends in sea surface temperature (SST) have been relative weak over the past 50 years, Knutson & Tuleya (2004) have argued that CO2-induced tropical cyclone intensity changes are unlikely to be detectable in historical observations and will probably not be detectable for decades to come. Although there is no clear linear trend in the Atlantic hurricane number (see Fig. 1), there may nevertheless be other indications which may suggest that the tropical cyclone statistics are changing.


Fig.1: Number of storms every 5-year interval since 1850 divided in 6 different categories of severities, with 'Tropical Storm' as the least and \'Category 5\' as the most powerful cyclones. The whole column indicates the total number of tropical cyclones over the given 5-year interval.

We know for sure that the number of Atlantic cyclones is affected by the presence El Ninos (the frequency is almost halved) and La Ninas (many hurricanes). Hence, we know that there must at least be something that influences the hurricane statistics. We know that one necessary condition for Atlantic cyclones to be spawned is that the sea surface in parts of the tropical Atlantic must be higher than a threshold value (~27deg C). The state of the atmosphere must also be favourable, that is, the winds should not change too much with height, as a wind shear might 'tear' the growing structures.

GCMs tend to be too coarse to resolve cyclones, but high-resolution regional models for storm studies exist. Knutson & Tuleya (2004) have studied how the hurricane activity may respond to increased CO2 levels, given known physical laws, and found a deepening of the central pressure and more intensive rainfall. If we choose to look at other statistics from the same data as in Fig. 1 - say the maximum wind speeds or minimum pressure associated with the Atlantic hurricanes - there appears to have been a trend after all (Fig. 2).


Fig.2: Indications of maximum wind speed and minimum pressure of the most severe tropical storm per season.

The analysis in Fig. 2 indicates a systematic tendency where the strongest winds associated with tropical storms have been in the most recent decades. The graphic also shows a deepening of the minimum pressure over time, i.e. an indication of increased severity - at least if the data represents what actually happened in the past. This trend is consistent with the model results of Knutson & Tuleya (2004) and with a recent statement made by Trenberth, disputed by Chris Landsea. On the one hand, the data presented in Fig. 2 seem to speak for themselves, suggesting there has been a trend in cyclone severity. Fig. 1, on the other hand, suggests there is little indication of a trend. The impression we get from the two figures seem to be contradicting, and we should rightly ask: Are the data consistent, and is their quality reliable? Before 1950, the data is scarce, especially with respect to pressure observations. The wind data is probably less accurate in the early part of the record also. Question is, was there a trend before 1950, or were the minimum pressure values more stable then?

In an independent study, Gettleman et al. (2002) noted a 0.4 degC/decade trend in the dew-point temperature in the 1958-1997 radiosonde observations from the Tropics. They found indications that the Convective Available Potential Energy (CAPE) may have increased. The CAPE and dew-point temperature are independent measurements that can give us clues about cyclone trends, but they are also a potential indicator of climate change.

World-wide trends?

The media attention on tropical cyclones was not been limited to the U.S., south Atlantic or Japan. Australia, has recently been ravished by Tropical Cyclone Ingrid, which was "unusual in that it is the only cyclone in recorded history to impact, as a severe tropical cyclone, on the coastline of three different States or Territories".

Other storms

The issue of storm trends and global warming has also received attention in Europe. A fairly recent devastating storm over Fance (December 26-27, 1999) made the headlines. Mid-latitude storms, such as those sweeping over western Europe, are distinct to tropical cyclones (more about this later). Some researchers have found that the frequency of mid-latitude storms may have dropped slightly over the European continent, but there have also been indications that the frequency of storms has increased elsewhere (the North Atlantic storm track - Iceland/Norwegian sea). Geng & Sugi (2001) noted an upward trend in the number of North Atlantic storms, albeit with strong decade-to-decade variations. The storm activity around Scandinavia during the most recent winter (2004-2005) was unusually high, and there was a great deal of media attention on the storm trains. These systems were responsible for a mild winter over northern Europe for most of the winter (in fact, March was colder than the other winter months, which is unusual) as well as extensive damage in southern Sweden.

Theoretical considerations

Why do we think that tropical cyclones would change because of a global warming? We have some basic physics-based principles that may possibly provide some insight. The main driving mechanism for all storms is the condensation of water vapour releasing energy/heat. Temperature differences cause instabilities and drive winds, and unstable disturbances grow into powerful storms. Tropical cyclones differ from mid-latitude storms in their primary driving mechanism. So-called 'baroclinic instabilities' are more important for the mid-latitude storms, whereas plain convective instabilities are the primary cause for the tropical cyclones - such as an increase in the CAPE.

Some climate models suggest that a global warming may be favourable for more intense mid-latitude storms. There are also a couple of physics-based considerations that can provide an indication of the main features: (i) As the surface and atmosphere warm up, more energy (heat) becomes available in the form of water vapour (more evaporation and the air manage to hold more moisture), but (ii) one factor which may act as a moderating influence, is that a global warming is expected to warm the polar regions faster than the lower latitudes, hence reducing the meridional (north-south) temperature differences (gradient). Mid-latitude storms tend to form where there are sharp temperature gradients, where conditions for instabilities are favourable. There tend to be a sharp temperature drop poleward of the polar fronts, and it is no coincidence that this is the same regions where the storm tracks lie. The mid-latitude storms play an important role in the climate system, as they facilitate the poleward heat transport. Hence, we would expect to see a relationship between the number of storms and the pole-equator temperature differences. Perhaps part of the reason why the poles warm more strongly than the lower latitudes (in addition to reduced sea-ice) may be an increase in storm activity along the storm tracks?

If the present high frequency of tropical cyclones prevails, that in itself would be a climate change per definition (i.e. a change in cyclone statistics). It is (virtually) impossible to prove that a single (or a small number) freak weather event(s) is (are) caused by increasing CO2, but when a large number are observed, it may be possible to spot a pattern of change - yet it is difficult to attribute such a change to a cause unless the mechanisms are understood. So, we can only tell for sure with hindsight whether the high number of tropical cyclones are an indication of a climate change (let's hope it's not!). It is nevertheless legitimate to ask whether these have been a consequence of natural variations of if a global warming can have increased the risk for more intense storms - and many have done so (e.g. in Eos 2004). And rightly so - when there are emerging signs that can point to a change in the storm statistics, there ought to be a discussion about these results. In my opinion, a disagreement between experts at present stage on this issue is therefore part of the normal scientific discourse.

Technicalities

For the sake of transparency and reproducability - two pillars of science -, the figures presented here are derived through an analysis using the R-script hurricanes.R. It is important to note that the analysis presented here is based on the re-analysis data from NOAA Hurricane Research Center, which has been taken at face value, and that it is indeed important to know the limitation of the data. For instance, the accuracy of the data may be less in the early period due to less observations at the early stage.

References:

Geng & Sugi (2001), Variability of the North Atlantic Cyclone Activity in Winter Analyzed from NCEP-NCAR Reanalysis data, J. Clim., 14, 3863-3873

Knutson & Tuleya (2004) Impact of CO2-Induced Warming on Simulated Hurricane Intensity and Precipitation: Sensitivity to the Choice of Climate Model and Convective Parameterization, J. Clim. 17, 3477-3495

Gettleman et al. (2002), Multidecadal trends in tropical convective available potential energy, JGR, doi:10.1029/2001JD001082

Chan & Liu (2004) Global Warming and Western North Pacific Typhoon Activity from an Observational Perspective, J. Clim. 17, 4590-4602


[source]

[Menydh]

Some recent updates

Real Climate
June 17, 2005


A couple of recent papers in Science this week relate to discussions we have had on RealClimate recently. The first by Curry and Mauritzen relates to the Gulf Stream Slowdown? post and describes the amount of fresh water that has been added to the North Atlantic over the last few decades (calculated from a database of salinity measurements) and what impact that has had on density and overturning circulations. There is a press release available at WHOI which is quite informative (hat tip to Joseph O'Sullivan for the link).


The second is related to the Storms and Climate Change (see post #1) post and is a perspective by Kevin Trenberth on the potential for a hurricanes and global warming link. The NCAR press release is available here.


These will undoubtedly not be the last word on the subject, and so we will probably be revisiting these topics at some point soon...



[source]

[Menydh]

Storms & Global Warming

RealClimate
August 5, 2005


Earlier this year, we posted two discussions on the association between climate change and storms: Storms and Climate Change (see post #1) and Some recent updates (see post #2). I will use the abbreviation TC here in the loose meaning of a tropical cyclone. These posts discussed the high number of TCs during the previous hurricane season and an essay on the relation between TCs and climate change. The uncertainty surrounding trends in storminess was underlined, and a point was being made about this subject being controversial.

Whereas some meteorologists believe that there is not a link between gobal warming and storms, there has also been a recent model study on how the storm statistics can be affected by a global warming that we failed to mention in the previous discussion. Furthermore, a new analysis published in Nature by Kerry Emanuel (\'Increasing destructiveness of tropical cyclones over the past 30 years\') points to a high correlation between the power of the TCs and the sea surface temperature (SST). This paper is also discussed in The New Scientist. He suggests that a likely consequence of a future warming is an upward trend in the destructive potential of TCs. Emanuel argues that a net power dissipation is a better indicator of TC threat than their frequency or intensity alone.


From NOAA

The conclusion of Emanuel at first sight appears to contrast with the a recent statement from NOAA that a 20-30-year reoccurring cycle is the dominant climate factor that controls Atlantic hurricane activity and that any potentially weak signal associated with longer-term climate change appears to be a minor factor. But, in fact, Emanuel acknowedged the lack of a trend in the frequency of the TCs, but finds a trend in their their destructive power which is related to their intensity and duration.

There are also some further differences. Chan and Liu (2004) argue that current models are not yet sufficiently good for addressing the question regarding global warming and typhoons (A typhoon is technically the same as a hurricane, the difference being that they form over the western Pacific or the Indean Ocean). But the GFDL study is based on a state-of-the-art high-resolution model that is more appropriate for hurricane studies and provides important evidence suggesting that climate change may have an effect on the TCs. Chan and Liu (2004) point to a lack of positive correlation between SSTs in the western part of the tropical pacific and the typhoon activity in the western North Pacific. They argue that the typhoon activity is related to El Ninos and that higher moist static energy provides conditions favourable for TCs. The atmospheric flow will also have an influence, as strong vertical wind shear can inhibit cyclogenesis (spawning of cyclones). I will not draw their analysis into doubt other than that I believed that they incorrectly framed the question. Therefore, I believe that there could be a different interpretation of their results. The SSTs they examined were from the 'warm pool' - the region with the highest temperatures in the world (~30degC) and well above the critical threshold of ~27 degC. SSTs in this region are not strongly positively correlated with ENSO. However, the region of high temperatures expands into central and eastern parts of the tropical Pacific during an El Nino resulting in an increased area of SST higher than the critical threshold value, and this aspect is in my opinion crucial to the interpretation. The tropical Typhoon frequency may be sensitive to the area with high SST (above ~27deg C). The clear seasonality in TCs ("hurricane season") with highest activities during the summer is one of the strongest pieces of empirical evidence that higher temperatures give more favourable conditions for tropical cyclones (After all, TCs only form in the warm tropics...).

One argument is as follows: as the globe warms, the area with high temperatures will increase, increasing the area on which tropical cyclones can spawn. However, climate model studies differ in their account on the tends in TC frequencies. I'm not aware of of any study which correlates the TC activity with the area of high temperatures, but such an analysis would perhaps be more appropriate than just correlating with the SST within one specific region. The notion of more intense TCs with higher temperatures is nevertheless supported by model studies from GFDL.


From NOAA

There have been TCs already in this hurricane season. According to Gerry Bell from NOAA\'s, seven TCs during June and July is a new record. NOAA also forecasts a high hurricane activity for the remainder of the season, and time will show if this season will match the 2004 in terms of number of tropical cyclones. But we do not yet know if this constitutes an emerging trend - only hindsight after decades will tell.

Thus, the bottom line so far seems to be that SSTs play a role for the TC statistics (destructive power). The most dominant change in TC activity is related to natural cycles (on inter-annual and inter-decadal time scales), which we would call 'noise' in a trend analysis in which the trend would be the 'signal'. There is little evidence of a trend in the TC frequency, however there are some indications of a trend in the destructive capability of the TCs. Conlusions can differ depending what you look at: frequency, intensity, duration of storms (life times) or destructive potential. Emanuel suggests that the recent upturn in the tropical cyclones' destructiveness can have two explanations: increased level of intensity (consistent with discussions on Storms and Climate Change [see post #1] and Some recent updates [see post #2]) or longer cyclone life times.

Note that when it comes to mid-latitude storms, there is a different story because they are affected by different types of instabilities (baroclinic) and different conditions.



References

Chan and Liu (2004), Global Warming and Western North Pacific Typhoon Activity from an Observational Perspective, J.Clim.,
17, 4590-4602

Kerry Emanuel (2005), Increasing destructiveness of tropical cyclones over the past 30 years, Nature, online publication; published online 31 July 2005 | doi: 10.1038/nature03906


[source]