Contact:

Building: Research Hall
Office: Room 120
Mail Stop: 2B3
Phone: 703-909-6570
E-mail: asrivas3@gmu.edu

Research Abstract

Climate models suggest that ceratin structures of internal variability such as the Atlantic Multidecadal Oscillation (AMO) and the Pacific Decadal Oscillation (PDO) are potentially predictable on decadal/multidecadal timescales. Accurate predictions of these natural climate variabilities are important from a societal, political and economic viewpoint because they have widespread impacts on weather and climate of the surrounding continents. The precise mechanisms for this decadal variability are not clear, and although the dynamics of ocean circulation are widely believed to play a major role, some studies have questioned the necessity of ocean circulations to generate such low-frequency variability.

We intend to improve understanding of mechanisms of low-frequency climate variability and improve decadal predictions. In a recent work, we have identified the most predictable components in 2m air temperature in the fully coupled and slab ocean models involved in the phase 3 of the Coupled Model Intercomparison Project (CMIP3). We found that the predictable patterns, predictability timescales and forecast skills derived from slab and coupled models are remarkably similar, suggesting that interactive ocean dynamics do not play a dominant role in many components of decadal predictability identified in coupled models and in observations. Instead, the dominant mechanisms of decadal predictability appear to involve only atmospheric processes and thermodynamic air-sea coupling (Srivastava, A. and T. DelSole, 2016: Decadal Predictability Without Ocean Dynamics, PNAS, Early edition, doi: 10.1073/pnas.1614085114.)

The above results suggest that realistic decadal predictability can be captured by an atmospheric system coupled to a slab ocean model. Accordingly, we have built a new stochastic model based on the linearized primitive equations for the atmosphere, a slab mixed-layer model for the ocean, a gray radiation scheme for radiative effects, and a diffusive scheme for vertical turbulent eddy fluxes. Temperature is randomly excited in midlatitudes and all variables except surface pressure are damped artificially with a 1-day time scale. The atmospheric model alone produces realistic seasonal mean eddy variances and fluxes in midlatitudes despite the absence of moisture, clouds, moist convection, topography, and zonal asymmetries in the background state. Because the atmospheric eddy statistics are realistic, we can argue that coupling these eddies to a mixed-layer model will produce more realistic low-frequency variability than the traditional Hasselmann model in which the atmospheric stochastic forcing is imposed by fiat. We have shown that such coupling does indeed generate peaks in the low-frequency power spectrum that otherwise would not occur in the absence of coupling. We are comprehensively analyzing the mechanisms of these low-frequency peaks, exploiting the fact that the model is purely linear. We further want to introduce a series of modifications to understand the process that enhance decadal predictability.