Using decadal climate predictions to constrain climate projections

What is it about?

Recently, large efforts have been made toward developing predictions up to 10 years. These predictions are initialized using observations and have shown to be skillful in predicting the observed variability of key variables as, for example, temperature. However, these simulations are not sufficient to meet the increasing demand for seamless climate information up to 50 years. Only climate projections (which are not initialized using observations and are in turn less powerful to predict shorter‐term variability) are available to cover these long time scales. To obtain most valuable climate information for the next 1–50 years, this study assesses in how far it is possible to combine information carried by both sources. This is done by choosing for each start date during 1960 and 2005 an ensemble of climate projections, which shows to have the best match with the decadal prediction mean for the next 10 years. Basic skill metrics suggest that it is possible to obtain more skillful predictions using such a framework compared to just using climate projections, even beyond 10 years when no information from decadal predictions are available any longer. However, the value of this methodology heavily depends on the added value of initialization.

Featured Image

Why is it important?

We have developed and applied a technique to constrain uninitialized climate projections using initialized decadal predictions. The constrained ensemble is mores killful than the unconstrained ensemble, even after 10 years, over the North Atlantic Gyre region. The potential of the proposed technique crucially depends on the value added by initialization