Enhanced regime predictability in atmospheric low-order models due to stochastic forcing.
Philosophical Transactions A: Mathematical, Physical and Engineering Sciences
Royal Society, The
© 2014 The Author(s) Published by the Royal Society. All rights reserved.
Regime predictability in atmospheric low-order models augmented with stochastic forcing is studied. Atmospheric regimes are identified as persistent or metastable states using a hidden Markov model analysis. A somewhat counterintuitive, coherence resonance-like effect is observed: regime predictability increases with increasing noise level up to an intermediate optimal value, before decreasing when further increasing the noise level. The enhanced regime predictability is due to increased persistence of the regimes. The effect is found in the Lorenz '63 model and a low-order model of barotropic flow over topography. The increased predictability is only present in the regime dynamics, that is, in a coarse-grained view of the system; predictability of individual trajectories decreases monotonically with increasing noise level. A possible explanation for the phenomenon is given and implications of the finding for weather and climate modelling and prediction are discussed.
This is the author accepted manuscript. The final version is available from The Royal Society via the DOI in this record.
Vol. 372, Article no. 20130286
Place of publication