| Summary: | In order to test data assimilation approaches adapted to paleoclimatology, we conduct experiments with the three-dimensional Earth system model of intermediate complexity LOVECLIM using particle filtering to reconstruct past climate states over the last 150 years. We consider two particle filtering methods: sequential importance resampling and an extremely efficient particle filter. We compare results of the simulations obtained using LOVECLIM only, LOVECLIM constrained by sequential importance resampling, and LOVECLIM constrained by the extremely efficient particle filter. In the last two simulations, surface temperature from a twin experiment is assimilated over southern hemisphere. As a mean of comparison we choose reconstructions of surface temperature, sea ice extent and sea ice concentration in the southern hemisphere.
|
|---|