ClimaTea: GS Chris Horvat

Date: 

Tuesday, February 25, 2014, 3:00pm

Location: 

HUCE Seminar Room

Graduate student Chris Horvat will lead a discussion of a recent Cryosphere paper by Notz entitled, "Sea-ice extent and its trend provide limited metrics of model performance."

The paper can be found here.

Chris summarizes the paper as follows:

"(Most of) the point in doing climate modelling is accurately representing reality. To determine just how well models do this, modellers compare their models to an assortment of representative observable benchmarks. In the modeling of Arctic sea ice, the most common and most publicized benchmark is sea ice extent, in particular summer sea ice extent, which is the total area of all model cells with concentration greater than 15%. Sea ice extent is frequently chosen because it is more easily retrieved than sea ice area, and closely related. Notz shows that various grid-independent and grid-dependent sources of uncertainty, including synthetic biases from grid selection, limit the quality of the use of sea ice extent as a model benchmark. He proceeds to compare the sources of uncertainty to those of internal variability, finding that while some of the model-reality deviation may be tied to internal variability, for a large subset of CMIP models, the model-observation spread in sea ice extent and area cannot be explained by internal variability alone."