Scientists from the National Center for Atmospheric Research in Boulder, Colorado, however managed to narrow the range down by bringing the lower limit up to 2.5 °C.
According to John Fasullo and Kevin Trenberth, large portion of the uncertainty is due to the inability of most models to take into account the influence of clouds during simulation. Fasullo and Trenberth, however, managed to tackle this problem by analyzing satellite records of relative humidity, responsible for cloud formation.
Although humidity is unevenly distributed throughout the year, the scientists tested climate models with ranging sensitivities in order to assess their performance in regards to these variations.
Trenberth is proud to present the new metric for evaluation of climate models that they have found. He points out that the models with sensitivities in the high end of the scale, 4°C and above, performed best, although 3°C is still viable.
Reto Knutti from the Swiss Federal Institute of Technology in Zurich comments that the team has made a huge step, however the study should be repeated by others in order for the results to be fully accepted.
Via: New Scientist