UCL Department of Geography


UCL Geography banner



    UCL Home / Geography / PMIP Var / Individual Models

    Paleo Data-Model Comparisons

    This pages contains a several figures comparing reconstructions of paleovariability to the simulated response of the same mode. They currently represent just a few examples that I've made to what the sort fo thing that is possible. Please feel free to email me requests or preferably go to the data webpage to for the data and scripts to make your own comparison.


    Emile-Geay et al (2013, J. Clim.) estimate the SST in the Nino3.4 region from multi-proxy reconstructions. Here I compare the running standard deviation in this dataset to that for the annual average simulated by models. The reason for the drop in estimated variability during the gap between the two PMIP3 experiments eludes me - but does appear to be real feature of the reconstruction.

    The change in ENSO variance in the Central Pacific for the period 7.5-5.5ka has been estimated by Emile-Geay et al (2016, Nature Geosci.). Their estimate of 66% reduction, with a 5-95% confidence interval of -5% to 92% is marked on the figure with horizontal lines. The change in standard devation of the Nino3.4 index in each of the PMIP3 models is shown as bars.

    Southern Annular Mode

    Abram et al (2014, Nature Climate Change) reconstructed the annual mean Southern Annular Mode from a compilation of proxy records that encompass the full mid-latitude to polar domain across the Drake Passage sector. I've chosen here to compare the 30 yr running standard deviation of this reconstruction (after normalization) to the running standard deviations simulated by PMIP3 in the past1000 and historical experiments.

    Arctic Oscillation / Northern Annular Mode

    D'Arrigo et al (2003, GRL) reconstructed the Arctic Oscillation since 1650 using tree-rings. This data only represents the warm season, so I've compared its 30 yr running standard deviation to the JJA state fo the Northern Annular Model seen by PMIP3 in the past1000 and historical experiments.