Globalwarming awareness2007 - » Climate modelling

Filed under: Climate Science, Climate modelling, IPCC — globalwarming awareness2007 @ 12:13 am

Halldór Björnsson, William Connolley and Gavin Schmidt

Late last year, the UK Treasury’s Stern Review of the Economics of Climate Change was released to rapturous reception from all sides of the UK political spectrum (i.e. left and right). Since then it has been subject to significant criticism and debate (for a good listing see Rabbett Run). Much of that discussion has revolved around the economic (and ethical) issues associated with ‘discounting’ (how you weight welfare in the future against welfare today) - particularly Nordhaus’s review. We are not qualified to address those issues, and so have not previously commented.

However, as exemplified by interviews on a recent Radio 4 program (including with our own William Connolley), some questions have involved the science that underlies the economics. We will try and address those.

Unlike an earlier report by the House of Lords, Stern spends no time quibbling, and essentially takes the science from the IPCC report, though somewhat updated by more recent work. Most of the science is flipped through fairly quickly within chapter one, and casual readers familiar with the IPCC report will find little to surprise them with sections including statements such as “An overwhelming body of scientific evidence indicates that the Earth’s climate is rapidly changing, predominantly as a result of increases in greenhouse gases caused by human activities” etc. However, the scientific possibilities in Stern are weighted slightly differently than in the IPCC reports since, as he states, “policymakers need to take into account the risks of greater dangers, as well as central expectations, because the consequences if these risks were to materialise would be very serious” (Stern reply to Byatt et al).

There are three strands to the science in Stern: the climate sensitivity, future emissions of greenhouse gases and the impacts of any particular level of change (scaled to the global mean temperature anomaly for convenience).

The climate sensitivity (as discussed here previously) was given a likely range of 1.5 - 4.5 C in IPCC TAR, and with a range of 2 - 5 C in the models used in that report. However, the probability of higher values plays a significant role in the report. Specifically, Meinshausen (2006) that there is “between a 2% and 20% chance that climate sensitivity is greater than 5C” but in the key message section of chapter 1 this is distilled as: “Several new studies suggest up to a 20% chance that warming could be greater than 5C”. This is true, but the report neglects to mention other new studies (Annan and Hargreaves; Hegerl et al) that suggest a negligible probability of CS greater than 5 C.

Uncertainty about future warming is not just the uncertainty about sensitivity, but also about the future greenhouse gas levels (GHG). There is a wide range of scenarios and estimates of future GHG levels that are used in the IPCC reports. The scenario used by the Review is the A2 one, but in this scenario GHG in the latter part of the 21st century is higher than in say, the A1b scenario. The point here is not that A2 is less sound than the A1b scenario, but simply that the Review chooses to work with one of the “high emission” scenarios. Additionally, the report also acknowledges the highly uncertain (but not clearly quantifiable) the possibilities of positive feedbacks in natural CO2 and CH4 emissions.

For impacts of climate change the story is similar: many of the impacts mentioned possible but their likelihood is debatable. For example, the weakening of the THC under 1 degree of warming, a risk of collapse for 3 degrees, risk of irreversible melting of the Greenland Ice sheet at 2 degrees warming, sea level changes of 5 - 12 meters over several centuries, - these eventualities are debatable, and should certainly be viewed as the “adverse tail” of possible impacts.

In conclusion: Stern gets the climate science largely right, though he strays on the high side of various estimates and picks the high side to talk about in the summary. This high-end bias lends the Review open to charges of “alarmism”. The report does make the fair point that the damages and their cost grows disproportionally with increasing temperature change and so, given that asymmetry, policymakers are correct in taking note of them. However, it looks like the major criticism of his work will be directed (in other fora) at the economics.

NB. Rather predictably, some of the usual contrarian suspects have also attacked the science in Stern. It is, however, a measure of their fundamental lack of seriousness that when there really are important uncertainties (i.e. the likelihood that climate sensitivity is higher than generally thought), they ignore them in favour of making the same repetitive uninteresting and incorrect claims they always make.

*Meinshausen, M. (2006): ‘What does a 2C target mean for greenhouse gas concentrations? A brief analysis based on multi-gas emission pathways and several climate sensitivity uncertainty estimates’, Avoiding dangerous climate change, in H.J. Schellnhuber et al. (eds.), Cambridge: Cambridge University Press, pp.265 280.

Comments (0)

Filed under: Climate Science, RC Forum, Climate modelling — globalwarming awareness2007 @ 10:00 pm

This is just a pointer to a ‘Quick Study’ guide on The physics of climate modelling that appears in Physics Today this month, and to welcome anyone following through from that magazine. Feel free to post comments or questions about the article here and I’ll try and answer as many as I can.

Comments (0)

Filed under: Climate Science, Climate modelling — globalwarming awareness2007 @ 6:26 pm

Statements often appear in the media about suggesting that more extreme mid-latitude storms will result from global warming. For instance, western Norway was recently battered by an unusually strong storm which triggered many such speculations. But scientific papers on how global warming may affect the mid-latitude storms give a more mixed picture. In a recent paper by Bengtsson & Hodges (2006), simulations with the ECHAM5 Global Climate Model (GCM) were analysed, but they found no increase in the number of mid-latitude storms world-wide. Another study by Leckebusch et al. (2006) showed that the projection of storm characteristics was model-dependent. (Note that the dynamics of tropical and mid-latitude (often called ‘extra-tropical’) storms involve different processes, and tropical storms have been discussed in previous posts here on RC: here, here, here, and here).

The factors that control this are often confounding and so make this a tricky prediction. Simple arguments based on the expected ‘polar amplification‘ and the fact that the surface temperature gradient between the tropics and the poles will likely decrease would reduce the scope for ‘baroclinic instability’ (the main generator of mid-latitudes storms). However, there are also increases in the upper troposphere/lower stratospheric gradients (due to the stratosphere cooling and the troposphere warming) and that has been shown to lead to increases in wind speeds at the surface. And finally, although latent heat release (from condensing water vapour) is not a fundamental driver of mid-latitude storms, it does play a role and that is likely to increase the intensity of the storms since there is generally more water vapour available in warmer world. It should also be clear that for any one locality, a shift in the storm tracks (associated with phenomena like the NAO or the sea ice edge) will often be more of an issue than the overall change in storm statistics.

I believe that the jury is still out on the extra-tropical storm issue because the climate models are still limited in their ability to represent them adequately. For instance, wind speeds are not well captured by the models (Leckebusch et al., 2006), and modelled key characteristics of the cyclones were sensitive to the models’ spatial resolution: Work by Jung et al. (also published in Quart. J. R. Met. Soc. (2006), vol 132, p. 1839-1857) suggested that several key characteristics of extra-tropical cyclones in the global ECMWF numerical weather model are highly sensitive to the horizontal resolution. This is also acknowledged in a recent paper by Wernli & Schwierz (2006; J. Atm. Sci., vol 63, p. 2486). However, for some regions, Jung et al. noted that model problems were insensitive to the horizontal resolution employed in their model experiments. Ulbrich (EMS/ECAC06) also found a dependency of the storm statistics in re-analysis with different spatial resolution (the picture from GCMs was similar to the re-analysis, provided the re-analysis was carried out with similar spatial resolution). It was also concluded that the different models analysed gave a similar large-scale picture of how extra-tropical storms respond to a global warming: the frequency of weak storms decline and the strong storms are projected to become more frequent. The sensitivity to resolution is understandable, because while an entire storm system can be very well resolved (they can be 1000 miles across), there are very sharp features at the fronts (the comma shaped clouds) which are a challenge even for weather forecast models to get right. Secondary ‘cyclogenesis’ (where a new storm is ’spun off’ from an existing storm) is also something that improves markedly as resolution increases.

One can try and address that by using a high-resolution regional climate model (RCM), forced by simulations from a coarser GCM at its boundaries (a process called ‘nesting’). The RCMs provide a similar description of the minimum sea level pressure (SLP - a parameter related to wind storms and the cyclone depth) as the GCMs, irrespective of their spatial resolution (The KNMI scenarios 06 Fig. 6-3). RCMs, however, are not completely free to do their own thing, but must follow the GCMs, at least on the larger scales. So should we really expect an RCM to produce a different storm climate? What implications would a substantially different cyclone climate in the RCM have for the larger-scales and the energy transport? Cyclones play an important role in the poleward energy in the mid-latitudes (’eddy-transport’), which ultimately has a bearing for the large-scale circulation. Since cyclones involve significant parts of the hydrological cycle, such as evaporation, moisture transport, condensation and precipitation, a different cyclone climate in an RCM and GCM would presumably present inconsistencies for the water budget. Furthermore, a paper by Peng et al. (2006) suggests that eddy forcing may be responsble for large-scale response to changes in the sea surface temperatures. In other words, the cyclone climate affects the large-scale circulation, and a widely different behaviour in the RCM and the GCM would imply an inconsistency.

One robust result among most GCMs is a poleward shift in the position of the storm tracks (Bengtsson & Hodges , 2006; Yin ,2006). It is important to keep in mind that for the local communities concerned, it is changes in the position of the storm tracks that is most important, rather than the global number of storms. Another robust result is that the NAO in the models tends to shift more towards its positive phase (stronger westerly winds) as greenhouse gases rise, tending to increase winter storms coming ashore in Northern Europe, and decrease them around the Mediterranean (Miller et al, 2006).

A conceptual picture of processes affecting mid-latitude is: One, that latitudinal variations in the temperature and air flow give rise to ‘baroclinic instability’; Second, the humidity of the air also plays a role as the latter influences the energy budget. An analogy for the two can be a sloping surface: the former is the how steep the slope is and the latter the height of the drop. Sharp spatial temperature contrasts and horizontal wind shear favour an unstable growth of the storm system.

As we mentioned above, global warming generally implies a ‘polar amplification’ (stronger warming near the North Pole), and so the average poleward temperature gradient is expected to diminish, leading to less unstable conditions on average. On the other hand, a warmer Arctic may imply less sea-ice and a greater heat loss to space, which must be balanced by heat transport from the lower latitudes, a poleward heat transport which may involve the mid-latitude storms (ice insulates the ocean from the atmosphere and keeps the temperatures down). Increased temperatures also implies higher humidity, and thus a higher capacity for energy conversion through condensation - the energy fuel of convection. So it isn’t a simple picture and one should be wary of simple statements on the topic.

Comments (0)