We have recently carried out climate simulations at high (T63L and T127L) and ultra-high (T191L) resolutions using an experimental semi-Lagrangian version of CCM3. The main objective of this series of simulations is to determine which simulated climate parameters improve with resolution and what simulation errors are not reduced with high-resolution.
Severe problems in the simulation of the hydrological cycle have been diagnosed at the standard resolution (T42 resolution) over these two regions. Over South America, simulations at T42 resolution show that north of the equator, the model simulates a double peak where observations indicate a single peak, and south of the equator, the model has too pronounced a dry season. Over the Southwest U.S., while the simulation of wintertime precipitation appears to improve with increased resolution, the southwest summer monsoon of Arizona and Northern Mexico is consistently underrepresented.
This presentation first describes the validation of the hydrological cycle produced by these simulations over the Amazon and Southwest U.S. regions. Observational data from standard climatology as well as blended gauge-satellite-model products are used in the validation. Second, the determination of an index based on precipitation and circulation quantities is used to examine the changes in onset and end of the wet seasons as a function of model resolution in these two areas