The variability that is unexplained by forcing provides a new, long estimate of climate variability generated through interactions within the climate system. A comparison to model simulated internal variability is made to evaluate coupled climate model variability (see Hegerl et al., 2007).
To arrive at a spatially complete description of the ocean-atmosphere state consistent with a coupled climate model, TITAN constrains a large ensemble simulation with HadCM3 to follow the signal recorded in available data using a particle filter technique (von Leeuwen 2009; Goosse et al., 2012). 50 simulations of the model HadCM3 are spawned from 4 existing simulations of the last millennium, starting from 1800. Simulations include the effects of external forcing.
The simulations are compared with available data, and simulations that are unlikely to agree with the data are discontinued. The remaining simulations are spawned to 50 simulations again (see Goosse et al., 2012). The result is an ensemble of dynamically consistent simulations of the coupled climate state as captured by observed SST data (Brohan et al., 2006; Morice et al., 2012), without the challenges and drift due to nudging. It enables, for the first time, to identify oceanic states that are consistent with observed sea surface temperatures. This is a very novel in approach for coupled climate models, but has been shown to work for simpler models.
This WP addresses questions i) and vi).