Collaborative Research: Regional and Global views of CGCM biases

Scientists at the University of California Los Angeles (UCLA) and Rosenstiel School of Marine & Atmospheric Science, University of Miami (RSMAS) propose to perform collaborative research in support of the NOAA’s Earth System Science (ESS) Program. The work proposed aims to improve the understanding of tropical Pacific processes, climatology, and model biases. Therefore, it is relevant to the Climate Variability and Predictability (CVP) competition.

The present proposal focuses on the tropical Pacific and the fundamental subject of error generation and evolution in the simulation of the regional climate by coupled atmosphere-ocean general circulation models (CGCMs). The overall hypothesis is that addressing the problem of CGCM biases in the simulation of the tropical Pacific climate requires both regional and global approaches. The working hypotheses are (1) Evaluating fast error growth before longer-time scale feedbacks develop will contribute significantly to identify parameterization deficiencies, (2) A better understanding of the local processes that determine the sea surface temperature (SST) in the region will be a major step towards the solution of tropical errors, and (3) Investigating the links between different biases, particularly those that are affected by remote and local processes, will lead to important insight into parameterization deficiencies. In view of these hypotheses we have designed a three-pronged research strategy based on the analysis of (1) the four-dimensional (space-time) structure of the error fields, (2) the asymptotic (climate) limit of the error fields, and (3) the contribution to the errors from locations that may be far away from the tropics. It will be argued that parameterizations have to be improved, but they may not be just those that refer to key processes in a particular region.

Specifically, it is proposed to examine the initial error growth in initialized seasonal climate forecasts from the US National Multi-Model Ensemble (NMME) phase II data set. An novel set of flux override experiments will explore the contribution of net surface shortwave radiation and wind stress errors to the fast SST error growth by first imposing observed fluxes and then allowing the atmosphere and ocean to recouple and redevelop their SST bias. Further experimentation will contrast flux override results from low- and high-resolution NCAR CESM runs to assess the contribution of their differing thermocline representations. To examine the processes that determine the SST under the stratus clouds in the SEP (and SEA) we will use a very high-resolution regional model (ROMS6) in upwelling regions running both uncoupled and embedded in a full CGCM. To examine the links between CGCM errors far away from the tropics and those in the tropics we will investigate why not all CGCMs that have severe warm SST biases in the tropical Pacific experience the double ITCZ problem.

BROADER IMPACTS: The CGCM biases in the tropical Pacific greatly compromise the models ability to predict climate variability and change. We expect this project to contribute significantly to eliminate these errors by providing new insights on parameterization shortcomings and thus suggesting future research on model improvements. In shorter time scales, decreasing the biases will improve the simulation and prediction of climate variability including El Niño.