climpred.stats.dpp(ds: Union[xarray.Dataset, xarray.DataArray], dim: str = 'time', m: int = 10, chunk: bool = True) Union[xarray.Dataset, xarray.DataArray][source]#

Calculate the Diagnostic Potential Predictability (DPP).

DPP_{\mathrm{unbiased}}(m) = \frac{\sigma^{2}_{m} -


Resplandy et al. 2015 and Seferian et al. 2018 calculate unbiased DPP in a slightly different way: chunk=False.

  • ds – control simulation with time dimension as years.

  • dim – dimension to apply DPP on. Default: "time".

  • m – separation time scale in years between predictable low-freq component and high-freq noise.

  • chunk – Whether chunking is applied. Default: True. If False, then uses Resplandy 2015 / Seferian 2018 method.


ds without time dimension.


  • Boer, G. J. “Long Time-Scale Potential Predictability in an Ensemble of Coupled Climate Models.” Climate Dynamics 23, no. 1 (August 1, 2004): 29–44.

  • Resplandy, L., R. Séférian, and L. Bopp. “Natural Variability of CO2 and O2 Fluxes: What Can We Learn from Centuries-Long Climate Models Simulations?” Journal of Geophysical Research: Oceans 120, no. 1 (January 2015): 384–404.

  • Séférian, Roland, Sarah Berthet, and Matthieu Chevallier. “Assessing the Decadal Predictability of Land and Ocean Carbon Uptake.” Geophysical Research Letters, March 15, 2018.