85th AMS Annual Meeting

Tuesday, 11 January 2005: 5:30 PM
Model Uncertainties Correlated with Spatial Variability of Prognostic Variables
Kun Yang, University of Tokyo, Tokyo, Japan; and T. Koike, K. Tamagawa, and P. Koudelova
Poster PDF (842.8 kB)
The Coordinated Enhanced Observing Period (CEOP) project provides an integrated, globally covered dataset. The dataset obtained in CEOP buildup phase 1 includes in-situ data at 16 reference sites, model outputs at two numerical weather prediction centers as well as satellite products that cover the period from July to September, 2001. Based on this dataset at the reference sites, we indicate that gaps between prediction and observation are less for some variables (air temperature, humidity, and net radiation) than for other variables (shortwave radiation, and longwave radiation, sensible heat, latent heat). These gaps are not only caused by observing errors and modeling errors, but also by the footprint mismatching. Through the comparison, we suggest that downward shortwave radiation is generally overestimated, and downward longwave radiation is underestimated. However, the differences between observations and model output cannot be simply deemed as model errors in most cases. Instead, their differences may be related to the representativeness of in situ observations. Model intercomparisons suggest that the representative scales may be correlated with model uncertainties.

Supplementary URL: