Using the 1997–1998 El Nino as a test case for climate model response

The 1997-1998 El Niño was one of the most intense in the past century, and provides a rare opportunity to evaluate climate model response to forcing from a well-observed event. The large-scale circulation and associated cloud systems in the eastern and western Pacific exhibited a large response to changes in sea surface temperature. In contrast to observations, many models used in the IPCC 4th Assessment Report simulations archive do not show a credible, consistent atmospheric response when subjected to observed sea surface temperatures (AMIP style). For example, observations show a complete collapse of the Walker Circulation, while some models fail to reproduce the observed changes in the tropical overturning circulation.

On much smaller scales, the event was well observed by the CERES instrument which was launched in late 1997. It has been demonstrated that the CERES cloud-object data set can be effectively used to characterize the geophysical signal of the 1997-1998 El Niño. Since the cloud objects represent, potentially, a process-level diagnostic to investigate the mechanisms of the atmospheric response to the sea surface temperature, we are currently using these observations to evaluate AMIP style GEOS-5 GCM simulations at a range of resolutions. In addition to analysis of these simulations, we also compare CERES cloud object data with fields produced by the NASA Modern Era Retrospective-analysis for Research and Applications (MERRA). Since the GEOS-5 used in the MERRA was constrained by the data assimilation system to be consistent with observed temperature, moisture, and winds, this case serves as a robust baseline test of the model physical parameterizations and of the realism of the analyzed hydrologic cycle.

This presentation will suggest several ways to study the 1997-1998 El Niño event on different time and space scales. The basic predicament remains that if climate models are to be used in regional climate prediction they should display a believable response to large scale sea surface temperature forcing. We suggest that process-level tests of climate models are a necessary but not sufficient test of model sensitivity and cloud feedback.