An Alternative Initialization Strategy for Present-Day Transient Climate Simulations

Wednesday, May 14, 2014 - 07:00
Add to Calendar

One approach to enhancing the realism of Earth System Models is to broaden the range of explicitly resolved physical processes that contribute to climate change and variability beyond those that are currently resolved by standard resolution climate models. The use of fine resolution meshes however comes with a significantly increased computational overhead relative to standard simulations, which necessitates reconsidering the protocol used to carry out climate simulations. Here, we test an alternative protocol for the initialization of present day coupled climate transient ensemble members that significantly reduces the number of integration years required to produce these simulations. Rather than initializing the transients from a pre-industrial control simulation, we use global ocean and sea-ice realizations, taken from an atmospheric reanalysis-forced global coupled ocean/sea-ice model in the latter half of the 20th Century, to initialize an ensemble of present day transients. We hypothesize that by initializing the ensemble of transients using ocean and ice states of minimal spread, selected just prior to the strong warming of the late 20th century, from a reanalysis forced coupled ocean/ice simulation, very low frequency variability will be precluded from our PD ensemble and we will only resolve variability associated with the latter half of the 20th and early 21st Century. Since our projections will only cover near-term climate we do not need to capture the very low frequency variability associated with centennial climate change projection. We start by testing this alternative protocol at standard climate resolution using the Community Earth System Model 1.0 (CESM1.0) where a Eulerian spectral dynamical core with triangular spectral truncation at 85 wavenumbers (T85) is used in the Community Atmospheric Model 4 (CAM4). The ocean and sea ice models: the Parallel Ocean Program (POP) and CICE, respectively, are on a one-degree grid. A preindustrial control (circa 1850) simulation was run for 300 years; ocean model drift and the change in the top of the atmosphere balance over the run are presented. The veracity of the present day ensemble is determined by comparisons of its climatology, trends, and variability with those from late 20th and early 21st Century observations. Particularly, we compare trends of sea surface temperature, upper ocean heat content anomaly, and September Arctic sea ice extent. Using the same coupled model configuration, the computation of a present day ensemble is underway whose members are initialized from the pre-industrial control. If available, comparative results from the two ensembles will be presented.

Funding Program: