User:William M. Connolley/Stern

My Stern Posts

 * http://scienceblogs.com/stoat/2006/10/le_stern_nouveau_est_arrive.php
 * http://scienceblogs.com/stoat/2006/11/a_bit_more_gnawing_on_the_ster.php
 * http://scienceblogs.com/stoat/2006/10/more_stern.php

From Sterns Summary (http://www.hm-treasury.gov.uk/media/8A8/C1/Summary_of_Conclusions.pdf)
''Using the results from formal economic models, the Review estimates that if we don’t act, the overall costs and risks of climate change will be equivalent to losing at least 5% of global GDP each year, now and forever. If a wider range of risks and impacts is taken into account, the estimates of damage could rise to 20% of GDP or more.''

But if you look at the damage-to-2100 from fig 6.5 (http://www.hm-treasury.gov.uk/media/8AC/CC/Chapter_6_Economic_modelling.pdf) *none* of the scenarios get to 5% mean damage by 2100. And if you take "at least" to mean "lower 5%-ile" then 1% damage is more plausible. And "now and forever" can only come from low discount rates.

So... there is stuff in there I don't really understand. The summary doesn't obviously match the text. Wading through the thing to find out exactly which numbers are used would be an awful lot of work.

Key points

 * The main problem with the Stern report is economic: the use of hard-to-justify very low discount rates. The science problems are secondary. Having said that, Stern appears to constantly err on the high side.
 * Stern is quite hard to disentangle. Exactly what are used for inpput to the economics model, and how that translates into outcomes, is hard to find.
 * Tim Plamers talk on CP.net http://scienceblogs.com/stoat/2006/12/climate_sensitivity_factoid.php
 * The use of A2 appears to be buried at http://www.hm-treasury.gov.uk/media/8AC/CC/Chapter_6_Economic_modelling.pdf (box 6.1 on page 154). There is a lot of talk of "BAU" elsewhere but the crucial use of A2 is buried deep and not (afaik) mentioned elsewhere. This is... dubious. Using A1B instead of A2 (everyone else uses A1B) reduces T inc by about 1 oC at 2100 (http://www.grida.no/climate/ipcc_tar/wg1/figspm-5.htm). IPCC TAR avg is probably about 2.5 oC with quite a few at 2 oC; Stern shows virtually no chance of 2oC.
 * The use of high climate sensitivities. Even using A2 they tune their PAGE model to 2.4-5.8 by 2100 - and I can't see where they get the rather high 5.8 from (see the A2 bar on fig c of http://www.grida.no/climate/ipcc_tar/wg1/figspm-5.htm). Annan and Hargreaves (GRL 2006; http://www.jamstec.go.jp/frsgc/research/d5/jdannan/GRL_sensitivity.pdf) say "We cannot assign a significant probability to climate sensitivity exceeding 6 oC without making what appear to be wholly unrealistic exaggerations about the uncertainties involved." This is important because Sterns damage estimates are highly non-linear in T. See also http://www.realclimate.org/index.php/archives/2006/03/climate-sensitivity-plus-a-change/
 * The wording: e.g. "Several new studies suggest up to a 20% chance that warming could be greater than 5°C." - but omits to say that these new studies are uncertain. Doesn't point out that lower numbers are far more probable.
 * Adding in extra carbon cycle feedbacks: probably fair enough, though very uncertain.
 * Use of low discount rates allows him to add in damage at 2200 which is highly uncertain, and again allows higher T to come in. I think this is very important to his high cost numbers.
 * Figure 6.2 clearly shows that some models of economic effects show -ve damage for low T rise: up to 2 oC for one model. But by figure 6.5, all T levels lead to +ve damages. So some of these econ models have been thrown away. On what grounds?

Key Messages
An overwhelming body of scientific evidence now clearly indicates that climate change is a serious and urgent issue. The Earth’s climate is rapidly changing, mainly as a result of increases in greenhouse gases caused by human activities.


 * I don't think that WGI stuff tells you the problem is serious or urgent, and nor could it. It tells you what is happening and what may happen in the future. Is it rapid? Depends on what you compare it to. To the previous millenium, yes.

Most climate models show that a doubling of pre-industrial levels of greenhouse gases is very likely to commit the Earth to a rise of between 2 – 5°C in global mean temperatures. This level of greenhouse gases will probably be reached between 2030 and 2060. A warming of 5°C on a global scale would be far outside the experience of human civilisation and comparable to the difference between temperatures during the last ice age and today.


 * I don't know where 2-5 oC comes from (OK, I do, see section lower down. 2-5 turns out to be from QUMP, which isn't "most models"). 1.5-4.5 would be more obvious. Is this just rounding error? As to when we get to 2x... in terms of just CO2, 2030 sounds rather early. 2060 does too. But maybe if you use CO2 equivalent its better (fig 1.1 supports this). 5oC would be quite a lot. But then, 2 oc would be rather less... and lets not forget the difference between equilibrium and transient.

Several new studies suggest up to a 20% chance that warming could be greater than 5°C.


 * Probably twaddle. See http://www.jamstec.go.jp/frcgc/research/d5/jdannan/believe_grl.pdf

If annual greenhouse gas emissions remained at the current level, concentrations would be more than treble pre-industrial levels by 2100, committing the world to 3 – 10°C warming, based on the latest climate projections.


 * Why does 3* get you up to 10, when 2* got you up to 5? Odd. And whats this "commit" stuff?

Some impacts of climate change itself may amplify warming further by triggering the release of additional greenhouse gases. This creates a real risk of even higher temperature changes.
 * Higher temperatures cause plants and soils to soak up less carbon from the atmosphere and cause permafrost to thaw, potentially releasing large quantities of methane.
 * Analysis of warming events in the distant past indicates that such feedbacks could amplify warming by an additional 1 – 2°C by the end of the century.


 * Possible. The level of feedback is very uncertain, however. But the possibility of the "airbourne fraction" going down is worrying.

[cut]

As the world warms, the risk of abrupt and large-scale changes in the climate system will rise.
 * Changes in the distribution of heat around the world are likely to disrupt ocean and atmospheric circulations, leading to large and possibly abrupt shifts in regional weather patterns.
 * If the Greenland or West Antarctic Ice Sheets began to melt irreversibly, the rate of sea level rise could more than double, committing the world to an eventual sea level rise of 5 – 12 m over several centuries.


 * "likely" is too much. "might" is better.


 * 5-12m over... a millenium, perhaps.

The body of evidence and the growing quantitative assessment of risks are now sufficient to give clear and strong guidance to economists and policy-makers in shaping a response.


 * Well thats up to them I suppose.

Climate sensitivity
Over weights the possibility of high sensitivity.

Feedbacks
The idea of feedbacks is fair enough. But the range given is 20-200 ppmv (ie trivial to big) and the T effect 0.1 to 1.5 oC (ditto).

Preliminary estimates suggest that these “positive feedbacks” could lead to an addition rise in temperatures of 1 - 2°C by 2100.
Says Stern, quoting Torn+Harte and Scheffer. But T+H say: 1.5–4.5�C associated with anthropogenic doubling of CO2 is amplified to 1.6–6.0�C which is 0.1-1.5 in my book. And S says suggest that the feedback of global temperature on atmospheric CO2 will promote warming by an extra 15–78% on a century-scale. which is about 1.7-8, ie +0.2-3.5. Where *does* Stern get his numbers from?

=== If annual emissions continued at today’s levels, greenhouse gas levels would be close to double pre-industrial levels by the middle of the century. If this concentration were sustained, temperatures are projected to eventually rise by 2 – 5ºC or even higher. ===

Aha, here we find the source for the 2-5 oC stuff (table 1.1) which turns out to be the hadley QUMP ensemble. This is probably defensible, though maybe not right. And then the 1.2-9.1 from "11 studies" is then not comparable, because the Hadley number is a confidence interval, whereas the 11 is max/min across the range. Stern does note that the T change to 2100 would be lower because of non-equilibrium at this time.

In general, higher latitudes and continental regions will experience temperature increases significantly greater than the global average.
True, and well worth emphasising.

Climate change could weaken the Atlantic Thermohaline Circulation, partially offsetting warming in both Europe and eastern North America, or in an extreme case causing a significant cooling.
Over-emphasised, as the short following paragraph shows. THC changes might help keep Europe a bit cooler is the mean projection.

Sea levels will continue to rise, with very large increases if the Greenland Ice Sheet starts to melt irreversibly or the West Antarctic Ice Sheet (WAIS) collapses.
True, though not as exciting as "...with only fairly small increases unless Gr/WAIS does something exciting". Doesn't really commit to anything different from the IPCC TAR line.

And thats it for chapter 1. Summary: Fairly close to IPCC; probably over-emphasises high sensitivity.