Earlier models have looked at stabilizing carbon dioxide levels at 450 - 550 ppm: this would need an investment of 0.5 -1% of global world product in mitigation. As a guideline, GWP in 2004 was roughly $41 trillion.

"More important than reaching a concentration target is reaching a temperature target, which is closer to most of the impacts," Hermann Held of the Potsdam Institute for Climate Impact Research told environmentalresearchweb.

But it's not certain what effect stabilizing carbon dioxide levels will have on global temperatures as a number of crucial parameters in the climate system, such as climate sensitivity, are as yet unclear. In order to overcome this problem, Held and colleagues used a constraint optimization model, calculating the investment stream needed to create a 3:1 chance that the 2 °C cap on temperature increase is met. The model aims to recommend the best investment strategy for maximizing global welfare.

"The economic models are set up in such a way that the investment streams are control parameters," said Held. "It's an intertemporal optimization problem - you can decide how much you'd like to invest in each year. If you want a high probability of observing the temperature target you need high mitigation."

The IPCC regards it as likely that climate sensitivity does not exceed 4.5°C. According to Held, a lowering of the upper limit on climate sensitivity by 1 ° would require the allocation of 0.5% less GWP for mitigation under a precautionary scenario. Work in Held's group has indicated that palaeo data on the last glacial maximum may provide such information about climate sensitivity.

Held believes this puts an economic value on palaeo information for the first time. "It would be rational for governments to spend more money on climate change research - orders of magnitude more," said Held. "We need information very quickly. These additional investments should however not be at the expense of the immediate massive investments in carbon-free energy sources that are needed whatever the value of climate sensitivity may turn out to be."

The new uncertainty-based approach indicates that emissions reduction must start almost immediately from 9 Gigatonnes of carbon per year and result in about 2 Gigatonnes in 2020, leading to a GWP loss of 1%. Such fast reductions appear logistically almost unfeasible, says Held.

"That's the reason my colleagues and I suggest reducing the radiative forcing of other greenhouse gases as well, to three-quarters by the end of the century," said Held. "This would leave more time for the above-mentioned carbon dioxide emission reductions until 2050 and reduce GWP loss to 0.5%."

The team's model provides a conservative estimate by looking at the costs of mitigation without including the resulting savings that will arise from reduced climate change.

"Traditional economic cost-benefit analysis models trade mitigation costs versus impact costs," said Held. "We believe the uncertainty for impact costs is much higher than for mitigation."

But what type of mitigation could the investment go into? Held suggests that land-use changes could help to decrease emissions of greenhouse gases such as methane, while investment in renewable energy, use of fossil fuels in combination with carbon capture and sequestration schemes, and increased energy efficiency could all help reduce carbon emissions.

According to Held, the team's model assumes a larger drop in the price of renewable energy over time than other models. That's because investment to increase installed capacity and boost mass production, as well as in research and development, will act to lower the cost per unit of installed renewable energy capacity.

Held reported his work at the European Geosciences Union meeting in Vienna.