Are Numerical Weather Prediction Models Getting Better?

Numerical weather prediction has undoubtedly improved during the past decades, but how much of this gain has been produced by improved models compared to improvements in data assimilation, observations, and resolution? Numerical weather prediction models have seen major increases in complexity, with the addition of additional processes and physics. But does increasing complexity alone ensure improved model realism and forecast skill?

This presentation will explore this issue by examining the results of a forecast experiment whereby the evolving WRF ARW model, representative of the varying state of the modeling system over the past decade, is driven by identical initial and later boundary conditions. The presentation will also describe the changes in the skill of NOAA/NWS modeling systems over the past several decades. The talk will end with an examination of current approaches to model improvement and whether the discipline needs a more coordinated, holistic approach to model development.