June 7 (UPI) -- In a new double-blind study, different groups of climate scientists working with the same data produced climate reconstructions with key differences.
The experiment -- detailed Monday in the journal Nature Communications -- was designed by researchers at the University of Cambridge to highlight the influence of human decision making on climate reconstructions and global warming projections.
To start, Cambridge researchers recruited respected climate scientists from research institutions around the globe.
Ulf Buentgen, a professor of environmental systems analysis at Cambridge, wrote to scientists, asking them via email to help generate "independent evidence to asses the role of subjectivity, data robustness and methodological scrutiny in climate science."
Each scientist's contributions, he told them, would be anonymized and "treated equally as an ensemble member."
Those who agreed to participate were given raw tree-ring data and asked to reconstruct temperature changes over the past 2,000 years.
Their work showed differences in how each group selected and organized data -- as well as the analysis techniques they deployed -- influenced the idiosyncrasies of their final reconstruction.
While all of reconstructions were in general agreement about the trajectory of temperature changes over the last 2,000 years, they showed key differences in variance, amplitude and sensitivity.
Each reconstruction disagreed on exactly how much global temperatures rose during the Medieval warming period. The reconstructions also disagreed on the levels of global cooling following major volcanic eruptions.
"Despite notable differences in variance, amplitude and sensitivity, which can be attributed to decisions made by the researchers who built the individual reconstructions, each of the reconstructions clearly showed that recent anthropogenic warming is unprecedented in the past two thousand years," Buentgen told UPI in an email.
Every climate model features some level of uncertainty. Buentgen said he hopes the latest experiment can help climate modelers pinpoint the decisions that yield divergent reconstruction results.
If climate scientists are able to isolate important decisions, they could potential consider multiple options, produce divergent reconstructions and split the difference, he said.
"This is the first time the proxy community -- for example, dendroclimatologists -- have done such an experiment," Buentgen said.
Buentgen suggests climate modelers could simply recruit other research groups to analyze the same data sets and then combine their reconstructions.
"For the climate proxy community, we recommend the routine use of ensemble reconstruction approaches to provide a more consensual picture of past climate variability," Buentgen said.