Putting models to the data test: a case study using Indian Ocean CFC-11 data

A method is presented for comparing time-averaged model output with snapshot data. The focus is on a comparison between the cholorofluorocarbon CFC-11 simulated in the Parallel Ocean Program (POP) model, and the near-synoptic Indian Ocean World Ocean Circulation Experiment (WOCE) CFC data-set collected in 1995. In the thermocline, the simulated fields fairly closely approximate the observations. We seek to distinguish between regions in which there is room for improvement in the model physics. and regions where the mismatch is the result of comparing time-averaged model fields with snapshot data. CFC variance is estimated using an eddy-permitting Miami Isopycnic Coordinate Ocean Model (MICOM) simulation. The results indicate that the widespread method of visual comparison of vertically-gridded data with model output might lead to misattribution of model deficiencies in regions of high oceanic variability. It is suggested that consideration of error-bars obtained from eddy-permitting model simulations in future model-data comparisons could help avoid such potentially erroneous conclusions. (C) 2004 Elsevier Ltd. All rights reserved.