At present, such model validation is quite problematical, owing to the uncertain quality of existing reference data sets. To account for these observational uncertainties, DSP 12 is comparing the AMIP continental simulations against several estimates of "ground truth" obtained by different techniques, e.g. satellite-derived products, reanalyses, and off-line runs of selected LSS that are driven by observed forcings. We also are adopting a variety of analysis techniques to highlight regions and seasons where the AMIP simulations deviate most from the various reference datasets and/or from one another.
Preliminary results of this validation exercise will be presented for available AMIP II simulations.