The simplest way to evaluate a chef is to taste the food. So when thinking about climate science, the simplest way for the public to get a feel for the reliability of future projections is to see how past temperature projections have held up so far. A real evaluation of climate modeling would be (and is) much more involved and rigorous than that, but that’s a more challenging meal that most people have time for.
Many modeling studies result in projections about the future, but the consensus projections made in the Intergovernmental Panel on Climate Change reports are the highest profile. A pair of recent papers has gone back to those reports to see how well they match the trends we’ve observed since their release. It can take a couple of decades for climate trends to become clear, but the first IPCC report was published in 1990, so there’s now enough data to make a comparison worthwhile.
It’s been said that prediction is difficult—especially about the future. That’s certainly the case for climate, where unpredictable variability—from volcanoes, the El Niño Southern Oscillation, and subtle changes in solar output, for example—are collectively a large determinant of year-to-year changes. On top of that, trends in human emissions depend on unforeseeable socioeconomic circumstances. When IPCC reports project future changes, they don’t try to guess at these things. They project average trends for multiple scenarios of human emissions.
This means that the comparison between projections and observed temperatures is not quite as simple as it sounds. First, you’ve got to figure out how well the scenarios given in the reports match the actual amount of carbon dioxide we’ve emitted. Then, you’ve got to account for the natural year-to-year variability somehow. Only then can you have an apples-to-apples comparison.
The IPCC temperatures are on target
A paper in Nature Climate Change checks in on the projections from the first IPCC report, published in 1990. That report projected simple trends based on greenhouse gas emissions through 2030, a period we’re just over halfway through. The most frequently cited projection estimates 0.7–1.5°C of warming between 1990 and 2030, which means we would see an increase of about 0.35 – 0.75 °C through 2010. (The range of values is a product of uncertainty about the exact sensitivity of climate to greenhouse gases.) The observed temperature trend through 2010 is about 0.35–0.39°C, depending on the dataset.
So, is it as simple as saying the projection was (barely) correct, but overestimated warming? Not really. The first thing to do is account for natural variability. The researchers chose to address this by running many climate model simulations in a “stable” configuration with no drivers of warming or cooling. Ninety percent of the natural variability fell within a range of ±0.19°C. If you apply that as a measure of potential noise around the signal of the underlying trend, the projected warming by 2010 becomes 0.28 – 0.81°C, which includes the observed trend a little more cleanly.
It’s also important to note that both the projection and the observed trend lie above the estimated range of natural variability. That means the difference between the observed warming and natural variability is statistically significant.
There’s still the matter of emissions scenarios, though. That oft-cited 1990 projection is actually based on a higher level of emissions than we actually produced. The emissions scenario closest to reality gives a projected trend of 0.16–0.63°C through 2010 (using the researchers’ estimate for natural variability). The observed warming of 0.35–0.39°C is right in the middle of that range.
Our comparison might be down to apples and apples at this point, but it’s Granny Smith and Red Delicious. To really assess the IPCC projections, the researchers used a very simple climate model like the one used for that IPCC report and calibrated it to make sure it would yield the same results as the 1990 model. Then they fed in actual emissions data and added in the volcanic eruptions that have occurred, as well.
The result, including natural variability, was a predicted warming of 0.29–0.67°C by 2010. Twenty-two years ago, that’s where the report predicted we’d be at today.
Another paper, this one published in Environmental Research Letters (Open Access), takes a different approach. The researchers build on previous work in which they attempted to adjust global temperature data to remove most of the natural variability, dampening some noise and bringing out the signal. Comparing this data to the projections in the third and fourth IPCC reports (2001 and 2007) shows solid agreement.
Sea levels, less so
The researchers also compare the sea level projections associated with those reports to the observed sea level rise since their publication. Here the projections are less accurate. Sea level has very slightly exceeded the upper bound of the uncertainty range for the projections. The reports’ best estimate of projected sea level rise was about 2.0 millimeters per year—it has actually increased at a rate of 3.2 ±0.5 millimeters per year.
This isn’t a surprise, as the projections are considered to be conservative by many researchers. Uncertainty about how quickly Greenland and Antarctica will lose ice seems to be largely responsible for the cautious estimates. The researchers note that the difference between projected and observed sea level rise to this point (and what we know about ice sheets) “suggests that the 21st Century sea-level projections of the last two IPCC reports may be systematically biased low.”
The reality is that the IPCC projections have stood up well in some ways, and work is ongoing in others. This is the nature of scientific progress.
Each IPCC report is a summary of the science existing at the time. If there were no questions remaining or progress to make, climate scientists would have moved on to something more interesting. But underlying all the complicated details is a very simple relationship—adding greenhouse gases to the atmosphere raises temperatures and changes climate. That’s what the science predicted, and that’s what we’re seeing play out.
Nature Climate Change, 2012. DOI: 10.1038/NCLIMATE1763, Environmental Research Letters, 2012. DOI: 10.1088/1748-9326/7/4/044035 (About DOIs).
reader comments
356