The United Nations Intergovernmental Panel on Climate Change (IPCC) says ECS is likely between 1.5 and 4.5 degrees Celsius, but it can’t be more precise than that. Which is too bad, because an enormous amount of public policy depends on its value. People who study the impacts of global warming have found that if ECS is low — say, less than two — then the impacts of global warming on the economy will be mostly small and, in many places, mildly beneficial. If it is very low, for instance around one, it means greenhouse gas emissions are simply not worth doing anything about. But if ECS is high — say, around four degrees or more — then climate change is probably a big problem. We may not be able to stop it, but we’d better get ready to adapt to it.

So, somebody, somewhere, ought to measure ECS. As it turns out, a lot of people have been trying, and what they have found has enormous policy implications.

To understand why, we first need to delve into the methodology a bit. There are two ways scientists try to estimate ECS. The first is to use a climate model, double the modeled CO2 concentration from the pre-industrial level, and let it run until temperatures stabilize a few hundred years into the future. This approach, called the model-based method, depends for its accuracy on the validity of the climate model, and since models differ quite a bit from one another, it yields a wide range of possible answers. A well-known statistical distribution derived from modeling studies summarizes the uncertainties in this method. It shows that ECS is probably between two and 4.5 degrees, possibly as low as 1.5 but not lower, and possibly as high as nine degrees. This range of potential warming is very influential on economic analyses of the costs of climate change.

The second method is to use long-term historical data on temperatures, solar activity, carbon-dioxide emissions and atmospheric chemistry to estimate ECS using a simple statistical model derived by applying the law of conservation of energy to the planetary atmosphere. This is called the Energy Balance method. It relies on some extrapolation to satisfy the definition of ECS but has the advantage of taking account of the available data showing how the actual atmosphere has behaved over the past 150 years.

The surprising thing is that the Energy Balance estimates are very low compared to model-based estimates. The accompanying chart compares the model-based range to ECS estimates from a dozen Energy Balance studies over the past decade. Clearly these two methods give differing answers, and the question of which one is more accurate is important.

Climate modelers have put forward two explanations for the discrepancy. One is called the “emergent constraint” approach. The idea is that models yield a range of ECS values, and while we can’t measure ECS directly, the models also yield estimates of a lot of other things that we can measure (such as the reflectivity of cloud tops), so we could compare those other measures to the data, and when we do, sometimes the models with high ECS values also yield measures of secondary things that fit the data better than models with low ECS values.

This argument has been a bit of a tough sell, since the correlations involved are often weak, and it doesn’t explain why the Energy Balance results are so low.

The second approach is based on so-called “forcing efficacies,” which is the concept that climate forcings, such as greenhouse gases and aerosol pollutants, differ in their effectiveness over time and space, and if these variations are taken into account the Energy Balance sensitivity estimates may come out higher. This, too, has been a controversial suggestion.

A recent Energy Balance ECS estimate was just published in the Journal of Climate by Nicholas Lewis and Judith Curry. There are several features that make their study especially valuable. First, they rely on IPCC estimates of greenhouse gases, solar changes and other climate forcings, so they can’t be accused of putting a finger on the scale by their choice of data. Second, they take into account the efficacy issue and discuss it at length. They also take into account recent debates about how surface temperatures should or shouldn’t be measured, and how to deal with areas like the Arctic where data are sparse. Third, they compute their estimates over a variety of start and end dates to check that their ECS estimate is not dependent on the relative warming hiatus of the past two decades.

Their ECS estimate is 1.5 degrees, with a probability range between 1.05 and 2.45 degrees. If the study was a one-time outlier we might be able to ignore it. But it is part of a long list of studies from independent teams (as this interactive graphic shows), using a variety of methods that take account of critical challenges, all of which conclude that climate models exhibit too much sensitivity to greenhouse gases.

Policy-makers need to pay attention, because this debate directly impacts the carbon-tax discussion.

Full post